The URL can be used to link to this page

Home My WebLink AboutBabcock & Wilcox Tech. Reports 1985 ' Performance Technical Paper optimization of a North Dakota lignite fired utility boiler S. Libby Operations Superintendent Minnkota Power Coop. B. A. Browers, P.E. Performance Engineer Minnesota Power & Light Z. J. Frompovicz Fuels Specialist P. G. Whitten, PE. Product Manager Power Generation Group Babcock & Wilcox Barberton, Ohio Presented to The Thirteenth Biennial Lignite Symposium Bismarck, North Dakota May 21-23, 1985 INC» Babcock & Wilcox BR-1275 a McDermott company Performance optimization of a North Dakota lignite fired utility boiler S. Libby Operations Superintendent Minnkota Power Cooperative B. A. Browers, PE. Performance Engineer Minnesota Power & Light Z. J. Frompovicz Fuels Specialist P. G. Whitten, PE. Product Manager Power Generation Group Babcock and Wilcox Barberton, Ohio Presented to The Thirteenth Biennial Lignite Symposium Bismarck, North Dakota May 21-23, 1985 Abstract PGTP 85-20 The use of North Dakota lignite in utility boilers typically creates operating problems and reduces generating capacity due to severe slagging and fouling. An extensive test program directed at maintaining unit design capacity while controlling ash deposition was recently conducted by Minnkota Power Cooperative at the Milton R. Young Station, Unit 2, located near Center, North Dakota. Key parameters evaluated included the effects of off-line boiler cleaning, combustion performance, and variables such as excess air, gas recirculation, gas tempering, and furnace wall sootblowing for furnace exit gas temperature control. In addition, efficient use of convection pass sootblowers, combined with low flue gas temperature, resulted in minimal fouling in the tightly spaced reheater. Babcock and Wilcox’s Boiler Performance Diagnostics - SYSTEM 140 was used to monitor surface cleaniness and boiler and plant performance throughout the test period. For the conditions and fuels tested, sustained operation at near full load was achievable with full superheat and reheat steam temperatures, with ash deposition controllable at reduced sootblower rates. Introduction Common to almost all lignite-fueled boilers are the problems associated with severe slagging in the fur- nace and fouling throughout the convection pass. Minnkota Power Cooperative’s M. R. Young Station, Unit 2, has experienced operating problems since in- itial operation. Many utilities have experimented with various additives injected at different locations to alleviate the problems. Each user has reported his own results which, of course, vary from good to bad. Additives’ behavior is fuel dependent. Similar ex- periments with injection additives had been con- ducted at the Young Station. In April 1984, Minnkota decided to initiate an extensive test program in an attempt to better understand the slagging characteristics and behavior of this fuel. The goal of the test was to restore lost generating capacity while controlling and maintain- ing ash deposition in the boiler. This unit had been forced to take a 60 MW derate to avoid forced outages resulting from severe slagging in the fur- nace and pluggage problems, particularly in the reheater. The test got underway in early June following an outage where convection pass surfaces were washed and sandblasted clean of ash deposits. Extensive testing was conducted over an approximate six-week period. The purpose of this paper is to report some of the more pertinent results and findings. Unit description M. R. Young, Unit 2 is a Babcock & Wilcox Carolina-type, radiant boiler designed to burn high moisture, high slagging/fouling, North Dakota lignite coal. Nominally rated at 3,050,000 lb/hr, this unit is a cyclone fired, balanced draft, pump-assisted circu- lation boiler. Maximum design capacity of the unit is 3,200,000 lb/hr at the 5 percent overpressure con- dition (2620 psig main steam pressure). Main steam and reheater steam temperatures are controlled at 1005 F by a combination of gas recirculation and iil A in oa | [LAs aa GAS-, RECIRCULATING PORTS \ CYCLONE 3" TEURNACES Figure 1 Minnkota Power Cooperative Milton R. Young Unit 2. spray attemperation. The steam is supplied to a Westinghouse turbine/generator with a rated capacity of 438 MWe gross. The unit began com- mercial operation in May 1977. A sectional side view is illustrated in Figure 1. At the time of testing, Young Station, Unit 2 was equipped with a total of 213 sootblowers. In the fur- nace, sootblower coverage is concentrated in the area between the lower furnace slope and the gas tempering ports. Two control circuits divide the 113 furnace sootblowers into two zones: upper and lower walls. Three rows of blowers, totaling 36, service the lower wall zone, while 71 blowers installed through- out eight elevations service the upper zone. In addi- tion to the steam IR’s, there are six water lances located above the furnace arch on the sidewalls. Eighty-four sootblowers service the convection pass region with 50 IK’s primarily covering the secondary superheater and the remaining 34 concen- trated in the reheater section. Each bank has an in- dividual sootblowing circuit which enables indepen- dent operation of blowers. This region is also ser- viced by 16 water blowers which are used in con- junction with steam blowing. The downpass region is serviced by three groups of blowers on one circuit. In the present configuration, a total of six soot- blowers can be in simultaneous operation in the unit: lower furnace, upper furnace, secondary (steam), secondary (water), reheater, and primary and economizer. Using design information from Diamond Power and known travel speeds, the time to clean the secondary superheater is approximately 197.5 minutes (steam only) as compared to approximately 227 minutes required for one sootblower sequence in the reheat section. Effects of off-line washing During the May ’84 outage, Minnkota completely removed all ash deposits from the reheater and downpass surfaces. Waterwashing of the furnace had been performed in previous outages, but this was the first time since initial operation that a thorough cleaning of these components had been performed. In the reheat superheater, waterwashing was inadequate and sandblasting was employed to obtain bare tube surfaces. In this area, concrete-type ash was bonded to the tubes and was approximately one-half-inch thick, particularly on the leading edges. At the onset of operation following the outage, it was evident that waterwashing greatly improved boiler performance. The unit was able to achieve design steam temperatures at near full load opera- tion. Flue gas and boiler exit gas temperatures were within design margins. However, as the unit con- tinued to operate at these high loads, ash deposition rates began to increase as a slight ash coating roughened the surface providing a mechanism by which ash would be attracted and accumulated. At that point, test engineers began to focus on furnace exit gas temperature (FEGT) and sootblower effectiveness. Data collected in the early stages of the test were compared to data collected prior to the outage. Assuming that unit operation was constant, the cleaned convection pass heating surfaces resulted in a net boiler efficiency gain of 1.84 percent. Unit heat rate decreased by 200 Btu/kW-hr. The unit avail- ability and capacity factor increased as ash deposi- tion was controlled. Maintaining clean conditions throughout the boiler was established as the No. 1 priority by the operators. Test set-up The test was conducted by a combination of Minnkota Power, Minnesota Power and B&W test engineers, familiar with boiler performance and operation. B&W’s Boiler Performance Diagnos- tics-SYSTEM 140 was used to provide the test engineers with on-line, real time data acquisition and performance analysis. The SYSTEM 140 is a deriva- tive of B&W’s proprietary in-house boiler design and performance analysis program, P140, and was modified and configured to accurately simulate the physical configuration of Young Station, Unit 2. The software model performs the detailed heat transfer analysis around each bank and cavity within the unit and determines such operating characteristics as boiler gas temperatures, component absorptions, surface cleanliness, and efficiency.' The software is configured to run on a Hewlett Packard desktop computer. Data was obtained from a combination of existing plant instrumentation and temporary test instrumen- tation. A second desktop computer, dedicated to data acquisition, was interfaced to the plant com- puter and B&W’s test equipment. Additional in- strumentation installed for the test included pressure transmitters for key measurements, an ex- tensive gas sampling system to measure flue gas constituents such as excess air, carbon monoxide, carbon dioxide, and nitrous oxide, and thermo- couples for measuring vital fluid temperatures and boiler exit gas temperatures. Periodic calibration of the test instrumentation was performed during the test to assure correctness. The plant computer was programmed to transmit data every two minutes while the temporary test instrumentation was scanned every 90 seconds. The raw data was col- lected over a ten minute period, averaged, and then directed to the SYSTEM 140 for performance analysis.” Fuel characteristics The fuel supply utilized at the Young Station is a locally-mined (Oliver County) lignite and is classified as a low rank coal. Coal from this mine can vary considerably in Btu, ash and sodium content. Typically, this fuel has a heating value of around 6600 Btu/lb; ash content ranges from 6 percent to 9 percent with ash sodium content ranging from 2 per- cent to as high as 12 percent. For this fuel, ex- perience has shown that ash sodium content is in- dicative of coal quality, which would appear reasonable since the mine excavates along one seam and only supplies one plant. Since the boiler operating problems are fuel related, the method and location of coal sampling is paramount to accurately determine the represen- tative fuel quality entering the boiler. In order to minimize errors associated with storage times and coal stream mixing determination, it was decided to collect the coal samples at the silo discharge prior to entering the feeders. This would ensure that the ac- tual sample taken was from the coal entering the boiler system during testing periods. Samples were collected from each of the twelve silo outlets and analyzed by both Minnkota and B&W laboratories. The major coal characteristics that were monitored via the daily samples included Btu content, percent ash and percent sodium in the ash on an as-received basis. These three properties were chosen since they could be used to determine the mass flow of ash and total sodium entering the furnace environment. Dur- ing the testing period, the general quality of the coal was better than desired for testing purposes as sodium content was relatively low. Btu content re- mained relatively constant, ranging from a low of 6378 to a high of 7299 Btu/lb. Ash content tended to increase during the test, beginning at around 7 percent, and increasing to 8.5 and 9.0 percent during the latter stages. The overall sodium content varied over a relatively wide range, from 1.8 percent early in the test, to a high of 4.55 percent near the end. Sodium content during the test was considerably lower than desired for purposes of evaluating worse case coal. Sodium levels for this coal typically run in the 3 to 6 percent range. Utilizing the coal characteristics of Btu, ash and sodium content, mass inputs for ash and sodium during the test period were calculated on a constant heat input basis (million Btu) and are provided per Figure 2. The mass inputs followed the percentages as both sodium and ash increased during the test. A sample collected prior to the test showed .66 to .72 Ibs sodium occurred with 9.01 Ibs of ash. This com- pared to a maximum of .64 lbs sodium with 14.61 Ibs ash. This difference is very significant since it is suspected that lignite with high sodium and low ash is more troublesome than high sodium, high ash lignites. This is based on the speculation that the fluxing and sintering characteristics of sodium may be diluted by the higher ash quantities. Unfortu- nately, the high ash, low sodium lignite did not materialize during the test period. However, the higher sodium values near the end of the test did in- crease the deposition rate in the secondary super- heater, albeit the deposit was easily removed by sootblowing, indicating minimum sintering. = =—=> i. Figure 2 Coal quality trends. Ash deposition and sootblower effectiveness Early in the test, gas temperatures through the unit were within normal limits. As the test proceeded, slight coatings on the furnace walls began to appear. FEGT increased as expected. Consequently, the sur- face cleanliness factors for the convection pass com- ponents began to decay. Gas temperature limits were immediately established in an effort to control deposition. Through a combination of visual observa- tions and the monitoring of the surface cleanliness factors, recommended temperature limits were set for the gas at the furnace exit and reheater inlet at 2000 F and 1500 F, respectively. Throughout the test, these limits were easily accomplished by proper use of a combination of gas recirculation, gas tempering, excess air and sootblowing patterns. For the most part, furnace slag build-up generally consisted of light coatings in the lower region and crusty patches in the upper area. The deposits had a dry base with occasional wet surfaces. The dry base and ease of removal indicates that no sintering was occurring. This was due to: 1. Condition of furnace and tube surface cleanliness 2. Relatively low sodium content of coal (1.18 per- cent to 4.55 percent) 3. Control of gas temperatures 4. Efficient sootblowing Figure 3 provides a trend chart for sootblower steam consumption for the furnace and convection pass sootblowers. Prior to the test, continuous sootblowing was required to avoid severe slagging and fouling. At this level, sootblower steam con- sumption totaled almost 42,800 lb/hr or approxi- mately 1.6 percent of main steam production. From Figure 3 it can be seen that even on the worst day (June 20) sootblower steam consumption was only 25,500 lb/hr or 60 percent of normal operation at continuous blowing. During the test, sootblower steam consumption typically ranged from 7,000 to 19,000 lb/hr. This reduction in steam consumption resulted in an annual fuel cost savings of approxi- mately $200,000. The amount of sootblowing required to maintain cleanliness factors varied with coal quality changes. Comparing Figures 2 and 3 confirms the problems associated with lignite. Note the increasing trend in sootblower steam consumption as ash began to accumulate on the bare tube surfaces. As mentioned earlier, increased sootblowing was required near the end of the test as deposition rates increased with higher ash content. Also note that as coal quality Legend: @ Convection Pass @ Furnace Walls Continuous Convection Pass Sootblowing Steam Consumption (lb/hr X 1000) Continuous Furnace Sootblowing Figure 3 Sootblower steam consumption. improved, as experienced during July 9-11, deposi- tion rates decreased as expected. These data and observations were then used to establish the appropriate sootblowing schedule to control and maintain surface cleanliness throughout the boiler. Each sootblower was cycled to determine its effectiveness in improving overall surface cleanliness. Excess air in concert with gas recircula- tion and gas tempering were also evaluated and the effects on deposition rates were quantified. The ap- propriate sootblower sequence was then developed. The test demonstrated very little effect on the fur- nace cleanliness factor from operation of the upper and lower furnace blowers. In the secondary super- heater region, only the lower elevation blowers in Group 1 (secondary superheater inlet leading edge) and the middle elevation blowers in Group 2 (secon- dary superheater inlet cavity) were effective in recovering only partial cleanliness. Group 3 (secon- dary superheater outlet leading edge) blowers were all effective. The reheater remained relatively clean and only minimal blowing was required. Figure 4 provides representative trend charts of the surface cleanliness factors for each bank with sootblower operation noted. Since July, the plant has operated below these temperature limits and has been able to Furnace SSH In SSH Out Reheat Pri-1 Econ Legend: LF - Lower Furnace WB - Water Blower UF - Upper Furnace G = - Group ee ee ee Sootblowing and Absorption Recovery ok 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 1 2 3 4 5 6 7 Figure 4 Surface effectiveness factors. maintain clean surfaces in the upper furnace and convection pass. Gas recirculation/gas tempering Young Station, Unit 2 uses gas recirculation and gas tempering to maintain optimum settings for flue gas temperatures entering the convection pass. Most utilities have sealed off this system because of the high level of maintenance required on the fans resulting from the associated gas temperatures and flyash in the gas. Minnkota has taken great pains over the years to maintain this system. Its opera- tion in the test proved to be of great benefit. Theoretically, the effect of gas recirculation is to alter the absorption patterns within the boiler with minimal effect on stack gas flow and overall boiler efficiency. Typically, total gas recirculation, as in- dicated by fan amps, should be maintained at relatively low levels to retain a reserve of control over the unit. When reheat absorption has decayed due to ash deposition, as evidenced by excessively low reheat outlet steam temperature, the amount of gas recirculation should be increased to enhance heat transfer in that section. Of course, sootblowing should be utilized to reduce ash deposits in the con- vection pass. Gas tempering is employed to cool gases at the furnace exit plane. The effect of these cooler gases on the secondary superheater is a reduction in heat absorption. This is of use when fouling of the fur- 2100 1200, 2050 1100 2000 1950 uw g = 8 a E S © E § 2 n c s = Furnace Exit Gas Temperature, F Hot Reheat Temperature, F 900 1850 1800 800 Key Hot Reheat Temperature Main Steam Temperature ls FEGT Tempering as % Gas LVG Furnace 4 5 6 7 8 9 10 Figure 5 FEGT control. nace is not adequately controlled by sootblowing. Gas tempering has a pronounced effect on secondary superheater absorption and a lesser effect through- out the remainder of the convection pass. This was proven in the test as shown in Figure 5. From Figure 5, the effect of increased gas tempering is most pronounced on FEGT and shows a 65 F drop for about a 6 percent rise in gas tempering at constant gas recirculation. Main steam temperature is decreased by 20 F and hot reheat temperature is decreased by 10 F. Conclusions The test proved that operation at a 60-to-80 MW higher load than previously encountered at Young Station, Unit No. 2, is possible, providing proper attention is paid to standard operating practices. By prudent use of gas recirculation, gas tempering, ex- cess air, slag shedding opportunities, and soot- blowing, prolonged generation at 400 to 440 MW is possible. Surface cleanliness for effective heat transfer must be maintained through discrete soot- blowing, coupled with visual observation of the boiler components. Periodic reductions in load should be coordinated with the dispatcher for slag shedding whenever gas temperatures exceed 2000 F at the furnace exit and 1500 F at the reheat inlet. It is very important to fully understand the fuel and its combustion behavior. The slagging and foul- ing characteristics of the coal ash play an important role in the operation of the boiler. Because of the variability of coal, ash and sodium content, daily sampling and monitoring of the fuel supply is essen- tial. In this way potential problem conditions are identified and correct operating procedures imple- mented. Continuous monitoring of surface cleanliness conditions and boiler gas temperatures helps sustain unit performance at high loads and avoids costly forced outages for slag removal. Acknowledgments The authors wish to acknowledge Michael A. Fatigati, Development Engineer, B&W, for his dedi- cation in conducting the test and in commissioning the first permanent installation of the SYSTEM 140. References 1. Heil, T. C., Nethercutt, R. M., and Scavuzzo, S. A., “Boiler Heat Transfer Model for Operator Diagnostic Information.’’ Presented to ASME/ IEEE Joint Power Generation Conference, St. Louis, Mo., October 1981. 2. Fatigati, M. A. and Whitten, P. G., “Methodology and Application of a Boiler Perfor- mance Diagnostic System to a Utility Boiler.” Presented to EPRI Power Plant Performance Monitoring Conference/Workshop, Washington, D.C., October 1984. | Technical Paper Pilot study of the BR-1276 sodium-based FGD system at the Jim Bridger Power Station D. W. Johnson W. Nischt Fossil Power Division Babcock & Wilcox Barberton, Ohio T. J. Flynn Research and Development Division Babcock & Wilcox Alliance, Ohio Presented to Second Annual Pittsburgh Coal Conference Pittsburgh, Pennsylvania September 16 - 20, 1985 a McDermott company Pilot study of the sodium-based FGD system at the Jim Bridger Power Station D. W. Johnson W. Nischt Fossil Power Division Babcock & Wilcox Barberton, Ohio T. J. Flynn Research and Development Division Babcock & Wilcox Alliance, Ohio Presented to Second Annual Pittsburgh Coal Conference Pittsburgh, Pennsylvania September 16 - 20, 1985 Abstract A pilot plant test program was conducted by Babcock & Wilcox (B&W) to demonstrate that its sodium-based FGD system would meet Pacific Power & Light's (PP&L) performance goals at its Jim Bridger Power Sta- tion. For PP&L, local abundance of waste soda liquor from the beneficiation of soda ash, absence of reagent preparation, and low operating costs make sodium-based SO» absorption more cost effective than a lime/imestone based process. During the pilot program, B&W’s patented counterflow absorption tray was optimized for the Jim Bridger design conditions. The system’s functional performance was characterized over a range of operating conditions such as load, SOz2 concentration, liquor spray flux, and pH. Correlations were developed for pressure drop and SO» removal efficiency as a function of operating conditions. The effects of scale formation on performance were also studied and characterized. The pilot plant program produced a com- mercial design that was low in both capital and operating costs. Nomenclature Introduction A = Contact Area Pacific Power and Light Company (PP&L) and A; | = Correlation Constant Idaho Power Company contracted with Babcock & D; = Diameter of Droplet i, Microns Wilcox to build a flue gas desulfurization (FGD) f = SO, Absorption Efficiency, Fractional system for Unit 2 at the Jim Bridger Power Station Ky = Overall Mass Transfer Coefficient located near Point of Rocks, Wyoming. Unit 2 is an L = Effective Spray Flux, gpm/ft? existing, steam/electric generating station with a Na_ = Total Mass Transfer nominal capacity of 552 MW. This is the second of Ny = Number of Transfer Units four duplicate pulverized coal fueled units. A typical n = Number of Droplets analysis of the coal is shown in Table I. Vv = Pseudo Hole Velocity, ft/sec The FGD system (Figure 1) consists of three iden- X = Sauter Mean Diameter, Microns tical 1/3 capacity wet absorber modules. These will y = Inlet SO, Concentration, ppm be connected to the existing induced draft (I.D.) fans y* = Equilibrium SO, Concentration, ppm and stack. The design SOz efficiency (86.5%) is “Pp = Dry Tray Pressure Loss, in wg based on an allowable outlet sulfur dioxide concen- Py = Total Tray Pressure Loss, in wg tration of 0.3 pounds per million Btu input. APw = Working Pressure Loss, in wg Though most FGD systems use lime or limestone, this system utilizes the washings from a soda ash beneficiation process as the reagent for SOg removal. The active reagent is therefore sodium carbonate, NagCO3. However, the resultant solution from the washings, soda liquor, also contains the impurities that were removed during the beneficiation process. These impurities, principally silica, contribute to pro- cess problems in the FGD system. An analysis of the soda liquor is shown in Table II. Prior to constructing the full-scale unit, a pilot study of the sodium carbonate-based FGD system was performed. The main objectives of this test pro- gram were: ¢ To demonstrate that the B&W system would meet contract specifications ¢ To identify areas in which the design could be improved TABLE 1 — Coal Analysis Ultimate Analysis, % Moisture 18.3 _ Carbon 56.9 69.7 Hydrogen 3.8 46 Nitrogen 0.89 1.09 Sulfur 0.33 0.41 Ash 88 10.8 Oxygen (Difference) 10.98 13.40 100.00 100.00 Figure 1 FGD system © To determine the extent of scale formation due to impurities in the reagent and make-up water streams ¢ To optimize system design parameters with respect to economics and performance Background Currently, most operating FGD systems use lime or limestone as the primary reagent. For PP&L, local abundance of soda liquor makes soda scrubbing cost effective. The use of sodium carbonate reduces the stoichiometry, system pressure differential, and liquid to gas ratio (L/G) below the levels typical of calcium-based systems. This sodium-based wet absorption system is designed to remove the sulfur dioxide from the flue gas. In a pure sodium carbonate system, the sulfur dioxide is reacted with sodium sulfite (NagSOs) in solution to form sodium bisulfite (NaHSO3). Sodium carbonate, NagCO3, is then added to neutralize the NaHSOs3 and replenish the supply of NagSO3. The sulfur compounds formed remain in the solution and the ‘‘scrubbed” gas passes out of the system to the atmosphere. The absorption process is divided into three pro- cess flows: gas flow, liquor flow, and water flow. Flue gas enters the tower, expands, and turns up- ward. As it enters, the flue gas is quenched by a recirculated liquor spray. The quenched gas flows up through the counter flow absorber tray, which evenly distributes the gas. On the tray, intimate contact between the liquor and the SOz in the flue gas occurs. Upon leaving the tray, the scrubbed gas flows upward through the main absorber spray, through the mist eliminators, and out the stack. Spent reaction products are sent to an evaporation pond. Unlike a pure sodium carbonate system, the use of a soda liquor presents some unique problems due to TABLE II — Chemical Analysis of Soda Liquor Concentration Compound Formula Typical Maximum Sodium Carbonate Na2CO3 27% 33% Sodium Bicarbonate NaHCO3 0 3% Sodium Sulfate Na2SO4 1.4% 2% Sodium Chloride NaCl 0.76% 0.8% Silica SiO2 0.44% 1.0% pH 11.7% 12 the impurities present in the soda liquor. The primary problem is due to silica compounds which exist at concentrations as high as 10,000 ppm. The solubility of the silica compounds in the soda liquor decrease dramatically with decreasing pH. Therefore, when soda liquor at pH above 11 is added to ab- sorber recirculation liquor at a pH of 7.0, the silica compounds become insoluble. Deposition of silica can occur if the solution becomes supersaturated with respect to amorphous silica. This will usually happen on solid surfaces when the surface bears hydroxyl groups. Also, if conditions are right, polymerization of silica will occur. This leads to the formation of colloidal silica particles, which can coagulate to form larger particles, precipitate from solution, and adhere to surfaces. The design of the absorber must there- fore accommodate the insoluble silica compounds. Water is added to the flue gas desulfurization system to replenish water lost to evaporation and water which is removed from the system with spent reaction products. This make-up water enters the system as blowdown from the cooling tower. Since cooling tower blowdown (CTB) water can be ecologically undesirable and since every gallon of blowdown water requires an additional gallon of fresh make-up water, it is in the utility’s interest to minimize that blowdown. This fact provides motiva- tion to use blowdown water as a principal source of water in the FGD system. During normal operation of the cooling towers, water is cycled through the heat exchanger equip- ment several times before leaving with the blow- down or as water vapor. During this time, im- purities are concentrated by evaporation. As these concentrations increase, the potential for problems associated with deposits is enhanced. Calcium deposits are the most common. These are controlled in the cooling tower by maintaining the calcium con- centration below 2000 ppm (as CaCOs), maintaining the pH between 7.4 and 7.8, and by using a chemical scale inhibitor. Using CTB water as the primary source of make- up water also poses a problem. The relatively high concentration of calcium in untreated CTB water would precipitate as calcium sulfate or calcium sulfite in the absorber. The water softening system reduces the potential for scaling by reducing the calcium concentration to less than 150 ppm as CaCO. A major concern of users of FGD systems is to minimize capital investment and operating costs. For PP&L, this was accomplished through the use of soda liquor, which eliminates the need for much of the equipment common to calcium-based FGD systems. This includes the reagent preparation system, dewatering system, and sludge handling equipment. Operating parameters such as liquor spray flux and gas side pressure differential are reduced over those levels used in calcium-based FGD systems. Also, 100% reagent utilization can be achieved in a sodium-based system. Chemistry Soda liquor was used in two ways. First, the soda liquor was used as the source of alkalinity for re- action with SO, in the absorber. Secondly, the liquor was used as a softening agent to reduce the calcium, Ca**+, concentration in the CTB water. The major reactions that occur in the absorber are gas absorption, solution neutralization, and solution oxidation. Sodium-based absorption differs from the calcium-based process in two major ways. First, in a sodium-based system, the SO2 absorption rate is primarily controlled by the gas phase SOz diffusion rate. Liquid phase mass transfer resistance is minimized in the high total dissolved solids (TDS) solution when sufficient alkalinity is present. This is due to the higher solubility of the sodium com- pounds as opposed to the calcium compounds. Secondly, the active absorbent is sodium sulfite. The absorption of SO, from flue gas is enhanced by chemical reaction with the sodium compounds in the liquid phase. SO absorption begins when SOz is dissolved in water to form sulfurous acid (Equation 1), which dissociates according to Equation 2. SO, + H,O — H»SO3 (1) H)SO3, — H* + HSO; (2) Since sulfurous acid is a diprotic acid, it will dissociate further into H* and SO3. However, dissociation stops with Equation 2 since sulfite quickly reacts with the ionic products of Equation 2 by: SO; + H* — HSO; (3) The SO2 absorption reaction, Equation 4, occurs in- stantaneously when SOz is dissolved in the scrub- bing liquor: SOs + SOo (aq) + HyO — 2HSOZ (4) In addition to the absorption reaction, a portion of the sulfite and bisulfite is oxidized to sulfate via reactions in Equations 5 and 6. SOF + 1/2 O, — SOF (5) HSO; + 1/2 0. — SOF + Ht (6) The resultant solution consists of three primary components in dynamic equilibrium: sodium sulfite (NagSO3), sodium bisulfite (NaHSO3), and sodium sulfate (NagSO,). This solution will efficiently absorb SOz given adequate contact between flue gas and the solution. As the availability of the active absor- bent, NagSOx, is depleted, the equilibrium of the system shifts toward the acidic sodium bisulfite component. Addition of carbonate replenishes sulfite by the following neutralization reaction: CO; + 2HSO; — 280; + CO, 1 +H,O (7) In addition to sodium carbonate, the soda reagent supply contains significant levels of sodium bicar- bonate (0% to 3%). Sodium bicarbonate, like sodium carbonate, neutralizes sodium bisulfite. The major reaction in the water softening system is: Ca** + COF — CaCO, | (8) Ionic forms of calcium in the CTB water include sulfate (SO; ), chloride (Cl), and bicarbonate (HCOs ). Calcium carbonate precipitation begins at a pH above 8. At a pH of 10.0, most of the calcium car- bonate precipitation is complete. However, complete removal of calcium cannot be achieved in sodium carbonate softening. In addition to the precipitation of calcium car- bonate, magnesium and silica are removed in the water softening system. Because the solubility of magnesium salts is higher than that of calcium car- bonate, precipitation of significant amounts of magnesium does not occur until the pH reaches about 9.5. Silica is removed with the magnesium. This, however, is not a stoichiometric reaction. Rather it is an adsorption reaction which typically occurs such that two magnesium ions are removed for each silica ion. The precipitation products in the water softening system are removed by settling or clarification. Recycling settled solids to the reaction compartment improves settling by increasing the size of the crystals. Additional improvement in clarification is facilitated by the addition of flocculant. Pilot plant facilities The pilot plant facility at the Jim Bridger Power Station included most of the elements common to the full-scale FGD system. This included an existing 10,000 ACFM pilot absorber tower owned by PP&L. Some modifications were made to the existing facility for this test program. These included install- ing B&W internals (counterflow tray, spray headers, mist eliminators, etc.) and modifying the inlet flues. Also, the fan, pumps, piping, and controls were replaced. An SOz vaporizer/injection system was in- stalled to permit variability in SO, concentration in the flue gas. Finally, a water softening system was provided to reduce the calcium concentration in the CTB water used for the make-up water streams to the absorber. A schematic of the FGD pilot plant is shown in Figure 2. The absorber is a round tower with a flow area of about 11.5 ft”. The tower is expanded in the mist eliminator section at the top to reduce the gas velocity. There were three liquor spray headers in the absorber. These are the quench, under spray, and upper spray. All headers are supplied by a single absorber recirculation pump. The pilot water softening system consisted of a multi-compartmented treatment tank. The softening system is classified as a high-density, solids-contact clarifier. The pilot program was supported by a fully equip ped portable laboratory which had the capability to perform all the required analyses. Test program The approach to the test program was to model the proposed FGD system as closely as possible and to make every effort to improve upon the expected per- formance levels. This was achieved by dividing the test program into tasks. The primary tasks included: Absorber Nozzle Characterization, Determination of Optimum Tray, Characterization of Optimum Tray, Scale Control, 60-Day Endurance Test, and CTB Treatment. The Absorber Nozzle Characterization was the first task involving the absorber. The purpose of this task was two-fold. First, in preparation for the remaining tests, the effect, if any, of nozzle pressure on SO», removal had to be determined. If nozzle pressure was found to affect performance, then it would be necessary to change nozzles as flow rates were changed to eliminate nozzle pressure as a variable. The second purpose was to generate a data base from which nozzles for the full-scale unit could be selected. Four different nozzle arrangements and designs were used for the first part of the Absorber Nozzle Characterization. Other variables included spray flux and gas flow. The remaining parameters such as tray geometry, nozzle type, pH, density, etc., were constant. The second part of this task included com- parison of 14 nozzle arrangements in which nozzle pressure, design, spray pattern (full cone and hollow cone), height, and spray angle were varied. These nozzles were subjected to rigorous testing in the ab- sorber in which such parameters as gas flow, tray geometry, and SO, concentration were varied. Some testing was also performed at an outside laboratory to characterize droplet size and distribu- tions of various nozzles and determine the properties of nozzle design that affect spray quality. From this testing the optimum pilot nozzle was scaled to full size. The second task was to establish an optimum tray design that would meet the required SO, removal ef- ficiency in the most cost-effective manner. Deter- mination of Optimum Tray consisted of a set of ex- periments to find the most favorable combination of percent open area and hole size on the tray. The op- timization was performed in three parts. First, several tray geometries were tested using a “rotatable experimental design’’ pattern. Second, the results of these tests were correlated to SO removal and tray pressure drop, and a cost evaluation per- formed to select the optimum tray. Finally, the op- timum tray was tested to verify its performance. The purpose of Tray Characterization was to com- plete the performance characterization of the opti- mum tray as a function of inlet SO, concentration and pH (stoichiometry) as well as liquor and gas flow. Correlations to be used for performance testing on the full-scale unit were generated from these test results. Scale Control testing was performed in three parts. First, scale formation was controlled suffi- ciently to permit the successful completion of the previous tasks. Secondly, scale samples were col- lected and analyzed to determine scale composition and identify formation mechanisms. Finally, informa- tion from the first two subtasks was used to formu- late scale control strategies that were implemented in the 60-Day Endurance Test. The 60-Day Endurance Test demonstrated the viability of the previously developed scale control strategies. The absorber was operated as the com- mercial unit would be operated. Periodic inspections indicated the acceptability of the absorber operation. The variables selected for each task involving the absorber are shown in Table III. For most of the controlled variables, at least three levels were tested. These were uniformly spread over the full range that might be experienced even under extreme conditions. The normal number of levels and ranges of the con- trolled variables are shown in Table IV. There were, however, many occasions when tests were performed at additional levels and outside the normal range. For example, the level of gas flows tested included increments of 5% from 30% to 110%. Also, the factorial test plan for Tray Optimi- zation and Characterization of Optimal Tray was changed such that every test condition within the matrix was performed, i.e., a full factorial was tested. Additional tests, including replicates, were CTB From Customer cTB Softening b+ Treated Water Soda Surge Tank Liquor Y To Waste Liquor Storage Tank Waste Liquor Disposal Tank Figure 2 Process schematic Waste Liquor Storage Tank Flue Gas From Unit One Moisture Separator Absorber Flue Gas From Unit Two Sump Waste Liquor Pump Sump SO, Injection performed as necessary. This information was used to evaluate trends and, as a quality assurance func- tion, to determine the repeatability and experimental error. CTB treatment testing was conducted for three reasons. The primary objective was to ensure that the CTB Treatment System design would meet the required process guarantees. Secondly, the effect of certain operating parameters, such as stoichiometry, pH, and concentration of chemical additive on ef- fluent water quality were investigated. The final ob- jective was to maintain the effluent water quality within the guaranteed range of calcium hardness (less than 150 ppm as CaCOs) and turbidity (less than 20 NTU) to prevent scaling in the absorber. In- formation obtained at the pilot would be used as base line information in the design and operation of the full-scale treatment system. Results and discussion Initially, 35 absorber nozzle characterization tests were performed. The results of these tests are shown in Figure 3. This indicates that SO2 removal efficiency improves as nozzle pressure increases at a constant flow, for similar nozzle designs. However, substantial increases in SO2 removal as a result of increased nozzle pressure does not occur until the nozzle pressure is increased several times over the TABLE II] — Variables Characterization 60-Day Absorber Nozzle Tray of Optimum Endurance Parameters Characterization Optimization Tray Test Liquid Spray Rate Vv Vv Vv Cc Gas Flow Rate Vv V Vv Cc Tray Design Cc Vv Cc Cc Stoichiometry Cc Cc Vv Cc Inlet SO2 c Cc V Cc Waste Liquor TDS Cc Cc Cc Cc CTB Treatment Cc c c Cc Spray Nozzle Vv Cc Cc Cc Configuration Spray Nozzle Type Vv Cc Cc Cc Inlet Gas Temperature U U U U Oxygen Concentration U U U U Fly Ash Contamination U U U U Feed Liquor U U U U Composition V = Controlled variable C = Constant U = Uncontrolled variable TABLE IV — Range of Variables Range Number Variable of Levels Low High Units Liquor Flux 5 0 12 gpm/ft? Gas Flow 3 40 100 percent pH 3 6.7 7.7 Inlet SOz 3 400 1200 ppm Nozzle Pressure 3 5 15 psi Tray Hole Size 5 375 1.5 inch Tray Open Area 5 17; 35 percent base condition. A portion of the increase in SOg removal efficiency for each nozzle in Figure 3 is a result of increased contact on the tray. Subsequent nozzle tests resulted in the selection of the optimal pilot nozzle based on economic and full- scale constraints. This nozzle was then compared to a number of full-scale (contract) nozzles. Droplet size was selected as the best parameter to relate nozzle pressure and SO2 removal. Droplet size was deter- mined as Sauter mean diameter (X) or volume- surface mean. By definition, this is the diameter of the droplet whose ratio of volume to surface area is equal to that of the entire spray sample. This is ex- pressed by: n lle He epee ami ELL (9) The Sauter mean diameter for each nozzle decreased as nozzle pressure increased. This is shown in Figure 4. The contract nozzle ultimately selected produced the same average droplet size as the pilot nozzle, but required a lower nozzle pressure than the pilot nozzle to produce the desired droplet size. 98 97F 4 96+ 95+ 94 93 92 91 90 85 Nominal nozzle size ° 3/4" o1" 4 1-1/4" 4A 1-1/2" Load: 100% (12.5 ft/sec) 80 SO, removal efficiency, % 70 60 507 40F 30+ 20 10- guUe41 1 a i Increase —> Spray flux Figure 3. Effect of spray nozzle flow on removal The next task, Tray Optimization, was repeated three times during the test program. The first two trays selected satisfied desired performance levels but were rejected due to results of subsequent activities. The results of the economic evaluation of the original trays are shown in Figure 5. The data points shown represent the relative power require- ment at the spray flux and pressure differential (slanted lines) required to obtain a minimum SO, ef- ficiency of 86.5%. The relative power was based on evaluation factors for pump power and fan power. The dashed line connects the trays with a constant hole size and the no tray (spray tower) case. For this hole size, the most economical geometry is about 28% open area. It is seen that at constant removal efficiency, as the open area increases, the pressure differential decreases even as the spray flux is in- creased. Though the spray tower had the lowest pressure differential, it was not the most economical arrangement. A tray with 26% open area and 39/32-inch holes was the most economical tray of those tested. However, the statistical analysis in- dicated that there is, in fact, a large number of tray geometries that yield similar results to this tray. This occurs due to the interactions of under spray, upper spray, hole size, and open area. Another observation that was made with respect to tray geometry is the relationship between hole size and pressure differential. For each case in which trays with the same open area but different hole sizes were compared, the tray with the larger holes required a higher pressure differential at a lower spray flux than trays with smaller holes required to achieve 86.5% removal efficiency. For example, the 4 Contract nozzle A © Contract nozzle B oO O Pilot nozzle i N N N No 0. Bae N hes 3 OLN aoe an E 2 x 3 < sb 3 Eg g8 ge BEE See ee Increase Nozzle pressure Figure 4 Comparison of contract and pilot nozzles relative power for three 22% open area trays was constant. However, the tray with the largest holes operated at a higher pressure differential than the trays with smaller holes. Trays with open areas below 22% were uneconom- ical. Addition of only a small amount of liquor above these trays resulted in generation of sufficient surface area to achieve 86.5% SO, removal. The working pressure loss, \Py, due to the liquor column or froth height combined with the dry un- recoverable pressure loss, \Pp, was substantially higher than that for larger open area trays. By the equation: APr = APp + APw (10) it can be seen that increase in either dry or working \P increases the total tray pressure loss, \Prp. The dry pressure loss for the trays with open areas below 22% was significant. Therefore, the fan power was high, the relative power was high, and the trays were uneconomical. The optimum tray arrangement is therefore the tray which minimizes dry pressure loss and spray flux. In the third absorber task, tray characterization tests were performed on all three trays selected during Tray Optimization. These tests examined the effect of stoichiometry (pH), SO, concentration, liquor flow, and gas flow on the performance of the absorber. The data was analyzed according to the individual tasks. Statistical analysis was performed for the Tray Optimization and Characterization of Optimal Tray tasks. In performing these analyses, number of Relative Power Increase ——> spray flux Figure 5 Tray optimization transfer units (Np) was utilized. This is related to SOp» absorption (f) by: Np = —In (1-f) (11) The final correlations for both pressure drop (A\P) and number of transfer units was in the form of: AP or Np = Ao + AiV + AgL + AgLV + AqgL? +A5V? +A6L2V + A7LV? +AgL?V? (12) In addition, the correlation for number of transfer units included a term for SO, concentration (Agy). A separate relationship was generated to correct the Nr correlation for variation in pH. The effective spray flux term, L, in Equation 12 accounted for the interaction of the upper and under sprays. The liquor in the upper spray was intimately contacted with the flue gas both in the spray and on the tray. A portion of the liquor in the under spray passed through the holes in the tray and performed similar to that liquor in the upper spray, i.e., it increased the froth height, which resulted in an increase in both SO» removal efficiency and tray AP. The effect due to the under spray, though significant, was less than the effect due to the upper spray. Therefore, to determine the effective spray flux, the upper and under spray fluxes were weighted proportionally to their respective effects on AP or Nrp. The trans- formation of L from upper spray flux to effective spray flux improved both the R? coefficient and standard error of the correlation. Pseudo hole velocity, V, is the absorber tower velocity divided by the square of the fraction open area. Use of pseudo hole velocity as opposed to hole velocity, like the use of effective spray flux, improv- ed the R? coefficient and standard error of the correlation. The relationship for the number of transfer units as a function of stoichiometry is shown in Figure 6. Though this relationship changes as the amount of contact surface varies, the trend remains consistent. Below a stoichiometric ratio of 1.0, the efficiency drops. Under these conditions, the alkalinity is depleted and the equilibrium shifts toward bisulfite. Both the liquid film resistance and the SO vapor pressure increase as the alkalinity is depleted. This can be further clarified by the mass transfer relationship: Na = Ky A ly-y*) (13) From this relationship, it can be seen that a change in the contact area results in a proportional change in mass transfer. Also, as the SOg vapor pressure in- creases due to the depletion of alkalinity, the mass transfer decreases. As the stoichiometry increases toward 1.0, y* approaches zero. A further increase in stoichiometry therefore has no net effect on mass transfer. Figures 7 and 8 illustrate the effect of inlet SO», gas load, and spray flux on SO absorption. There is a slight decrease in SO, removal as the gas flow approaches 60% to 80%. This can be contributed to an “unloading” effect on the tray. As the hole velocity decreases, the froth height and subsequently the contact surface on the tray decreases. At 60% to 80% load, the ratio of gas flow to surface area is therefore less than that ratio at higher loads. At lower loads, the surface area generated in the spray zones becomes significant and the efficiency again increases. The SO2 removal efficiency decreased slightly as inlet SO, concentration increased. From Figure 7, it can be determined that the effect due to inlet SO» concentration was less than 2% removal as the SO, concentration was tripled from background levels. Theoretically, this would not occur without liquid film resistance. As the driving force for SO» absorp- tion increases, then the total mass transfer from Equation 13 would increase proportionally and there would be no change in SO2 removal efficiency. However, during substoichiometric operation, the equilibrium SO, concentration becomes significant. This results in the observed decrease in SOo removal. If the stoichiometry was further decreased, the effect in inlet SO: concentration on removal would be more dramatic. A reduction of spray flux, as shown in Figure 8, 95 ———— % oo 8288 ve T T ° @ SO» efficiency, oO a T 2 So T ~ a T ~ 3 T n i 08 0.9 10 Stoichiometry 1 Figure 6 Effect of stoichiometry on SO removal results in a decrease in SOg removal efficiency. Again, looking at Equation 13, it is seen that the overall mass transfer will decrease if the contact area decreases. The reduction of the spray flux decreases the contact area both in the spray zones and on the tray. This results in the decrease in SOo removal efficiency observed in Figure 8. The primary types of scale observed during the operation of the pilot plant included noncrystalline silica scale and calcium scale. Growth rates were established during the 60-Day Endurance Test. The amount of scale observed was not considered serious. The amount of cleaning required of the full- scale unit should be less frequent than once a year. The desired performance of the CTB Treatment System was obtained by adding enough soda liquor to achieve an effluent pH of 9.8 to 10.0. To further reduce the calcium concentration, the pH was in- creased. The minimum effluent calcium concentration was approximately 20 ppm as CaCO3. This level was obtained at a pH of about 10.4. This can be seen in Figure 9. At this pH, approximately 50% of the magnesium in the CTB water was removed. Also, approximately 80% of the total silica present in the system was removed. During the early operation of the CTB treatment system, flocculants were not used to aid the coagula- tion of solids. The water quality was good (less than 20 NTU), however, the system was sensitive to Be 92+ So ob Sat ° Jie 90 ONO b= ore Thee oe 6—4—,—— 85 F & > 3 5 80h = 3 3 Legend £ 70 - © Background SO, 2 O 700 ppm SO, ~ fo} 4 1000 ppm SO. ? 6h 7 50, 40b 30 20 0 EO yh TT 0 10 20 30 40 50 60 70 80 90 100 gas load, % Figure 7 Tray performance versus load and inlet SO> changes in operating conditions. The use of a floc- culant desensitized the CTB treatment system. Many flocculants were examined including ferric sulfate and anionic, nonionic, and amphoteric poly- electrolytes. Anionic polyelectrolytes in concentra- tions between 1.75 and 2.25 yielded the best results. Summary and conclusions Many parameters had an observed effect on SO removal efficiency. The SO2 removal efficiency was increased by increases in contact area generated by addition of a tray, change in tray geometry, increas- ed liquor spray rates, and increased nozzle pressure. Variation in the gas load, pH, and inlet SO, concen- tration all had an effect on SO, removal efficiency. The absorber AP was influenced by tray geometry, liquor spray rates, density, gas load, and under spray nozzle position. The optimum arrangement for this system was not the case with the lowest pressure differential, the no tray case. Instead, addition of the counter- flow tray reduced the liquor spray rate at a suffi- ciently low increase in AP that a tray was more economical. This is consistent with the results of 92 Nn 91 4. am gob ~~ NG = SS 85 & > N\, om § 80+ Se = © 8 — 70L Legend © a © Spray flux 50% of design 3 © Spray flux 100% of design 60 - 4 Spray flux 120% of design 50 40 - 30 20 10 0 1 4 L 1 1 tL 1 L 1 1 0 10 20 30 40 50 60 70 80 90 100 Gas load , % Figure 8 Tray performance versus load and spray flux 10 trays used for calcium-based systems. The additional hold-up provides both surface area and residence time which results in increased mass transfer at reduced liquor flux. The economics for trays with lower open areas (below 22%) was not as favorable because the dry tray pressure differential was excessive. The effluent calcium concentration from the water softening system reached a minimum of 20 ppm as CaCO3 at a pH of 10.4. The use of an anionic poly- electrolyte at a concentration of 1.75 to 2.25 ppm improved the water clarity by increasing floccula- tion. The use of the softening system contributed to the acceptable scale growth rates. The pilot plant program produced a commercial design with a low L/G and a low system pressure drop especially when compared to lime/limestone systems. This system is capable of meeting the design requirement of 86.5% SOz removal using soda liquor as the reagent. These factors, low L/G, low system pressure drop, and use of soda liquor, com- bine to minimize both the capital and operating costs for the FGD systems at the Jim Bridger Power Station. 300 ° Legend © Unsteady state data 20 NTU turbidity 250 + O Steady stata data 20 NTU turbidity Ny S 6 —— 150 Effluent calcium (ppm as CaCO3) 8 T 50 [ Ho ome [ ago oO 0 stiri ti ptiriitiiin 9.0 95 10.0 10.5 11.0 115 Treated water pH Figure 9 Effluent calcium concentration Technical Paper Operational experience with chromized boiler tubes and steam leads A. J. Blazewicz J. M. Tanzosh Fossil Operations Division Babcock & Wilcox Barberton, Ohio R. N. Deardoff Dayton Power and Light Company Dayton, Ohio Presented to EPRI/TVA Solid Particle Erosion of Steam Turbines Symposium Chattanooga, Tennessee November 13-15, 1985 Babcock & Wilcox BR-1277 a McDermott company Operational experience with chromized boiler tubes and steam leads A. J. Blazewicz J. M. Tanzosh Fossil Operations Division Babcock & Wilcox Barberton, Ohio R. N. Deardoff Dayton Power and Light Company Dayton, Ohio Presented to EPRI/TVA Solid Particle Erosion of Steam Turbines Symposium Chattanooga, Tennessee November 13-15, 1985 Abstract PGTP 85-33 After extensive development and long-term testing, Babcock & Wilcox (B&W) in 1978 commercially intro- duced chromizing as a solution to the turbine solid particle erosion (SPE) problem. The B&W chromizing process produces a chromium rich, metallurgically-alloyed surface that is virtually oxidation and exfoliation resistant when exposed to a high temperature steam environment. Based on long-term laboratory oxidation tests, the expected life of the chromized surface was estimated to exceed 80 years at 1050 F steamside metal temperatures. In 1978, Dayton Power and Light Co., after extensive studies and evaluation of several alternatives to alleviate the severe SPE problem at their J. M. Stuart Station, decided to control future SPE problems by chromizing. For their new Killen unit, Dayton Power and Light Co. decided to chromize all surfaces exposed to 900 F-and-higher steam temperatures including the main and hot reheat steam lines. This paper will review some of B&W’s testing and evaluation of the chromized layer, including exposure in an operating unit for 15 years. It also reports on the Dayton Power and Light Co. chromizing experience, including projected economic evaluations based on operating experience. Introduction Turbine damage due to solid particle erosion (SPE) has been a long term industry problem that in- creases operating cost and decreases unit availability and efficiency. The source of the problem is the oxide scale that is formed on the inside surfaces of superheater and reheater tubes, headers, connecting pipe, and steam leads. When the oxide scale be- comes thick enough, the stresses generated by the difference in the thermal expansion coefficients of the scale and metal substrate will cause the oxide scale to exfoliate, or spall off. It then becomes entrained in the steam flow to the turbine thus causing erosion of turbine components. B&W’s solu- tion to SPE has been to eliminate the source of the problem by chromizing surfaces exposed to high temperature steam. This chromizing process results in a chromium diffusion surface that is virtually oxidation and exfoliation resistant in high temperature steam. In 1978, after extensive laboratory studies and evaluations, and long term exposures in operating units, Dayton Power and Light Co. accepted Babcock & Wilcox’s proposal to chromize the inside surfaces of superheater and reheater tubes, headers, connecting pipe and steam leads for the 600 MW cyclic (or two shift) operation unit then being de- signed for the Killen Station. This paper describes B&W’s chromized surface, laboratory and field ex- perience, Dayton Power and Light Co.’s operational experience with chromizing, and the resultant savings achieved due to elimination of turbine effi- ciency losses from SPE. Babcock & Wilcox chromized layer The chromizing process, developed by Babcock & Wilcox and the Alloy Surfaces Co. of Wilmington, Delaware, is the metallurgical alloying of chromium to the surfaces of boiler components. The process provides a distinctive layer that consists of an outer, thin chromium-rich carbide adjacent to a thicker chromium-diffused zone possessing a columnar ferrite microstructure. A thin decarburized zone is left in the adjacent base metal substrate. The micro- structure of a typical B&W chromize layer is shown in Figure 1. The chemistry of the layer including the transition into the substrate base metal of a typical chromized SA213T22 (Croloy 2%4*) tube is shown in Figure 2. The chromized layer is generally 2 mils thick consisting of a thin chromium-rich (about 80% " +—Mounting Media ~~Carbide Zone +—Columnar Ferrite Diffusion Zone ~ ~~ Decarburized ae . : fay Bee: z SAS Figure 1 Chromized Croloy 2-1/4. 400X 04 08 12 16 19 29 39 49 Mils 100} Iron Weight 80 Percent 60] 40] * Chromium 0 rp Surface Layer Diffusion Base Carbide Zone Material 7.0 Weight Percent 5.0} 3.0 Carbon 9 eS ee 0 10 20 30 40 50 75 100 125 Microns Figure 2 Electron probe analysis of chromized Croloy 2-1/4 tube. *Croloy 2% is the Babcock & Wilcox trade name for 2%Cr-1Mo steel. chromium) carbide, and a thicker columnar ferrite diffusion zone (16-20% chromium) beneath the car- bide. The decarburized zone in the base metal substrate typically ranges from 7 to 15 mils deep. A microhardness traverse on a piece of chromized Croloy 2% tube, starting at the chromium carbide edge through the decarburized zone, is shown in Figure 3. The results show that the chromium car- bide zone is significantly harder than the remainder of the material, with an approximately equivalent value of Rockwell C (Rc) 58. Values in the de- carburized zone correspond to Rockwell B (Rp) 60-65, while those in the base metal and the colum- nar ferrite diffusion zone range around Rp 85. Oxidation resistance Chromium has long been recognized for its contri- bution to the resistance of alloys to corrosion and high temperature oxidation. Steam and air oxidation tests conducted at B&W’s Alliance Research Center and in field trials have proven that the chromium- enriched surface is highly effective in reducing oxidation and exfoliation under both cyclic and isothermal conditions. ¢ Laboratory Air Oxidation Test Chromized and unchromized Croloy 2% tubes were cyclically heated in air in a laboratory furnace. The total test duration was 20,000 hours with approximately 16,630 hours at the test temperature of 1200 F. The test included 116 heating cycles, 113 of six-day duration, and three w= 695 Hardness ia” === 180 oy 1/597 > os wapl23) * —= 128 ieee Oda? at Figure 3 Microhardness (100 gram load Knoop) traverse on chromized Croloy 2-1/4 tube. 70X td a 20 um ALLOY Me a Ta MOUNTING MEDIA } te Sie : ‘COLUMNAR FERRITE _ DIFFUSION ZONE Figure 4 Cross sections of tube surfaces after cyclic air oxidation test with 16,630 hours at 1200 F. 100X (a) Unchromized Croloy 2-1/4. (b) Chromized Croloy 2-1/4. of 13-day duration. During the cool-down periods, which lasted one day, the temperature dropped to approximately 135 F before the beginning of the next heating cycle. Figure 4 compares cross sections of the un- chromized and chromized surfaces after the air oxidation test. Figure 5 shows a cross section of the chromized surface at higher magnification. The total depth of oxide penetration is approxi- mately 0.6 mils. The penetration rate for the 16,630 hours at 1200 F is approximately 0.3 mils/yr. Laboratory Steam Oxidation Test Chromized and unchromized Croloy 2% tubing, severely bent chromized Croloy 2% tubing, and 304H stainless steel tubing were exposed to dry steam in a laboratory test autoclave. To accelerate oxide scale formation and exfoliation, and to better simulate power plant conditions, the ex- posure temperature was cycled weekly, with a cool-down period of one day. Total test duration was 6560 hours with 5480 hours at full tempera- ture and 1000 psi pressure. The effective metal average temperature for the hold periods was 1180 F. During the weekly 24-hour cool-down periods, the temperature dropped to between 450 and 520 F before the heating cycle resumed. The test temperature is sufficiently above the normal maximum anticipated service temperature for Croloy 2% tubing to achieve substantial accelera- tion of the oxidation process. Figure 6 shows unchromized Croloy 2%, 304H stainless steel, and chromized Croloy 2% tubes “MOUNTING MEDIA ALLOY 50 um Figure 5 High magnification of chromized surfaces after cyclic air oxidation test. 400X after exposure in the steam oxidation test. The unchromized Croloy 2% tube is covered with an approximately 70 mils thick oxide scale, part of which exfoliated during the test exposure. The 304H stainless steel tube is covered with multiple layers of thin scale, 0.5 to 3 mils, much of which exfoliated during the test exposure. The chromized tube is covered with a particularly thin, 0.02 to 0.2 mils, layer of oxide scale with no evidence of exfoliation. Field Experience Babcock and Wilcox chromized superheater and reheater tube samples have been, and are cur- rently, installed in operating units for periods that now exceed eighteen years. Figure 6 Tubes after steam oxidation test exposure: (a) Unchromized Croloy 2- 1/4. (b) TP304H Stainless Steel. (c)Chromized Croloy 2-1/4. Oxide Mounting Scale Media SA 213 TP 321, unetched 250X Figure 7 After 8 years exposure. 250X A. J. Blazewicz and M. Gold! reported on tube samples that had been removed for laboratory evaluation after 6, 28, 30, and 58 months, in addi- tion to a sample that was removed from Texas Power & Light Co., Valley 2 unit, after 8 years. In all cases, the chromized surface was reported to be in excellent condition and was judged to be able to effectively prevent steamside exfoliation for the life of the boiler. After eight years ex- posure, the chromized surface had shown better oxidation resistance than a SA213TP321H stain- less steel tube that was exposed to the same steam conditions. The high temperature oxidation resistance of the chromized surface can be seen in Figure 7 by noting the difference in scale thick- ness between the chromized Croloy 2% tube and the type 321 H tube. In 1983, another section of Oxide Mounting Chromium Diffusion Scale Media Carbide Layer Zone ane PP NN CaP hi oe] Chromized 250X Croloy 2'4, 5% Nital etch chromized tube was removed after 15 years ex- posure from the Valley 2 unit for laboratory evaluation. The results of the laboratory exami- nation showed that little, if any, degradation occurred in the 1000 F steam in the seven years since the last examination. As shown in Figure 8, the chromized layer on the tube ID surface of the tube continues to resist oxidation and exfoliation. © Chromized Layer Life Expectancy Based on the penetration rates that were observed during the laboratory steam and air oxidation test, and estimated penetration rates assuming an Arrhenius-type dependence of the oxidation rate on temperature, the maximum and minimum life expectancy was estimated for different tempera- tures. For 1050 F metal temperatures exposed to Oxide Mounting Chromium _ Diffusion Scale Media Carbide Layer Zone + a - LEE a Pe Figure 8 Chromized tube sample after 15 years exposure. 100X steam, the estimated life expectancy for a 0.6 mil thick chromium rich carbide layer was calculated to be from 80 to 500 operating years. At 1100 F the estimated carbide life expectancy was calcu- lated to be 25 to 150 years. It should be noted that this life expectancy only relates to depletion of the carbide layer. Beneath the chromium rich carbide is the columnar ferrite diffusion zone that will provide additional oxidation resistance, the equivalent of a 400 series stainless steel. The laboratory air and steam oxidation test, and the excellent condition of the chromized samples after long-term exposure in operating units, indicate that the chromize layer will provide oxidation and exfoliation resistance for the normal life of the boiler. Bond integrity The integrity of the metallurgical bond between B&W’s chromized layer and the base metal was ex- tensively evaluated in the laboratory and confirmed in field trials. ¢ Laboratory Evaluation of Bond Chromized Croloy 2% tubes were deformed by flattening to various OD strain levels (Figure 9). Microscopic examinations show that the normal cold and hot bending manufacturing operations (12% strain) will crack the outer carbide, and that with severe deformation (28% strain) the columnar ferrite diffused zone will also crack to the base metal. However, in no case was the chromized layer separated from the base metal, nor did any As Chromized \ 12% Strain 20% Strain 28% Strain singe nea “en 4 iy Figure 9 Chromized tube sections after deformation. localized spalling occur. The examinations also showed that the decarburized ferrite layer under the diffused zone blunts the cracks, preventing them from propagating into the base metal. The metallurgical bond between the chromized layer and the base metal and the effects of the chromized layer cracking were further evaluated in high temperature air and steam oxidation studies. Severely deformed chromized tube samples with the layer cracked to the base metal were exposed to air at 1200 F for more than 16,000 hours under cyclic and isothermal conditions, and in steam at an average temperature of 1180 F, under cyclic conditions, for 5480 hours at temperature. The test results reported by P. L. Daniel, et al.,” stated there was no loss of the chromized layer due to oxidation penetrating through the cracks and under the layer. It was also reported that the cracked chromized layer significantly reduces the local oxidation rate by blocking (reducing the cross section area available for) diffusion through the oxide scale. This is illustrated in Figure 10 where the oxide thickness in the area of particu- larly severe coating damage is compared with an unchromized Croloy 2% specimen. Dayton Power and Light Co., Killen Station Killen Electric Generating Station is located on a 2200 acre site along the Ohio River in Adams County, Ohio, approximately 100 miles south of Dayton. The station is jointly owned by the Dayton Power and Light Company and the Cincinnati Gas & Electric Company. Dayton Power and Light, which has 67% ownership, has the responsibility for the day-to-day operation of the plant. The steam generator is a Babcock & Wilcox balance draft, pulverized coal-fired unit with a maxi- MOUNTING MEDIA ALLOY a 500 um i MOUNTING MEDIA | oi : b 500 um ‘| Figure 10 Cross sections indicating the chromized layer impedes oxide formation. 50X. (a) Unchromized Croloy 2-1/4. (b) Chromized and severely deformed Croloy 2-1/4. mum continuous rating of 4,545,000 pounds per hour at 2620 psig superheat outlet pressure, and with superheat and reheat outlet temperatures of 1005 F. The turbine-generator is a General Electric tandem compound three-casing unit with two double-flow low pressure sections with 30-in. last stage buckets. The nameplate rating of the turbine is 612,574 kW at a throttle pressure of 2400 psig. The generator is rated at 734,000 kVA, 24,000 volt with a .90 power factor. In the initial design stages of Killen Station, it was decided that the Dayton Power and Light system’s electric load variations required that the unit be capable of cyclic (or two shift) operation. This means the station will operate with the in- tegrity and cycle efficiency of a base loaded unit, but is fully capable of efficiently and effectively being shut down every night and started up every morning, with shutdowns over the weekends. During the design of the Killen Station, it was also determined that the four 600 MW coal-fired base loaded units at J. M. Stuart Station were ex- periencing extensive solid particle erosion problems. The severity of the problem required major main- tenance outages to replace or rebuild turbine blades and diaphragms every five years. With the know- ledge that the Killen Station would be operated in the cyclic mode, which could cause an even worse SPE problem, the decision was made in 1978 to eliminate the source of the problem by chromizing. Dayton Power and Light decided to chromize the in- side surfaces of the superheater and reheater tubes, the headers, connecting pipe and the main and hot reheat steam leads exposed to 900 F-and-higher steam temperatures. Steam blow test One of Dayton Power and Light’s concerns in selecting chromizing, a previously untried product, was the ability of the chromized surface to with- stand the debris that is removed during the steam- blow cleaning prior to the start-up of a new unit. To address this concern, Dayton Power and Light requested that five ID chromized test sections be installed in selected areas of the temporary steam- blow lines for the Killen unit. After the steam blow cleaning of the unit, a 12 in. schedule 100 pipe elbow (90°) and a 18 in. schedule 80 pipe elbow (90°) were judged as being the ‘‘worst case’’ area for damage and selected for laboratory evaluation. The 12 in. pipe elbow was located downsteam of one of the main steam stop valves and was exposed to more than 127 steam blows. The 18 in. pipe elbow was located at one of the combined reheat valves and experienced more than 83 steam blows. Ring samples were taken for laboratory examina- tion from both elbows at the locations of direct steam impingement. The macro examination of the samples showed a large number of “‘crater-like”’ in- dents on the ID surface of the 12 in. pipe, obviously due to impact of debris during the steam blows. By comparison, the 18 in. pipe did not have any indents. Figure 11 shows the indents on the 12 in. pipe and the direction of the steam flow. Metallographic examination showed that the in- dents did not penetrate the chromized layer. Figure 12 shows a section through the deepest indent. (Note the mechanical twins below the indent caused by the rapid deformation associated with impact). Table 1 lists the microprobe analysis of the chromized layer at the deepest indent. The probe = 2 a € o 2 a = 3S < 2 3 2 & Figure 12 Chromized layer at deepest indent. 80X Note: Mechanical twins below indent. analysis revealed that the chromium levels were similar to those at the unaffected areas; most im- portantly, that the chromium carbide layer was intact. The laboratory examination of the chromized test samples showed that the steamblow test produced only minor damage and did not change the chemical make-up of the chromized layer. Calculated savings due to chromizing It is generally accepted that the exfoliation of the oxide scale that is formed on the surfaces exposed to high temperature steam is the primary source for the turbine SPE problem. The SPE problem has a definite effect on turbine efficiency which results in increased unit heat rate and turbine maintenance cost. The potential savings due to chromizing the Table 1 — Microprobe analysis of chromized layer on 12” OD ring at location of deepest indent Distance Cr Fe Mn Si Mo (mils) (wt %) (wt %) (wt %) (wt %) (wt %) Surface 69.6 19.5 0.19 0.04 2.70 0.3 37.1 616 015 044 0.64 1.4 27.7, 71.0 0.24 041 0.92 Diffusion Zone 24 21.0 77.7 0.29 0.34 0.91 3.9 15.3 83.5 0.33 0.37 0.94 5.4 12.8 85.6 0.34 0.35 0.86 6.4 12.0 85.9 0.37 0.34 0.98 7.5 8.0 85.9 0.42 0.29 0.98 Decarb Zone 8.3 2.6 95.6 0.47 0.29 1.05 9.5 2.2 968 0.46 036 091 Killen unit to substantially reduce the SPE problem was analyzed by Dayton Power and Light Co. and published in the EPRI ‘“‘First Use’’ paper, 3323D (RP644-1). The benefits that are expected include: ¢ Minimum efficiency losses due to turbine erosion. ¢ The elimination of extensive repairs to the tur- bine rotor and diaphragms every five years. ¢ Substantial reduction of turbine maintenance cost and unit outage time. The economic analysis that was performed by Dayton Power and Light Co. for the ‘First Use”’ paper is shown in Table 2. The analysis indicates a levelized annual savings of $216,000 due to chromizing the Killen unit. Table 2 — Calculated savings due to chromizing Cumulative Investment Savings Fixed Year ($000) Charges O&M Total Estimated Savings ($000) 1982 (1535) (307) 10 (297) 1983 (1535) (307) 39 (268) 1984 (1535) (307) 117 (190) 1985 (1535) ~— (307) 241 (66) 1986 (1535) (307) 1801 1494 1987 (1535) ~—-(307) 190 (117) 1988 (1535) — (307) 257 (50) 1989 (1535) ~—-(307) 395 88 1990 (1535) (307) 591 284 1991 (1535) (307) 2806 2499 Present value of total estimated savings ($000) Levelized annual savings ($000) $ 216 The following assumptions were used in the calcula- tions for Table 2. ¢ Cumulative investment savings are based on the following: 1. Cost of chromizing high temperature boiler tubing and steam piping = $1,535,000 (1982 $). ¢ Fixed charges = levelized fixed charge rate of 20% times the cumulative investment savings. ¢ O&M savings are based on the following: 1. Turbine efficiency losses (Btu/kWh) avoided by chromizing in each of the first ten years of operation are: 1.8, 6.6, 18.1, 33.1, 48.1, 21.8, 27.3, 38.8, 53.8, and 68.8 Btu/kWh. 2. Fuel cost estimates ($/M Btu) for the first ten years of operations are: 1.96, 2.14, 2.34, 2.64, 2.90, 3.16, 3.41, 3.69, 3.98 and 4.32. 3. The Killen Station capacity factor is 45%; the boiler efficiency is 90%. 4. Example calculation of first year losses: (1.8 Btu/kWh) (630,000 kW) (8,760 h) ($1.96/M Btu) (0.45) (1/0.9) = $9,735. 5. Turbine HP/IP rotor and diaphragm rebuild which will not be required in 1986 and 1991 = $800,000 (1982 $). 6. The avoided one-week additional outage costs in 1986 and 1991 are: Maintenance cost = $140,000; replacement cost at $0.005/kWh = $227,000 (1986 $). 7. The above calculation assumes a cost escala- tion factor of 7% per year to reflect inflation. © 1982 present value calculations assume a 12.5% discount rate. Killen Station operational experience The Dayton Power and Light Co., Killen Station began commercial operation in June 1982. Since the start of operation the unit has been on line for 12,134 hours through August 1985. The cyclic oper- ation of the station is demonstrated by the 691 start-ups that were experienced by the unit during the 38 months of operation. By comparison, the base-loaded units at the Stuart Station averaged about 30 start-ups per year and for a comparable period of time would have been on line for about 20,000 hours. The disparity between the operating hours and the number of start-ups between the Killen and Stuart Stations makes it difficult to directly compare their historical turbine efficiency data. J. A. Haberman and H. Keyton’ reported that, based on observations at the Stuart Station, very little erosion damage was found in the first two to three years of turbine life, a time during which most of the rubbing and wear of the seals and packing probably takes place. They assumed that 70% of the observed turbine efficiency loss during this period was due to increased leakage with the remainder due to SPE. SPE was then assumed to account for 80% of the losses observed in years four and five. They also reported that, bas- ed on damage to the turbine bypass valve stems and the fact that no real damage was evidenced to the stop and control valves, most of the damage oc- curred during start-up. They concluded that fre- quent start-ups are a major contribution to the tur- bine SPE problem. After 12,134 hours of operation and 691 start-ups, the annual enthalpy drop test for the Killen unit showed turbine efficiency losses of approximately 1.9% for the High Pressure (HP) and 1.5% for the Intermediate Pressure (IP) turbines. Without open- ing the turbine, it is impossible to accurately quantify the slight degradation of efficiency as being caused by turbine blade buildup, turbine blade damage, or to the increase in internal stage clearances. Dayton Power and Light Co. intends to defer the opening of the turbine until such time that the annual enthalpy drop test shows efficiency losses that will dictate the opening of the turbine for in- spection. In the meantime, we will closely monitor the Killen operating hours, particularly between 14,000 and 34,000 hours, when a sharp drop in tur- bine efficiency was observed on the base-loaded Stuart units. At that time, a more definitive answer will be made as to the effectiveness of chromizing on turbine efficiency losses. References 1. A. J. Blazewicz and M. Gold, ‘‘Chromizing and Turbine Solid Particle Erosion’ presented at the ASME/IEEE/ASCE Joint Power Generation Con- ference, Dallas, Texas, September 10-14, 1978. B&W paper BR-1124. 2. P. L. Daniel, et al., “Steamside Oxidation Resistance of Chromized Superheater Tubes”’ presented to the National Association of Cor- rosion Engineers, Corrosion/80 Conference, Chicago, Illinois, March 3-7, 1980. B&W paper BR-1158. 3. J. A. Haberman and H. Keyton, ‘Considerations Leading to Selection of Chromized Boiler Internal Surfaces by the Dayton Power and Light Com- pany,” presented to American Power Conference, Chicago, Illinois, April 26-28, 1982. B&W paper BR-1217. Technical Paper Furnace-wall corrosion in refuse-fired boilers J. L. Barna Materials Engineer Power Generation Group Babcock & Wilcox Barberton, Ohio J. D. Blue Manager, Industrial Boiler Design Power Generation Group Babcock & Wilcox Barberton, Ohio P. L. Daniel Severe Environment Corrosion Supervisor Materials Performance Babcock & Wilcox Alliance Research Center Aliliance, Ohio Presented to ASME Twelfth Biennial National Waste Processing Conference Denver, Colorado June 1-4, 1986 Babcock & Wilcox BR-1279 a McDermott company Furnace-wall corrosion in refuse-fired boilers J. L. Barna Materials Engineer Power Generation Group Babcock & Wilcox Barberton, Ohio J.D. Blue Manager, Industrial Boiler Design Power Generation Group Babcock & Wilcox Barberton, Ohio P. L. Daniel Severe Environment Corrosion Supervisor Materials Performance Babcock & Wilcox Alliance Research Center Allliance, Ohio Presented to PGTP-85-40 ASME Twelfth Biennial National Waste Processing Conference Denver, Colorado June 1-4, 1986 Abstract Modern refuse-fired boilers dispose of refuse while extracting energy for generation of steam and elec- tricity. One of the factors that must be considered in the selection and design of these units is corro- sion. This paper describes and discusses furnace wall corrosion, citing specific experience in a refuse- derived fuel (RDF) boiler. Tubing from the RDF-fired boiler was examined in the laboratory to identify corrosive species in the furnace. High concentrations of chlorine, sodium, potassium, and zinc were associated with areas of corrosion. Laboratory tests simulating the furnace environment were conducted to establish rela- tive corrosion rates of carbon steel and higher alloys. It was concluded that HCl and NaClare not sufficiently corrosive to explain observed furnace-wall corrosion rates while mixtures of ZnCly and NaCl readily do so. Alloy 625 cladding on furnace- wall tubes was found to provide good resistance to chloride attack in laboratory short-term screening tests, and should also resist out of service chloride pitting and chloride stress corrosion cracking. This resistance to chloride attack has been tentatively confirmed in a six-month trial of test overlays in an operating RDF unit. Laboratory and field tests are in progress to further evaluate this and other means to address fireside corrosion problems in refuse-fired boilers. Introduction Many refuse-fired boilers are in operation continued development toward units operating throughout the world. These units provide an more efficiently and at higher steam temperatures attractive means for disposal of refuse while and pressures. extracting energy for generation of steam and However, this progress is limited by the occur- electricity. There remains economic pressure for rence of fireside corrosion problems. Refuse-fired boilers generally suffer more fireside corrosion than do other boilers. Alkali metals, heavy metals, and chlorides in the fuels corrode furnace- wall and convection pass superheater tubes. The problem is exacerbated by the inherent heteroge- neity of the fuel which makes furnace control dif- ficult and by high fouling rates which in some designs require frequent boiler outages. While these problems by no means threaten the viability of the modern boiler design, they do increase maintenance costs, decrease unit reliabilities and efficiencies, and limit design and operational flex- ibilities. Corrosion problems occur on supports, superheater, and furnace water wall tubing. Cor- rosion problems on higher temperature compo- nents are described in a previous paper (1). This paper uses experience with one particular unit to illustrate the furnace-wall corrosion problem, how corrosion relates to other boiler characteristics, and how the corrosion problem can be addressed. The subject boiler was built by The Babcock & Wilcox Company. The unit is designed to burn 950 tons/day of 4800 Btu/lb refuse derived fuel (RDF). The boiler is designed to produce 250,000 Ibs/hr of 750°F (399°C) steam at a pressure of 650 psig. The saturation temperature is approximately 500°F (260°C). The furnace-wall tubes are 2.5” OD x 0.165” wall (63.5 mm x 4.2 mm), SA-178A carbon steel tubing, in a membrane wall construction. The unit configuration is shown in Figure 1. Furnace-Wall Corrosion Corrosion of furnace-wall panel tubes with fire- side metal temperatures of 500°-600°F (260-316°C) SSS eew_ LJ Figure 1 Drawing of the RDF- fired boiler. in refuse-fired boilers is frequently localized, occurring above the fuel feed points, along sides, or in corners of the furnace. The severity of the corrosion generally increases with increasing proximity to the flame zone. The corrosion kinet- ics are usually linear with the corrosion rate showing little tendency to decrease with the pas- sage of time. Corrosion rates increase rapidly with increasing tube metal temperatures (2). Experience indicates strong correlation of these wastage problems with a number of boiler operat- ing conditions. Conditions that aggravate the problems include: e Refuse fuel rich in PVC plastics and, hence, in organic chlorides, e Higher fuel Btu-values and higher gas temperatures, e Higher tube temperatures and waterside deposi- tion which interferes with heat removal, e Flame impingement and stratification of air flow, e Frequent and thorough fireside deposit and scale removal. Several approaches have been taken to amelio- rate the furnace-wall corrosion problems: e Covering the wall with silicon carbide, ¢ Controlling the fuel chemistry (e.g., burning a mixture of coal and refuse), e Lowering the steam pressure and thereby lower- ing the saturation and tube-metal temperatures, and e Decreasing the amount of flame contact with the furnace wall. Each of these ameliorative actions costs a penalty in terms of unit operating load, efficiency, and flexibility. Maximum steam temperatures and pressures, and boiler efficiencies are limited by the danger of corrosion; and boiler designers attempt to maximize these desirable unit charac- teristics while minimizing the potential for fire- side corrosion problems. The subject unit was not lined with silicon car- bide because its use impedes heat transfer. For the design fuel characteristics and metal tempera- tures and with good air and fuel distribution in the furnace, the unit was expected to operate within the envelope of safe operating conditions with respect to furnace-wall corrosion. However, once the unit went into operation, the heating value of the fuel delivered to the plant was 15-25% above the design heating value. If the design RDF throughput (lbs/hr) was maintained the heat input to the unit (Btu/lb) would increase corres- ponding to the heating value increase. The unit also experienced water chemistry upsets during the period of initial operation. The resultant high heat release rates and waterside deposition would raise the tube-metal temperature into the range where fireside corrosion becomes a serious prob- lem for localized areas of the furnace wall in the flame zone. In addition to raising the tube-metal temperature, excessive heat release rates and resultant steep thermal gradient across fireside wall deposits would accelerate diffusion of corro- sive species to the tube-metal surface. Ultrasonic thickness measurements indicate that corrosion rates of 30 to 80 mpy (0.8 to 1.8 mm/yr) may occur in localized sections of the lower furnace walls. Tube Examination A rear wall tube from the subject unit was exam- ined in the laboratory. The tube had been in ser- vice since the original unit startup in July, 1984. It was in service for approximately six months during which time it had suffered a 25%, 40 mil (1.0 mm) loss in thickness or a corrosion rate of 80 mpy (2.0 mm/yr). When received for laboratory analysis, the tube retained a sooty external de- posit. When one gram of deposit was dissolved in 100 grams of water, the resultant solution had a pH of 4.0. The tube surface was generally rough and pitted. Procedure Bulk deposit chemistry of the fireside deposit was obtained from x-ray diffraction and spectrogra- phic analyses. Optical metallography was used to determine the deposit and scale morphologies. To retain water-soluble constituents of the fireside deposits, kerosene was substituted for water dur- ing the metallographic preparation. Electron mic- roprobe analysis of the metallographic samples provided microchemical information about the fireside deposits. Results The chemical analyses of the fireside deposit on the rearwall tube are given in Table 1. The depos- its were composed primarily of iron, sodium, and chlorine. Large amounts of the following elements were also identified: silicon, aluminum, titanium, calcium, zinc, lead, carbon, sulfur, potassium, and oxygen. X-ray diffraction analysis identified sodium chloride (NaCl) as the major crystalline constituent and identified hematite (Fe203) and magnetite (Fe304) as minor constituents. Photomicrographs depicting the nature of the fireside corrosion and associated deposit mor- phology are shown in Figure 2. The deposits were Table 1 Spectrographic Semi-Quantitative Analysis (%)* Of Fireside Deposits Rear Wall Tube Silicon as SiO. 3.3 Aluminum as Al,03 13 lron as Fe203 Major Titanium as TiO2 3.0 Calcium as CaO 25 Magnesium as MgO 0.2 Sodium as Na20** 20.89 Nickel as NiO < 0.06 Chromium as CR203 < 0.06 Molybdenum as MoO3 < 0.06 Vanadium as V205 < 01 Cobalt as CoO < 0.06 Manganese as MnO» 0.06 Copper as CuO 0.4 Zinc as ZnO 2.3 Lead as PbO 12 Tin as Snug 0.2 Zirconium as ZrO < 0.06 Carbon as C*** 5.68 Sulfur as SO3 2./0t* Chlorine as Cl*** 22.7 Phosphorus as P20; **** NA Potassium as K,0** 6.14 X-Ray Diffration (Crystalline Constituents) Major NaCl Minor Fe203, Fe304 unidentified Notes: * The results of spectrographic analysis are re- ported by the Research Center as the oxides. This does not necessarily mean that the elements are Present as such in the sample. ** Flame photometer analysis. *** Wet chemical analysis **** ICP spectrographic analysis Major - > 20% NA - not analyzed porous, grey in color, and multi-layered. A layer of protective iron oxide scale, adjacent to the tube metal, was not always visually apparent. The rearwall tube deposits were generally adherent and were measured from cross sections to be approximately 20 mils (0.5 mm) thick. The x-ray scanning capability of the electron microprobe was used to determine the relative distribution of several elements in the fireside deposits. Resultant x-ray fluorescence maps, ISCALE| [a ALLOY 25um | Figure 2 Photomicrographs of the fireside deposit and scale on the rear wall tube. along with electron images of the area, are shown in Figure 3. Examination of the element maps in Figure 3 reveals several patterns: e Chlorine and zinc are distributed throughout the deposit. ¢ Sodium is more abundant in outer portions of the deposit. e Potassium is more abundant in the inner por- tion of the deposit near the tube surface. e There is a thin layer at the metal/deposit inter- face that is iron-rich. This layer does not appear to be associated with oxygen, but there is a def- inite association with chlorine. e Lead, while scarce, is concentrated near the metal/deposit interfaces. ¢ Oxygen is present in relatively low concentra- tions uniformly distributed throughout the deposit. e Relatively small amounts of sulfur are ran- domly distributed through the deposit. Laboratory Tests Laboratory tests were undertaken to measure cor- rosion rates in zinc chloride and sodium chloride salts at 600°F (316°C). The tests helped determine the corrosion mechanism and served to screen alloys for use as weld overlay materials to protect the wall tubes from further attack. The test tem- perature was 600°F (316°C), which is approxi- mately the maximum operating furnace wall tube metal temperature in this unit. 50 um Sn (a) ELECTRON IMAGE (b) IRON MAP 50 um (c) LEAD (e) POTASSIUM (f) OXYGEN (g) CHLORINE (i) SULFUR Figure 3 Element maps of rear wall tube deposit at the metal/deposit interface. This area is the same shown in the photomicrographs in Figure 2. The materials tested, which included carbon steel and higher alloys, are listed in Table 2. Iron oxide, Fe304, was included in the test to determine the extent to which the initially protective oxide, once formed on steel, would be attacked. Also tested, was a section of furnace-wall tubing (car- bon steel) clad with the commercial nickel-base Alloy 625, an alloy believed to possess superior corrosion and pitting resistance in the anticipated furnace environment. Each test specimen was a rectangular sheet coupon that was initially prepared by rinsing in deionized water and acetone followed by ultra- sonic cleaning to remove any foreign material. The specimen coupons were weighed and placed in separate crucibles and covered with the chlo- ride mixture with air as the cover gas. The chloride mixture used in the tests was of the two corrosive salts of particular concern, ZnClg and NaCl. The mixture was 84 wt% ZnCly and 16 wt% NaCl. At the conclusion of each of the tests, the samples were rinsed with deionized water and acetone, acid cleaned to remove any oxide forma- tion, and reweighed. Resultant corrosion rate data are presented in Table 2. Table 2 Corrosion Rates In Chloride Salts At 600°F (316°C) 84% ZnClp + 100% NaCly Time 16 wt% NaCl. Rate, mpy Material (hrs) Rate, mpy (mm/yr) | (mm/yr) Carbon Steel 168 1700 (43) 6 (0.15) Sample A Carbon Steel 336 910 (23) Sample B Croloy 2-1/4 168 410 (10) 1 (0.03) Fe304 168 750 (19) a Alloy 600 168 5 (0.13) 0.1 (0.003) Alloy 625 168 2 (0.05) 0.2 (0.005) Aloy 625 336 <10 (<0.25) clad on carbon steel Alloy 625 672 <12 (<0.30) clad on carbon steel Alloy 690 168 3 (0.08) 0.0 (0.000) Alloy 800 168 3 (0.08) 0.0 (0.000) Alloy 304 168 3 (0.08) 0.1 (0.003) Alloy 309 168 11 (0.28) 0.2 (0.005) Corrosion Mechanism Corrosion in the RDF-fired unit appears to be caused by chlorides which deposit on the furnace- wall tubes. Several different modes of chloride corrosion may occur: ¢ Corrosion by HCl in the combustion gas, e Corrosion by NaCl and KC] deposits on tube surfaces, e Corrosion by low melting point metal chlorides (mainly ZnCl»), and ¢ Out-of-service corrosion by wet salts on the tube surface. Corrosion by HCI in the Combustion Gas Chloride present in the fuel volatilizes largely as HCl. HCl reacts with steel to form FeCly. Short duration (1 day) tests indicate that the corrosion rate of carbon steel in dry HCl gas is approxi- mately 45 mpy (1.1 mm/yr) at 550°F (288°C) (3). Lower HCl partial pressures should result in lower corrosion rates. If the corrosion rate were proportional to the HCl partial pressure in the combustion gas, then 0.1% HCl would cause an 0.045 mpy (0.001 mm/yr) corrosion rate. Hence, while HCl in the combustion gas may corrode furnace-wall tubes, the HCl corrosion rate is probably not high enough to account for the cor- rosion rate observed in the subject unit. Corrosion by NaCl Deposits on Tube Surfaces Substantial amounts of sodium and potassium are present in paper, textiles, wood, and other components of refuse derived fuel. A portion of these alkali metals is volatilized, for example, as alkali metal chlorides and hydroxides. The hydroxides in the combustion gas react with HCl to form the alkali metal chlorides (e.g., NaOH + HCl — NaCl + H20) which deposit on the rela- tively cool furnace walls. Short duration (50 hours) laboratory tests by Miller, et al., indicate that, in oxidizing combus- tion gas atmospheres, sodium chloride deposits accelerate steel oxidation rates to approximately 35 mpy (0.89 mm/yr) at 800°F (472°C) (4). Based on this data point, the corrosion rate was expected to be substantially less but still significant at 600°F (316°C), perhaps several mpy. Our labora- tory test results indicate that the corrosion rate for carbon steel in NaCl at 600°F (316°C) is only 6 mpy (0.15 mm/yr). Hence, NaCl and KCl depos- its alone cannot account for the corrosion rate observed. Corrosion by ZnCl Zinc compounds are present in textiles, leather, and rubber. Zinc oxide is the most common of the zinc compounds, and it is relatively non-volatile and innocuous. However, in the flame, zinc oxide is readily reduced to metallic zinc, and it reacts with HCl to form ZnCly. Zinc chloride is volatile but readily condenses on the relatively cool furnace wall. Because of its high volatility, it moves through the surface de- posit and condenses at the tube surface. The zinc (and lead) chloride may be present on the tube surface as a liquid: for example, zinc chloride and sodium chloride form a eutectic which melts at 504°F (262°C). Zinc chloride destroys the protec- tive oxide film on steel (it is commonly used as a fluxing agent for soldering and brazing) and attacks the underlying metal. Furthermore, by disrupting the protective iron oxide on the steel, the zinc chloride facilitates rapid corrosion by HCl in the combustion gas. Laboratory tests by Miller, et al., indicate that the addition of 30% ZnCly and 15% PbCly to sulfate deposits on steel increases the corrosion rate to 120 mpy (3.0 mm/yr) at 600°F (316°C) (4). The sulfates alone had comparatively little effect on the corrosion rate. The aforementioned corrosion process occurs largely under reducing conditions. Oxidizing con- ditions convert ZnCly to ZnO and promote the formation of a protective oxide scale on the tube surface. Reducing conditions facilitate formation of ZnCly and PbCly and retard formation of the protective scale. Where conditions are interme- diate or alternating, some of the iron chloride cor- rosion product is subsequently oxidized and forms a non-protective iron oxide within the deposit. Under these conditions, displaced chlorine may cause further corrosion of the steel substrate. Our laboratory tests indicate that under oxidizing conditions, a molten zinc chloride, sodium chlo- ride mixture at 600°F (316°C) attacks carbon steel at a rate of several hundred mpy. Out-of-Service Corrosion Chloride-rich deposits on tube surfaces are hydroscopic and become moist during out-of- service periods. The pH of this moist surface de- posit is acidic (1.0 gram of deposit in 100 grams of water gave a pH of 4). Asa result, corrosion can proceed rapidly during out-of-service periods and may account for a part of the tube metal loss. Cor- rosion of carbon steel in high chloride acidic environments typically causes general wastage and pitting. Out-of-service corrosion is particularly danger- ous for austenitic stainless steel overlays and cladding. Where the overlays have residual stresses and the surface pH is acidic, the aus- tenitic stainless steels may be subject to stress corrosion cracking (5). Conclusions Of the chemical species identified on wall tubes removed from the unit, zinc chloride is the most aggressive. Laboratory tests described in this paper indicate that zinc chloride can account for the particularly rapid corrosion occurring in the subject unit. This mechanism depends on a number of factors: e Chloride in the refuse derived fuel, e Zinc (and lead) in the fuel, ¢ Sodium and potassium in the fuel, e Reducing conditions at the wall, e High tube metal temperatures, and ¢ Furnace-wall material susceptible to chloride attack. Reducing any of these factors is expected to reduce the corrosion rate. However, some of the factors may be beyond practical control. For example, the fuel composition can probably not be altered. Above about 500°F (260°C), corrosion rates generally increase exponentially with tem- perature (2). Hence, avoidance of waterside depo- sition which increases tube metal temperature is imperative. Promising weld overlay materials for protecting furnace tubes are Inconel 625 and similar high- chromium nickel-base alloys. Promising iron- base alloys are Incoloy 800 and 825. The greatest danger to austenitic stainless steel overlays is out- of-service chloride stress corrosion cracking. The greatest danger to nickel-base alloys (for furnace- wall application) is out-of-service pitting. The objectives of this investigation were to identify a possible furnace-wall corrosion mecha- nism, measure the corrosivity of chloride com- pounds, and evaluate potential weld overlay materials. Results show that ZnCly can account for the high in-service corrosion rates of furnace wall tubing. Other similar low melting point com- pounds may also be involved. The effects of ZnCl and NaCl on carbon steel tubing at upper limit furnace-wall temperatures are severe. However, weld overlay of carbon steel furnace-wall tubing with Alloy 625 may provide a viable solution to the furnace-wall corrosion problem. The viability of Alloy 625 overlay will be deter- mined by long-term field tests. Alloy 625 overlay has been applied in the furnace of the RDF-fired boiler discussed in this report, and after six months of operation the overlay shows no evi- dence of corrosion (as determined by ultrasonic thickness measurements). 1 References . S. F. Chou, A. J. Pramik, M. E. Scott, and P. L. Daniel, “High-Temperature Corrosion of Tube Support and Attachment Materials for Refuse-Fired Boilers,” Presented at the 1985 Joint Power Generation Conference in Milwaukee, WI, October, 1985. ASME Paper 85-J PGC-Pwr-41. . F. Norwak, “Corrosion Problems in Incinerators,’ Combustion, November, 1968, pp. 32-40. . D. A. Vaughan, H. H. Krause, and W. K. Boyd, “Chloride Corrosion and Its Inhibition in Refuse Firing,” Ash Deposits and Corrosion Due to Impurities in Combustion Gases. Hemisphere Publishing Corporation, Washington, 1977, pp. 455-472. . P. D. Miller, H. H. Krause, J. Zupan, and W. K. Boyd, “Corrosion Effects of Various Salt Mixtures Under Combustion Gas Atmospheres,” Corrosion, Vol. 28. June 1972, pp. 222-225. . H. H. Krause, D. A. Vaughan, and P. D. Miller, “Corrosion and Deposits from Combustion of Solid Waste, Part 2 -Chloride Effects on Boiler Tube and Scrubber Metals,” Transactions of the AIME, July, 1974, pp. 216-222. Technical Paper Considerations for the design of refuse-fired water wall incinerators J. D. Blue Manager, Boiler Development J. R. Strempek Product Specialist Babcock & Wilcox Barberton, Ohio Presented to Energy from Municipal Wastes Conference Opportunities in an Emerging Market POWER Magazine SynFuels Waste-to-Energy Report Washington, DC. October 24-25, 1985 Babcock & Wilcox BR-1280 a McDermott company Considerations for the design of refuse-fired water wall incinerators J. D. Blue Manager, Boiler Development J. R. Strempek Product Specialist Babcock & Wilcox Barberton, Ohio Presented to Energy from Municipal Wastes Conference Opportunities in an Emerging Market POWER Magazine SynFuels Waste-to-Energy Report Washington, D.C. October 24-25, 1985 Abstract PGTP 85-30 The combustion of refuse for volume reduction, heat recovery, and power generation has a long history. Recently, there has been considerable interest in burning refuse on a larger scale in high pressure/temperature boilers for increased power generation and waste disposal. These generating facilities require high availability, reliability and, in some instances, complex flue gas clean-up equipment to meet environmental requirements. Of significant importance is the boiler design which affects efficient combustion of refuse while, ideally, pro- viding availability, ease of operation and maintenance. This paper discusses design considerations that reflect the latest technology in refuse-fired water wall steam generators. Introduction The method of incinerating refuse fits into two large categories; namely, prepared Refuse Derived Fuel (RDF) and mass-fired Municipal Solid Waste (MSW). Within these two broad categories, several tech- nologies have been used for the incineration of refuse, some of which are spreader stoker, recipro- cating grate, rotary combustors, kilns, gasifiers, fluid bed, etc. These categories are distinguished by the mode of refuse preparation. In the mass burn technique, the refuse is utilized in its as-received state, and is fed directly into a large storage pit from a tipping floor. Large objects, noncombustibles, and hazardous materials are removed either manually from the tipping floor or remotely from the refuse pit prior to burning. Mixing of the various constituents may oc- cur in the refuse pit at the crane operator’s discre- tion. The refuse is then fed into a reciprocating grate stoker, the combustible portion is burned and the residue is dropped into an ash pit for reclama- tion or disposal. In the prepared refuse technique, as-received material is processed in any number of alternate schemes to yield a high-quality shredded refuse fuel and other salable or recyclable byproducts. Hazar- dous and large, bulky materials are removed prior to the processing system. The RDF is then fed into a furnace to burn on a traveling grate stoker, fluidized bed or other suitable means. Refuse is a highly volatile fuel, and is readily com- busted with any of the above listed technologies. However, as a fuel, it is heterogeneous, difficult to handle, and contains variable water and ash quanti- ties. In addition to the concerns for hang-ups and abrasion in the fuel handling, fuel feed and ash remov- al systems, refuse incineration produces significant potential for corrosion, erosion, slagging and fouling, which presents a challenge to the boiler designer. Design considerations Unit sizing Refuse unit sizing is generally expressed in tons per day of refuse burning capacity. Of more significance, however, is the heat input of the refuse to the unit. Since the as-fired Btu value of refuse can vary signi- ficantly, proper evaluations of the Btu and tonnage ranges are important in order to arrive at a properly sized unit. The heat input is set by the design refuse rate and heating value. Since the furnace is a Btu machine, the fuel burning capacity of the fur- nace will be inversely proportional to changes in the refuse heating value. Proper consideration should be given to sizing so that the contracted quantity of refuse can be processed without overloading the refuse unit. To further complicate the selection, the Btu value of refuse has been steadily increasing and is expected to continue to do so. In fact, projections for future RDF heating values are in the 7000 to 7500 Btu/lb range with 5200 to 5500 Btu/lb fore- casted for mass refuse. If known in advance, the boiler designer can, within limits, incorporate features, or provisions for future changes, recogniz- ing the projected trends in heating value. Failure to do so may result in a unit that, in the future, is too small. In this condition, the tendency is to overload the unit. Past experiences show that prolonged over- load operation can have a major impact on avail- ability and cost of operation.' Refuse feed/stoker sizing The design of the feed system to the boiler is a major consideration for a refuse-fired unit, and is critical to the successful overall design and future operation and maintenance of the unit. For either the mass or RDF approach, the feed system must be capable of supplying a steady, controlled and uninterrupted quantity of refuse to the stoker with provisions for proper fuel distribution. Figure 1 is an illustration of a typical mass-fired unit. Overhead cranes equipped with grapples are generally used to move the refuse from the storage pit to the charging hopper. A hydraulic ram at the bottom of the charging hopper is used to feed bulk refuse to the stoker-furnace. The ram design is rug- ged due to the nature of the refuse being handled. Water cooling may be supplied to the ram and parts of the charging hopper that are exposed to furnace radiation to keep their operating temperatures below certain limits to reduce maintenance. The ram speed is controlled by demand for steam, furnace tempera- ture or refuse inventory. In this concept, the ram is controlled independently of stoker speed which pro- vides flexibility in the control of the combustion pro- cess as well as the ash depth retained on the stoker for protection from the furnace radiant heat. Past experience has shown that refuse feed and control is critical to the successful performance of a mass-fired unit, particularly to the control of steam demand, slagging and reduction of corrosion potential. Once on the grate, the refuse dries, is heated to ignition temperatures, burns and is discharged as ash from the opposite end. Stoker width and length must accommodate this type of feed. Stoker widths are normally set by the quantity of refuse feed, and typically do not exceed 30 tons/day/ft. Stoker length is then set to allow sufficient time for fuel burn-out with reasonably low carbon residual. Stoker length typically varies from 27 to 40 feet. Many variations of feed arrangements and devices have been used on RDF units, several of which are derivations of those used to feed wood and bark- fired units. In actuality, the feed system has to be designed to accept reasonable quantities of wire, rope, rags, streamers, sand, bulky materials, ferrous, non-ferrous and glass materials. Due to these physical characteristics, many of the early RDF systems were unreliable and troublesome, required high maintenance and, ultimately, modifications. Further experience with the earlier systems showed that consideration should be given to a feeder system that minimizes significant storage, keeps the RDF moving, and provides low maintenance and maximum availability and reliability. After considering several designs, the best concept seems to be the overfeed-type supply system. In this concept, the RDF is distributed to a bank of feeders located across the unit width in sufficient numbers to satisfy the full load requirements. The excess refuse is discharged to a conveyor and is returned to the RDF storage area. Individual feeders, each hav- ing approximately 3-5 minutes of storage capacity, utilize a variable speed ram to feed an inclined con- veyor on level demand. The conveyor is also variable speed and supplies RDF on steam demand. Each in- clined feeder can be independently controlled to pro- vide the capability to bias and control refuse distribution side to side. These feeders, shown in Figure 2, have demonstrated the capability to modulate load on 100% RDF firing. Each feeder supplies RDF to an air-swept spout. Pulsed air is in- troduced through the spout in combination with under spout air nozzles to provide front to rear depth of throw and fuel distribution. Experience has shown that the air-swept spouts should be sized desirable to keep wear at a minimum, ash content large enough to avoid pluggage. A minimum of 30 and the limits on stoker speed typically set the inches wide and 18 inches high is preferred. stoker width requirements. Stoker length is deter- Stoker widths are determined by the number of mined by the ability to distribute the RDF uni- feeders required across the unit width to feed the formly front to rear. Typically, the length has not RDF and the grate speed that is required to convey exceeded 17% ft. Overall grate size is set so as not and discharge the ash. Since slow grate speeds are to exceed 750,000 Btu/ft?-hr of effective grate area. i _m 1 ha e —— i 1 ii | > + + 1+ aa —$—— ao G L a Tr o| | |e} jo ie ; Economizer i — | Top of Refr. 4 | Topo Sidewall & Front wall ! Refer. ! ; : oS] ae: Refuse In 1 | i i (eam id I A ty \ | cate FAH oo TI ie V\ ! — Overfire Air | 1 a Bas Feed | roi i \ : ae Lys ' Grate cdee74 cs wo PRA : au if; 2@ @ o ' | ' 1 Bt | + = 3) | ¢ Ash 1 5.C.A.H Z Sitti A ‘ieee au ¢ F.D. Fan Figure 1 Mass - fired unit. Overfire Air Ram Feed Overfire Air Inclined Feeder L Figure 2 RDF - fired boiler with feeder. In the combustion of refuse, aluminum melts. In both mass and RDF firing, this aluminum can solidify on the grates, plugging air holes and requir- ing periodic outages to clean. Good stoker designs recognize this potential and are provided with the ruggedness required to shear the fused aluminum at parts moving relative to each other and at seals and tuyeres. The design of the stoker should also pro- vide enough flexibility to control ash bed thickness of sufficient depth to allow molten aluminum to solidify within the ash bed instead of on the grate. The design and arrangement of the RDF feed system may be affected by the RDF processing system’s product. Interface between the designers of the RDF processing system, feed system, boiler, and stoker is a must. RDF sizing and quality must be agreed to in advance so that the overall system design can be coordinated and be successful. The boiler-stoker designer needs to set specifications on RDF sizing. Typical top sizing might be 99% less than 6 in. X 6 in. with a maximum of 12 in. streamers. Fines should be minimized. Due to the potential for boiler slagging and erosion on material handling equipment, it is desirable to remove the glass prior to shredding. Furnace sizing The furnace is a water-cooled membrane wall enclosure which contains the combustion process and provides for sufficient water-cooled surface to cool the products of combustion to temperature levels suitable for passage into the convective sur- faces, and to provide residence time for burn-out of entrained materials. Existing units have been sized to operate at furnace exit gas temperatures up to 1600°F for RDF firing and 1400°F for mass firing. Furnace volumetric heat release rate is indicative of residence time. Units are operating at volumetric heat release rates as high as 20,000 Btu/ft*-hr. Since the design to a specific furnace exit gas temperature is predominant in the ranges discussed above, the volumetric heat release rate will be a lower, more conservative value. For mass and RDF units, these values typically range from 8,000 to 16,000 Btuw/ft*-hr. Furnace exit gas temperature is basically fixed by the furnace size for a given capacity. Significant changes can only be effected by changing the fuel input. Small changes may be effected by deviations in excess air from the design value. These, however, will be limited due to the potential for increased cor- rosion, slagging or erosion at higher values. The interface between stoker and the lower fur- nace design is of great importance. The physical shape and arrangement of the lower furnace varies between mass- and RDF-fired units and is predomi- nantly dictated by the stoker dimensions. The lower furnace includes the overfire air system, thus the design and arrangement of both must complement each other to promote adequate mixing for rapid burn-out of combustion products (to be discussed later). This assures that the remaining furnace sur- face will be effective in reducing the gas tempera- ture to design levels and will provide an oxidizing environment that will reduce the potential for slagg- ing and corrosion in the upper furnace. Furthermore, the need to visually observe combus- tion and slagging conditions in the furnace cannot be over emphasized. View ports of a sufficient design and quantity should be strategically located to permit an unobstructed view of the stoker, the furnace walls and the superheater. Combustion air requirements Experience has confirmed that mass-fired units are typically designed to operate between 80-100% ex- cess air. Earlier designs of RDF units were based on 35% excess air to the stoker. Experience indicates that higher levels of excess air may be required to control furnace slagging. To provide this flexibility, RDF units should be designed to permit operation up to at least 50% excess air levels at the stoker. Air leakage and cooling air are additives to the stoker air requirements and will result in higher ex- cess air values at the unit outlet. Overfire air re- quirements may vary depending on the particular stoker supplier. Experience shows that overfire air systems have operated at air quantities in the range of 40% of total stoker air requirements. Based on this experience, the overfire air system should not be sized for less than the 40% for good combustion burn-out and slagging control. Flow modeling of overfire air systems may be desirable to guide the design. To improve effec- tiveness, arches may be added to reduce the overall depth of penetration required, and provide improved fuel-air mixing for complete burn-out. Figures 1 and 2 are typical arrangements of mass- and RDF-fired units with arches designed and arranged to maxi- mize the effectiveness of the overfire air system. Furnace corrosion protection Refuse fuels contain significant quantities of chlorine which can lead to corrosion of pressure parts. Chlorine contents of up to 1.8% by weight on a dry basis have been observed in refuse analyses. In the combustion process, this chlorine may exist as gaseous hydrochloric acid that can cause significant corrosion when in contact with carbon steel tube Operating experience on mass-fired units has dic- tated that the lower furnace tube metal be protected from the corrosive flue gas. Designs utilizing pin studs and silicon carbide refractory material coatings have effectively eliminated corrosion of the lower carbon steel walls over the range of today’s operating units. The design and arrangement of the lower furnace - combustion system is important when defining the extent of refractory coverage. Typically, the area of coverage will encompass all four walls up to approximately 30 ft. above the grate where there is reasonable assurance that oxi- dation zones are predominant. It is very important that the quality and physical characteristics of the silicon carbide refractory be maintained through proper application and curing. Lack of control during installation will result in spalling, deterioration and increased maintenance. It is desirable that the refractory material have high thermal conductivity rates to minimize the reduction in effectiveness of the water-cooled surface it is pro- tecting. However, such characteristics may reduce its resistance to erosion as experienced along the grate line due to the scrubbing action of the refuse fuel and ash as it moves along the grate to the ash discharge. Increased erosion-resistant silicon carbide materials are available that are suitable for use in these zones. They do, however, have lower thermal conductivities. Figure 3 shows a typical arrangement of stud pattern and size that has been used success- © 3x 2 Pattern Figure 3 Typical stud arrangement anchoring silicon carbide. fully to retain the silicon carbide refractory. Main- tenance and refractory life may be affected by the refractory surface temperature in operation. Higher operating surface temperature will increase the potential for slagging and may reduce refractory life due to spalling; higher temperatures may also result in the refractory contents chemically reacting with flue gas constituents to reduce the material inte- grity. Stud design, density and application can also have a significant effect on surface temperature. This is illustrated in Figure 4 showing the relative relationship between stud density in studs/ft of tube, stud diameter, and face temperature. RDF units now in service are operating at pressures that range from 250 to 1200 psig. With one exception, these units were supplied without lower furnace wall protection, although one did have curtain air capability.? The lower pressure unit (250 psig) has operated for many years without signifi- cant furnace wall corrosion. Corrosion has been reported in various degrees on some of the units operating at 700-750 psig and 1200 psig. A unit was retrofitted with studs and silicon carbide refractory on the lower furnace wall panels. It has been reported that the refractory has stopped the corro- sion but the furnace is still subject to slagging.” The one unit originally equipped with refractory suffered severe slagging and was subsequently removed.® 1" Long All Studs = 1/2" Dia. Stud 3/4" Long ~>~-- Stud Silcon Carbide Surface Relative Face Temperature of ‘| ———~ Increasing Temperature 20 25 30 32 35 40 45 4850 (Studs/Ft) | Figure 4 Relative face temperature vs. stud length and density. Stud Density In summary, it is apparent that there is potential for corrosion in the unprotected lower wall panels of RDF-fired units, just as there is in mass-fired units, if the lower furnace is not protected with refractory. Experience indicates that the traditional mass-fired studs and refractory protection may not be a prac- tical solution to the corrosion potential for RDF units due to the potential for slagging. Field work has identified, in one instance, the lower furnace corrosion mechanism to be due to chlorine. Laboratory analysis indicated that this cor- rosion is in part due to HC? acid attack with some evidence of molten salt (ZnCf./PbC?,) attack.* Corrosion rates may vary and be inconsistent around the periphery of the furnace. Thicker wall carbon steel tubes with adequate corrosion allowance will undoubtedly suffice to provide reasonable life in a major portion of the lower walls in the medium-to- low pressure units. It may be desirable, however, to consider corrosion protection for these units in localized areas which have typically shown higher corrosion rates and also in the higher pressure units with their correspondingly higher saturation temperatures. Development work is under way to identify alloys that resist corrosion in the lower furnace wall area. One unit has been overlaid with a chrome/nickel alloy material. This selection was based on laboratory corrosion studies and the results of field test panel work which showed the alloy to be very effective in resisting chloride corrosion.’ Since the corrosion has only recently been found, work con- tinues, with efforts to better understand the mech- anism and to find solutions of a more cost-effective nature. Field experience further shows the importance of water quality as it relates to tube corrosion. Poor water quality can result in internal deposition on tube surfaces, which insulates the tube and results in higher than normal tube metal temperatures. Recently, in an operating refuse unit, internal deposition, ranging up to 20 gm/ft? resulted in a 0.165 wall carbon steel tube corroding to failure in the equivalent of about six weeks operation on refuse. This is extremely rapid corrosion and dramatically highlights the fact that the corrosion rate is metal temperature-dependent as evidenced by laboratory studies of HC? corrosion versus temperature shown in Figure 5. Operating pressure determines the time-proven standards for feedwater and boiler water quality. We do not expect these standards to be any more stringent for refuse fired units than for other fuels. We do expect, however, that the maintenance of boiler and feedwater quality, in adherence to those standards, will be more critical on refuse-fired units due to the corrosive nature of the products of combustion. 110 100 90 80 70 60 50 40 Corrosion Rate, Mils Per Month 30 20 bi Figure 5 Corrosion of carbon steel in chlorine and hydrogen chloride. 700 800 900 1000 1100 1200 Metal Temperature, F Superheater Superheater design for refuse units must provide consideration for slagging and corrosion. Domes- tically, superheaters are in service at 750°F and 830°F. Neither the 830°F nor the 750°F super- heaters have experienced significant corrosion or slagging problems. Evidence of localized metal loss at the sootblower elevations has been found. The cause is likely to be a combination of moisture in the steam and corrosion due to the removal of pro- tective oxides attributed to the over-cleaning that occurs in the proximity of the sootblower. Superheater design must be based on conservative parameters. Inlet gas temperatures should be kept relatively low, typically not to exceed 1600°F for RDF and 1400°F for mass. Advances in combus- tion-furnace systems and, in particular, the effec- tiveness of overfire air systems assures that all com- bustion is complete prior to entering the super- heater. This reduces the potential for corrosion that may otherwise be accelerated by cyclic oxidation/re- ducing conditions. Conservative furnace exit gas temperatures usually result in larger furnaces which provide longer residence times for burn-out of en- trained material and reduce the potential for slagg- ing. This is not necessarily the case where furnace screen surface is used rather than furnace wall sur- face, to obtain the same furnace exit gas temperature. Lower superheater inlet gas temperatures require larger superheaters to effect the same performance. Surface duty is reduced, meaning that for the same steam temperature, tube metal surface temperatures will be lower. The use of a parallel superheater arrangement whereby the cooler steam is located in the higher inlet gas temperature zone, as shown in Figure 6, results in lower peak metal temperature as compared to a counterflow arrangement. Steam side mass flows should be higher than for the typical non-refuse industrial unit. This results in higher steam side pressure drops which forces more uni- form steam side distribution, and thereby reduces the metal temperature gradient between tubes across the width of the unit. Higher steam side mass flows do result in higher design and operating pressure and, therefore, higher feedpump horse- power. This is more than offset by the benefit of lower peak tube metal temperature, reduced poten- tial for corrosion and, consequently, better reli- ability/availability. A typical 750°F design temperature superheater includes only carbon steel material. Higher steam temperatures result in higher metal temperatures that will require chrome/nickel alloys to resist the corrosion. In our opinion, this is currently the only design option to employ since refractory coatings, etc. have not been developed to the point of pro- viding satisfactory service in high steam tempera- ture applications. Boiler banks Fouling, corrosion and erosion are the concerns for the design of boiler banks on refuse-fired units. Ex- perience has shown that corrosion can occur in the boiler bank where excessive sootblowing is used which increases the exposure of protective oxides to the corrosive flue gas constitutents. The situation can be relieved by proper design and arrangement of surface. Conservative side and back tube spacing coupled with ample space for retractable sootblowers and low design inlet gas temperatures reduce the potential for boiler bank fouling. A typical gas temperature limit entering the boiler bank is 1400°F with a 2% inch clear side spacing. To preclude ero- sion, gas side velocities should not exceed 30 ft/sec in cross flow applications. Economizers Few problems have been encountered with economizers. However, two areas the design should recognize are inlet water temperature to avoid low metal temperatures at or below the dew point of the flue gas, and gas side velocities not to exceed 30 ft/sec to reduce the potential for erosion. Rotary sootblowers have proven effective in cleaning the surface. Figure 6 Typical parallel flow. Superheater. Air heaters Air heaters may be employed on refuse boilers for two reasons: to supply preheated air to assist in drying and igniting the refuse on the stoker, and/or to increase thermal efficiency where high feedwater temperatures preclude designing to lower exit gas temperatures with economizers. In those designs having lower feedwater temperature, steam coil air heaters may be used in lieu of tubular or regenera- tive, and the thermal efficiency would be retained with additional economizer surface. RDF-fired units have typically used air heaters to preheat the combustion air to the 300-350°F range to assist in drying and igniting the fuel for the higher range of specified moisture content. Alter- natively, a steam coil air heater could serve the same purpose. Mass-fired units have typically been supplied with ambient air temperature combustion systems. Re- ductions in refuse firing rates can result when high moisture fuel is being used. Steam coil air heaters can be supplied to provide preheated air to assist in drying and igniting the fuel during instances of high moisture. Tubular and regenerative air heaters have been used successfully in refuse applications. Due to leakage and fire potential, regenerative types have been limited to the outlet side of hot electrostatic precipitators where the flue gases are relatively clean. Where air heaters of either the tubular or regen- erative type are employed, the design and arrange- ment should recognize and minimize the potential for low end temperature corrosion and fouling. To some extent, the arrangement of the surface in a tubular air heater will maintain adequate protection. It is wise to supply steam coil air heaters at the air inlet, on either type, to preheat the incoming air in order to maintain metal temperature above acid dew points. Auxiliary burners Conventional circular oil and/or gas burners are used mainly for startup and usually are sized for some nominal 20-30% of rated load heat input, although a few operating units have full load capabilities on auxiliary fuel(s). In the former case, the burners are usually located where they can serve the startup and refuse combustion support role. Auxiliary oil/gas burners are used to maintain furnace temperature during startup and upset conditions since operation at low furnace temperatures could result in the in- complete destruction of volatile organic compounds. A requirement for full load capability on auxiliary fuel can have significant impact on the refuse unit design and sizing. This is also true for conventional fuel-fired boilers with the requirement for the co- firing of refuse above 20% of rated input. Refuse in- puts in excess of 20% may require that the overall design be set to the design parameters of refuse fir- ing rather than to that of the primary fuel. Relative to oil and gas firing, these design limits are conser- vative and will result in a significant increase in unit size. The unit design should recognize where signifi- cant co-firing of refuse and auxiliary fuel is required, and where compromises are necessary, the refuse parameters should prevail. Auxiliary burners require cooling air to protect the burner parts from overheating due to exposure to the furnace radiation and to prevent debris from col- lecting in the burner and windbox during idle periods. This cooling air will add to the total excess air leaving the unit, because it is not supplied to the furnace in an area to be included as effective refuse combustion air. Further, if excessive in quantity, some tempering of furnace temperature will occur. To reduce the need for cooling air, some refuse units have been equipped with retractable burner features which permit the installation of gates to isolate the burner from the furnace during out-of-service periods. These gates are effective but require routine maintenance. Auxiliary oil/gas systems should be equipped with all the control and flame safety devices in accor- dance with standard practices to assure a safe and reliable system. Gas side cleaning To maintain the effectiveness of all convective heating surface and to prevent pluggage of gas passages, it is necessary to provide some means of removing ash and slag depositions from external tube surfaces. Steam or air sootblowers are most commonly employed for this purpose. Saturated steam is preferred for its higher density and better cleaning ability. However, one disadvantage of soot- blowing is that localized erosion and corrosion can occur in areas swept too clean by the blowing medium. This problem can be mitigated by installing tube shields for localized protection. In the superheater, where high temperature corro- sion is a prime concern, mechanical rapping is a suitable alternative for sootblowing. In this system, a series of anvils strike designated pins to impart an acceleration through the superheater tube assembly, which knocks off ash and slag deposits. This tech- nology has been used extensively in Europe and has recently been applied successfully to high tempera- ture/high pressure units in the United States. The mechanical rapping system may be backed up with a conventional sootblower system. The decision to use a mechanical system is an economic one and should be made early on in the project to avoid modifications later. It is important to realize that gas side cleaning re- quirements vary with each boiler. Left unattended, ash and slag depositions may grow, sinter, and even- tually bridge the gas flow path between tubes. This can lead to erosion failures where flue gases find channels through partially-bridged tube lanes. To best determine sootblowing requirements for a par- ticular installation, a state-of-the-art computer con- trol system should be used. This system would monitor thermal performance for each section of the boiler on a continuous basis, determine the cleaning requirements, and select the optimum sequence and frequency of blowing. The frequency of operation should be on an as-required basis only, rather than on a routine cycle. Summary The use of refuse for energy conversion has presented many challenges. Nonetheless, progress has been made in developing the technologies to res- 10 pond to those challenges while providing reliable, maintainable facilities with good availability. This progress has been demonstrated by the performance of recent projects, which gives us confidence that the use of refuse as fuel for energy conversion pro- jects will be firmly established in our future. References 1. Hestle, J. T., Jr., et al., “The Importance of Proper Loading of Refuse-Fired Boilers,” Pro- ceedings of 1984 National Waste Processing Con- ference, Orlando, Fla., 1984. 2. Blasius, G. F., “Municipal Wastes Used for Large Scale Cogeneration,” CEP, March 1985. 3. Nollect, A. R. and Greeley, R. H., “Startup and Shakedown of Albany, New York, Solid Waste Energy Recovery System,” Proceedings of 1982 National Waste Processing Conference, New York, New York, 1982. 4. Daniel, P. L., et al., “Furnace Wall Corrosion in Refuse-Fired Boilers,” Proceedings of Twelfth National Waste Processing Conference, Denver, Colorado, 1986. 5. Chon, S. F., et al., “High-Temperature Corrosion of Tube Support and Attachment Materials for Refuse-Fired Boilers,’ Presented at ASME/IEEE Joint Power Generation Conference, Milwaukee, Wisconsin, October 1985. Technical Paper Process recovery boiler design considerations J. A. Barsin Manager, Industrial Projects Babcock & Wilcox Barberton, Ohio Presented to Kraft Recovery Operations Seminar Orlando, Florida February 10-14, 1986 Babcock & Wilcox BR-1281 a McDermott company Process recovery boiler design considerations J. A. Barsin Manager, Industrial Projects Babcock & Wilcox Barberton, Ohio Presented to Kraft Recovery Operations Seminar Orlando, Florida February 10-14, 1986 Abstract PGTP 85-41 The fundamental design principles from the first successful Kraft Recovery Furnace developed by Babcock & Wilcox and G. H. Tomlinson, which were commercially applied in 1929, are reviewed with specific emphasis placed upon present day applications and B&W’s development plans for the future. Introduction The first section of this paper gives an overview of the major changes to the Kraft Process Recovery Boiler. Design principles are detailed in the following sections, entitled Fuel, Combustion System, Water Side Circulation, Circulation, Convection Pass and Deposition, and Future Considerations. Background The Kraft recovery process initially evolved in Danzig, Germany some 25 years after the soda process was developed in Britain in 1853. In 1907 the Kraft process was tried in North America and from its inception, a variety of furnace types, including rotary furnaces and stationary furnaces, all competed for a successful commercial design. During the late 1920’s and early 1930’s, significant developments in furnace design were achieved by G. H. Tomlinson, working in conjunction with B&W engineers. The Tomlinson design evolved with a technique of spraying black liquor onto the walls of the furnace. The liquor is dehydrated both in flight and as it builds up on the wall, at which time pyrolysis begins with release of volatile combustibles and organically bound sodium and sulfur. The resulting mass causes it to break off and fall to the hearth where pyrolysis is completed and the char is burned, providing the heat and the carbon required in the reduction reaction. A smelt consisting primarily of NagS and NagCOs3 is produced and con- tinuously drained off to a dissolving tank. The world’s first Tomlinson recovery system was installed in 1929 at the Canada Paper Company, Windsor Mills, Quebec plant. This first black liquor recovery boiler, supplied by B&W Canada, had a water-cooled roof with refractory furnace walls. The early refractory furnaces proved costly to maintain and the amount of steam generated was much less than the amount theoretically possible. Tomlinson decided that the black liquor recovery furnace should be completely water-cooled with the tube sec- tions forming an integral part of the furnace. This new concept boiler was designed in cooperation with The Babcock & Wilcox Company and was installed at Windsor Mills in 1934. This water-cooled type was a complete success and, in fact, this first unit LL Zee Figure 1 Windsor Mills Unit, 1929. continues in operation today (Figure 1). The first USS. sale occurred in 1935 to the Southern Kraft Company for their Panama City, Florida Mill, and the first sale external to North America occurred in 1936 to Mo och Domsijo AB Mill in Husums, Sweden. The Kraft Recovery Boiler has a dual mission: 1. Recovery of the sodium and sulfur from the spent pulping liquor in forms suitable for regeneration of the cooking liquor and, 2. Efficient heat recovery from the burning of the liquor to generate steam for process use. Over the years, B&W designers have been challenged to increase the chemical recovery; decrease the cycle dead loads by increasing the reduction efficiencies, obtaining more chemical in the useful form of NaS; and to generate more steam at higher pressures and temperatures to permit in- creased cogeneration of electrical power from black liquor fuels. Figure 2 1500 tons. Design changes Furnace size The dominant influence on the physical size of the furnace is the black liquor solids processing capa- bility specified by our clients. The unit for Canada Paper Company at Windsor Mills in 1929 was capable of processing 60 air-dried tons/day (Figure 1) as compared in scale to our present design, which is capable of processing 1500 B&W Btu tons/day* (Figure 2). The need for furnace size increases to process greater quantities has been adequately demonstrated and is accepted by an industry that continues to obtain a return from applying eco- nomics of scale principles.’ The larger furnaces (1500 B&W Btu tons/day), required to achieve the dual missions, challenge the combustion system designer to obtain air/fuel mixing in a furnace now 38 feet wide by 37% feet deep (11.55 m X 11.4 m) and 162 feet high (49.25 m) as compared to a 20 ton/day unit, 54% feet wide (1.67 m) by 6% feet deep (2.05 m) and 29 feet high (8.82 m). Complicating the problem is the need to maintain the lower portion of the fur- nace (the hearth zone) in a reducing atmosphere, i.e., the oxygen level must be controlled to provide an overall hearth zone stoichiometry well below that re- quired for complete combustion. This insures that the reduction of NagSO,4 can be maximized. Furnace shape The question of having two distinct combustion zones, isolated one from the other, such as the NSP”, is not necessary to achieve predicted percent reduction, percent chemical recovery or carbon utilization performance. The present rectangular shape with separate combustion zones created by segmented air zones is cost effective and does pro- vide the desired performance. However, it is recognized that as the unit tons processed per day increase, the difficulty of maintaining good air/com- bustible mixing at an acceptable power penalty, will increase. A concept that is currently undergoing flow model testing would utilize lower furnace arches to (1) contain the reducing zone, (2) improve secon- dary and tertiary air penetration of the flame cone, (3) reduce the possibility of black liquor droplet car- ryover (entrainment), (4) increase the lower furnace temperatures, and (5) maintain the consistently high reduction efficiencies traditionally obtained. This concept has been successfully applied to our high moisture bark-fired power boilers and offers a cost effective way to achieve the goals without utilizing a totally separate furnace (Figure 3). *B&W has established, as a unit of capacity, a heat input of 19,800,000 Btu (20,890 MJ) in 24 hours. This unit, known as the B&W Btu ton, corresponds to the heat input from 3,000 pounds (1,362 kg) of solids (approximately equivalent to one ton of pulp produced) having a heating value of 6,600 Btu/lb (15,337 kJ/kg) of solids. Figure 3 Double arched lower furnace concept. Corrosion protection The application of pin studs in the recovery boiler reducing zone does protect that zone from corrosion as long as the studs are maintained. An application history is available.? As operating pressures were in- creased to levels above 1,000 psig (6.89 MPa cor- responding to a furnace tube outside metal tempera- ture of 600°F (316 C) ), furnace tubing corrosion became a problem. Consequently pin stud size, appli- cation densities, and application zones are row all designed to minimize corrosion. Composite tubing (bi-metallic or duplex) provides an alternative, with a six-year history in the U.S.A. That experience in- dicates that the bi-metallic approach is an alter- native to the pin studs and provides corrosion pro- tection and reduced maintenance (no pin studs). The furnace construction has evolved over the years from refractory only, initially, to tube and tile construction using 3% inch (8.26 cm) tubes on 6-inch (15.24 cm) centers with flat studs, to today’s 2""/s, inch (7.62 cm) tubes on 4-inch (10.16 cm) centers with the membrane furnace. Air infiltration is greatly decreased, permitting air to be placed where it is required, and total air levels reduced to more closely parallel requirements. Field evaluations are currently underway to insure that this construction does provide adequate cooling margins at 1500 psig (10.34 MPa) and above. Operating pressures to in- sure membrane fin temperatures are below high cor- rosion potentials. Environmental effect A second major impact upon the size and design of a Recovery Boiler installation is the effect of environmental and economical forces applied during the past 15 years. The direct contract evaporator was applied to evaporate water from the black liquor by utilizing the latent heat in the flue gases. The stripping of total reduced sulfur (TRS) from the liquor in a direct contact evaporator was an en- vironmental problem due to the odor released into the atmosphere. Additionally, it was desirable to im- prove the thermal efficiency of the boiler by generating more steam and utilizing exhaust (waste) steam to concentrate the black liquor. These desires led to the development of the Low Odor Recovery Boiler, which was first applied commercially in 1969 at American Can’s Halsey Oregon Plant, now owned by the James River Corp., which utilizes a cross flow economizer as the last heat trap to maximize steam generation. The economizer is a large struc- ture that has radically changed the side sectional elevation picture of a recovery boiler. Competitive pressures Finally, competitive pressures play a major role in the shaping of a recovery boiler. Active investment in development by several suppliers and many users insures that the technology will progress. In the electric power industry during the 1960’s, power plants were purchased strictly on first-cost bases and designers were forced to remove conservatism to stay competitive. Specifications were met but some margins were removed if client acknowledge- ment and evaluation of the “conservatism” could not be obtained. A similar phenomenon has been oc- curring in the pulp and paper industry since the late 1970’s. Competitive pressures are forcing optimized designs with diminished conservatism since the con- servation cannot obtain a dollar evaluation. Unit solids processing capability at 150 percent of the nameplate Maximum Continuous Rating (MCR) was not uncommon in the past, in fact, overloads have been the expected norm for the majority of our units placed in service since 1931. Today’s units re- quire improvements in operation to minimize fume carryover, and improvements in combustion, and in hearth temperatures, in order to obtain capacity ad- ditions over nameplate, at least until circulation limitations are reached. Specific design principles Fuel The prime factor which affects all recovery perfor- mance parameters is the specific black liquor being fired and the large deviations which may occur with this fuel. Black liquor is a unique fuel in that no other can match its ash content of approximately 25-35 as-fired percent and its moisture, of 30-40 percent. We have recently been involved with developing a coal water combustion system, directly firing a slurry containing 30 percent H2O. Un- supported flame stability was a major concern with this suspension-fired fuel but with preheat to 225°F, (107.2 C) top size droplet control and high recircula- tion rates in the flame, acceptable stability has been obtained and adequately demonstrated.‘ The coal- water mixture is manufactured to meet a fuel specification while the black liquor results from a process. The fuel varies with the type of wood being pulped, whether carried by salt water or fresh, if ox- idized or not and, as a result, demonstrates variable solids, viscosity, pH, organics, and various chemicals relating to the tightness of the cycle. Both in the wood and the make-up salt cake, all of the following can ride along and wind up in the black liquor: potassium, aluminum, iron, silicon, manganese, magnesium and phosphorous. Our standard practice in designing a specific unit is to calculate a Kraft Recovery unit’s air and gas weight on the basis of an elemental analysis and gross heating value furnished by the client. When no analysis is provided by the client, we utilize a liquor analysis which corresponds to the mill’s geo- graphical location and the wood species being pulped at the mill in question. This analysis is reviewed with the client whose agreement must be obtained to use the analysis as the basis for the design. This analysis is “salt cake free’ with the NagSO, make- up and precipitator hopper recycle excluded. The design analysis heating value (Bomb Calorimeter) is then corrected as required for: Effect of oxidation and heat of reaction to arrive at the heat of combus- tion, resulting in the burning of black liquor in a furnace. This is referred to as the heat available (HA). The heat available is then corrected for losses in the process, such as: ¢ The latent heat required to evaporate water formed from combining Hg in solids. ¢ The latent heat required to evaporate water added to the black liquor. © One-half the calculated boiler radiation losses. ¢ Heat required to reduce the salt cake feed to NaoS. ¢ Heat lost in smelt and the heat of fusing smelt. ¢ Heat of reduction of the sulfate in the liquor. ¢ Heat absorbed in the furnace. The application of these corrections generates the heat available for superheater absorption. Furnace absorptions are affected by gas temperatures, soot formation, percent solids fired, wall cleanliness, droplet carryover, excess air and sensible heats. The black liquor viscosity is an important variable as it affects droplet size and, therefore, rate of de- hydration. Viscosity is affected by the solids concen- tration, temperature, pH and the organic-to-inorganic ratio. Solids concentration and temperature are the variables that we fix when designing the liquor feed system — these are the dominant variables, but pH has been identified as an important factor in some circumstances. Extensive experience obtained on the many Kraft liquors utilized have provided the design flexibility which permits fairly wide variations in the fuel pro- perties to be processed, while still meeting the specified performance. The capacity of a recovery unit should be based upon its ability to burn completely, in 24 hours, the dry solids contained in the liquor recovered in the pulp produced in 24 hours. B&W measures the capacity of a recovery unit as we do with all boilers of our design, i.e., using heat input to the furnace rather than a solids processing capability. B&W’s Btu tons are constant and the tons per day of capacity specified by clients are converted by us to B&W Btu tons. This is achieved by cor- recting for solids recovery lb/ton of pulp, for heating value of solids Btu/lb, and pulp output of mill, and tons/24 hours (Figure 4). LIQUOR COMPONENTS HARD | SOFT CARBON 39 426 | Cc HYDROGEN 34 3.6 | Ho | Design OXYGEN 316 31.7 | 0) | Bases SULFUR 43 3.6 Ss Analyses SODIUM 215 18.3 | Na TOTAL 99.8 99.2 HEATING VALUE (Btu/LB) 6200 6600 SOLIDS 607 63.0 1 H20 39.3 37 ° Btu/Ib 3850 | 4160 Famece Figure 4 Analysis of black liquor - hardwood/softwood. Combustion system The low Btu, high ash and high moisture Kraft black liquor fuel, plus the heat absorbing chemical reaction involved, require a two-stage combustion process. The first stage prepares or conditions the fuel while the second stage consists of the actual combustion. The first stage includes concentration, increasing the solids to at least 60 percent, and heating the mixture to at least 220°F (104 C); both of which are accomplished external to the furnace. Preparation and delivery The fuel is sprayed on the walls by means of a splash plate nozzle and oscillator. ‘As fired’’ solids have gradually increased from about 60/61 percent to new systems now going on stream providing 70 percent solids. In addition, falling film type heaters can now heat the liquor to 240°F (116°C) without excessive scaling causing frequent clean- ings. These improvements in fuel preparation tend to increase bed stability and enhance the combus- tion process. Formerly, with the low solids, direct black liquor heaters and low firing temperature, it was absolutely critical to dehydrate the liquor in flight and on the walls, but not on the bed, to aid in maximizing the bed temperature. Not all the liquor droplets end their flight on the walls; some fall directly to the bed and some are entrained in the flue gases rising in the furnace. It is the designers’ intention that the dehydration and pyrolysis should occur in the lower furnace under reducing conditions. If the black liquor viscosity was controlled, the droplet size could be controlled, and the number of droplets entrained could be reduced. T. Herngren, et al.® have made a study of a control system and an online measurement and control of droplet index (DIX) that has demon- strated performance improvements, relative to droplet size variation on a recovery unit in Monsteras, Sweden. The control of droplet size is not as critical using wall drying as it has proven to be with total bed-only dehydration, i.e., ‘‘sta- tionary firing.” Tertiary eumerzone YY | Figure 5 Two-stage combustion process. Wall dehydration and pyrolysis The liquor that lands on the walls (the major por- tion) is dehydrated, removing heat from the fireball. When dehydration is complete, the droplet temperature increases and some volatile organics evolve. These become part of the gas stream and are ignited and burned. The particle volume increases during both the dehydration and pyrolysis phases. The temperature further increases and pyrolysis reactions occur producing volatile combustibles from complex organic compounds. The pyrolysis phase is important because it is a major factor in the release of sulfur gases, generates volatile combustibles, and provides the fixed carbon and hydrogen that is burned off. Hearth Eventually, the mass of the built-up char on the walls causes it to break off and fall to the furnace hearth in a highly reactive state. As particle volume is increased during dehydration and pyrolysis, porosity increases due to the volatile evolution. This porosity greatly increases the total surface — inter- nal and external of the particle — and, thereby, in- creases reactivity. During the char-burning phase, some of the carbon is utilized in the reduction re- action and the excess gasified, forming volatile com- bustible compounds such as CO. The reduction of NagSO, to NaeS occurs on the char bed and the atmosphere is controlled to maintain a reducing condition. High temperatures are important to the rate of char burning and it is desirable to have only fixed carbon and inorganics landing on the char bed (Fig- ure 5). The fuel, carbon, is on and in the bed but in a high porosity state to which air is discretely directed, using relatively low head (4 inch H2O [100 mm)]). If the porosity of the bed could be further in- creased, more carbon and hydrogen could be burned in the bed. The oxygen concentrations in the gas boundary layer next to smelt in the char bed must be high enough to burn the carbon and low enough to provide an adequate atmosphere for reducing NagSO,4 to NagS. Experimental evidence indicates that additional primary air would be beneficial to in- creasing bed temperatures as long as the amount of oxygen supplied does not exceed the requirements for burning the amount of char landing on the bed”*. The references further indicate that if carbon is in the smelt, reduction will be continued even in an oxidizing atmosphere. Sufficient heat is liberated in the hearth to support the reduction reaction and generate gases hot enough to aid in the dehydration phase. The floor of the hearth is sloped to provide positive and continuous drainage of smelt from the char bed and minimize the inventory of smelt within the furnace. Molten smelt is drained from the fur- nace through spout openings located in the lowest part of the floor to assure positive smelt removal. A removable water-cooled smelt spout assembly, designed to match the V-shaped wall opening, is ex- ternally mounted over the opening and inhibits air infiltration into the furnace. The tight-fitting smelt spout is designed to prevent smelt from infiltrating in the dead space between casing and tube wall, which should reduce maintenance in this normally- high maintenance area. Primary air The reducing condition is maintained by introducing up to 50 percent of the total air (including excess) required for complete combustion through the primary air ports. The primary air ports’ size and location are critical to attaining the high reduction percentages we have successfully demonstrated. They are approximately rectangular in shape, sized 2 inches wide X 12°/, inches high (5 X 8.6 cm) follow- ing the bent tube geometry. They are located on all four walls surrounding the periphery of the char bed and are vertically located 0.9 m above the furnace floor, following the smelt drainage slope. The wind- box supplying the primary air is sized to obtain ex- cellent distribution around the full furnace periphery and each set of primary air ports is equipped with a damper to permit group adjustment of air flow. Formerly, the port was pointed downward and im- parted a downward deflection to the air stream. The port on new units is pointed both up and down. The air interacts with the carbon at the outer periphery of the bed with a maximum penetration of three meters. The depth of bed is variable, but the active smelting region is on the top. Weyerhaeuser’s fur- nace imaging system indicated that the thickness of the hot part of the bed was less than 5 cm which, in turn, indicates that the active layer may be less than 5 cm thick*!°. The primary air provides the oxygen for the carbon burnout which provides the heat for the reduction reaction to proceed. The multitude of small ports producing discrete air streams is similar to our approach for burning heavy fuel oils. The fuel in that case is atomized in- to discrete streams of droplets to enable the oxygen to surround the individual droplet and permit combustion. The primary ports are subjected to pluggage and, once plugged with smelt, the ideal air distribution is lost. This results in loss of activity in that area and a cooling of the bed. In addition, the restriction on the air path changes distribution all around the primary zone which gets worse as more ports plug. Port pluggage is a fact of life and it, as well as the hand rodding required to clear it, is disruptive to the bed, steady state reduction and oxidation. Steady state conditions and bed stability have been greatly improved by the retrofit of automatic port rodders. They have now been retrofitted to six units in the U.S.A. with measurable benefits. We offer this improvement on all our new recovery unit pro- posals because they do an excellent job. Secondary air One meter above the highest primary air ports are the secondary air ports, providing up to 40 percent of the total air required for combustion. They are located there to control the top of the char bed as the momentum is sufficient to provide heat and tur- bulence across the top. These ports are evenly spac- ed around the periphery on all walls and provide the air required to burn the volatiles to provide the heat for aiding in the dehydration of the droplets. The secondary air ports are larger than the primary ports and are 2 X 14 inches (5 X 35.5 cm) on the 500 ton and larger units. The static pressure avail- able is double that utilized for the primary system and each port is fitted with its own damper control. The full penetration of these secondary air jets into the bulk furnace gas is critical to completing burn- out low in the furnace; to break up the combustion gas cone; to assure mixing of the air with the vola- tile gases rising from the char bed, and to utilize the furnace effectively. Steam coils provide preheat to both the primary and secondary air streams. Popular cycle pressures and maintenance concerns, in the past, have kept steam coils at fairly low pressures which have limited resulting air temperatures to about 320°F (160°C). Air temperatures up to 400°F (204°C) have been utilized when poorer quality fuel is specified. Temperatures in the lower furnace could be en- hanced by increasing combustion air temperatures which would aid dehydration and reduce bulk gas temperature losses caused when the combustion air is brought up to gas temperature in the furnace. Water coil air preheaters, utilizing recycled economizer feedwater, have been beneficial in permit- ting higher air temperatures while reducing feed water temperatures to the drum — a positive benefit on some units where it is desirable to increase the saturated water head. Tertiary air Located above the black liquor spray nozzles are the tertiary air ports, which provide up to 30 percent of the total air required for combustion. They are located on only two opposing walls either front/rear or side/side. The sizing varies with solids processing capability and the largest is 5 X 31 inches (12.7 X 79 cm) at 1,500 tons. Individual dampers at each port provide control and the static pressure is more than double that available to the primary system. The momentum developed at this level is also critical to complete combustion of the volatiles, to complete the break up of the flame cone and insure oxygen availability throughout the upper furnace to eliminate TRS. This stream can be cold or hot air, but the momentums must be great enough to insure success. Each air flow stream: primary, secondary and tertiary are metered individually and can be separately controlled to insure the furnace is opti- mized for the fuel actually being processed." Furnace construction Floor and wall construction of furnace. The lower half of the furnace, Figure 6, is operated as a chemical retort. Partial combustion of the char, in a reducing atmosphere at the surface of the porous bed in the furnace, releases carbon monoxide and elemental car- bon which act as reducing agents to convert the sulfate in the smelt to sulfide. The heat evolved melts the inorganic sodium compounds of the smelt. The molten smelt filters through the char bed to the sloped furnace floor, and drains to spouts, from which smelt flows into the dissolving tank. To withstand the erosive and penetrating charac- teristics of the smelt, furnace walls and floor are constructed to assure tightness. The surface of the floor and wall tubes must also be protected against the corrosive potential of the smelt in the reducing atmosphere of the hearth. Chilled smelt acts as a refractory, cooled and retained by pin studs on the water-cooled tubes and prevents molten smelt from contacting the tube metal. Pin stud construction is used for the floor and walls in the retort zone up to the tertiary air ports (Figure 6). The length of the studs and thickness of refractory are limited to that which can be effectively cooled. A positive mem- brane barrier at the centerline of the tube prevents penetration beyond that point. The pin studded sur- face, even when covered by char, continues to per- mit some heat transfer due to the conductivity of the pin studs. Furnace wall tubes above the pin-stud zone are bare, with membrane closing the space between tubes. The construction provides a gas-tight, fully water-cooled metallic surface, forming a barrier to furnace combustion products and air infiltration. The application of composite tubing — stainless outer layer, carbon steel inner pressure carrier has also been successfully demonstrated. Corrosion in the primary air ports, on the cool or windbox side, resulting in loss of the stainless wrapper, has been reported on one unit in Sweden and several units in North America. We are removing all flat studs from the primary and secondary port openings to elimi- nate these hideout areas where susceptibility to cor- rosion exists. The stainless outlet layer has been lost at a rapid rate (four years) but confined to a localiz- ed zone. Reported observations are similar to two cases, reported by S. Ingeuald, and F. Bruno,’? Bi- metallic tube smelt openings, supplied by two dif- ferent boiler manufacturers, are reported to have demonstrated cracking in the stainless layer but the majority presently in service are performing at least as well as the carbon steel/stack stud design. Oscillator Oscillator and spray nozzle. Oscillator burners, located in the center of the front and rear walls (and on all four walls of larger units) emit a flat, sheet spray of coarse droplets from the spray nozzle (Figure 7). The nozzles are continuously both rotated and oscillated to cover a wide band on all walls above the hearth. The degree of movement is adjust- able to cover a greater or smaller area of wall sur- face as required to compensate for variation in the solids concentration of the atomized liquor and the Stellite Material Oval Splash Plate Figure 7 Typical black liquor spray nozzle. consequent residence time that is needed for dehydration. The temperature, pressure, viscosity and solids content of atomized liquor are important to Kraft Recovery Furnace operation. It is desirable to create a large particle of atomized liquor. This maximizes the amount of liquor sprayed on the wall to mini- mize entrainment of liquor particles and mechanical entrainment of sodium chemicals in the combustion gases passing to heat absorbing surfaces. A liquor temperature of approximately 240°F (115°C) and pressure of about 25 psig (172 KPa), with solids as high as practical — 68 to 70 percent — would aid in maximizing lower furnace temperatures. The object continues to be the largest droplet, major spray pattern on wall for dehydration with some liquor at the high solids, and high temperature directly on the bed. Units, having a capacity from 500 to 1000 tons, usually utilize two spray nozzles, one in the front and rear wall; whereas, smaller capacity recovery units have a single oscillating burner in the center of the front wall distributing liquor on the side and rear walls. Units larger than 1000 tons have four oscillating burners, one in the center of each of the four walls. Considerable variation in the quantity of liquor introduced through a nozzle location is accom- plished by varying nozzle sizes. Currently, many of our larger units are successfully combining wall spraying (major) with in-flight drying and bed spray- ing. The bed spraying alone would not have permit- ted high reduction efficiencies in the past with the low solids and low fuel feed temperatures. Addi- tional nozzles have been provided and a common ar- rangement on several 800 ton units is to have two front wall nozzles, providing typical full wall spray, and two rear wall nozzles, providing partial wall and partial bed spray. Efforts are underway to evaluate the improvement that this mode of firing might provide. Water circulation The natural circulation recovery boiler is designed with circulation adequate to at least 110 percent MCR when firing oil. The auxiliary fuels such as oil and gas have the highest spot absorptions and, if a circuit is limiting under that extreme condition, it would be modified. The unit is rechecked at MCR on black liquor to insure all circuits meet our criteria. Gas temperatures leaving the boiler on a new unit are limited by the need to maintain sufficient satu- rated water head to insure circulation in the boiler bank, and provide water to the supply circuits. The large steam drum is fitted with sufficient steam separators, our most efficient cyclone separators, to insure minimum liquid particle entrainment in the steam to the superheater. Our experience permits calculating superheater pressure drop with good accuracy. In the retrofit case, where solids capacity is in- creased, all the above circuits, steam separating capacity and superheater pressure drop are potential choke points and circulation must be rechecked. Heat input limits for black liquor have been em- pirically established over the years and relate to: Plan Area Index: Hearth heat input — 900,000 Btu/ft? (10.221 MJ/M?”) of furnace floor plan. Average Tube Absorption: Average heat flux (studded tube zone) 40 to 60,000 Btu/ft? (454 to 681 MJ/M?) of wall flat projected surface. As heat inputs are increased, the bed temperature increases, the heat flux will increase and the air flow splits will have to be adjusted to contain the fireball. As solids are increased above the typical 3500 Ibs dry solids/day ft? (17,100 kg/M’), and exceed the plan area index, the potential for increased carryover does exist if mixing or firing is not optimized. Fume will probably increase and establish the next choke point which usually is high temperature surface deposition. Convection pass and deposition The furnace volume per unit input is a measure of residence time and is calculated as the cubic feet of volume up to the level of the furnace arch. Another empirical index in common use is the volume index composed of furnace volume (ft’) per B&W Btu tons/day (TPD). Values are established and range from 65 ft3/TPD (9,940 M’/B&W J TPD) suitable for an 800 TPD unit to 80 ft*/TPD (2,388 M*/B&W J TPD) suitable for a 1200 TPD unit. Exceeding these indexes will increase solids carryover which will accelerate pluggage of convection surfaces if all other conditions are not optimized (hot bed, etc). Staying within the index insures that sufficient residence time is available for combustion, and suffi- cient heat transfer surface is available to cool the gases to below 1700°F (927°C) entering the superheater. The furnace arch baffle serves two important func- tions. First, the nose baffle shields most of the superheater from the radiant heat of the active burning zone of the furnace, and the high- temperature steam loops at the rear of the superheater are completely protected. Secondly, penetration of the nose into the furnace distributes the gas to enter the furnace screen (if used) and superheater at a uniform temperature and velocity. An eddy above and behind the tip of the nose causes the gas to recirculate in the superheater tube bank, preventing hot gas from bypassing superheater surfaces. Convection heating surfaces In the recovery unit shown in Figure 2, the flue gases leaving the furnace are not cooled by a fur- nace slag screen. In our newest units, no screen at all is provided and the reduction in FEGT required to minimize deposition is obtained by increasing fur- nace volume. When a screen is provided or exists, the extent of screen surface is determined by the quantity of heat which must be absorbed in the superheater. Where a large superheater for high steam temperature is used, the screen must be small. The superheater is arranged for parallel flow of gas and steam whenever permitted by final tempera- tures specified. Saturated steam enters the front tubes of the superheater in contact with the hot gas and flows through successive loops, so that the final tube with the hottest steam is in contact with the cooler gas. There is a dual advantage with this arrangement: First, cooling of the gas is most rapid at the front of the superheater, where the need for cooling of the ash is greatest; second, the parallel- flow arrangement results in a lower average tube metal temperature at both the high-steam-tempera- ture and high-gas-temperature end of the super- heater. Careful coordination of alloys for superheater tubes permits final steam temperatures up to 950°F (510°C). Fireside deposits are a variable mixture of in- organic components. As a result, they melt over a broad temperature range.’* Eutectic points (first melting point of a mixture) of a solid mixture con- taining NagSO4, NagCO3 and NaCL decreased from 620°C to about 520°C as the percentage of NaCL was increased. As the NaCL, potassium and sulfide contents in the deposit increase, the eutectic point decreases. The example is still above the Top temp- erature of a clean superheater tube and the mixture should remain a solid. However, if deposits are already on the tube, the depth might be great enough to have the deposit surface at or above 520°C, which would permit melting to occur, sticki- ness to be created, and additional deposit added, ad infinium. The designers’ defense is to limit gas temperatures entering the superheater, reduce fume carryover, widely space the superheater and apply sootblowers on conservative centers. We have evaluated the relative worth of platen type surface (close back spacing) versus pendant four inch backspacing from a cleanability standpoint on a southern Kraft Mill. B&W’s original design was a pendant type superheater. The rebuild unit has a platen superheater, and we cannot discern any difference in cleanability. In the past, superheaters were designed on 6-inch (0.15 m) side spacing. At present, we are supplying new units on 12-inch (0.3 m) side spacing. When potassium or chlorine are in the black liquor in a total greater than one percent, the drastic effect upon the eutectic points is recognized by limiting the gas temperature to approximately 1150°F (621°C) entering the 5-inch (12 cm) boiler bank side- spaced tubes. The boiler bank spacing can be in- creased to 10-inch (25 cm) side spacing to relieve that choke point if firing/air adjustments cannot con- trol fume carryover. As the desire for better cogeneration heat rates gets stronger, the recovery unit will be pushed towards higher pressure cycles and higher tempera- tures. This progression ended in the electric utility industry about 15 years ago with the 3500 psig (24.11 MPa), 1000°F (538°C), 1050°F (566°C), 1050°F cycle. It became necessary for designs to go to wider and wider convection side spacing to per- mit continuous operation on slagging coals. Pendant superheaters are presently on 1.5 m side spacing centers and finishing superheaters are on 60 cm side-spacing centers. The successful strategy, applied to utility units, hinders the deposits’ ability to bridge. Then they would fail in shear and return to the ash hopper. We have proposed superheaters on 18-inch (46 cm) side spacing on retrofits and believe that it is an ef- fective way to increase solids - throughput cap- ability by moving the deposition choke point out further along the increased solids processing capa- bility load line. Gas velocities are designed to be low at design MCR. We utilize a limit of 25 fps (7.6 m/s) in the convection pass’ tightest flow zone. Another requirement for maximum superheater life is to structurally tie the superheater sections together to maintain alignment while leaving enough flexibility to allow for differential expansion of the sections. This is accomplished with a unique arrangement of front-to-rear and side-to-side ties utilizing both D-link and TG-link ties (Figure 8). The additional benefit is reduced side-to-side swing which reduces the possibility of premature failures due to fatigue near the roof penetrations. The boiler section is of a single-pass design without gas baffling, thus providing maximum cleanability of boiler surfaces. The boiler bank is provided with two cavities through which sootblow- ing and maintenance access are accomplished. An additional design feature is that the furnace rear wall tubes are carried into the steam drum instead of the lower drum, forming a screen section in front of the boiler bank. A boiler hopper is located below the lower drum providing positive removal of ash from all areas of the boiler bank. Two large downcomers supply water to the lower furnace cir- cuits. The emergency shutdown rapid drain connec- tions are located in these downcomers for fast, positive drainage to recommend water levels during emergency situations. Economizer The economizer used is either the vertical bare-tube type, generally baffled to establish crossflow of gases or the long flow, with no internal baffles and finned tubes (Figure 9). T. King and B. R. Blackwell’® indicate that in the boiler bank and economizer the dominant mechanism for deposit for- mation is the diffusion - driven deposition of chemical fume particles (major) and condensed vapors (minor). The deposits are enriched with potassium and chlorides. They expect the enrich- ment of NaCL in deposits to increase as flue gas temperatures decrease. Deposits in these regions of the boiler are usually soft and fluffy and easy to 12" "12" 12" 12” D-Link TG-Link Side to Side Superheater Tie Front View Side to Side Ties Tubes Bent Out of Line to Provide Rigid Tie and Maintain Wide Tube acca Front to Rear Superheater Tie Plan View Figure 8 Superheater structural ties. AN A Figure 9 Economizer - long flow. remove with sootblowers. A design velocity of 35 fps (10.64 m/s) is our limit for this heat trap. The economizer gas outlet design temperature is subject to a practical limitation because of the sulfur con- tent in the products of combustion from Kraft black 10 liquor. To prevent rapid corrosion and failure, as a result of condensation and formation of dilute sulfurous acid, the metal temperature of ordinary steel in contact with the gas must not be below the dew point of the gas. This means that the tempera- ture of the feedwater to the economizer should be above the dew point temperature of the gas. To minimize external corrosion, experience in recovery unit operation indicates that feed temperature should not be less than 275°F (135°C) and the associated gas temperature should not be less than 350°F (177°C). Sootblowing The buildup of chemical ash on recovery-unit heat- transfer surfaces is related to design and operation. Entrainment of ash in gases ascending the furnace is affected by gas velocity, air distribution, and liquor properties. Care is taken in the design of all recovery-unit heating surfaces to assure that such surfaces are arranged for sootblower cleaning. Cavities are left at optimum locations in super- heater, boiler, and economizer banks to provide for insertion of sootblowers. Gas temperatures are calcu- lated to make certain that velocities and particular tube patterns are compatible with good cleaning characteristics. Steam sootblowers are universally used in modern recovery unit installations. Soot- blowers are arranged for automatic sequential opera- tion, controlled from the main recovery unit panel board. When a modern recovery unit is operated at or near its rated capacity, no hand lancing is required to keep the gas passages open. As load is increased on a unit, however, mechanical entrainment of ash and volatilization of sodium compounds increase and invariably lead to cleaning problems. In addition to excessive quantities of ash in the flue gas, velocities and temperatures at all points in the unit are in- creased, and ash deposits become more difficult to remove. Future considerations Plans to further improve our Kraft Recovery Boiler by performing additional field evaluations will: A) Quantify “hot lower furnace” effects. We plan a field program to evaluate higher secondary/pri- mary air temperatures, higher solids firing and firing at a higher liquor temperature. It is planned to utilize an acoustic temperature detec- 11 tor (evaluating it) to measure hearth zone, tertiary and furnace exit temperatures, as well as a furnace imaging system to enhance our ability to detect changes. Hot high velocity ther- mocouple (HVT) probing will be done to verify the acoustic detector accuracy. B) Increase momentum of various air streams. The objective would be to increase the mixing in the furnace which would be a follow-up to Program A. C) Assist in the evaluation of water sootblowing. Ex- tremely effective on high sodium deposits generated by lignite firing on utility boilers. Perhaps initiate a program to determine its ef- fectiveness on black liquor. D) Quantify “combination” firing effects where wall spraying and bed spraying are combined in one furnace. A follow-up to program described in A. E) Develop ceramic smelt spouts. Presently, one test with evaluations ongoing for every shutdown. Preliminary findings are encouraging. F) Minimize manual operation by applying the tools now available to automate the unit. And in the home office: G) Design a membraned closure for the generating bank area, because our clients are requesting it for two reasons — The present design is flat stud and refractory, which permits water when water washing to seep into the refractory. The second reason is a desire to close the last major air infiltration point on the boiler. H) Continue the investigation into the mechanism of the cold side corrosion observed on composite tubing. I) Continue investigating improvements in smelt spout assembly cooling by additional flow modeling. Summary The pulp and paper industry represents a major market for B&W’s expertise, technology and hard- ware. We are committed to that industry in continu- ing to invest in developmental activities that will further improve the performance, increase the avail- ability and reduce the maintenance on our steam generators for both power and process recovery service. References i. Blue, J. D., Dudek, R. F., and Suda, S., “Some Considerations in the Use of Kraft Recovery Boilers for High Temperature and Pressure Application,” TAPPI Engineering Conference, Atlanta, GA, September 1981. Bjorklund, H., et al., “The NSP Cyclone Fur- nace for Black Liquor Reductive, Combustion Status Report,” International Chemical Con- ference, New Orleans, LA, 1985. Dickinson, J., Murphy, J. A., and Wolfe, W. C., “Kraft Recovery Boiler Furnace Corrosion Pro- tection,’”” TAPPI - Engineering Conference, Atlanta, GA, 1981. Barsin, J. A., “Coal Water Burner Develop- ment,” TAPPI - Engineering Conference, Atlan- ta, GA, September 1985. . McGillivray, S., Harris, L. E., and Blackwell, B. R., “Black Liquor Evaporation,” Project Memo V535017; ‘Dead Load Reduction in the Kraft Pulping Process,” Environment, Canada, Ottawa, 1984. Herngren Torbjorn, et al., “Control of Black Liquor Supply and Furnace Conditions,” Inter- national Chemical Conference, New Orleans, LA, 1985. 12 10. 11. 12. 13. Grace, J. M., Cameron, J. A., and Clay, D. T., “Role of the Sulphate/Sulphide Cycle in Char Burning — Experimental Results and Implica- tions,”’ International Chemical Recovery Con- ference, New Orleans, LA, 1985. MacCallum, C., Blackwell, R. B., “Modern Kraft Recovery Boiler Liquor-Spray and Air Systems,” Sandwell and Company Limited, International Chemical Recovery, New Orleans, LA, 1985. Harrison, R. E., and Ariessohn, P. C., ‘“‘Applica- tion of a Smelt Bed Imaging System,” Inter- national Chemical Recovery Conference, New Orleans, LA, 1985. Blackwell, B. R., and King, T., Chemical Re- actions in Kraft Recovery Boilers, Sandwell & Company Ltd., Vancouver, B.C., Canada, 1985. Lange, H. B., Pierce, D. P., Kisner, J. W., “Emissions from a Kraft Recovery Boiler, the Effects of Operational Variables,” TAPPI, Alkaline Pulping Conference, Atlanta, GA, October 1973. Ingevald, S., and Bruno, F., ‘Forty Years Fight Against Corrosion in Recovery Boilers,” Inter- national Chemical Conference, New Orleans, LA, 1985. Barynin, J. A., Dickinson, J. A., ‘Considerations for the Updating of Recovery Boilers,” Inter- national Chemical Conference, New Orleans, LA, 1985. | Technical Paper Reducing so, emissions from coal-firing utility furnaces J. A. Barsin Manager, Industrial Projects Domestic Fossil Operations Babcock & Wilcox Barberton, Ohio Presented to American Institute of Mining Engineers New Orleans, LA March 2-6, 1986 BR1282 Babcock & Wilcox a McDermott company Reducing SO> emissions from coal-firing utility furnaces J. A. Barsin Manager, Industrial Projects Domestic Fossil Operations Babcock & Wilcox Barberton, Ohio Presented to American Institute of Mining Engineers New Orleans, LA March 2-6, 1986 Abstract Stationary air pollutant sources established prior to the New Source Performance Standard (NSPS) are not, at present, subject to federal emissions regulations. However, proposed legislation could require many existing stationary sources to reduce emissions. This paper will report upon the laboratory work undertaken to date by B&W to commercialize a limestone system for sulfur capture both in pulverized coal and cyclone coal-fired utility boilers. Updates will be provided on the US EPA/State of Ohio/Ohio Edison/B&W sponsored pulveriz- ed coal full-scale limestone injection demonstration, and on the EPRI/Baltimore Gas & Electric, Atlantic Elec- tric/B&W sponsored laboratory-scale cyclone coal limestone injection investigations. Introduction The increased public attention associated with air pollutants produced by stationary sources has resulted in clean air legislation in many states and on the federal level. State and federal standards, such as the New Source Performance Standard, have established maximum allowable emission levels for gaseous pollutants from new sources. These stan- dards are reviewed periodically to insure that they reflect the best available control technology. Sta- tionary sources established prior to the NSPS are not, at present, subject to federal emission regula- tions. However, proposed legislation could require many existing stationary sources to reduce emis- sions. This paper will focus on the projects to reduce coal combustion SOx emission, either demonstrated in the field or in the laboratory, and their applica- bility to existing sources. Background SO2 Reduction In August 1973, the Tennessee Valley Authority (TVA) issued a report on the demonstration of a full-scale desulfurization of stack gas by dry limestone injection.! That demonstration was carried out on a pre-1960 B&W-designed steam generator, located at TVA’s Shawneee Steam Plant. Dry limestone injection was selected for the full-scale demonstration because of process simplicity, low cost, and existing knowledge. Laboratory scale demonstration (late 1960s) had indicated a sulfur renewal potential of up to 80 percent using dry limestone injection. The process was thought to have potential for retrofit applicability to older medium-sized power plants, with limited remaining life.? The process concept was to pulverize limestone (CaCO3) and inject it into the combustion chamber of the boiler; flash-calcine it to lime (CaO) and, as it calcined, the lime would start to react with sulfur dioxide (SOg) and oxygen (Og) to form calcium sulfite (CaSO) and/or calcium sulfate (CaSO4). The capture of sulfur was disappointingly low in the full- size field trial at full load, 20 percent (calcium-to- sulfur molar ratio of 2) and increased to 36 percent at half load (at a molar ratio of 2) utilizing marl limestone. The field demonstration results indicated that SOg removal depended greatly upon stoichiometry and residence time and, to a lesser ex- tent, upon limestone feed particle size. The optimum injection location on this demonstration was the up- per furnace, at an average temperature level of 2050°F. Extensive reheater fouling that could not be removed with existing sootblowers, and loss of col- lection efficiency in the electrostatic precipitator, were associated observations. The authors concluded from this demonstration that the dry limestone in- jection process would not play an important role in controlling SO2 emissions from power plants. RWE Experiments Researchers continued to investigate the sorbent activity, mixing, residence time, temperature, ash and particle feed size effects on laboratory scale and full-size experiments. Excellent successes were reported in 1977 by researchers at RWE in West Germany, experimenting with brown coal (a German high-moisture lignite) and a dry sorbent that was fed with the coal.? RWE postulated a layer-type reaction after using a scanning electron microscope analysis of an additive particle, following its passage through a lignite coal-fired combustion chamber. The element calcium was evenly distributed over the total cross section. The distribution of sulfur showed an enrichment at the outer edges. Examining the structural changes of the sorbent subjected to dif- ferent temperatures revealed that, when heated to 2012°F, the single sorbent particles were strongly interlinked, and the outer surface showed a rounded shape that reduced specific surface area. Additional heating to 2462°F resulted in a stronger inter- connection between particles and an additional decrease in available surface area for reaction. This phenomenon is referred to as “deadburning,” and was evidenced at Shawnee where average temperatures at the point of sorbent injection, i.e. below 2200°F, increased SO capture. The brown coal, because of the high moisture and associated low heating values, plus the non-turbulent combus- tion system utilized, has low flame temperatures when compared to bituminous coal. In addition, the combustion chambers are usually sized for low heat release rates, providing long residence times, and the convection heat transfer surfaces are widely spaced because of fouling and erosion design concerns with the low grade lignite. RWE has successfully demonstrated a Dry Additive Process for SO2 emis- sion control on a 60 MW steam generator, and now has two years’ evaluation of the process of a 300 MW brown coal-fired steam generator. Recent (1985) legislation enacted by the Federal Republic of West Germany requires 90 percent reduction of sulfur and will force the addition of stack gas scrubbers. The Energy and Environmental Research Corporation Experiments The Energy and Environmental Research Corpora- tion (EERC), under funding by the U.S.A. Environmental Protection Agency (EPA), has reported NOx and SOx reductions on a laboratory furnace (100x106 Btu), utilizing a low NOx burner (the EPA Distributed Mixing Burner) firing Utah bituminous and Illinois bituminous coals.* The sorbents utilized were limestone and dolomite. EPA researchers reported that coal ash has serious poisoning effects upon sorbent activity, confirming RWE postulations; however, the dolomite proved to be more reactive with coal and is somewhat immune to the poisoning effect. Particle feed sizes were im- portant and 11m (50 percent less than) proved op- timal for their system. They confirmed that sorbent activity (surface area) is most important and in- dicated that rapid calcination produced greater sur- face areas for limestone, which was observed to frac- ture in an area ratio of 4 to 1; dolomite at a ratio of 8 to 1, but higher temperatures decreased the crea- tion of surface area when Ca(OH) was the sorbent. However, in a typical steam generator furnace, ob- tainable surface areas top out at 15m7/gram. Calcin- ing in separate furnaces (not in steam generator flames), under a nitrogen blanket, have produced sorbents up to 90m?/gram surface area. This material should be superactive but production is not yet commercialized. The EPA continues to fund experiments to develop and commercialize super-sorbents. Steinmuller Corporation Experiments The Steinmuller Corporation in 1983, utilizing their low NOx burner, identified the optimum sorbent in- jection point to be the area around the flame (rather than in) as the best location to avoid ‘“‘deadburning”’ when applying the system to bituminous coals.° Steinmuller has carried out single-burner experi- ments at the International Flame Research Founda- tion (IFRF) using natural gas, coals and heavy fuel oil. Gas was doped with HgS. Using Ca(OHg) as the sorbent, 50 percent SOg reductions were achieved at Ca/S molar ratios of 2. They completed a two-burner trial (of twenty-four burners) on the 700 MW Weiher Unit 3, owned by Saarbergwerke A. G. in West Ger- many. The objective of the four-week trial was to in- vestigate slagging in the burner zone, in the furnace, and in the convection pass. The evaluations were made after injecting a total of 300 tons of slaked lime and 60 tons of powered limestone. The client, following this evaluation, granted permission for a full-scale demonstration to proceed with all twenty- four burners injecting limestone, which was schedul- ed for early 1985. Stein Industries — Althstom Stein has conducted limestone (CaCOs) injection tests in France in conjunction with Electricité de France on two units of CDF (Coal Mining Com- pany). These are mine mouth plants of 160 MW and 600 MW. Test results on the 600 MW plant, at a molar ratio of 2.5, yielded 62 percent sulfur reduc- tion on a consistent operating basis. They achieved similar results using a hydrated lime (Ca[OH]g) at a molar ratio of 3. Babcock & Wilcox Experiments The B&W Limestone Injection In-Furnace DeSOx Program was initiated in 1969 and restarted in July 1983. It utilized our commercial low NOx Dual Register Burner, and our newest test furnace, the Small Boiler Simulator (SBS), which became opera- tional on June 1, 1983. The 1983/84 program ex- amined such variables as sorbent injection point, particle size, residence times, oxygen levels, sorbents, two staging-low NOx modes, slagging and fouling. The results correlated with those of other researchers, in that the highest-capture injection point was found below 2300°F in the upper furnace (Figure 1). The mean sorbent particle size of 1lum and residence times of two seconds combined could capture greater than 50 percent SOg at molar ratios of less than 3:1 (Figure 2) utilizing a Kemco limestone.® B&W experiments indicated that different sorbents have different activities; burner designs can influence NOx and SOx destruction and capture; in- jection of the sorbent with the fuel (bituminous coal) provides the lowest SOg capture; and injection of the sorbent in furnace gases requires excellent mixing and proper temperatures to maximize SO2 capture.’ LIMB Program Current B&W Experiments Under U.S. EPA/State of Ohio/Ohio Edison and Babcock & Wilcox sponsorship, a program started in October 1984 has had the technical objective of developing a limestone injection multi-stage burner (LIMB) and low nitrogen oxides burner technologies for both retrofit and new applications, which could lead to commercialization of wall-fired boilers.® This program could also lead to a full-scale retrofit of combined low NOx/LIMB at Ohio Edison’s Limestone . injection Convection pass Dual register burners Figure 1 Test Furnace P.C. SBS 5x 106 Btu/hr Coal-Ohio No. 6 60} Sorbent-KEMCO limestone Stoichiometry, 116.2-119.4% Injection location: O - Upper furnace 50 ~ - Between the burners O- With coal 40 30 $O2 capture (%) 20 10 0 1 2 3 4 5 Calcium-to-sulfur molar ratio Figure 2. Sorbent Injection Location 5x 10® Btu/hr GAS OUTLET Mm Tl i|| FORCED Hi} DRAFT FAN i | | i | a 25 BRATS x = TT HN Mu | . i ry TTT ° _ mi Tet; i , . . | ° SECONDARY II|| IO H 0 ‘SUPERHEATER | REHEATER lll; s ° @ 4 $1 Hi PRIMARY: SUPERHEATER ate } i Hi COAL BUNKER ECONOMIZER || out.er weaver 127-3" / ECONOMIZER : : INLET HEADER i Oo: U - a eee oy Gas RECIRCULATING 7 Be Duct PULVERIZERS oO ; bh ae TTT TTT Fi] | TEMPERING AIR PRIMARY i AIR FAN i t 1 T ne-251.800 | + 40%0" —+ 31-9" ! 44-0" +| Figure 3 Ohio Edison Edgewater Unit No. 4 Edgewater Unit 4, a 105 MWe B&W-designed pulverized coal-fired radiant boiler (Figure 3). Preliminary work is currently underway and will provide the basis for a decision as to whether to go ahead with a full-size demonstration. Three of B&W’s Low NOx Burner Designs have been tested on the federal EPA’s large water tube simulator (LWS) by EERC at inputs of 60106 Btu/hr and 120X106 Btu/hr on several different coals and sorbents. These inputs represent a scale-up factor of 12 and 20, respectively, from our initial 1983 laboratory experiments. The Ohio Edison, Edgewater retrofit burners would be designed for 78X106 Btu/hr and, therefore, the laboratory work is full scale. The results obtained to date reinforce and correlate with those obtained during our 1983 5X106 Btu experiments (Figures 4, 5 & 6). Additional development work per- formed by others during the past two years rein- forces our earlier observations, and some new infor- mation has been developed. The efforts of many researchers evaluating sorbents have identified commercially-available sorbents that in both our LIMB and cyclone programs have demonstrated superior performance. Pressure or atmospherically- hydrated calcitic lime (Ca[OH]g) has repeatedly per- formed well on the LIMB program with various coal ashes, and does not incur the penalty of increasing the ‘‘dead load’’ — an inert magnesium component present in dolomitic lime. This easily surpasses the 25-30 percent removal efficiencies exhibited by the limestones. The work on developing super-sorbents continues but for the next five years hydrated calcitic lime will the best for LIMB. The ‘‘quench rate,’’ or the rate at which the temperature falls through the sulfation window, has always been recognized as a major influence in the kinetics of the reaction. The Edgewater unit has been characterized under the baseline tests to permit actual temperature and velocity patterns to be plot- ted under three load conditions, 50%, 75%, and 100% maximum continuous rating (MCR). B&W in- itiated a supporting study using a proprietary mathematical model of the boiler. The objectives of the modeling were to predict the velocity and temperature distributions throughout the Edgewater 70 60 50 $02. 40 Capture (%) 39 Operating conditions Load = 60 x 106 Btu/hr SRs = 0.68 SRr = 1.20 Illinois Coal : Hydrated san [vn W/Coal Tert. Ports 4 Level 8 Level Figure 4 Sorbent Injection Location 60X 106 Btu/hr 70 Hydrated Lime - 4’ Level SO2 Capture (%) O Illinois O Utah ~ Comanche © Wyodak Ca/S Figure 5 Effect of Coal Composition 60x 10® Btu/hr 60 Illinois Coal 50 40 SO2 Capture (%) 30 O' Cx(OH)2 20 | ca/s =2 © CaCos Estimated temperature at injection location F 1800 2000 2200 2400 2600 ° a c 1000 1100 1200 1300 1400 Figure 6 Thermal Effects 60X10 Btu/hr furnace. The actual measurements and the predictive measurements have permited projections to be created that will guide the sorbent injector design and location to insure optimized mixing. Supplemen- ting the above effort was the construction of a cold- flow scale model to test various injectors and locations. Another major area of concern was the effect the calcitic lime would have on flyash resistivity and resulting electrostatic precipitator performance. B&W, under this program, has supplied Southern Research Institute (SRI) with calcitic flyash and SRI measured resistivities up to 10'* from the base coal ash at 10°. SRI, under contract to EPA, has recommended several flyash treatments including the humidification flue gas treatment to improve the collectability of this ash. The U.S. EPA plans to research this effect further. Precipitators that handle flyash with resistivities as high as 10'* have been provided commercially and are meeting NSPS emis- sion requirements on new units. Retrofitting or upgrading an existing precipitator would be a site specific problem. Cyclones There are 23,000 MW of cyclone furnaces, located primarily in the MAIN and ECAR National Electric Reliability Council (NERC) regions. Proper cyclone fuels consist of crushed bituminous coal having low ash fusion temperatures to insure good slag tapping (but not low enough to permit the barrel slag coating to be destroyed). The design coal is low fusion temperature coal, which is usually high in sulfur and iron and, therefore, emits large quantities of SOg. The high NOx levels are a result of the high-turbulence hot combustion required to max- imize total carbon conversions and are an integral effect of the burner design. Concepts are available to apply to the cyclone furnace to reduce SOx by limestone injection and reduce NOx by applying afterburning, but the associated risks have not yet been explored in any depth. The effects upon reliability and availability, the additional costs, and the percent-emission reductions that might be attain- able are unknowns that are now being investigated in laboratory-scale screening studies, which must be followed by a full-scale demonstration. A proposal made to establish a consortium of cyclone users, encouraging the start of a cyclone developmental effort in parallel to the wall firing U.S. EPA effort, has been accepted by a group of cyclone-owning utilities; namely, Baltimore Gas & Electric and Atlantic Electric, the Electric Power Research In- stitute and The Babcock & Wilcox Company. That work is currently underway and is summarized below.® The B&W small boiler simulator’s pulverized coal firing system was removed and retrofitted with an 18-inch diameter cyclone furnace in 1985 (Figure 7). During July through November of 1985, the cyclone firing system was characterized for deSOx potential utilizing limestone injection. This facility was the same as utilized in 1983 for our pulverized coal LIMB characterization studies. Two coals were utilized: an Ohio No. 6 sulfur at 2.31 percent and a West Virginia bituminous sulfur at 1.85 percent, cor- responding to that presently utilized by Baltimore Gas & Electric (BG&E). Several sorbents have been utilized and are provided in Table 1. The cyclone is a scaled down version of our com- mercial cyclone, having two unique features — a secondary air system that permits sequential addi- tion of air and a preheat system that permits up to 800°F (vs 600°F) secondary air preheats. The need to scale down to a laboratory experiment changes the surface-to-volume cyclone relationship from that present in a commercial system. To obtain perfor- mance typical of a commercial installation, i.e. the turndown, carbon utilization and slag tapping, higher than normal secondary air preheats were designed in the laboratory system. Previous P.C. work indicated that temperature windows between 2250°F and 2300°F were at the injection point. A cold flow model, 1/3 scale of the SBS, was constructed to permit tracing gas flows 28” x 48” Inside Dimensions Economizer 216 Tubes 18 Deep 7'-3-1/2" 12 Wide 5’-3-1/2" ‘ Primary Secondary Air Suplerhaeter Reheater 100 Tubes 10 D 40 Tubes Te 4 Deep Primary 10 Wide Air and Coal Aa AMIEL Limestone Superheater Injection Ports 28 Tubes 4 Deep Tertiary 7 Wide Air Furnace Arch Slag Collector Figure 7 Small Boiler Simulator Cyclone Configuration Table 1 Cyclone Sorbents Utilized A) KEMCO B) Marianna Limestone C) Longview Hydrated Lime D) Corrosion High Mg Lime E) Corrosion Pressure Hydrated - 90% less than 20pm, 50% less than - 90% less than 53pm, 50% less than - 90% less than 74ym, 50% less than 5-6ym with - 90% less than 9pm, 50% less than - 90% less than 9pm, 50% less than 8ym with 5-7% MgO 8um with 2% MgO 1% MgO 4um with 32% MgO 4um with 29% MgO Table 1 Cyclone Sorbents Utilized to locate injectors to optimize mixing of sorbent with the gas stream. Injection location predicted by the model was not as effective as the alternatives tested under actual combustion conditions. Five side wall injectors were utilized (three on one side, two on the opposite side) located at the furnace outlet. The best SOg capture results obtained thus hydrated dolomitic lime. Approximately 52 percent capture at a Ca/S = 2 was achieved (Figure 8). This was a very encouraging result and indicates that this sorbent is less sensitive to cyclone firing condi- tions than alternatives. The Longview hydrated calcitic lime did produce CaO with higher porosities and greater surface area than that measured with far have utilized the Corson, an atmospheric the Corson, but SOg capture was lower than that Variation of SO2 capture with sorbents Side wall injectors @ 6 MBtu/hr, 2250 F. 70 o——— Corson high Mg hyd. lime mewecenn +=) Longview hydrated lime 604 +----- —¢ Kemco limestone ye ———-——* Marianna limestone 50 Sulfur Capture 40 (%) 30 20 10 30 40 50 60 70 80 90 100 110 120 Sorbent mass (Ib/hr) Figure 8 Cyclone SOp Capture — Sorbent Effects obtained by the Corson sorbent under equal opera- tional conditions. The sorbent activity selected is known to be temperature sensitive and the hydrated dolomitic, in this case, proved to be less temperature sensitive than the hydrated calcitic lime. The TVA Shawnee demonstration had indicated the potential for increased slagging and fouling while injecting limestone. The SBS as configured for the Cyclone Program did have a convection pass simulating that found on typical utility steam generators. Baseline sootblowing cycles were established using a 5 to 8 percent loss in convec- tion pass heat absorption. Figure 9 indicates the thermal cycling that was observed during the West Virginia bituminous coal/Corson atmospheric hydrated, dolomitic lime at a Ca/S of 2. It was observed that SOg capture is sensitive to the amount of tube ash deposits, i.e. time interval be- tween sootblowings. SOg capture did increase with time due to a dual effect: (1) The time temperature environment changes as the heat transfer is reduced, and (2) the amount of sorbent reactant on the tubes increases — increasing the surface available for the reaction. This is the first time we had had a simulated convection pass in a laboratory demonstration and while the effect had been expected, it had never before been quantified. This phenomenon should be inherent in any full- scale demonstration. The baseline blowing (no sorbent injection) re- quired a 60 psig blowing pressure to return convec- tion pass surfaces to a commercially-clean condition. The sorbent effect on fouling required an increase in blowing pressure from 60 psig to 100 psig with blowing on the same time cycle to return surfaces to original absorption levels. This increase represents a significant change in deposition removability and 60 Sulfur aun Blowing capture a (%) lowing 20 0 50 100 Variation of SO2 capture with soot blowing cycle BG&E coal with high Mg hydrated lime @ CA/S=2 80 Time (minutes) Blowing No Sorbent Feed 150 200 250 300 Figure 9 Cyclone SO Capture — Sootblowing Effects correlates with the Shawnee experiment. The successful laboratory demonstration of SOg capture in a cyclone furnace now requires a full-scale field demonstration. In the interim, we have pro- posed a deNOx program to EPRI, utilizing this laboratory facility and applying the In Furnace NOx Reduction techniques, already successfully demonstrated on pulverized coal. That deNOx pro- posal and the full-scale deSOx cyclone field demonstration proposal are presently being reviewed by EPRI and the EPRI deSOx Cyclone Advisory Committee. At this writing, there is no approval to proceed. What Was Learned About In-Furnace SOy Reduction During the past ten years, we have learned the following about In-Furnace SOx Reduction: ¢ Sorbent particle feed size should be 50 percent less that 11m. ¢ Dolomite is less susceptible to ash poisoning than CaCO3. e Residence times in the 2300°F to 1400°F region are critical and required about two seconds to maximize SOx capture. e There are ways to increase sorbent activity. One is increased surface area, but commercialization is several years away. ¢ Temperature environments above 2300°F will “deadburn’”’ the sorbent, however, some sorbents are more temperature sensitive than others. For instance, hydrated calcitic lime prov- ed to be more temperature sensitive than dolomitic in our cyclone trials. e Burner design can affect sorbent surface area generation and activity. e Heat transfer will be reduced in the furnace and the dust loading to the electrostatic precipitators will increase. ¢ Flyash resistivity will increase requiring an upgraded electrostatic precipitator. ¢ Sorbent mixing with the flue gas is critical to maximize capture. e All sorbents are not equal. The above, if known and applied at the original Shawnee Plant demonstration, probably would have increased capture to 50 percent. Retrofits All the reported work to date indicates that sorbent injection for deSOx holds even greater promise now than it did for the TVA researchers in 1972 and 1973, but unknowns continue to require more research on the kinetics involved before this 10 technology can be considered commercial. The sor- bent to SOQ mixing is critical and modeling work continues — both physical and mathematical, with more experiments planned by B&W under the U.S. EPA program. The residence time required is also critical for capture to proceed. In retrospect, and based upon the latest experiments, TVA’s Shawnee unit did not have sufficient residence times, while Weiher Sta- tion, the site of the next full-size demonstration (1985), had almost double the residence time that was available at Shawnee. The retrofit cases all have fixed resident times and, unless they are derated, an increased sorbent activity level will be required to permit meaningful capture. The use of atmospheric hydrated calcitic lime appears most ef- fective in the pulverized coal case and atmospheric hydrated dolomitic lime in the cyclone case appears very attractive when compared to efficiencies ob- tained using limestone. NSPS units have somewhat greater residence times but, in most cases, also have a wet scrubber or are burning compliance coal. The ash-poisoning effects must be mitigated, and research is proceeding in that area. The effects upon increased fouling in the convection pass and degradation of the electrostatic precipitator perfor- mance must be quantified. The latest technology developed for severe Dakota-lignite fouling in con- vection passes, such as the unidirectional water sootblower, will be evaluated for cleaning sorbent retrofitted convection passes. The deNOx and deSOx in-furnace techniques share a common need; namely, that the greatest reductions occur at second stage temperatures of 2300°F or below. A retrofit scenario, as presented in Table 2, is identified as ““demonstrated”’ or ‘not demonstrated.”’ The standard unit is a 500 MWe pre-NSPS design firing 3.5 percent sulfur pulverized bituminous fuel in the traditional manner. Required reductions in NOx and SOx are not yet known so the maximum expected potential for each control technology is given. Precipitator upgrades will range in cost with the high end or worst case utilized in this paper. The case shows a relative cost for the nonexistent standard unit. The fluid bed, while not discussed in this presentation, is a possible second retrofit alternative on smaller units. We are currently performing 4 fluid bed retrofit on a 100 MW unit!®, but are not yet sufficiently confi- dent to suggest a 500 MW retrofit as practical for the following reasons: a) Costs can vary widely — even with a “standard unit.” Table 2 500 MWe DeNOx deSOx Retrofit Costs 50% de NO. 50% deSO. @ Retrofit Low NO. Burners @ Limestone Injection System @ Precipitator Upgrade @ Additional Sootblowers existing electrostatic precipitator Risks — Uncontrolled slagging, fouling, deactivation of Technology Not Demonstrated | Demonstrated $ 2.50/kW $ 19.00/kW $ 35.00/kW $ 3.00/kW $ 59.50/kW Table 2 500 MWe DeNOx deSOx Retrofit Costs b) Promising cost-effective technology has not yet been demonstrated sufficiently, at this time, to provide alternatives to scrubbers for the 500 MW case. c) Actual costs are site specific. Summary The government’s present policy supports ac- celerated research to determine how acidic com- pounds are formed in the atmosphere and the rela- tionship between air emissions and acid deposition effects. Funding has also been increased for research on the control of NOx and SOx emissions from man-made sources; this, coupled with the public’s interest in acid rain, perhaps has made the subject the nation’s most controversial environ- mental issue. The application of New Source Performance Stan- dards to utility stationary sources has been effec- tive in reducing SO2 emissions by 22 percent from 1972 to 1982 (EPA data). The possibility of reduc- ing NOx and SOx emissions, utilizing novel in- furnace techniques, offers potentially-attractive 11 emission reductions that are cost-effective. Babcock and Wilcox will continue their develop- ment, both in the laboratory and in field demonstrations, to aid in commercializing this technology. Acknowledgments In summarizing several of the major in-house deSOx programs, the author was assisted by the principal investigators, researchers, program and technical managers involved, who provided valuable data and updates on their programs. Specifically, the contributions of Al LaRue and Mike Acree of B&W’s Combustion Systems, for the LIMB Program 60 and 120X106 Sorbent and NOx Trials; Anita Lainge and Larry Rodgers of B&W’s Research & Development Division, for LIMB Gas Flow Modeling/Quenching and Time/Temperature Histories; G. Maringo of Combustion Systems for the EPRI Cyclone deSOx Program, and Paul Nolan, Technical Manager of the EPA/State of Ohio/B&W sponsored LIMB program updates, are recognized as contributors and are thanked. References :. Gartell, F.E., “Full Scale Desulfurization of Stack Gas by Dry Limestone Injection” by TVA, Chattanooga, TN. Project Office — R.D. Stern, Control Systems Laboratory, Research Triangle Park, N.C. Attig, R.D., “Additive Injection for Sulfur Diox- ide Control in a Pilot Plant Study,’’ Order 4078-01. B&W report to HEW, March 27, 1970. Glasser, W., Hein K., Kirchen, G., “Further Research into the Reduction of SOg Emission from Brown Coal-Fired Boilers,” paper presented to Members Conference, 1983 International Flame Research Foundation, Noordwijik, Netherlands, May 1983. Report to the Advisory Workshop on LIMB by B. Martin, U.S. EPA, and M. Heap, EERC, Washington, July 1983. Chughtai, M., Michelfelder, S., ‘“Direct Desulfurization Through Additive Injective in the Vicinity of the Flame,” paper presented to 8th Symposium on Flue Gas Desulfurization EPA/EPRI, New Orleans, LA, Nov. 1-4, 1983. 12 10. Yeager, K., Electric Power Research Institute “Control Alternatives and the Acid Rain Issue,” paper presented to First International Con- ference on Acid Rain, Washington, D.C., March 1984. J.A. Barsin, “Options for Reducing NOx and SOx Emissions During Combustion” presented to First International Conference on Acid Rain, Washington, D.C., March 27 & 28, 1984. Hurst, T.B., Nolan, P.S., “Limestone Injection Multistage Burner Demonstration”; U.S. EPA Order No. 68-02-4000 Task 2 Report, “Background for the Process Design,” EPA Pro- ject Officer, R.V. Hendricks. Maringo, G.J., ““SOg Reduction by In-Furnace Sorbent Injection in Cyclone-Equipped Coal Fired Boilers,’’ Monthly Reports, EPRI RP2533-5. News Release: ““Montana/Dakota Utilities Choose B&W to Upgrade Lignite Stoker-Fired Steam Generator.” Technical Paper Nondestructive oxide thickness measurement in superheater and reheater tubing D. W. Bonin Life Extension Services Babcock & Wilcox Barberton, Ohio Presented to EPRI Fossil Plant Inspections Workshop San Antonio, Texas September 9-11, 1986 Babcock & Wilcox a McDermott company BR-1284 Nondestructive oxide thickness measurement in superheater and reheater tubing D. W. Bonin Life Extension Services Babcock & Wilcox Barberton, Ohio Presented to EPRI Fossil Plant Inspections Workshop San Antonio, Texas September 9-11, 1986 Abstract PGTP 86-33 Steam-carrying superheater and reheater tubes in fossil-fired boilers operate at high levels of stress and temperature that cause failure by creep rupture. Creep is accelerated by the growth of a steam-side oxide layer which inhibits heat transfer through the tube wall, increasing the tube metal temperature. Estimation of the remaining life of these tubes requires the measurement of this internal scale which, until now, involved the physical removal and laboratory examination of specimens. To overcome the disadvantages of this method, Babcock & Wilcox designed and built the ultrasonic Nondestructive Oxide Thickness Inspection System (NOTIS™)*. This portable system is capable of measuring the thickness of a tube’s internal oxide without-removing the tube. These oxide measurements are then entered into a computer program, along with tube design and operating data, for calculation of remaining creep life. Consequently, recommendations concerning specific areas of the superheater can be formulated based on the plotted remaining life estimates of these tubes. Introduction Steam-carrying superheater and reheater tubes operating at temperatures above 900°F are subject to failure by creep-rupture. Creep is the process by which metal under stress at high temperatures over a long period of time will fail. The expected creep life of a tube can be estimated from tabulated creep data, provided the applied (hoop) stress and the temperature of operation are known. When the tube enters service, the tube metal in contact with the internal steam begins to form a *Patent pending layer of magnetite (Fe,O,) scale. As the tube’s service life progresses, this I.D. oxide gradually grows in thickness at a rate dependent on temperature. This scale acts as a barrier to heat transfer from gas side to steam side and causes an increase in the tube metal temperature. This augmentation in temperature is a function of scale thickness and, therefore, also gradually increases with time. The magnitude of this increase can be typically as high as 20°F and can have a marked effect on the creep life of the tube. Background In the past, the remaining creep life of a superheater tube was determined by a simple calculation. The tube was assumed to operate at a single temperature throughout its service life. The tube creep life was determined by the Larson-Miller Parameter (LMP) which is dependent on the tube’s operating stress. The calculation of remaining life, as it is performed today, is greatly complicated by the temperature-raising effect of the internal oxide. Internal oxide thicknesses are obtained by the physical removal of tube samples from the superheater (until late, this has been the only method available). The oxide thicknesses are measured in the laboratory from polished metallographic specimens. This method, however, is fraught with disadvantages. For instance, removal of tube samples is difficult, and access to the most desirable tubes, deep within the bank, is often nearly impossible. Also, tube sample removal and replacement is costly, as is the lab analysis. Accordingly, decisions to replace the superheater have often been based upon a limited (and unrepresentative) amount of samples. Nondestructive oxide measurement In order to circumvent the problems associated with laboratory analysis of tube remaining life, B&W designed and built the Nondestructive Oxide Thickness Inspection System (NOTIS™). This is a portable ultrasonic inspection system capable of measuring the thickness of the tube’s internal oxide without physical removal of the tube. The technique is similar to standard ultrasonic wall-thickness measurements except that this computer-enhanced system provides the high resolution needed to detect and measure the I.D. scale. An ultrasonic frequency transducer is coupled to the tube metal surface using a standard couplant, and then a short pulse of ultrasound is directed inwards. The reflections from the metal-oxide interface and the oxide-air interface are displayed on an oscilloscope. The two pulses are overlapped, and the difference in arrival times is displayed digitally. This number is converted to an oxide thickness by means of a reference table developed from calibrated standards. The NOTIS equipment is also capable of standard ultrasonic wall-thickness measurements. The use of the NOTIS method for nondestructive internal oxide measurement has distinct advantages over tube sample removal. The time and costs associated with tube sample removal and replacement with dutchmen are eliminated since these steps are no longer necessary. Also, the NOTIS equipment is so portable that all the necessary equipment can be easily carried by a single individual. Further, large numbers of tubes can be measured for a fraction of the cost of lab analysis. In an eight-hour workday, a two-man team can measure in excess of one hundred tubes for internal oxide thickness. Decisions regarding future replacement of superheater tubing can now be based upon a much larger and more representative number of samples. Inspection locations In order to develop a NOTIS inspection plan for a given unit, drawings of the superheater are first studied. Several criteria are taken into account in order to choose the most critical tubes for testing. The majority of NOTIS testing is usually performed in the SSH and RSH outlet banks, where tube operating temperatures are hottest and oxides are expected to be thickest. Because carbon and carbon- molybdenum steels do not grow oxides of appreciable thickness at normal operating temperatures, components containing a high percentage of tubes of these steels, such as primary outlets or secondary inlets, are usually only spot- checked. At least ten elements equally spaced across the superheater should be examined; the density of inspected banks can be increased in the vicinity of known hot spots. The spacing between elements and access doorways are noted in order to determine the number of tubes in each element that can be reached for testing. The chosen elevation in the element is usually that which lies before a material transition weld, where a given steel grade of tubing is expected to run the hottest. Once the most critical locations for NOTIS testing have been identified, the tubes are labeled. B&W utilizes a standard numbering system which eliminates the possibility of confusion. Typically, all oxide thickness measurements in the superheater are taken in the same plane lying normal to the tubes (i.e., the same elevation in a vertical pendant). This plane is referred to as the plane of inspection. The intersection of the superheater with the plane of inspection is a grid as shown in Figure 1. The x-axis is the number of the element or section, counted from the left-hand sidewall of the unit. The y-axis is the number of tubes in the element, counted from the front of the unit to the rear (or from the bottom to the top for horizontal tubes). The location of the tube may, therefore, be described by the coordinates within the grid. Figure 1 shows an example of this numbering system. An oxide thickness measurement is Rear of Unit { (cue woe ee tre ee eae) ea ore ia are € Fie Cea Ci ee = 4O0-O-O-0-0-0-0-—6-0—0—0-—0-0-0-0-0-0 Ba ae ee ee eae 3 O-O—O—O0-O0-0-O0-O-0-0-0-0-0-0-0-0-0 AP PD 0 A SP SGP We le A Vaal ae be PRI err iel Palate fe Ide 1 O-O—O—O0-—O0—0-—0-0-—-0-—-0-—-0-0-—-0-—-0-—-0-0-0 1 2 3 4 5 6 7 8 9 10. (11 12 130 14 150 16 17 Element Number ms Section A-A Inlet Outlet PT Header Header Inlet Header Rear of Unit -_ Outlet Header “An Horizontal Superheater Pendant Superheater Figure 1 Elevation views and section of typical pendant and horizontal! superheater, indicating standard tube numbering system assigned the same coordinates as the tube from which it is taken. An elevation may be attached to the tube coordinates for greater precision; this is especially important when the same element is inspected at two different heights. Surface preparation After the desired test locations have been selected and labeled, the outer surface of each location must be prepared for testing. A small area of the tube is cleaned free of scale and deposits by wire brushing, sandblasting or grinding. The exposed metal surface is ground until free of ridges or pits. Finally, a finish of 100 grit fineness is applied with 3 sandpaper discs or flapwheels. A nickel-size area of metal so prepared is sufficient for the NOTIS measurement, although a larger area is usually desired to allow freedom of movement if a reading is questionable. Estimation of tube remaining life The oxide thickness measurement is made and recorded, and is assigned the coordinates of the tube for identification. The wall thickness of the tube is also measured in this location. These data are converted, utilizing known operating and design conditions, into remaining life estimates by use of a specially designed and proprietary computer program. An oxide scale technique is used together with an engineering life fraction model. When actual design or operating temperature data is unavailable, the tube temperature is determined by means of the Larson-Miller Parameter, which is derived from the oxide thickness using B&W data. Since the operating time for the tube is known, the effective operating temperature of the tube can be calculated over time from the LMP equation and compared to the original design tube metal temperature (normal base loaded operation) to validate the method. The effects of steamside oxide growth are then taken into account. The oxide acts as a barrier to heat transfer through the tube, elevating the tube metal temperature with time and thereby shortening the tube remaining life. This increase in temperature with time is roughly proportional to the oxide thickness. The gradual growth of the oxide (and the subsequent gradual increase in tube metal temperature) is applied to each interval of the life fraction analysis. Wall thinning due to wastage is also taken into account. Wall loss will result in increased stresses in the thinned areas, which in turn will cause premature creep failures. The following assumptions are used for the life- fraction analysis: ¢ Creep, or stress-rupture, is the primary ultimate failure mode. e At time zero, I.D. scale thickness is equal to zero, and the tube metal temperature is near its design temperature. e Steam temperature remains constant through time. e Tube metal wastage rates will be linear with time. Results and recommendations Once the remaining lives of each tube within the component have been calculated, they are entered in the NOTIS Data Management System (DMS). Each remaining life, retaining the coordinates of its tube, is entered into an array, analogous to Figure 1. The readings are then put into a remaining life band, according to pre-selected ranges, and plotted, Babcock & Wilcox 8/1 1 1 2 2 3 2 21 1 «21 7/1 12 1 2 2 3 3 21 «21 «21 6/1 1 2 2 4 3 3 2 2 2 g@5];1 1 1 2 3 3 3 2 21 21 H4/2 2 3 5 5 5 5 5 3 3 3/1 1 2 2 2 2 2 2 2 «21 2/1 2 2 2 2 2 2 2 «21 «21 Yy}lioao2.2121 2 2 2 21221 0 5 10 15 20 25 30 35 40 45 Element Number Secondary SH Outlet Bank Date of Test Band Sym 1000 Hrs 0286 A 5 1- 50 B 4 51- 75 Cc 3 76-125 D 2 126 - 180 E 1 181 - 200 Figure 2 Example of a full component plot of the remaining lives of superheater tubes in the plane of inspection. Each numerical symbol denotes a range, in thousands of hours, into which each tube remaining life falls. using a number-color code, each on the coordinates of its tube. The result, called a full-component plot, is shown in Figure 2. This enables the reader to view all the remaining lives taken at the same elevation in the component. It is also helpful in identifying and targeting areas of the superheater in need of replacement. When necessary, linear plots showing remaining life of each tube row across the unit, and bar graphs showing remaining life at each tube within a particular element, can be generated. Finally, the remaining life data is studied and recommendations are formulated for each examined component. Such a recommendation might be the replacement of certain superheater elements, or a reinspection of components with the NOTIS method at specific intervals. Technical Paper Economizer header BR-1286 cracking problems due to cyclic boiler operations J. Peet Advisory Engineer G. T. Dunker Manager, Field Engineering Babcock & Wilcox Canada Cambridge, Ontario, Canada Presented to The Canadian Electrical Association Thermal and Nuclear Power Section 1986 Fall Meeting Edmonton, Alberta, Canada October 5-8, 1986 Babcock & Wilcox a McDermott company Economizer header cracking problems due to cyclic boiler operations J. Peet, Advisory Engineer G. T. Dunker, Manager, Field Engineering Babcock & Wilcox Canada Cambridge, Ontario, Canada Presented to PGTP 86-34 The Canadian Electrical Association Thermal and Nuclear Power Section 1986 Fall Meeting Edmonton, Alberta, Canada October 5-8, 1986 Abstract Utilities have experienced an increase in the incidence of economizer inlet header cracking in fossil boil- ers. The increase in these failures is associated with the change to cyclic or “two-shift” operation of the boilers. This paper will address the failure mechanism, its causes and cure, for drum-type utility boilers. Introduction | 1c The changing role of large fossil-fuel-fired steam Thermocouple ye generators in electrical power generation has Ser een increased the emphasis on two-shift operation. and a ers Plants which formerly would be base loaded are Legs now required to operate in a middle load function due to decline in load growth, increase supply by nuclear plants, or fuel cost changes. The routine start-up and shutdown of drum-type fossil units causes stress to several components whose design or operating procedures are marginal with respect to the total cyclic life. This is particularly true of units that were designed and built prior to the change in operating philosophy and thus have var- Tc ying tolerance to cyclic operation within the unit. In addition, certain operating procedures and controls which were provided for base load operation have become inadequate for cycling duty. Three major areas of concern have been identified for a cycling boiler (Figure 1) as follows: 1. Boiler sub-cooling during off-line periods TC Waterwall Inlet Headers Thermocouple (Figure 2). Locations 2. Tube leg flexibility in vestibules and penthouses Tc TC (Figure 3). TC = Thermocouple Locations 3. Economizer inlet header thermal shock. The failure mechanism associated with the above is Figure 1 Test boiler from thermal fatigue (with creep interaction for This paper will address Item 3 (above), econom- high temperature material) which is exacerbated by izer header cracking, by describing the failure the combined effect of the alternating stress and the mechanism, its cause and cure for drum-type number of such stress events. boilers. 316 600 0900 1000 1100 1200 1300 8 Drum Temperature Nec ce ren eeneee LoL Mee nee es, sen. ee Smee, Rear Waterwall and hl Flow Estab. 260 500 Oo we KS 2 s > 8 8 ov o a Q £ 204- £& 400 FD and ID 4 & Fans Off All Boiler Right Waterwall Fires Out Inlet Header 13.5 Inlet Header MW 149 300 | FW to Econ 127 L 260 Left Waterwall Inlet Header Front Waterwall Inlet Header lst Burner in Service \ Synchronize (2% MW's) Purge Air Turbine 1400 1500 1600 1700 1800 1900 2000 2100 Time Figure 2 Cycling test - furnace subcooling. 538 1000 482 900 © 427 u 800 2 2 S > s s Average SSH ~~ g 371F 3 700 Outlet Tube —€ E€ Leg Temperatures o o e e 316 600 Full Load Turbine and 260 500 75MW's | 204 400 0800 0900 1000 1100 1200 1300 Average SSH Outlet Header Temperature Saturation Temperature lst Burner Synchronize in Service Turbi FD and ID ‘\ Roll \ Full Load Fans On \ 1400 1500 1600 1700 1800 1900 2000 2100 2200 Time Figure 3 Secondary superheater outlet header temperatures. Background Before discussing the failure mechanism, a review of the operating procedure which has been used historically on drum-type utility boilers is warranted. Consider a typical drum-type utility boiler (Fig- ure 1) during a typical start-up. Prior to firing, the water level must be visible in the gauge glass (per ASME Code) and a purge of the gas side must be completed. Therefore, the furnace water walls, downcomers, supplies and risers are filled with water as is the lower portion of the drum (above low level trip elevation). Once the fires have been ignited, some steam is generated in the water walls which causes the cir- culation in the unit to commence. Since the drum pressure in general is low at start-up, the circula- tion ratio will be much higher than at full load (Figure 4) and consequently the steam quality leav- ing the water walls will be low. A change in the fluid within the boiler is taking place as steam is generated in the walls. Steam carried into the ris- ers is released into the drum steam space by the separators. If a photograph of the boiler were taken during the start-up and compared with the initial condi- tion, we would find that the total required volume of fluid would be much larger than the original, based on the equations given in Figure 5. The above operations assume that no mass transfer had taken place (i.e., feed flow to the unit equals steam flow from the drum). A numerical example is given, showing the total mass change in a typical boiler between cold condi- tion and 700 kPa with the water level at normal in both cases (i.e., for the same volume). The results of this analysis are given in Figure 6. To maintain the same water level, 59.5 tonnes of water (or steam) must be removed from the system. This represents approximately one hour at 5 per- cent steam flow for the unit. This is obviously much greater than the vented steam flow possible with no feedwater flow. This example indicates the problem with drum boilers on start-up (“drum-swell”) showing that no feedwater is required initially and drum level must be drawn down to avoid flooding the separator. However, prolonged firing with no feed flow has the potential to boil the economizer dry and, therefore, a means of protecting the economizer during start-up is needed. 140 120 S._ Variable Pressure ~ ss iseewe - 100 ite Percent of Full Load Circulating 80 Flow Constant Pressure 60 40 20 20 40 60 80 100 Percent of Full Load Figure 4 Effect of load on circulating flow. il Figure 5 Relationship volumes to mass. 6 Volume Ap Fluid inventory constant 6 Volume = Mr Ve + Mg Vg — (Me + Mg)Vo = drum level change Mass of water at time t = t, Specific volume of saturated water at time t = ty Mass of steam at time t = t, Specific volume of saturated steam at time t = t, Specific volume of subcooled water at time t= 0 Area of water surface in drum [ Xrop (SBW%) Where 1/CR, = Avg. Xrop CR = Foire x CRp x LR Foirc = Circulation factor for load and pressure (Fig. 4) CRp = Design circulation ratio LR = Load ratio Typical values CR, = 1.5x 3.75x 20= 112.5(at 700 kPa) Avg. Xrop = 0.9% SBW Mass (Tonnes) 700 kPa(t=t,) Cold (t = 0) Drum (Vol = 11.5 cu. m) 10.3 ‘1 Dc (Vol = 39 cu. m) 34.8 | 156.3 Supplies (Vol = 10.6 cu. m) 95 Water Walls (Vol = 76 cu. m) 36.6 Risers (Vol = 19.3 cu. m) LL. 5.6 Total 96.8 156.3 Figure 6 Sample calculation of mass change The protection system in use consists of a recircu- lation valve and piping between the downcomers and the economizer inlet header (Figure 7). Thus, when the unit is being fired and no feed- water is being fed to the economizer, the recircula- tion valve is opened to maintain a full economizer. When feedwater flow is initiated, the recirculation valve must be closed to prevent the injection of cold fluid into the downcomers. Failure Mechanism Figure 8 plots the average economizer inlet temper- ature during a warm start-up for a typical unit employing the recirculation valve. These data were taken from tests performed on a 75 MW oil-fired unit. After tripping the unit and stopping feedwater flow to the unit, the economizer inlet header temperature gradually increases to the surrounding temperature. As the pressure in the unit decays, the water level in the drum drops to a point where the addition of water is required. When a flow of cold water is introduced to main- tain acceptable drum level, the economizer header experiences a rapid cooling. The gradual increase in temperature towards ambient is again seen after the drum filling operation. At the beginning of the start-up, with the recirculation valve open, the econ- omizer inlet header experiences a significant ther- mal shock as saturated water is drawn into the header from the downcomer. Cycling between the recirculation valve and feedwater flow during the start-up causes the header to be subjected alter- Orum = Drum Press Level ~—— > Continuous Blowdown ¥ Valve Economizer O Feedwater Flow | _O Feedwater Temperature Economizer Circulating Valve Feedwater Control Valves Figure 7 Boiler schematic. nately to saturated water and cold feedwater (no high pressure heaters in service). It is evident that one start-up can impose several fatigue cycles on the header. Figure 9 shows a typical economizer header arrangement which was modelled by finite element techniques in order to quantify the stresses which 260 500 ; A-10 MW's and E - Boiler Feed J-Turbine Roll Decreasing Pump Tripped K - Synchronize B - Turbine and F-FDand 1D ti Gen. Trip Fans in S Service = C - FW Flow Off S Boiler Feed G- 1st Burner 204 = 400 Pump Tripped in Service oO 2 D- Boiler Feed H-2 Burners 2 oD Pump in in Service 2 = Service; FW = = : 1- BFP in 3 g Fed to Unit ica E E o 3S e < o 149 & 300 ov OD ° o $ x 93 200 0900 1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000 2100 Time Figure 8 Cycling test - economizer inlet header temperature with recirculation valve operation. Economizer Inlet Header Tube Stub Tubes Header Stress Concentration Points Figure 9 Finite element.model of economizer header 5 would occur during a thermal shock. Within the header, the fluid temperature changes as the feed- water valve or recirculation valve is opened. The fluid film coefficient at the header I.D. is a function of the mass flux and total fluid temperature - i.e., the degree to which the water at the two tempera- tures mix prior to contacting the metal surface. The stresses generated are a function of time as the heat transfers through the surface of the metal and into the bulk metal (Figure 10). The fatigue life is expressed as the number of cycles before crack initiation, determined from the amplitude of the alternating stress for the material and its operating temperature. The asymmetrical nature of the header and tube stubs produces a complex heat path which, in turn, produces significant transient AT’s within the material and thus high localized thermal stresses. The local discontinuity stresses from pressure and mechanical loads may also be higher in these regions (Figure 11). The thermal fatigue cracks start in the region of the tube stub on the inside surface, as shown in the photographs of header samples removed from an operating unit (Figure 12). Since the AT’s, and con- sequently the stress, vary between locations and with time, it is impossible to monitor with any degree of confidence. The rate of change of a ther- mocouple on the header O.D. or partially drilled into the header wall will give very little information concerning the rate of temperature change at the header bore. In addition, the thermocouple will bias the result due to its presence and its thermal inertia. Analysis While thermocouple measurements cannot be used to determine the magnitude of the thermal stress produced, it is known that the maximum thermal stress is governed by the total change in fluid temperature to which the header is exposed (assum- ing an infinite film coefficient). Therefore, thermo- couples can be used to establish the equilibrium condition of the header and the temperature of the fluid prior to the transient. Control of this easily measured temperature differential to values con- sistent with a design number of cycles can assure the component cyclic life. Simple techniques such as the exemption from fatigue analysis in Section VIII and Section III of the ASME Code may be applied to establish maxi- mum header-to-fluid temperature differentials 10° Temperature < t, Alternating Stress Amplitude (Ksi) 10 10° 10° Temperature < t, 10 Temperature < t3 Number of Cycles, N Temperature gradient vs. time 10* 10° 10° Figure 10 Design fatigue strength. Contour map of temperature time t = 9 sec Contour Ident —-— 125.7°C — 180 seco 234 --— 288.4 Contour map of temperature for time t = 11.8 sec Contour Ident =-= 125.7 °C —180 © 234 == 288.4 Contour map of effective stress for time t = .9 sec Contour Ident —-— 101 MPa — 297 493 Contour map of effective stress for time t = 11.8 sec Contour Ident --— 124.4 MPa — 3404 --- 556.4 a= 772.4 Figure 11 Finite element analysis results Figure 12 Typical fatigue es eek (a) cracks in an economizer header Press. MPa Temp. °c Press MPa Temp °c Design 19.4 Operating cycle C to F repeated 3 times. 17.9 89 48 Design 370 240 180 tt 11 AB CD EF G HI Total number of cyle A through K = 1000 17.9 48 240 Total number of supplementary cycle H-H = 3000 Results of analysis per ASME code Section VIII Division 2. AD 160.2 « Allowable temp. difference between adjacent points< 115°C for cold or warm starts (N = 1000 cycles max.) « Allowable temp. difference between adjacent points< 78.3°C for hot starts (N = 4000 cycles max.) Figure 13 Economizer inlet header design operating cycles. which could be tolerated for the required number of cycles. These techniques are very conservative and, pro- vided the boiler operator can maintain these values without significant impact to operations, no further effort should be needed. A typical example of applying Section VIII rules is given, Figures 13 and 14. Solution The analysis has shown that limiting the change in fluid temperature will prevent premature failure. The operating data shows that the major problem involves the difference between saturation tempera- ture and the feed temperature during start-up (when heaters are not in service). If the recirculation sys- tem is not used then the major cause of thermal stress can be eliminated. The procedure recommended for cycling units is to discontinue the use of a recirculation valve and maintain a trickle feed of water to the economizer to protect against boiling dry. Adoption of these tech- niques is shown in Figure 15 for the same unit as shown in Figure 8. Comparison of these two curves shows dramatic improvement in header tempera- ture change as well as the rate of change. The minimum controllable feed flow which can be achieved and the equipment available to draw down water from the unit will obviously affect the degree to which drum swell is a problem during the initial phases of the start-up. The use of a low flow, control valve on the feedwater line is recommended to enable the trickle flow to be adjusted to match the drum level shrinkage during off-line and maintain a full economizer during re-start. To control drum swell, a manifold connected to the downcomers (Figure ,16) is now provided on B&W units so that drum level can be lowered safely and effectively, rather than 180 tm = ta + (totale Range of Econ. Header 140 Temperature 120 ao o Temperature tr (°C) Range of 40 Feed Temperature 2 4 6 8 Time T (hours) 10 Figure 14 Expected economizer inlet header and feed temperature. 204 400 A — Turbine and Gen. Trip B—Trickle FW Flow Setup 149 300 Block Valve Bypass to FW Flow (Greater) Temperature C 200 Average Economizer Inlet Temperature F 38 100 1000 and Controlled Through Low Load FW Control Valve C— Slight Adjustment 1100 1200 1300 1400 1500 1600 1700 O — Slight Adjustment to FW Flow (Less) E — Fans in Service F — 1st Burner in Service G— FW Flow Increased H — Turbine Roll | — Synchronize 1800 1900 2000 2100 2200 Time Figure 15 Cycling test—economizer inlet header temperature. 9 being limited by the capacity of the blowdown line, in achieving this function. Conclusion The control of thermal fatigue in economizer inlet headers can be achieved by modifications to equip- ment and, especially, operating procedures during cycling operation. The final solution will differ between units depending on the equipment and configuration of the plant. Testing to identify the magnitude of the problem, followed by the development of the changes to be made and then final testing to dem onstrate its effectiveness, is necessary. Inspection of headers to identify the accumulated damage and determine the need for replacement should also be part of the program for existing units. On new units, recognition of potential thermal fatigue will allow the incorporation of the equip- ment required and the operating procedure to elimi- nate the problem. 10 ¢ Boiler Manifold Drain fam Down Acid Cleaning +<+— Pipe Cap Valves Locked Closed ~———— Run to Waste Section “A-A” (Typical at Downcomers) Downcomers Figure 16 Blowoff valve schematic Technical Paper Considerations in the design of high temperature and pressure refuse-fired power boilers M. P. Hepp Manager, Engineering Signal Environmental Systems, Inc. Hampton, New Hampshire R. M. Nethercutt Principal Engineer Boiler Diagnostic Testing Power Generation Group Babcock & Wilcox Barberton, Ohio J. F. Wood Manager, Power Systems Operations Power Generation Group Babcock & Wilcox Barberton, Ohio Presented to ASME Twelfth Biennial National Waste Processing Conference Denver, Colorado June 1-4, 1986 Babcock & Wilcox BR-1289 a McDermott company Considerations in the design of high temperature and pressure refuse-fired power boilers M. P. Hepp Manager, Engineering Signal Environmental Systems, Inc. Hampton, New Hampshire R. M. Nethercutt Principal Engineer Boiler Diagnostic Testing Power Generation Group Babcock & Wilcox Barberton, Ohio J. F. Wood Manager, Power Systems Operations Power Generation Group Babcock & Wilcox Barberton, Ohio Presented to ASME Twelfth Biennial National Waste Processing Conference Denver, Colorado June 1-4, 1986 Abstract PGTP-85-46 This paper outlines current efforts to advance existing technology of high temperature and pressure mass-refuse-fired boilers by gaining an in-depth understanding of design and operating requirements. To accomplish this, computer software was written to obtain long-term data collection and perfor- mance analysis as well as special metallurgical and ultrasonic evaluations. These programs have resulted in the opportunity to increase design limits on steam temperatures and pressures in this type of refuse-fired boiler. Introduction The combustion of refuse in waterwall mass-fired boilers has been done successfully for over thirty years. The practice that became commonplace in Europe and Japan is now used worldwide as the major alternative to land disposal of solid waste. Refuse-to-energy facilities are designed to provide reliable refuse disposal, state-of-the-art environ- mental control, efficient energy recovery, and reduction of landfill requirements. For systems designed to generate electricity, steam conditions are selected to provide good energy recovery and minimum boiler downtime. The combustion of refuse in mass burning appli- cations presents both the designer and operator with problems of tube corrosion from the products of combustion, tube erosion from ash laden gases, and tube fouling and slagging from the high ash fuel. However, by carefully studying the operation and performance of mass-fired boilers, methods can be employed to prevent unscheduled down- time resulting from tube fouling, wastage and erosion. In an attempt to achieve this goal, Babcock & Wilcox in cooperation with Signal Environmental Systems, Inc. (SES) embarked on a data acquisition and product development pro- gram. Some of the results of that program are given in this paper. Background Signal Environmental Systems, Inc. entered the refuse-to-energy business with the completion of its Saugus RESCO facility in 1975. Steam condi- tions for that facility were designed for 690 psig (4,757 kPa) and 850 F (454 C) in order to match the conditions of the existing equipment at the General Electric Company in neighboring Lynn, Massachusetts. Since these steam conditions were not viewed as severe by the system design engi- neers, serious problems with tube corrosion were not anticipated. However, in just 2000 hours of operation, the SA-T13-T22 tubes in the high temperature superheater (SH) began to experience random failures. Laboratory examination of the failures indicated that high temperature corrosion was the predominant failure mechanism. A task force was established to analyze the boiler design and operation in order to resolve the problem of SH failures. The resultant changes, most notably the use of high nickel alloy tubes (Inconel 825) in the high temperature superheater, have enabled Saugus RESCO to reliably produce high tempera- ture steam for power generation. 600 psig 752 F 250 F 450 kWh/ton (4137 kPa) (400 C) (121 C) (497 kWh/tonne) 900 psig 830 F 300 F 525 kWh/ton (6205 kPa) (443 C) (149C) (580 kWh/tonne) Above based on: Turbine back pressure 2.5” Hg (63.5mm Hg) Refuse Calorific 4500 Btu/Ib HHV (2170 kCal/kg LHV) In-Plant Power 65 kWh/ton (72 kWh/tonne) Table 1 Electrical Output Comparison | Steam Steam — Feedwater Power Pressure Temp. Temp. Sold In 1981, construction commenced on a new 2250 ton/day (TPD) (2036 tonnes/day) facility in Peekskill, New York in Westchester County. Armed with the Saugus experience, SES selected B&W to build three 750 TPD (679 tonnes/day) boilers operating at 900 psig (6,205 kPa) and 830 F (443 C). In addition, construction of a sister plant in Baltimore, Maryland, utilizing three 750 TPD (679 tonnes/day) B&W boilers of the same design was completed in 1984. As shown in Table 1, the ability to increase electrical output is a driving force to depart from the traditional steam conditions of 600 psig (4,137 kPa) and 752 F (400 C). The Westchester facility provided the opportunity to fully analyze the design and operation of a modern high tempera- ture and pressure mass-fired refuse boiler. The testing programs described below provide an accurate assessment of boiler performance through the automatic acquisition of operating data, metallurgical analyses of a variety of alloy materials for corrosion protection, and periodic analysis of tube thicknesses through ultrasonic testing. Data Acquisition System Considerable effort was expended on the design of these units to ensure a state-of-the-art, low main- tenance mass-burning refuse boiler. In order to provide the project with continuous performance analysis capability, while simultaneously provid- ing an “early warning” system for performance abnormalities, it was decided to install a remote, unattended data acquisition system (DAS) to automatically collect and transmit operating data. This system was installed by B&W at the Westchester facility in May of 1984 and at the Baltimore facility in October of 1985. Both sys- tems have been providing performance data since installation. System Configuration and Operation The DAS not only collects and stores boiler oper- ating data, but when requested, transmits these data for remote analysis. The system draws upon two sources for data. Interfaced with the plant computer, the system acquires flow rates, pressures and temperatures from permanent plant instrumentation. Addi- tional data is collected with the aid of a voltmeter internal to the DAS. The two sources jointly pro- vide about 250 data points including gas tempera- ture thermocouples (TCs), SH outlet tube metal temperatures, chordal TC temperatures, numerous steam temperatures, excess oxygen, and draft indications. System Equipment/ Peripherals (Figure 1) The heart of the DAS is a small Hewlett-Packard desktop computer (model HP85). This computer receives hourly portions of its data from the plant computer control system via an RS-232 connec- tion to an unused printer port. The HP85 collects additional data with the aid of an HP 3497/3498 digital voltmeter system. Data is collected hourly by the test computer and is stored on floppy disks. When requested, the HP85 communicates with B&W’s engineering facilities through a telephone modem. System Operation Stand-By Mode The DAS system normally is in a stand-by mode, monitoring action on the printer port and “listen- ing” for incoming calls. A request in this mode will send the test computer into either a data col- lection or data transmission mode. Data Collection Mode Incoming data from the plant computer will send the DAS computer into the data collection mode. The test computer.carefully monitors the incom- ing logs for an initialization command. Detection of this command starts the transfer of appropriate incoming data into the test computer’s memory. This transfer ceases when a termination code is received. Once required information is received from the plant computer, all remaining data is taken from Field Data Input B&W Performance Diagnostics Data Link Utility Engineering Office Computer System un Telephone Modem Data Acquisition Controller Figure 1 System equipment. r 80 4 x 10k Excess Oo, 5 L 1 1 1 1 1 ! L L i 1 7 175 A~ EONS aoe wo Main Steam aad z= 150 = So a 50r Spra 25L Pray 0 4 1 1 a 4 L 1 am 1 1 iJ 200 LL Pf ee ° 1004 J \ ml Air Outlet 2 Ecos Cee nens! Rm = 1 2 3 4 8 9 10 ll 12 Figure 2 Transmitted data 1985. the test system voltmeter. This mode is now com- plete and the collected data is stored, after which the test computer returns to the stand-by mode. Data Transmission Mode A call from B&W’s engineering office will shift the test computer into the data transmission mode. The on-site test computer will then automatically receive instructions remotely regarding data transmission and resumption of data collection. This mode is automatically terminated in the event of telephone system abnormalities. Performance Evaluation (Figure 2) Performance calculations include boiler effi- ciency, convection bank absorptions, flue gas weights and air weights. Key items are then aver- aged and plotted as both daily and weekly aver- ages. These long-term trends provide a major tool for overall performance evaluation. They are generated monthly and distributed to appropriate personnel within SES and B&W. Metallurgical Evaluations Higher temperature-pressure cycles can improve project economics and the likelihood that a given project will go forward. Material selection is one of the most critical design criteria for refuse-fired boilers. Because gas-side corrosion is more aggressive at higher temperature-pressure cycles, arrangements were made to carry out metallurgi- cal evaluations of SH tubes at the Westchester and Baltimore facililties. Based on prior studies performed at Nashville Thermal Transfer and the City of Hamilton (Ontario, Canada), B&W designed and installed a superheater test loop at the Baltimore facility and a composite tube test wall panel at the Westches- ter facility for the purpose of observing corrosion rates of different materials in the refuse-fired environment. Composite Tube Panel Test The boiler waterwalls at the Westchester and Bal- timore facilities are a gas tight membrane con- struction consisting of 2-1/2 inch (63.5 mm) diameter tubing on 3 inch (76.2 mm) centers. This construction is similar to the boilers supplied to Nashville Thermal Transfer. A prior test program conducted at that plant yielded data suggesting a cost-performance advantage for co-extruded stainless-carbon steel tubing over carbon steel. During fabrication of the sidewall panel for one of the Signal boilers, a ten-foot (3 m) long, ten-tube wide panel insert constructed from tubing jack- eted with 304 stainless steel (Figure 3) was installed. A chordal thermocouple tube section was installed in the middle of this insert, in a position of strategic value relative to the location of the furnace arches and overfire air ports. Base ultrasonic testing of the composite tube panel was completed in January 1984, after installation but prior to initial operation. Follow- up ultrasonic testing was done during periodic outages in order to track corrosion progress. Within the first six months, it was visually apparent that the stainless steel was corroding at a rate in excess of the Nashville experience. Since exposed carbon steel tubing in the vicinity of the panel insert was not suffering measurable corro- sion, it was decided to leave the insert in place and observe the corrosion phenomenon after the base carbon steel tubing became exposed. Chordal thermocouple data were charted to track heat input to the panel. Corrosion of stainless steel continued in a patt- ern approximating the gas flow leaving the lower front wall furnace arch. Corrosion was not arrested nor diminished when the stainless steel jacket had completely corroded. In May 1985, a tube failure occurred in the panel insert and the entire test panel was removed for metallurgical studies. All other furnace tubing remains in satis- factory condition. Superheater Test Loop A long-term superheater corrosion study had been completed previously on the boilers of the City of Hamilton, Ontario. These boilers burn refuse- derived, semi-suspension fuel. The results of the study were helpful in formulating test strategies for a similar study at the 2250 TPD (2036 tonnes/ day) mass-fired facility in Baltimore. A special single-flow superheater loop, constructed of nine different materials placed in various locations, was fabricated and installed subsequent to plant commissioning (Figure 4). Steam flow to the loop in the Baltimore plant is controlled separately. Temperatures (gas side, steamside, and tube metal) are monitored by thermocouples embedded or pad welded to tubes in several locations. The entire test loop can be isolated in case of tube fail- ure so that plant availability is not jeopardized by a potential tube leak. All tubing received an ultra- sonic test to baseline tubing thickness. The test Billet Components [sree [esse] (——1 Carbon Steel [== Stainless-Carbon Steel Extrusion Billet Extrusion Figure 3 Composite tube panel. loop, installed since October 1985, is being oper- ated at metal temperatures exceeding 1025 F (552 C) and at a steam temperature exceeding 925 F (496 C). The installation of a similar data acquisition system on this boiler greatly facilitates data col- lection and analysis of this test loop as well as integration of its performance with other boiler data. This system also permits real time observa- tion and analysis of operating data at B&W’s engineering offices in Barberton, Ohio. The test program is continuing and preliminary data indi- cate good correlation with the Hamilton test data, which suggests that higher alloys perform better in the refuse combustion environment. The results of this test program will enhance future super- heater metal selection on a cost versus perfor- mance basis. Ultrasonic Testing Routine non-destructive examination of pressure parts is an important aspect of the preventive maintenance program for refuse-fired boiler oper- ation. Tube thickness measurements by means of ultrasonic testing (UT) alert the operator to poten- tial trouble spots in the boiler setting and indicate a need for more frequent examinations, changes in operating procedures, or replacement/shielding during scheduled maintenance outages. Table 2 Ultrasonic Testing Profile Westchester Unit 1 Number of Location Levels* Level Front Wall 1l Every 5 feet Right Sidewall 11 Every 5 feet Left Sidewall ll Every 5 feet Composite Panel 8 Every 1 foot Center of Nose 1 esees Superheater Face 12 Every 2 feet *Every tube at each elevation was subject to Ultrasonic Testing tion of the furnace and the superheater. The com- plete set of readings was done prior to first firing (baseline). A second set of readings was taken at 2500 hours of operation and a third set, represent- ing 12,500 hours of operation, was taken in Feb- ruary 1986. These data points are carefully recorded and color plotted to illustrate a clear pic- ture of overall pressure part soundness. System engineers and operators carefully review the data for signs of areas with excessive metal loss so that corrective measures can be implemented. Intepretation of this data by both operators and engineers is the key ingredient which turns 17,000 tube thickness measurements into a meaningful analysis. This study readily locates areas of con- cern, their likely cause, and possible corrective actions. By carefully studying the location of the thinning on the tube, this process helps to identify the mechanism of tube wastage as a result of cor- rosion, gas side erosion, or sootblower impinge- ment to provide assurance that unscheduled downtime can be minimized. Superheater Test Loop Existing Superheater Composite Tube Panel Conclusion SES and B&W have developed an optimum arrangement of heat absorbing surface and air flows to achieve reliable combustion and heat re- covery from unprepared refuse. An overall per- formance and materials test strategy (including installation of data acquisition and analysis equipment) was chosen during the design of these boilers. This process has provided an excellent Figure 4 Test section locations. In order to fully evaluate the performance of the 750 TPD (679 tonnes/day) boilers at Westchester, an extensive UT program has been conducted on Unit 1. As shown in Table 2, extensive coverage of the unit provided a detailed profile of the condi- understanding of boilers and materials perfor- mance, allowing system designers to make informed decisions on refuse boiler design optimi- zation with the confidence that overall plant reli- ability will not be compromised. THE AMERICAN SOCIETY OF MECHANICAL ENGINEERS 345 E. 47 St., New York, N.Y. 10017 from ASME for fifteen months after the meeting. Printed in USA. 86-JPGC-Pwr-19 The Society shall not be responsible for statements or opinions advanced in papers or in dis- cussion at meetings of the Society or of its Divisions or Sections, or printed in its publications. Discussion is printed only if the paper is published in an ASME Journal. Papers are available Operating Experience is Reflected in Nashville’s New Refuse Boiler R. M. McNERTNEY, JR. The Babcock & Wilcox Company Barberton, Ohio C. W. CHAMBLISS |. C. Thomasson Associates, Inc. Consulting Engineers Nashville, Tennessee J. T. HESTLE, JR. Nashville Thermal Transfer Corporation Nashville, Tennessee ABSTRACT The operating history of Nashville Thermal Transfer Corporation's resource recovery plant is briefly chronicled. This history, which dates from June 1974 through 1985, shows how the facility has evolved from being a pioneer in resource recovery to being one of the most successful refuse-to- energy plants in the world. This paper will concentrate on the recent plant expansion with special emphasis on the new boiler system. Highlighted will be the major changes and enhancements of the new boiler and associated equip- ment when compared with the original units. These new features are a direct result of the more than ten years of operating experience obtained prior to the addition of the third refuse-fired boiler system. INTRODUCTION Nashville Thermal initiated a trend when, in mid-1974, they became the first district heating and cooling plant to provide uninterrupted service using municipal solid waste as the dedicated fuel. Since that time, favorable operation and improved econom- ics have led to the recent plant expansion. In- cluded in this expansion is the addition of a 400 ton per day (363 tpd) refuse-fired boiler system. This boiler is similar in design to the orig- inal two units but slightly larger in capacity. In spite of the similarities, there are many salient features of the new boiler which set it apart from the original two. These enhancements are a result of both advancements in mass burning technology and the unique operating experience gained over the past decade at Nashville. The purpose of this paper is to highlight these features, each of which has pro- vided solid contributions to the state-of-the art in resource recovery. BR-1290 BACKGROUND Nashville Thermal Transfer is a non-profit cor- poration which built and now operates the plant. The purpose of the facility is twofold: . To convert the energy in municipal solid waste (MSW) to steam and chilled water for uninter- rupted year-round district heating and air con- ditioning of 35 buildings, representing 7.5 million square feet (0.697 square km) of occu- pied space. To process refuse and thus extend the life of the existing landfill. The residue from the plant is less than 10 percent of the volume of the raw incoming refuse and represents a sanitary landfill material. Initially, the plant included two 360 TPD (327 tpd) municipal solid waste incinerator boilers, Units 2 and 3, designed to fire solid waste with a range of higher heating values up to 6000 Btu/1b (3336 kcal/kg). Also included was a standby auxil- iary gas/oil fired package boiler, Unit 1, rated at 125,000 1b/hr (56,750 kg/h) of steam. Chilled water was produced by two steam turbine-driven centrifugal chillers with 13,500 refrigeration tons of water- chilling capacity. Severe problems with the air pollution control equipment were encountered during the first two years of operation which resulted in replacing the origina} jyet scrubbers with electrostatic precipi- tators. These precipitators made a dramatic change. Recent particulate emission tests indi- cated an average of 04022 grains per dry standard cubic foot (5.4 mg/Nm’) corrected to 12 percent co, compared to the U.S. EPA standard of 0.08 (19.6). After the precipitators were installed in 1977, the plant underwent a five-year period when mainten- Presented at the Jt. ASME/IEEE Power Generation Conference Portland, Oregon — October 19-23, 1986 ance was high and availability relatively low. Un- acceptable rates of tube wastage and breakages of the stoker components were chronic problems. There were several contributing reasons for these occur- rences but clearly the most significant one was the operation of boilers and stokers beyond their design ratings. This happened when the boilers were incorrectly rerated to 530 TPD (481 tpd) based on 4500 Btu/1b (2502 kcal/kg) refuse. With this rerating it was planned that one boiler (running at or near 530 TPD [481 tpd]) could meet the customer load for steam and chilled water. The second boiler would be taken off line for repairs so that only one refuse boiler would be operated at a time. This philosophy of operation did not prove prudent and was abandoned in the summer of 1982. At that time, the maximum continuous rating of each boiler was re-established at 80,000 1b/hr (36,000 kg/h) of steam when burning 360 TPD (327 tpd) of MSW with a heating value of 4500 Btu/1b (2502 kcal/kg). The new mode of operation adopted was to operate the two solid waste boilers simul- taneously. The average load on each boiler was projected to be approximately 50,000 1b/hr (22,700 kg/h). During periods of unavailability of one of the boilers, the remaining boiler would be tempor- arily loaded to 100,000 1b/hr (45,400 kg/h). This method of operation, i.e., reduced load on the boilers and simultaneous operation had quite a favorable impact on the plant's performance as shown in Table I. Notably, the overfire air sys- tem modification (to be discussed later) was added just prior to the change in operating philosophy. The effect of this new system was undeniable and most certainly influenced the dramatic improvement in plant performance. Table 1 NASHVILLE THERMAL TRANSFER CORPORATION 1974 to Date Summary of Performance FY Steam Sold Chilled Water Tons/Day Wune-May) (Ibs. x 1000) Sold (tons:hrs.) _ Tons Processed (Avg) % Refuse as Fuel 1974-75 161,812 6,669,890 72.658 199.0 N/A 1975 - 76 237,415, 19,484,412 115,355 3160 N/A 1976-77 236,415, 18,223,868 134,586 368.7 878 1977-78 231,670 19,108,618 140,973 386.2 938 1978-79 253,731 17,389,934 134,370 368.1 910 1979 - 80 230,811 16,292,787 118,730 325.3 912 1980 - 81 198,671 21,595,357 131,664 360.7 88.6 1981 - 82 183,657 20,282,018 120,889, 3312 899 1982 - 83 190,798 21,101,101 145,641 3990 96.2 1983 - 84 230,657 25,105,596 171,718 469.2 974 1984-85 207,459 26,547,926 173,206 4745 96.3 1985 - 86 126,297 15,633,810 102,423 476.4 90.74 (Thru January) i fl Totals 2,489,393 227,435,317 1,562,213 367.1 92.5% (Avg) (Avg. Thru FY 84-85) 1984 PLANT EXPANSION Studies completed by Nashville Thermal's engi- neers in 1982 provided justification for a major plant expansion. The objectives of this expansion were to provide more low-cost heating and cooling and to reduce landfill needs by burning more waste. Downtown Nashville has experienced consider- able growth recently as exemplified by the con- struction of a new convention center, hotel, and financial center. These new buildings and other existing ones provide the additional demand for steam and chilled water. The expansion will have a dramatic impact on prolonging the life of the existing landfill as the plant will handle approx- imately 918 TPD (833 tpd) or 66 percent of Nash- ville's burnable waste. This amounts to 88 percent more waste than the current level. SS} | - Figure 1 District heating, cooling and cogeneration plant fueled by municipal solid waste. Figure 1 is a schematic flow diagram which illustrates how the plant operates. An important aspect of the plant additions will be the 7,300 kilowatt steam turbine-driven generator shown in the schematic. Energy revenues will now be col- lected from the sale of steam, chilled water, and electricity, which will serve to keep the cost of incineration down. An indirect benefit from the turbine generator addition is the flexibility it affords. The amount of refuse processed will not be limited by the seasonal demand for steam or chilled water. As previously mentioned, Nashville Thermal expects to process 88 percent more waste than the current level with only a 56 percent in- crease in refuse-burning capacity. In summary, the following major equipment was in- cluded in the expansion: One 400 TPD (363 tpd) mass burning, waterwall- type incinerator boiler capable of generating 90,000 pounds per hour (40,860 kg/h) of 400 psig (28.12 kg/cm,), 600°F (315°C) steam. This unit is referred to as Unit 4. One 7,300 kilowatt steam turbine-driven gener- ator for cogeneration of electricity. One 6,600 foot (2,012m) distribution system, most of which is in a new 8.5 foot (2.6m) diameter tunnel to provide steam and chilled water to the new buildings under contract. Two 6,750 ton steam turbine driven centrifugal chillers. THE NEW BOILER SYSTEM Figures 2 and 3 show sectional side views of the original units and the new unit, respectively. At first glance, these units look identical except Figure 2 Units 2&3, 360 TPD with 4500 Btu/Ib refuse. for size. However, there are several major changes included in the new unit which are described below. The Reciprocating Grate Stoker As shown in Figures 2 and 3, the boilers at Nashville use the mass burn technique to incinerate incoming waste. "As-collected" refuse is dumped into a storage pit where it is mixed and fed to the incinerator hopper by high speed overhead cranes. Large bulky items are removed by the crane operator before they reach the hopper. The fuel is then fed from the charging hopper to the tri- sectional reciprocating grate stoker where the combustible portion is burned. The ash from the last grate section falls into a hopper where it is quenched prior to discharge into a truck parked below the hopper. A view of the stoker and lower furnace on Unit 4 is shown in Figure 4. Major enhancements of the stoker system applied to Unit 4 include the following: Charging rams. Hydraulically operated charg- ing rams, which feed the refuse onto the first burning section of the stoker, are a major addi- tion. There are two rams, each with dual hydrau- lically operated ram cylinders. The rams have adjustable frequency and length of stroke and are capable of simultaneous or alternate operation. Ram speed is controllable to permit changes in fuel flow. Grates. Grates are cast from a special high chrome/nickel alloy to resist wear, warpage and burnout. As shown on Figure 5, the grate bars are a single-piece casting, 6 inches (152mm) wide. These bars are a new generation design and are Figure 3 Unit 4, 400 TPD with 4500 Btu/Ib refuse. constructed with closely spaced ribs which provide added strength and minimized warpage. High resis- tance air metering openings located in the nose of the grate bars provide uniform distribution of air flow and promote effective fuel burning over the entire grate surface. There are alternating rows of moving and stationary grates. The moving grates are reciprocated across the face of the stationary rows and have a movement of approximately 6 inches (152mm). The resulting action promotes good com- bustion by conveying refuse evenly across the stoker. Fuel is also agitated, torn apart, and tumbled so that maximum surface exposure to the fire is achieved. Grate support. The grate surface support arrangement has been completely redesigned with the stationary grate support fastened to the in- cinerator side plates and the reciprocating mov- ing grates mounted on a dolly-type support. Each of the six grate sections is separately supported and driven. The stoker employs hydraulically operated mechanisms for movement of the grates, gate and feeders. The grate drive arrangement consists of twelve hydraulic grate drives (two per module) which are directly connected to the moving grate dolly. Ram speed, which is control- led by a signal from the combustion control sys- tem, is independent of stoker speed. Flexibility. The importance of flexibility cannot be overstated as it is crucial to success- ful operation of a mass-fired unit. On Unit 4 ram speed is controlled independently of stoker speed. ty aM \—~ N r Figure 4 Reciprocating stoker and lower furnace. Figure 5 Single-piece grate bars. This provides flexibility in the control of the combustion process as well as the ash depth re- tained on the stokep for protection from the fur- mace radiant heat. This also serves to control slagging and reduce the potential for corrosion. Siftings removal. The automatic siftings re- moval system is another innovation that resulted from operating experience on the original units. This system includes six hydraulic cylinder drives (one per module) with pusher rods and blocks. Siftings are continuously conveyed from underneath the ram and each grate section to the discharge end of the unit and into the residue hopper. Overfire air. The new overfire air system installed in 1981 improved performance on the first two boilers. The effects of this new system were far reaching and were seen in the extended life of furnace, superheater and boiler tubing. To a large degree, the same design and philosophy of this system was applied to the 400 TPD (363 tpd) unit. The original (1974) system was designed to operate at 80 percent above the theoretical air requirement with fans sized to provide 85 percent of this air under the grates (underfire air) and 15 percent as overfire air. Nashville Thermal operating personnel soon found that uniform com- bustion could only be achieved by reducing under- grate air flow to some value less than 85 percent of the total. There was no capability, however, in the original overfire air system to compensate for this reduction in undergrate air. The result was less than optimum combustion and reducing at- mospheres, as 80 percent excess air was usually not achieved. Furnace, superheater and boiler tubes were consequently exposed to an alternating oxidizing/reducing atmosphere which accelerated corrosion. To remedy this situation, new higher pressure, higher capacity overfire air fans were installed to provide 40 percent of the total air. Additional high velocity overfire air jets were installed in the sidewalls of both incinerators. Undergrate air. The new undergrate air system on Unit 4 has six air zone compartments below the grate for regulation of undergrate air. Each com- partment is provided with individual air inlet dam- pers which permit manual control of each air zone. Dampers are also provided in the main overfire air duct and in each individual branch. This system provides optimum control of air flow which facilitates good combustion of the MSW fuel. The Furnace Ultrasonic tests taken on the lower furnace tubes of both boilers after one year's time (April 1975) revealed that considerable thinning had occurred. At that time these tubes were replaced and the tubes in the lower part of the furnace from the grate line to a point approximately 30 ft. above the grate were studded and covered with silicon carbide refractory. The application of this refractory has remedied the problem of lower furnace corrosion for Nashville Thermal and was included in the new boiler. This covering is only effective with air cooling or high density stud patterns with maximum heat transfer. 3S ve 8 28 ft So gag o = §3 5 es 5 ee 8 5 OS E s o eee 23 w 2 85 8 3 « 2 5 £ All Studs=1/2" Dia. 20 25 30 32 35 40 45 50 Stud Density (Studs/ft) Figure 6 Relative face temperature vs. stud length and density. Corrosion occurred on the bare tubes of the lower furnace for several reasons but the two most prominent were the alternating reducing/oxidizing environment that the lower fireside was subjected to and the significant amount of chlorine that exists in refuse. The chlorine forms gaseous hy- drochloric acid which causes significant corrosion when in contact with carbon stee y tubing at the metal temperatures encountered. Flame impinge- ment on the furnace walls increases the corrosion rate. There were other contributors to corrosion seen in the units during this period. For instance, inadequate water quality increased the rate of at- tack because metal temperatures were raised due to waterside deposits. A more insidious vehicle of attack which may have affected the units was out- of-service corrosion. Tube surfaces can become moist during idle periods and this moisture com- bines with the chlorides in fireside deposits to form hydrochloric acid. Tube wastage can thus occur rapidly when units are off line in a moist environ- ment. Experience gained in the application of pin studding and refractory on process recovery (PR) boilers was applied to the refuse units at Nash- ville. It should be noted, however, that although there are similarities in the functions of the stud, they serve a different purpose on a PR boiler than a refuse boiler. For this reason, stud dia- meters, lengths, and density should be different. On a process recovery unit, the pin studs serve to cool molten smelt so that a layer of "frozen smelt" is built up on the furnace tubes. Over time the studs wear back due to a combination of erosion and corrosion which necessitates that studs ini- tially be long enough to allow for some wastage. On refuse boilers the studs serve to both anchor the refractory and keep the refractory cool enough to prevent slag buildup on the walls. The less refractory available to insulate the furnace wall, the lower the temperature of refractory will be and thus the likelihood of slagging will be reduced. Another benefit of lower operating surface temper- atures is reducing the tendency for spalling, which reduces refractory life and increases maintenance costs. Figure 6 shows the relationship between the stud density in studs/ft of tube, stud diameter, and face temperature. Unit 4 has a 2 x 2 pattern with 1/2 inch (12.7?mm) diameter studs that are 3/4 inch (19.0mm) long. The maximum depth of refractory is 1 inch (25.4mm) thick. Operating experience has also led to the use of increased erosion-resistant silicon carbide material along the grate line of the furnace. This zone of the furnace is subjected to a scrubbing action as the fuel moves over the grates to the ash discharge. Since this type of refractory has a lower thermal conductivity, it is only used along the grate line of the furnaces. The furnace volumetric heat release gate for Unit 4 will he between 10 - 11,000 Btu/ft~-hr (89000- 97900 kcal/mhr). The installed water-cooled sur- face is sufficient to cool the products of combus- tion to between 1300-1400°F (704-760°C) leaving the furnace. These conservative design guidelines are consistent with those used on the original units at the time of their rerating. They insure adequate residence time for burn-out of fuel and suitable temperature levels for passage into convection sur- face. Along with the precipitators, the furnace de- signs on all three units assist in minimizing stack emissions. Furnaces were designed and sized such that the volumetric heat release rates to theoret- ically determined average temperature planes of 1800°F (982°C) and 1500°F (815°C) do not exceed 30,000,Btu/ft~ hr (267 ,000,kcal/m hr) nor 15,000 Btu/ft~hr (133,500 kcal/m hr), respectively. The Superheater The superheater on Unit 4 is similar to the original two and has a parallel flow, single flow, single steam pass arrangement with a terminal attemperator to control steam temperature to 600°F (315°C). The superheater is of the continuous tube type and consists of a section of pendant tubes extending across the full width of the unit on nine-inch (228.6mm) side spacings. These are six rows deep in the direction of gas flow, and the first two rows of the superheater (those that receive direct furnace radiation) are 2-1/4 inches (57.1mm) OD high alloy tubes. The remaining four rows are 2-1/4 inches (57.lmm) OD SA210 Al-tubes. The corrosion rate of the superheater tubes on Units 2 and 3 has been slowed dramatically and is quite acceptable and manageable today. Main- taining the proper load on the boilers and improved operator savvy are the primary reasons for the im- proved performance. The other major reason is the ability of the boiler and stoker system with ample overfire air to complete combustion in a defined zone low in the furnace. Severe wall tube and superheater wastage occurred on Units 2 and 3 when flame from the fuel bed extended into the super- heater. This accelerated wastage stopped when the overfire air system was modified. Operating at proper excess air levels with combustion completed prior to entering the superheater eliminates the cyclic oxidizing/reducing condition and thus re- duces the potential for corrosion. Tube metal temperature was an important con- sideration in the design of the new superheater. Experience in burning refuse at Nashville and elsewhere shows that the corrosion rate for car- bon steel accelerates at metal temperatures above 850°F (455°C). Experience plus operating and test data on Units 2 and 3 have substantiated that both seamless (SA-210 Al) and ERW (electric resistance welded) SA-178A carbon steel tubes are adequate for use at metal temperatures less than 850°F (455°C). Metal temperatures should always be below 850°F (455°C) on Unit 4 because of the 600°F (315°C) steam temperature. The installation of two rows of high alloy tubes is a conservative measure to take into account the high furnace radiation received by the first tubes of the superheater. During fur- nace upsets, spot metal temperatures in the first two rows may rise for short periods. Field and laboratory tests show that the high alloy mater- ial installed has superior corrosion resistance. The Boiler Bank The boiler banks on all three boilers are similar and are comprised of three banks of tubes, arranged in the direction of the gas flow between an upper and lower drum. The bank is a single pass design and does not use gas baffles to in- crease the velocity to enhance heat transfer. Operating experience has shown that excessive sootblowing in a bank increases the likelihood of corrosion since the protective ash coating is con- tinually removed which exposes bare metal to the corrosive flue gas constituents. Sootblowing will be kept to a minimum (i.e., never more than once a shift, if that) on the new boiler because of the conservative tube pitch. Problems associated with sootblowing are mitigated by the ample cavities provided in addition to the conservative flue gas temperatures. However, sootblowing will be re- quired at times to maintain the surface effective- ness. Erosion in the bank is minimized because of the low design velocities (30 ft/sec [9.1 m/sec]). Miscellaneous Accessories Steam Coil Air Heater. A steam coil air heater was supplied on the new boiler to heat the undergrate combustion air and thus assist in dry- ing and igniting the refuse during instances of high moisture. Auxiliary Burners. Two combination natural gas and No. 2 fuel oil burners were installed in boiler No. 4. Unlike the original two units, these burners are installed high in the furnace. A suf- ficient amount of air flow is passed across the burners when they are idle to protect them against furnace radiant heat and the buildup of fly ash in the windbox. The burners were installed for startup and to provide additional assurance that the supply of heating and cooling would be unin- terrupted. Sootblowers. Reference has been previously made to the problems associated with operating sootblowers too frequently. Conservative tube spacings, gas temperature and surface arrangement on the new boiler will all contribute to the judi- cious use of blowers. Economizer. The design philosophy used on the first two economizers was followed on the new unit. Units 2 and 3 have operated relatively free of economizer problems. The feedwater tem- Underrate Furnace Main Overfire Furnace Air Steam Steam Aux. Fuel Aux. Fuel Oraft Air Press Temp 02 Flow Pressure Flow Air Flow Flow Oo © © ° OO O O 02/CO Steam Trim Pressure Controller Controller Future Steam Rate of Flow Change Controller Limit Controller T Steam Flow Air From Controller H/A Manual Automatic Controller Controller} f(x) —. Air Fuel Air Fuel Cross Cross I Limit Limit Tr H/A H/ H/A H/A Air Flow Fuel Flow Controller Controller f(x) Ram H/A H Speed f(x) Controller ; Grate Speed f(x) f(x) f(x) (Typ 6) f(x) f(x) 1D. Fan Overfire Air Undergrate Air Aux. Burner Aux. Burner Control Controller Flow Control Air Flow Fuel Flow Control Control Figure 7 Schematic of combustion controls. perature entering the economizer is 240°F (115°C) which is the minimum metal temperature. The new economizer is a horizontal, continuous bare-tube gineers contributed to the success of Nashville's facility. design, Major advancements in the area of stoker furnace construction and the combustion type with water flow counter to the gas flow. air system were achieved on the original units and applied to the new refuse boiler. These advance- Controls. Combustion controls for Nashville ments made a solid contribution to refuse-to-energy Thermal have been designed so that the boiler can technology as evidenced by their wide adoption be fired automatically to maintain steam pressure or a target steam flow. Undergrate air flow is controlled with a variable speed forced draft fan. Overfire air is controlled with an automatic dam- per on the overfire air fan. As shown on Figure 7, the control system has 0,/CO trim of excess air. The system also has the fiiture capability of using the furnace temperature override of air flow control. A drop or falling furnace temperature could override the normal undergrate 1 air flow control to increase air flow. Accord- 2. ingly, strong combustion would be re-established in the event of a sudden input of wet or hard-to- burn refuse. SUMMARY 3. Persistence and cooperation among the major equipment suppliers, the operators, and their en- throughout the industry. REFERENCES Hestle, J.T., Jr., et al, "The Importance of Proper Loading of Refuse Fired Boilers", Proceedings of 1984 National Waste Process- ing Conference, Orlando, Florida, 1984. Blue, J.D., et al, "Considerations for the Design of Refuse-Fired Water Wall Inciner- ators", Proceedings of Energy from Municipal Waste Conference, Washington, D.C., 1985. Daniel, P.L., et al, "Furnace Wall Corrosion in Refuse Fired Boilers", Proceedings of 12th National Waste Processing Conference, Denver, Colorado, 1986. Technical Paper Boiler design considerations with pulverized lignite A. D. LaRue W. J. Peet R. A. Clocker D. F. Levstek The Babcock & Wilcox Company Barberton, Ohio Presented to U.S.-Spanish Coal Combustion Symposium Zaragoza, Spain October 21-23, 1986 a McDermott company [ Babcock & Wilcox BR-1291 Boiler design considerations with pulverized lignite A. D. LaRue W. J. Peet R. A. Clocker D. F. Levstek The Babcock and Wilcox Company Barberton, Ohio Presented to PGTP 86-41 U.S.-Spanish Coal Combustion Symposium Zaragoza, Spain October 21-23, 1986 Abstract BéW’s pulverized coal technology has developed over several decades into efficient, reliable boiler systems capable of handling coals with a wide range of characteristics. Lignites present challenges due to their high quantity of inerts, moisture and ash, which depress the heating value of the fuel. Significant quantities of other impurities, such as sulfur or sodium, may be present in the lignite and require special considerations when designing equipment. Spain has large reserves of high sulfur black lignite, and other lignite deposits of varying quality, which can be used as utility boiler fuel. This paper describes boiler design considerations for firing lignite, in general, and for firing high sulfur Spanish lignite in particular. For a given calorific input, lignite fuel handling and pulverizer equipment must convey, dry, and grind twice the quantity of coal, often with four times the ash, as for their bituminous counterpart. Special measures are necessary to characterize grindability of lignite coals to insure proper pulverizer capacity. Pulverizer inerting and clearing systems have been developed to safely handle this reactive fuel during normal operation and following mill trips. Enhanced Ignition - Dual Register Burners have been designed to over- come ignition difficulties with lignite due to lignite’s high inerts and low heating values. NOx emissions are simultaneously controlled by regulating fueVair interactions at the burner to reduce the conversion of fuel bound nitrogen. Systems are available to provide uniform secondary air and fuel distribution among the burners. These systems are essential when burning high sulfur lignite in order to prevent reducing atmos- pheres in the furnace where fireside tube corrosion and slagging can reduce reliability and availability. Furnace design parameters must accommodate large throughputs of fuel and ash, with provisions to control deposition and maintain furnace heat transfer. Superheater heating surface and convection pass design must accommodate high temperature corrosion, slagging, fouling, and erosion while providing steam temperature control and reasonable service life. Bypass systems provide more rapid start-ups to minimize auxiliary fuel consumption. Wet scrubbers provide proven technology to control SO2 emissions. Introduction Knowledge of fuel properties is paramount in deter- mining the design of pulverized coal-fired boilers. A coals and their resultant blends should be accom- modated in the design stage of the equipment. thorough understanding of the coal’s characteristics is a requisite for setting capacity and achieving in- tended performance of the systems and components which comprise the boiler island. Variations of the These variations in coal properties will occur due to differences within seams and mining techniques from a given mine, and use of more than one mine to meet the fuel requirements of the plant over its lifetime. Knowledge of coal from a new source usually takes form from a battery of laboratory tests which define numerous coal properties from small samples of coal available prior to development of the mine. Most of these standard laboratory tests are based on extensive experience with bituminous rank coals. Numerous factors, political and economic, have led to increased usage of lower rank coals in many parts of the world. The behavior of these coals usually cannot be predicted using the standard tests based on bituminous coals. The greater the deviation of the coals’ properties from the database from which the test was derived, the greater the extrapolation and level of uncertainty. Misinterpretation can lead to inadequate boiler capacity, poor performance, and low reliability. Through years of research, development, and operating experience, The Babcock & Wilcox Com- pany has developed technology for using lignite type coals in utility boiler applications. This paper describes systems to handle lignite through fuel preparation, combustion, steam generator design, backend cleanup of exhaust gases, and ash collec- tion. Particular attention is given to methods and equipment to accommodate the high ash and sulfur content of the ‘“Black’’ lignite which is abundant in Spain. While it is generally apparent that lignite can suitably be fired by other means, for instance fluidized bed boilers, this paper strictly addresses pulverized coal applications in high capacity utility boilers. B&W’s experience with lignite-fired utility boilers is listed in Table 1. This experience includes pulverized coal (PC) and cyclone-fired units. Cyclone designs currently are not attractive in most situa- tions due to their NOx emission levels. The PC units range in size from 66 to 775 MWe. The smallest unit is 33 feet wide by 23 feet deep and 109 feet high, and is equipped with 8 burners arranged on the front wall served by 4 pulverizers. Texas Utilities owns the largest of these units (Figure 1). This unit is 90 feet wide, 57 feet deep, and 200 feet high; and has 70 low-NOx dual register burners served by 10 MPS type pulverizers. For B&W’s series of bituminous coal-fired 1300 MWe units, dimensions are 111 feet by 51 feet by 196 feet (WxDxH). Boiler efficiencies for the PC lignite units are in the 81-84% range, with around 10 points of efficiency loss associated with moisture (Hp and H20). Lignite Characteristics Coals classified as lignite by the American Society for Testing and Materials (ASTM) have heating values less than 8300 Btu/lb on a moist, ash-free basis and are non-agglomerating (1). The Class IV lignites are subdivided as Group A or B, with A having heating values from 8300 to 6300 and B less than 6300. Group A lignites are referred to as con- solidated in structure versus Group B as uncon- solidated. Group B lignites are also commonly re- ferred to as Brown coal. International Classification differentiates lignites and Brown coal in increments Table 1 Babcock & Wilcox Lignite Boiler Experience Capacity Fuel Properties Contract — Utility Type MW’s H20 Ash Sulfur Btu/Ib RB 398 Otter Tail Power PC 66 35.7 69 07 6900 RB 412 Basin Elec. Power PC 216 35.2 119 0.5 6320 RB 457 Minnkota Power Coop. CYC 235 34.6 91 0.8 6744 RB 489 Basin Elec. Power Coop. CYC 400 38.2 8.2 0.5 6719 RB 490 Montana-Dakota, Otter Tail CYC 400 40.8 8.0 0.9 6076 RB525 = Square Butte Elec. Coop. CYC 457 36.6 71 0.7 6800 RB 563 Otter Tail Power CYC 456 37.5 7.3 0.7 6994 RB 565 San Miguel Elec. Coop. PC 448 30.0 28.4 2.1 5000 RB580 SWEPCO PC 640 30.5 19.2 0.7 6200 RB590 SWEPCO PC 640 31.0 12.0 08 7200 RB612 Houston Lighting & Power PC 690 33.0 13.2 08 6535 RB 613 Houston Lighting & Power PC 690 33.0 13.2 0.8 6535 BWC Saskatchewan Power Corp. PC 300 36.5 16.5 0.5 5325 BWC Ontario Hydro PC 200 34.0 9.0 0.6 7050 UP 124 _ Texas Utilities PC 775 32.3 18.9 0.6 5740 UP 125 Texas Utilities PC 775 316 16.1 0.7 6637 of moisture content (ash free) and by low tempera- ture tar yield. In either case, Brown coal is distinguished by its very high moisture content. 002 — 092 | ______169° Figure 1 775 MW-PC lignite supercritical unit Texas Utilities Service Inc. Monticello Steam Electric Station - Unit 3 Brown Coal Conventional pulverized coal systems exemplified in Figure 1 are capable of handling coals with moisture contents up to approximately 40%. Limitations are set by pulverizer drying capability with reasonable quantities of preheated primary air and the subse- quent ability of the combustion system to ignite and burn partially dried coal. Brown coal, having moisture contents of 60% and higher, necessarily re- quires a totally different approach to drying and combustion as shown in Figure 2. The massive coal drying requirement is satisfied by extracting hot fur- nace gases from the upper furnace. Gases at 1900°F dry the coal ahead of, and during crushing, by the beater mills. This reduces the moisture content in the coal from 60% to 25%. The moisture-laden flue gas may be transported directly to the burners or separated from the coal and separately introduced into the furnace or vented. The high temperature flue gas used for drying has sufficient heat content Flue Gas Damper 2 1 Primary Air }-— Flue Gas Extraction Inserted Gate ™s . Overfire Air Core Air Brown Coa! Pulverized Fuel Feed Underfire Air DGS-Mill Figure 2 Brown coal firing system to vaporize most of the moisture in the Brown coal and transport it as steam. This system is not suitable for use with lower moisture lignite. The fuel handling, crushing, combustion equipment and fur- nace design are unique for Brown coals and require a separate discussion. However, the Spanish reserves of Brown coal are limited and are already highly utilized at the Puentes and Meirams Power Stations. Therefore, this paper will concentrate on Group A, or Black lignite, referred to hereafter simply as lignite. Lignite Lignites vary in type due to the nature and biochemical alteration of the original plant species (2). Geological classification helps establish which major vegetation forms were present at the start of the coalification process. During coalification the organic material transforms first into peat, then lignite, subbituminous, bituminous, and finally anthracite. The extent of coalification is a function of pressure (supplied by water and overburden), temperature, and time. Geological age does not by itself set coal rank. For example, lignites are in evidence from the Carboniferous Age (300 million years ago) because they were subjected to moderate pressure or temperature; while some much younger deposits from the Tertiary Age (12-60 million years ago) have coalified to anthracite. Wide differences in properties can be found in the bands or layers within a single seam, since the bands correspond to changes in the types of plant residues and im- purities contributed in sequence over a period of time. However, for purposes of classification, the bands are averaged. Lignites, by virtue of their classification at the lowest range of coal heating values, contain a large percentage of inerts, i.e., moisture and ash. Moisture in coal is commonly differentiated as inherent or sur- face moisture. Inherent, or bed moisture, refers to moisture intrinsic to the coal seam and for purposes of coal classification is equal to equilibrium mois- ture. Intrinsic moisture’s link with the coal requires energy transfer to bond or break and has a sub- normal vapor pressure, meaning it is more difficult to remove. Surface, or extraneous moisture, is the difference between total and inherent moisture, and exhibits normal vapor pressure. Surface moisture varies with exposed surface area of the coal and humidity and free water available to the coal. Ash in coal can also be classified as inherent or extraneous. Inherent refers to ash forming mineral matter that was part of the original plant material from which the coal developed, and is generally less than 2% of the total ash. Extraneous ash includes detrital material which settled in the deposit, crystalline materials carried by water into the deposit, saline deposits, plus material introduced through mining from the roof, floor, and partings. The range of ash content in lignites vary from less than 10% to over 40%. Table 1 lists B&W’s lignite experience including several coal parameters. Moisture content can be seen to be relatively high (30-40%) but descends to the teens in Spain and elsewhere. Table 2 lists analyses for Black lignite prevalent in the Teruel region of Spain. The heating value of this lignite is relatively low, less than 5000 Btullb, depressed by the 57.1% inerts. The moderate moisture content in this lignite, 17.5%, can readily be dealt with by conventional methods. The very high percentage of ash, 39.6%, calls attention to related issues of abrasion and erosion in coal preparation equipment, coal piping, the furnace back Table 2 Babcock & Wilcox Coal & Ash Characterization of Teruel Lignite Coal Source Country - Spain — Raw Coal and Ash Data (As rec'd. basis) — Coal Chemical Analysis Ash Chemical Analysis Proximate analysis SiO. 37.43 Alo03 23.50 Total moisture 17.50 TiO> 0.79 Volatile matter 22.10 Fes03 27.00 Fixed carbon 20.80 CaO 4.00 Ash 39.60 MgO 0.72 Na,O 2.18 Btu per Ib 4856 K,0 142 P20 0.43 Ultimate Analysis SO3 402 UD 0.00 Moisture 17.50 Carbon 28.50 Hydrogen 1.60 Oxygen 570 Nitrogen 0.40 Sulfur 6.70 Ash 39.60 — Calculated Results of Characterization Analysis — Sum of inerts. 57.1 Lignitic factor 0.17 FC-VM ratio 0.94 Total bases 35.32 Dulong HHV Btu/Ib 4967 Total acids 61.72 Lbs SO. per MKB 27.59 Base.acid ratio 057 Lbs Nitrogen per MKB 0.82 Dolomite ratio 13.36 Silica ratio 54.13 Lbs Na,O per MKB 178 Lbs Ash per MKB 8155 Total Si + Al 60.93 Results Interpretation Ash type Slag propensity Foul propensity Erosion (recommend) Eastern bituminous —severe— —severe— 45 ft/sec pass and airheaters. The low heating value of the lignite and high quantity of ash can lead to pro- blems with coal ignition, flame stability, and turn- down. In addition, the low heating value compounds problems with high ash loading. More than twice as much lignite is needed for a given calorific input compared to a typical bituminous coal with 12,000 Btu/lb. Such a bituminous coal with 10% ash has 8.3 pounds of ash per million Btu input, whereas the Teruel lignite has 81.5 pounds of ash per million Btu. This order of a magnitude increase in ash throughput justifies additional capital expenditures for conservatism to reduce associated problems. Ash characteristics dictate the abrasiveness of the coal during pulverization as well as its erosiveness (an exponential function of the coal transport velo- city). Slagging and fouling propensity of lignites, as with other coals, vary with the chemical make-up and mineral forms in the ash. Empirical correlations have been developed by boiler manufacturers to predict slagging and fouling tendencies as a function of ash chemistry and physical properties. These are discussed further in the section on furnace design. Fuel Preparation and Combustion Equipment With this background on lignite coal characteristics, attention is now directed to equipment designed to be compatible with the coal. System considerations start with coal preparation and follow through to back end clean-up equipment. Fuel Preparation Equipment Coal Handling A well-designed coal handling and feeder system is elemental to the successful operation of the pulver- izers. As with other coals, lignite needs to be clean- ed of the foreign material introduced during mining, transportation, storage and reclaiming. Magnetic separators will remove the majority of iron based materials. Other stray materials such as wire, cloth, and wood should be removed by screening as they can lead to feed interruptions, mill dribble, and fires. Crushers are used to size lignite to pulverizer re- quirements, typically a 1% inch top size. Fines must be minimized due to their high surface area and the hygroscopic nature of lignites. Fines with high sur- face moisture tend to segregate and form blockages in feeders and can impose an overwhelming drying burden on the mill during the wet season. Coal silos with steep angled (70° plus), stainless steel lined conical hopper are recommended to reduce feed dis- ruptions and corrosion (3). See Figure 3. Large diameter feed pipes to and from the feeder are used to minimize stoppages in these critical components. Pulverizer Section The functions of the pulverizers, or ‘‘mills,” are to generate surface area by crushing to a specified fineness and throughput, while drying the coal in preparation for combustion. Major issues related to pulverizer selection include type, sizing requirements, capacity limitations, wear life, reliability, and safety. Safety, although listed last, is of foremost concern with reactive coals like lignites. The substantial moisture content of the fuel requires elevated primary air temperatures (500-700°F) to accomplish drying. These high temperatures can be transferred through the mill to stagnant coal, either with the mill running or idle. As the coal is heated, gases are emitted which can reach flammability or explosive limits. Cases of fires, explosions, and personnel in- ¢ Bunker Outlet 5 Conical Outlet 72° min. - to be lined with _~ Stainless steel _- Downspout - 24 3/4” 0.D. ~~ Secoal Nuclear Coal Monitor —S-E-Co. Pressurized Coal Valve Dresser Coupling _— S-E-Co. Gravimetric Feeder Explosion Pressure Construction Feeder Outlet Hopper 1/4” Stainless Steel (reinforced for explosion) SE-Co. Air Coal Valve Downspout - 18” 0.D. — 1/4" Stainless Steel heh — Dresser Coupling - 18” dia. Pulverizer Inlet Figure 3 Bunker to pulverizer arrangement with S-E-Co. 24” gravimetric feeder and accessories juries/fatalities are too well known. These fuels re- quire a means to establish an inert atmosphere suffi- cient to smother a flame or smoldering coal follow- ing a trip (4). The mill should be isolated, inerted, and cleared of fuel prior to restart. Horizontal ball tube mills are susceptible to fires and explosions due to the inability of clearing the mill of lignite during normal or emergency shut- downs. Restarts force the introduction of air to a potentially explosive mixture of gases, or may re- quire feeding air to a smoldering fire. Consequently, ball tube mills are not recommended with lower rank coals. Roll and race or ball and race mills can be cleared of lignite to avoid these conditions. B&W’s guide specifications (for low rank coals with mill inlet Primary Air [PA] temperatures of 450°F or higher) call for complete isolation of the mill from all sources of air following a mill trip, or detection of a fire in a mill by CO monitoring equipment or elevated mill outlet temperature. See Figure 4. The isolation is accomplished without allowing an in- crease in air concentration, which could reach ex- plosive limits of gases in the mill. Complete isolation is accompanied by introduction of prescribed quan- tities of inert media (No, COg, or steam) for a timed period. This is followed by a water wash to clear the Local Manual —-—= Feeder Vent Air —-e= 1” NPT Local Manual 1-1" NPT Hot (Male) — Air Remote Auto Pulverizer Tempering re Air ir Hot a ' \ a Air _ Remote Auto —] Ie Auto hero a Tempering 1” NPT Remote >} Air Auto. (Male) Auto Water Wash Nozales — Inerting Control 60-80 psig Header Damper 4 1” NPT (Male) Water Wash Nozzles 60-80 psig Figure 4 MPS pulverizer isolation, inerting and clearing PA duct and windbox of coal through the pyrites system , first with the mill idle and then with the mill running to grind out its contents. Pulverizer Sizing Most current day mill sizing specifications tor new units stipulate a spare mill requirement. This criterion generally requires specified capacity and fineness with one mill idle, with the unit at maxi- mum continuous rating, and with use of the pulver- izer design coal. Many times the pulverizer design coal is an artificial combination of the negative fac- tors related to mill capacity based on a range of sample coals obtained from the mine(s), e.g., highest moisture, lowest heating value, lowest grindability. The actual coincidence of these factors in a single coal may never occur, but even if such a coal exists, it should be considered statistically for incidence of occurrence. Sizing to a more realistic ‘‘worst coal”’ will result in a better match between pulverizer per- formance and burner performance. As a rule, at full load with one mill out, the pulverizers should be at or above 85% of rated capacity on performance coal. This ensures reasonable turndown of the boiler without the need for flame stabilization by lighters or removing mills from service. Pulverizer capacity for a vertical spindle mill grinding a given coal will be set by one of three parameters: power input, coal bed level, or drying capacity (5). Power input refers to shaft horsepower provided for grinding and can be varied within mechanical limitations of the machine design. Coal bed level is linked to the circulating load of coal in the path above the grinding zone up to the classifier and back to the grinding zone. The coal bed level changes according to the feed rates to the mill. When feed rate exceeds grinding capacity, the cir- culating load increases, bed density increases and eventually collapses, resulting in a clogged mill. Drying capacity is set by the sensible heat of the primary air at the maximum flow and temperature permitted by the machine design. Grindability Capacity determinations are further complicated by the complex relationship between grindability and the moisture content of low rank coals. The familiar Hardgrove Grindability Index (HGI) was developed during the 1930’s using bituminous coals. ASTM D409 describes the apparatus and test method. The Continuous Grindability Index (CGI) was developed by B&W in the early 1970’s to simulate the characteristics of a continuous flow machine. Both methods related grindability to the amount of new surface created by particle fracturing as a function of the work performed on a coal sample. However, with high moisture low rank coals, the grindability varies unpredictably with the coal moisture as the coal is being dried. This is shown in Figure 5. Ex- of @ MPS-32 tt -32 apparen grindability @150 1407 @ ASTM HGI Se (air dried) Mississippi Lig. 120+ 4 100F 2 S 807 Onakawana 3s & 6 60+ Saskatchewan Power OT °%38 40 \ 27 207 Basin Mind 0 10 20 30 40 50 Moisture, % Figure 5 Hardgrove Grindability Index versus fuel moisture 4 Vent Raw Coal Bunker Classifier Motor Thayer Gravimetric Feeder Micropul Dust Collector Rotary Valve Rotary Valve Mill Motor 55 Gal. Drum Electric Scale ea Nat. Gas Burner T — Thermocouple P —Pressure Taps L Figure 6 MPS-32 test arrangement perience has shown that the true grinding character- istics of low rank coals are best determined by testing them in an actual pulverizer. B&W uses the MPS-32 for such tests (Figure 6). The MPS-32 is a scaled down version of the commercial mill and is equipped with a rotating classifier with variable speed motor to facilitate testing. Crushed coal is fed to the mill in series of six or seven brief tests in which fineness is varied with feedrate to determine capacity. Results, as shown in Figure 7, relate fineness, power, and capacity of the test mill for the coal of interest. The 70 percent passing 200 mesh point for feedrate and net power are then compared to B&W standard performance curves to derive 107+ apparent grindability. None of the methods, (HGI, CGI, or MPS-32), completely correlate to full scale field results for all coals. Difficulty in obtaining small representative samples and variations in HGI with moisture con- tribute to errors with the HGI and CGI methods. The 3000 pound sample used in the MPS-32 reduces representative coal sample problems but such a large coal sample can be difficult to obtain from new sources. The MPS-32 tests have proven to be valu- able in evaluating grindability of low rank coals and result in more accurate capacity determinations, Feed rate (as received), kg/hr Net specific power (as received), kWh/ton which in turn result in a better selection of 10 20 40 60 80100 pulverizer equipment for a given application. Percent passing through 200 mesh Pulverizer element wear life and grindability deter- arrangement which provides each roller assembly Figure 7 MPS-32 capacity test results minations are dependent upon differences in grind- ability of the various coal constitutents. The “pure coal” grindability varies with total moisture as previously discussed. However, portions of the ex- traneous ash can have radically different grind- abilities. The Teruel lignite ash, which constitutes 40% of the feed, displays high quantities of iron and sulfur. Iron, expressed as Fe2O3 represents 27.0% of the ash and the coal has 6.7% total sulfur. These elements are present largely in the form of iron pyrites, FeS2, which enter the mill as rocks of various size and display low grindability. (Note only a small fraction of the pyrites are rejected by the mill for such a coal.) The lower grindability com- ponents require more work to achieve a given fineness and consequently spend more time in the grinding zone and coal bed than the easier to grind portions. Numerous samples taken from commercial pulverizers have proven that FeS2, SiOg, and Al2O3 are present in disproportionate quantities in the grinding zone material compared to the parent coal analysis (6). Coal Abrasiveness and Grinding Element Wear An accurate determination of coal abrasiveness is re- quired to establish pulverizer element wear life and for predicting mill spare part and maintenance requirements. The Yancy-Gear-Price (YGP) test in- volves rotating four removable wear blades in a charge of coal for a fixed number of revolutions. The weight loss of the blades is used as an indicator of abrasiveness. In B&W’s experience (7), the YGP test has not proven to be a reliable indication of abrasiveness when compared to actual field results (Figure 8). The YGP method is sensitive to sample 120 100 80 60 40 ir YGP Index ~ MG/KG, Coal 20 10 20 30 40 50 60 70 80 MPS-89 Tire Wear Rate ~ tons/in* eae ee ea ae Figure 8 YGP Index vs. MPS-89 tire wear rate particle size, weight, and moisture, and to blade clearance and test duration. The MPS-32 wear test has proven to be a more realistic measure of abra- sion or wear as shown in Figure 9. In the MPS-32 wear test, a known quantity of coal is run through the mill and grinding elements are measured for wear. The roll assemblies are cleaned and precisely weighed under strict condition to assure accuracy. Weighings take place initially; once again after coal is introduced and the mill has reached steady condi- tions; and a last time after the sample is depleted. To enhance MPS-32 results, B&W uses relative quartz value determined from the raw coal and from the grinding zone of the mill, in comparison to other samples from which field data are available. As yet, the data base of quartz and pyrites concentration in grinding zone residence is not adequate for wear rate predictions. 70 60 50 40 30 20 MPS-89 Tire Wear Rate ~ tons/in® 10 10 20 30 40 50 60 MPS-32 Tire Wear Rate ~ tons/in? | Figure 9 MPS-89 vs. MPS-32 wear rates Grinding element wear life is important in any PC application, but crucial with abrasive coals. Wear life is directly dependent on usable wear material of the grinding elements of the pulverizer. The MPS-89 has nearly 3000 pounds per tire of through-hardened wear iron in its 4'%-inch-thick tread. The MPS design enables the owner to operate until the tread is completely worn through. The key to the high wear metal usage (40% of the total tire weight) is the MPS roll wheel pivot with freedom of movement in a direction radial to the mill table. This permits continuous correction of contact between the roller and ring as wear progresses. Highly abrasive coals necessarily reduce wear life for any pulverizer. This emphasizes the impor- tance of maintaining pulverizing capacity and fineness through the wear cycle. The MPS achieves this by virtue of the pivoting roll arrange- ment, and on-line adjustability of the spring loading system. The MPS's reliability is a result of conservative mechanical design, and features which reduce maintenance and improve availability. The triple reduction gearbox has high design margins and an excellent service record, yet can be removed as a separate component without pulverizer teardown. Ceramic wear protection is standard equipment in the classifier and recommended in other selected locations for erosive coal (Figure 10). Ceramic Lined Turret Ceramic Lined Classifier Cone Ceramic Lined Housing Liners Ceramic Lined Roll Wheel Bracket Wear Plates (Lower) Figure 10 Recommended ceramic application on MPS pulverizers Combustion System The primary functions of the combustion system are to combine the pulverized lignite and combus- tion air in a controlled manner and safely initiate the combustion process for completion in the fur- nace. Performance is measured by turndown, com- bustion efficiency (excess air and unburned com- bustibles), NOx emissions, and reliability. In addi- tion, downstream phenomena such as furnace slag- ging and gas side corrosion are influenced by the air/fuel distribution and combustion performance. Combustion of Lignite Ignition of pulverized coal is dependent upon several coal characteristics. For a particle of lignite, the process begins with radiative and con- vective heat transfer to raise particle temperature. The surface of this irregular, fibrous particle begins to burn, and the gases released from the particle ignite. The combustion of the volatile gases further raises the particle’s temperature and contributes to the heating of other nearby par- ticles. The non-agglomerating nature of lignite means the particle does not go through a molten plastic transformation during combustion. Rather the mass is reduced to noncombustible ash through the carbon burnout stage as oxygen penetrates the particle. The high volatile matter of lignite (on a moisture and ash free basis), a fuel ratio (fixed carbon to volatile matter ratio) usually near unity, and significant inherent oxygen, are indicative of the reactive, free burning nature of these coals. Burn- ing profile tests provide insight on relative com- bustion characteristics of various coals. The test involves heating, at a specified rate, a prepared coal sample while measuring its rate of weight loss as a function of temperature. Figure 11 shows burning profiles for coals typical of the various ranks. The burning profile for the lignite displays a relatively high peak at 100°C indicative of a high release of moisture. This is rapidly followed by a resumption in weight loss associated with ignition. The ignition point is considered the temperature at which weight loss reaches 2 mg/minute in the positive sloped curve following moisture release, e.g., 220°C in Figure 11. The peak rate of weight loss occurs at lower temperatures for lignite than the higher rank coals. This fact, plus the steepness in the preceding portion of the curve indicate rapid rates of combustion. Char burnout, again refer- enced to the 2 mg/minute rate, is completed at a lower temperature (610°C) for the lignite than the other coals. The non-agglomerating nature of lignite and high inherent surface area of the par- ticles enable the use of reduced fineness from the pulverizer. Even so, combustion efficiency tends to be very good with lignites. AS FIRED 7 VM FC ASH 47 B46 96 64 773 15.1 164 710 12.1 354 480 15.1 33.4 332 214 74 COAL _M_ Ww 12 04 15 12.0 ANTHRACITE LV BITUMINOUS HV BITUMINOUS SUB-BITUMINOUS LIGNITE 16 14 RATE OF WEIGHT LOSS, Mg/MIN 700 900 1000 FURNACE TEMPERATURE, °C load, which becomes a limiting factor in turndown. For wall-fired systems, the burner design issues center on high inerts loading in the fuel stream and recirculation constraints. The effect of recir- culation on the fuel jet depends on the ratio of the secondary air to primary air (SA/PA), the relative momentum of the two streams, and the near field temperatures around the burner. Pulverizers at maximum capacity tend to operate within a fairly narrow band of primary air/pulverized coal ratios (PA/PC), typically 1.5 to 2.0 lb PA/Ib PC. As the coal’s heating value drops, more and more coal must be supplied to the burners to reach a given thermal input. This is accompanied by proportional increases of primary air leaving less secondary air. Primary air can reach 40% of total air, resulting in a SA/PA ratio of 1.5 (See Table 3). Note most cir- Figure 11 Burning profiles for coals of different rank The burning profile test indicates the ease of burning lignite once it is ignited, but does not address the difficulty associated with generating conditions for ignition to occur. The high quan- tities of inerts, ash and/or moisture, in lignites interfere with coal particle heat-up by absorbing heat in this critical stage. In addition, the fuel stream of primary air/pulverized coal supplied by the pulverizer has about half of the heating value compared to a bituminous coal. As a result, many lignites are much more difficult to ignite and sus- tain during low load operation than higher rank coals with higher ignition temperatures and less inerts. Burner Design Wall-fired burner systems are intended to produce stable flames at each burner. Circular swirl-gener- ating burners induce recirculation of hot furnace gases back toward the burner to ignite and stabilize the air fuel mixture. Flame scanners monitor conditions at each burner and the flame safety system takes action when flame signals drop off. The advantage of the wall-fired system is that flame stability is primarily related to burner performance and less dependent on bulk furnace condition (load). This offers the potential for flame stability at lower boiler loads. By way of contrast, corner fired burner systems rely on flame support corner to corner around the furnace. This allows for delayed ignition with low rank coals while still providing flame support to the sequential corner burner. However, the flame ignition and stability are more closely linked to furnace temperature and 10 Table 3 Fuel Effects - B&W PC Systems PA Temp Fineness Fuel In Out PA SA/PA (%-200) Anthracite 600 200 17-20 4-5 85+ Bituminous 400 150 17-20 4-5 70 Subbituminous 600 140 21-25 3-4 65 Lignite (~40%M) 600-700 135 26-40 15-3 65 cular burners were developed using bituminous coal with SA/PA ratios of 4 to 5. With a SA/PA ratio of 1.5, much less secondary air is available to induce recirculation toward the fuel jet. The momentum ratio between the fuel stream and the recirculated gases can be controlled by burner design. Near field temperatures around the burner become very important. Delayed ignition tendencies associated with a given lignite must be compensated for such that ignition occurs within a short distance (one throat diameter) of the burner. This enables heat from the ensuing flame to be recirculated back to the throat to ignite incoming fuel. Stability is lost when near field (2-3 throat diameters) downstream temperatures fall so that recirculated gases have insufficient heat to ignite the fuel. Enhanced Burner Ignition B&W has developed the Enhanced Ignition burner (Figure 12) to fire lignites and other difficult to ignite fuels in a stable, efficient manner while limiting NOx emissions. Primary air and pulveriz- ed coal enter the burner elbow where centrifugal forces act to concentrate the pulverized coal along the outer radius at the exit of the elbow. The coal Secondary Air Variable Input NM CFS Lighter A Burner Elbow. A\ (Ceramic Lined) ~, Throat Pitot Outer Air Zone Inner Air Zone — (Coal Rich) (Coal Lean) Recirculated Furnace A side Gases, Flame Primary Air Damper Vanes| — \ Adjustable + | Outer Adjustable Vanes Inner Vanes Pulverized Coal Figure 12 Enhanced ignition burner stream is deflected into the conical diffuser (U.S. Patent 4,380,202) where it is dispersed into a coal- rich pattern along the inner perimeter of the coal nozzle with a coal lean region in the interior por- tion of the nozzle. The burner nozzle is designed for a low exit velocity to increase residence time of the fuel in the ignition zone. The fuel is not dispersed in the throat, rather it enters as a low velocity axial jet. Secondary air enters the burner through a bell- mouth opening. The air travels past an impact/suc- tion pitot grid which locally meters the quantity of air to each burner. With this information, the secondary air distribution is balanced among the burners by use of the adjustable slide damper. A portion of the secondary air enters the inner air zone and passes through adjustable inner vanes. These vanes generate high swirl resulting in a low pressure zone around the fuel jet. The majority of the secondary air travels through the two-stage outer vanes. The first set of outer vanes are fixed position curved vanes to initiate swirl, followed by a second stage of adjustable vanes to impart final swirl. This results in efficient generation of high swirl air and permits swirl adjustment. The outer air swirl results in the recirculation of hot gases and flame, from a region 2-3 throat diameters downstream, back along the outer boundary of the fuel jet. The fuel jet ignites along this boundary near the burner and the flame propagates along the fuel jet as the flow continues into the furnace. Flame stability is achieved by the Enhanced Ignition burner by virtue of a design which pro- duces a suitable set of conditions in the near flame field. The design is relatively insensitive to bulk furnace conditions, meaning stable flames are achieved at very low boiler loads. For example, the Enhanced Ignition burner was tested in a BkW Stirling boiler at our research center in Alliance, Ohio (8). This battery of tests included a variety of fuels from delayed petroleum coke to Texas lignite. Analysis of the lignite fuel is shown in Table 4. 11 Table 4 Texas Lignite Analyses for Enhanced Burner Tests Proximate Analysis, percent by weight As received Total moisture 35.2 Volatile matter 26.8 Fixed carbon 26.3 Ash 11.7 Ultimate Analysis, percent by weight As received Moisture 35.20 Carbon 38.40 Hydrogen 2.90 Oxygen 10.58 Nitrogen 0.64 Sulfur 0.58 Ash 11.70 Higher Heating Value, Btu/Ib 6520 Sum of inerts 46.9 Fuel ratio FC/UM 0.98 The burner produced high combustion efficiencies and was very stable over the load range, limited by the mill to 55% to 100%. To better evaluate flame stability under adverse conditions, the burner was tested during a cold start. Following an overnight shutdown the burner was brought into service with a PA/PC temperature leaving the mill of 129°F and with 90°F secondary air (primary air preheat was required to grind the coal effectively). The flame was quite stable and so the lighter was removed after 10 minutes. Flame stability continued throughout the cold start without secondary air preheat or lighter flame support with the Enhanced Ignition burner. To further evalute stability, steam was introduced in the coal piping ahead of the burner to simulate a higher moisture lignite. Flame stability was maintained up to 43% total moisture to the burner, under adverse start-up conditions without preheated secondary air. The self-stabilizing characteristic of the flame pattern was responsible for this success. By comparison, many convention- al burner designs (wall or corner fired) encounter stability problems below half boiler load, with a hot furnace and secondary air of 500°F plus. NOx emissions are controlled by the Enhanced Ignition burner in much the same manner as the B&W Dual Register Burner (9). Typically 80-90% of the NOx produced from pulverized coal is a result of oxidation of nitrogen bound to molecules in the fuel, i.e., fuel NOx. This NOx is effectively reduced by limiting oxygen available to the fuel, particularly during devolatilization of the coal. The preceding discussion described the manner in which air/fuel interaction is limited in the Enhanc- ed Ignition burner. Moderate NOx emissions are an added benefit of this approach. Secondary Air Control Uniform secondary air distribution is essential with high sulfur, high iron coals. Air deficiencies lead to a reducing environment in portions of the furnace which can result in furnace tube wastage due to H2S and FeS attack. The furnace sidewalls and tubing adjacent to the burners in the combus- tion zone are particularly vulnerable to wastage. Several North American utilities have had to replace major sections of furnace walls in these areas as a result of tube wastage. Some have resorted to aluminizing the tubes to reduce wastage. A more direct solution is to carefully con- trol the secondary air distribution to the burners. Since the early 1970’s, B&W’s boilers have been equipped with compartmented windboxes (Figure 13) to control secondary air. Previous to this, Compartmented Windbox Furnace Observation Doors Burner Secondary Air Control Dampers Burner Secondary Air Foils Figure 13 Pulverizer-burner system secondary air was metered and controlled on a boiler basis, i.e., the secondary air was metered downstream of the airheater and was then free to distribute according to natural flow patterns in open windboxes on the front and rear walls. The compartmented windbox partitions burners assoc- iated with each pulverizer into separate windboxes. Secondary air is then metered and controlled on a pulverizer basis, i.e., relative to the loading on each mill. A second major advantage of this sys- tem is reliability. Air flow control is accomplished by dampers which are not subjected to furnace radiation and associated binding problems. 12 Secondary air distribution has recently been im- proved further by the burner design exemplified in the Enhanced Ignition burner (Figure 12). With this design, secondary air flow distribution is accomplished on a per burner basis. Each burner has an impact/suction pitot grid at the entrance to the burner to measure air flow. Lab tests have demonstrated these pitot devices to be accurate within +5% over a wide range of settings. During commissioning the burner vanes are adjusted for optimum combustion and the slide dampers are used to balance air flow to all burners in each compartment. The vanes and sliding dampers are then locked in position and the compartment air foils and dampers are used to control flow over the load range. CFS Lighters The burner ignition system has an increasingly dif- ficult task to perform. First and foremost, it must be able to provide ignition and flame stabilizing functions on demand. This involves intermittent service of moving equipment in a hot, dusty environment. This by itself has proven to be a difficult assignment for many lighter designs. B&W has developed the CFS lighter (Figure 14) to meet Flame Detector Electronics HEI Power Supply \ J HEI Probe > Retract Cylinder Spark Tip Speed Control Valve Cc Flame Stabilizer ‘Sprayer Plate Integral Flame Detector Sleeve Assembly Quick Disconnect — NS OS porn va Oil Atomizer Air Iniet Lighter Retract Cylinder Oil Inlet - Proximity Insert Switch Figure 14 Variable input lighter type CFS/oil these requirements. This lighter is equipped with a reliable, high energy spark ignition system to ignite the lighter fuel (from a gas spud or oil atomizer). The CFS lighters are also equipped with an integral flame detector which is built into the lighter assembly and dedicated to proving lighter flame. The lighters are controlled in groups by a microprocessor for improved reliability and flexi- bility to different applications. Another duty of the ignition system is to limit the quantity of auxiliary fuel burned. The oil or gas fired in lighters is a precious resource which in many cases is difficult to supply (trucked in) and, in any case, expensive relative to lignite. Obvious- ly the lighters are only one factor in auxiliary fuel consumption. Burner flame stability and boiler cycling requirements are primary factors influenc- ing usage of lighters. But the fuel usage per event can be reduced by a well designed ignition system. B&W’s CFS lighter has variable input capability. Lighter input is increased to maximum for ignition purposes during the brief periods when burners (mills) are being taken in or out of service. However, for flame stabilizing functions, e.g., during periods of wet coal flow interruptions, lighter input can be reduced to a nominal value commensurate with conditions. Furnace Configuration and Surface Arrangement Slagging A primary factor influencing furnace design is the slagging propensity of the coal to be fired. The for- mation of slag deposits is caused primarily by the physical transport of molten or partially fused ash particles entrained by the gas stream. When the particles strike the wall or tube surface, they become chilled and solidified. The strength of their attachment is influenced by the temperature and physical contour of the surface, and by the direc- tion, force, impact, and melting characteristics of the slag. Coals with low ash fusion temperatures (i.e., coal ashes that are plastic or semi-molten at tempera- tures less than 2200°F), have a high potential for slagging. Some coals have significant iron content in the coal ash. As iron content increases, it signi- ficantly depresses the fusion temperatures of the ash in a reducing (oxygen deficient) environment. Slagging is normally confined to the radiant heat receiving surface of the boiler (Figure 15), but it also occurs in the convection superheater if gas temperatures exceed prudent levels. Slagging is a complex phenomenon that occurs because coal ash is heterogeneous and contains a number of chemically unrelated constitutents; and because of varying furnace conditions, such as gas temperature. The design of the boiler and its mode of operation definitely affect slagging. A boiler that is base loaded for long periods of time is more likely to slag than one that has varied boiler load. The flow pattern of the flue gas which may bring large concentrations of dry or molten ash into con- tact with boiler surfaces influences the build-up of slag deposits. Slagging Indices. Liquid slag in a boiler is not troublesome as long as it remains a true liquid, 13 ip WS x Rss QA A N N . Zp KL i, Or ti I, \ 7 Burner Zone OOMOWAVW] QW» Lon uu Slagging Figure 15 Typical coal fired unit with a viscosity below 250 poise, as it forms a thin layer of minimal thickness on the furnace tubes. The most troublesome form is plastic slag which arbitrarily is defined to exist in the region where the viscosity is 250 to 10,000 poise. An indication of where slag deposits may form in the furnace can be predicted by comparing gas temperatures in the furnace with temperatures defining the plastic slag zone. The plastic zone discretely defined enables placing wall blowers where they will do the most good. Slagging in the convection surfaces can be minimized by adequate furnace sizing. Slagging indices have been developed to deter- mine the probability that a particular coal will or will not slag. The base-to-acid (B/A) ratio determin- ed from a coal ash analysis is a rough indicator of the slagging potential for that particular coal ash. If the B/A ratio of a coal’s ash ranges between 0.5-1.2, there is a good possibility that slagging problems might be experienced. The slagging index (Rg) used for bituminous coals is B/A x S, where S is the percent sulfur in the dry coal analysis. This index is not valid for the low sulfur coals, such as subbituminous coals having lignitic ash characteristics. The slagging indices used for these coals are based on the initial deformation tempera- ture (IT), and hemispherical softening temperature (HT), or the plastic viscosity-temperature relation- ship of a coal ash. Fouling and slagging characteristics have been statistically related according to ash classification. This empirically determined relationship is pre- sented in an abbreviated form in Table 5. Table 5 Bituminous & Lignitic Ash (Eastern) (Western) * Defined by ratio of iron (Fe203) to sum of calcium (CAO) and magnesium (MgO) in ash « Bituminous type when Fe.0; > CaO + MgO Lignitic type when FeO; < CaO + MgO In establishing slagging indices for low rank coals based on fusion temperatures, the IT and HT temperatures are used because they indicate the temperature range of plastic slag for a coal ash. The lower the temperature within this range, and the greater the range, the greater the probability of slagging. A slagging index using the reducing and oxidizing IT and HT fusion temperatures of a coal ash has been developed and is expressed in the following equation (1): (Max. HT) + 4 (Min. IT) 5 Rs* = (1) Where Rs* is the slagging factor, (Max. HT) is the highest reducing or oxiding hemispherical softening temperature, and (Min. IT) is the lowest reducing or oxidizing initial deformation tempera- ture. The slagging factor, Rs*, is used to identify the types of slagging coal as shown in Table 6. Table 6 Slagging Classification versus Rs* Factor sequently the curves are expensive to obtain. A slagging index using the viscosity-temperature relationship has been developed that will identify the slagging tendency of coal in the same manner as Rs and Rs*. The index (Rvs) is applicable to bituminous, as well as to the lignitic ash coals. Fouling The volatile constitutents in coal ash (i.e, Nag SO4 or CaSO,) cause fouling, and can be used as an in- dicator of the fouling potential of a given coal. The constitutents condense on fly ash particles, boiler tubes, and deposit in areas where the temperatures are in a range where the constitutents remain liquid. These constituents react chemically with fly ash, other deposits, and the flue gas to form bond- ed deposits. Fouling Indices. Two factors that influence foul- ing are deposit hardness and rate of deposition. Deposit hardness is affected by the chemical com- position or ash temperature and, to some extent, time. The rate of deposition is dependent on the volatile constitutents of the coal ash and the amount of ash present in the coal. The sintering strength of a coal ash, as determined in the labora- tory, is an indication of how hard a deposit might become at different temperatures. Messrs. Attig and Duzy have related the fouling potential of a coal ash to sintering strength and the sintering strength to the chemical composition of the coal ash. Using this precept, a fouling index was for- mulated for bituminous coal (Table 7). Table 7 Fouling Index for Eastern Coals Re = B/AX Na20 Slagging Classification Slagging Factor Rs*, °F Medium 2450-2250 High 2250-2100 Severe less than 2100 Slagging index - Viscosity (Rvs) Fouling Classification Fouling Index, Rr Low less than 0.2 Medium 0.2-0.5 High 0.5-1.0 Severe greater than 1.0 The most accurate indicator of potential slagging for bituminous or subbituminous and lignitic coals is the viscosity-temperature relationship of the coal ash. Considerable information can be obtained from a viscosity curve since it clearly defines the plastic temperature range in both oxidizing and reducing atmospheres. The disadvantages in using a viscosity curve to predict slagging are that large coal ash samples are needed, only a few labora- tories have high temperature viscometers, and con- 14 The NagO content of coal ash and a measured deposition rate in a laboratory pilot scale furnace have been used to establish fouling indices for lignitic ash coals. Figure 16 shows the percent rate of vaporization for a CaSO4 mixture extracted from a North Dakota lignite. Figure 17 shows the relationships of the sodium content of coals with a lignitic-type ash to the volatile constitutents of the coal ash. As the sodium content increases, these volatile constitutents also increase, providing a 1.0 —- vo if on 2 = Oo o = So 2 © . | 01 CaSO, - NazSO, extracted from North Dakota lignite 15 mg sample TGA - air velocity part sample ~ 0.1 ft/second 001 | 1800 2000 2200 2400 Temperature, F Figure 16 Percent rate of vaporization of CaSO4-NazSO4 [2 ne — 10 - - 9 _ Wt % Naz0 in lignite ash 00 0.2 04 0.6 08 1.0 Weight % CaSOg - NagSOq Layer/Weight % Slag Figure 17 Increase of CaSO4-Na2S0,, layer with increase of Naz20 in North Dakota lignite ash basis for using the Na2O content of a coal with a lignite type ash as an indication of its fouling potential (Table 8). Table 8 Fouling Classification as a Function of NazO in Ash Fouling Classification % NaO Low to Medium less than 3 High 3-6 Severe greater than 6 Boiler design for high and severe slagging coals Mathematical Modelling to Predict Slagging, Fouling Empirical indices of the type discussed above have been developed to account for extensive experience gained from burning pulverized coal. These indices do not account for variations in fuel preparation, such as changing the degree of fineness during pulverizing, micronizing, slurrying, or benefici- ating. Variations in burners or atomizers, or in design details of coal-designed or oil-designed steam generators are not considered. In general, effects of unique proprietary design variations in combustion and heat transfer equipment are not considered by the indices. A further missing ele- ment is related to choice of operational details in- cluding load, the pattern for changing load levels, excess air, and frequency of sootblowing. Also, the empirical indices are only as good as the experi- ence base from which they are derived. A real-time prediction is necessary to describe the strongly time-dependent effects of fuel charac- teristics on heat exchanger performance in steam generators. An additional requirement is to develop a way to relate quantitative data from drop-tube furnaces, pilot-scale combustors, indus- trial steam generators, and utility steam generators. With the power of modern computer models, it appears possible to calculate, estimate, and/or cor- relate the changes in composition of entrained par- ticles and vapors that occur within the steam gen- erator due to separation of particles and vapors from gases. Key measurements will be required for input and verification. A new matrix of significant variables for evalu- ating fuel effects has been identified for B&W’s numerical model — the Generic Method concept. Corresponding measured data are required. Many of these generic variables have not been recognized nor treated as significant variables. Therefore, generic samples and measurements differ from some of those currently used for traditional empirical applications. Generic deposition calculations are divided into two parts. The first considers whether particles or vapors actually contact the surface of a specific heat exchanger. The second part considers whether or not particles stick upon contact and remain as part of the deposit. Three primary mechanisms are considered for contact: 1) inertial impaction, 2) thermophoresis, and 3) vapor deposition. These mechanisms act simultaneously and have important couplings. inertial impaction is predominant for particles larger than 10 micrometers. Thermophoresis usually predominates for particles smaller than one micrometer. In the size range between approxi- mately 1 and 10 micrometers, inertially-enhanced thermophoretic diffusion is important. An input for generic engineering calculations is provided by analyzing samples of gas-entrained particulates taken near the site of deposition. By working with these samples, details of the complex interactions among fuel characteristics, fuel preparation, and combustion within the flame envelope are all integrated into the production of the sample itself. Deposit strength can be measured using new quantitative on-line techniques as reported in Reference 14. Strength is expressed in terms of the level of heat flux that can be maintained with various intensities of sootblowing during pilot-or commercial-scale testing with the boiler generating steam at rated capacity. Penetrometer force measurements have proven to be unreliable indica- tors of sootblower requirements for powderous deposits. Strength of any fireside deposit can be quanti- fied as jet energy required (JER) to recover a certain percentage of heat flux. The corresponding “maintainable” heat flux (MHF) represents an operational range between sootblowing with an average value that is lower than the value for a new, completely clean tube. The average MHF value depends both on JER and duration of the time between sootblowings. Details of steam generator design, including the number of sootblowers and their locations and pressures, can be expressed as jet energy available (JEA) at the deposition site. Jet energy available is a function of peak impact pressure (PIP), jet dwell time, and tube bank penetration efficiency. Sootblowing will only be effective when local JEA is equal to or greater than current local JER. 16 Generic calculations combine heat and mass transfer effects. Thermal properties of deposits in place on heat exchanger surfaces can and have been measured in the furnace. Effective thermal conductivity and radiant emittance are strongly influenced by voidage within the deposits. With unfused, powderous deposits, the effective thermal conductivity may be only 1/10th of that usually given for the intrinsic particle property. The degree of sintering or melting within the deposit also produces a major effect on thermal properties (15, 16). Therefore, thermal properties are con- sidered in terms of gradients within the deposit. Gradients also are present for a wide array of chemical and physical properties. The first applications for generic calculations using the DEPO-F model developed by B&W have 100 ANGLE FROM STAGNATION LINE TTTTT1T T | 10 5 E\ \ T TTT PREDICTED DEPOSIT THICKNESS (mm) T 0.1L L VARIABILITY IN THE [~ MEASUREMENTS = ~ lo.o1L PREDICTIONS ASSUME DEPOSITS UNAFFECTED . f- DEPOSITS ARE REMOVED BY SOOTBLOWING + BY SOOTBLOWING [ EVERY 8 HOURS 1 al if 1 1 1 | 1 0 20 40 60 80 100 120 140 160 180 ANGLE FROM STAGNATION LINE (DEGREES) I Figure 18 Comparison of predicted and measured deposit thickness on simulated secondary superheater tube in small boiler simulator at end of 96-hour test — coal firing only ANGLE FROM STAGNATION LINE MEASURED VARIABILITY IN THE MEASUREMENTS DEPOSIT THICKNESS (mm) TTT 117 T PREDICTIONS ASSUME DEPOSITS UNAFFECTED 0.1 be >| [> DEPOSITS ARE REMOVED BY SOOTBLOWING + BY SOOTBLOWING + EVERY 5 HOURS D200 eee Drea Lee eee eee eR eee 0 20 40 60 80 100 120 140 160 180) ANGLE FROM STAGNATION LINE (DEGREES) 4 Figure 19 Comparison of predicted and measured deposit thickness on simulated secondary superheater tube in small boiler simulator at end of 96-hour test — with sorbent injection shown remarkable agreement with measurements. Predictions were accurate even with large quanti- ties of sorbent particles injected into an upper- furnace location for SOx reduction. The predictions included the effects of periodic sootblowing during 96-hour test periods. Pilot-scale measurements were used to validate site-specific model capabili- ties. Figures 18 and 19 compare predicted and measured deposit thickness distributions on a simulated secondary superheater tube for cyclone- fired coal and coal-plus-sorbent tests, respectively. Figures 20 and 21 compare corresponding changes in heat exchanger performance for predicted and measured heat flux during 96-hour test periods. Note that sootblowing was conducted every 8 hours for coal firing, and every 5 hours for coal- plus-sorbent firing. Higher peak impact pressures (JERs) were required for deposits produced with sorbent injection. Decreasing the time between sootblowings helped to minimize the average reduction in heat flux produced by the sorbent. 17 Average Heat Flux + 48.1 KW/m? < = xs 3 z= 5 x L Ebi ttt ttt ttt a Soot- s r Blowing 0 4 1 4 4 1 1 1 1 1 0 20 40 60 80 100 Time (hours) ea Average Heat Flux + 46.2 KW/m* “e 807 > ie 5 | xz 60 = = LI s 2 40F eer ti ttt tttrty 3 20 F Soot- a + Blowing 0 1 1 1 1 1 L 1 1 L 0 20 40 60 80 100 Time (hours) Figure 20 Comparison of predicted and measured heat flux on simulated secondary superheater tube in small boiler simulator — coal firing only Design for Slagging and Fouling Control — Spanish Lignite Using the empirical method mentioned above, and assuming that the coal analysis shown in Table 2 is representative of Spanish Lignite, the coal ash is classified as bituminous (Feg03 > CaO + MgO). Base/Acid ratio is 0.57. The slagging factor (B/A x S dry) is 4.6 and the fouling factor (B/A x Na OQ) is 1.24. Using these empirical methods, this coal ash is classifieu as having severe slagging and severe fouling potential. Of primary concern, from a slagging standpoint, is the iron content of the coal ash and how it reacts with the sulfur in the coal. The iron is present in pyritic form and, if locally reducing areas are present in the furnace, large quantities of iron sulfide could be formed. This phenomenon would significantly lower coal ash fusion temperatures and the ash viscosity would be in the slagging range at relatively low Average Heat Flux = 41.2 KW/m? 80+ 60 40 Measured Heat Flux (KW/m°) SUELEEEEEE EEE ET EE ET ° 20 | 10 noth 80 . ne g eo Average Heat Flux = 42.3 KW/m Sol ttttt ttt t treet ttt Time (hours) Figure 21 Comparison of predicted and measured heat flux on simulated secondary superheater tube in small boiler simulator — with sorbent injection gas temperatures wherever reducing conditions exist. The high ash content of the coal would also cause the ash accumulation rate to be quite high whenever the ash viscosity is in the plastic range. If an ash viscosity versus temperature diagram or the ash fusion temperatures (reducing and oxidiz- ing) were available, the slagging potential of this coal could be more accurately assessed. Based on the limited data available, the furnace should be sized so that the average gas temperature entering the first steam cooled tube bank on 36 inch side centers, or less, should not be greater than 2100°F. It should be noted that the very high sulfur content of this coal is an indication that the coal ash may have very high corrosive properties in the superheater and the reheater. Since the steam entering the above-described bank has pick- ed up a significant amount of superheat from the primary superheater (or from furnace platens, if present) it should be shielded from direct furnace 18 radiation by being placed over the furnace arch. As discussed previously, the furnace and firing system should be designed to minimize the poten- tial for the development of locally reducing zones. Excess air to the operating burners should not be allowed to fall below 20% at rated burner input. Burner clearances to the sidewalls, hopper incline, and bottom of the furnace arch should be conser- vative. These clearances are a function of rated burner input. As rated burner input increases, these minimum allowable clearances must be in- creased in order to avoid unacceptable deposition rates and possible flame impingement on the fur- nace walls. In order to provide sufficient insurance against the above conditions occurring with this very high sulfur lignite, it is recommended that input/furnace plan area be limited to no greater than 1,500,000 Btu/ft?/hr. From a fouling standpoint, convection pass side spacing should be conservative until the gas temp- erature is decreased to around 1400°F. Below this temperature level, the sodium compounds formed in the furnace cease to play a significant role in the fouling process and no longer interact with the other constituents in the coal ash to form high sintering strength deposits. The gas temperature limits listed in Table 9 are recommended to control fouling. Table 9 L Fouling Influences on Gas Side Design Gas Temperature Minimum Clear Side Spacing Range, F Inches —_ From ash ID temp. to 1850F 20” 1850 to 1700F 10" 1700 to 1400F 6" Coal Ash Corrosion All coals contain enough sulfur and alkali metals to produce corrosive ash deposits on superheat and reheater tubes. Coals containing more than 3.5% sulfur and 0.25% chlorine may be particularly troublesome. Sodium and potassium in the coal ash also play an important part in the corrosion process by combining with sulfur to form corrosive compounds which are molten at around 1000°F. Since the sulfur content of Spanish lignite is very high (up to 8.62% as received from the Elvira Mine) and sodium and potassium are present in significant quantities, the probability exists that coal ash corrosion would shorten superheater and rehater tube life unless suitable preventative design measures are employed. Pere eerie Corrosive Zone Gas Temperature, F 600 950 1050 1150 Metal Temperature, F 1250 Figure 22 Fuel ash corrosion. Stable and corrosive zones Experience has shown that deposit temperature adjacent to the tube surface is the dominant factor affecting the rate of corrosion. Deposit temper- ature is a function of OD tube metal temperature and bulk gas temperature (Figure 22). The data used to develop the empirical curve shown in Figure 22 did not include coals having the very high sulfur levels characteristic of Spanish lignite. For this reason, it is possible that high rates could occur at lower metal/gas tempera- tures than shown in this figure. It is recommended that austenitic materials be used where super- heater or reheater tubes are exposed to high gas temperatures and direct furnace radiation. If wide spaced furnace platens are present, the bottom portion of the platen loop and the leading and trailing edge tubes should be upgraded to stainless steel material. If long term operation still yields unacceptable service life for some of these platen tubes, consideration should be given to replacing them with tubes clad with Inconel 671. Tubes that handle high temperature steam should not be located in high gas temperature areas and should be shielded from cavity radiation. Consideration should also be given to lowering the allowable use temperature limit of SA213T22 material when selecting materials for tubes that handle high temperature steam. The high sulfur content of the coal must also be taken into consideration for the design of the regenerative airheater/steam coil airheater systems. 19 It is recommended that the regenerative airheater average cold end temperature not be allowed to fall below 200°F. With 80°F ambient air condi- tions, a reasonable MCR design exit gas temperature (uncorrected for airheater leakage) for the regenerative airheaters would be 320°F. Flyash Erosion The silica plus alumina content of the Teruel lignite ash is high, as is the ash content/million Btu input. B&W has developed an empirical cor- relation between these two indices and erosion potential of the coal ash. This empirical method in- dicates that average flue gas velocities in the con- vection pass and in the regenerative airheaters should not be allowed to exceed 45 feet per second. It is also recommended that primary air fans be placed upstream of the regenerative airheaters so that they are not required to handle dust laden air. B&W has also developed a test method for deter- mining flyash erosion potential that consists of blowing a measured quantity of prepared flyash into a target at a standard velocity, time duration, and angle of attack. This method provides a more accurate assessment of erosion than the empirical method. Bypass and Start-up Systems With the increased requirements of load-following and cycling operation placed on fossil-fired units, increased interest in bypass and start-up systems has been generated. Proper selection of a bypass system requires understanding of the fundamental characteristics of the system and the operating objectives of the unit being considered. One discriminating difference between bypass sytems is whether the steam bypasses both the superheater and the turbine or the turbine alone. Each system type has unique characteristics as defined in Table 10. Table 10 Bypass System Characteristics Superheater Turbine bypass bypass Flow capacity Low (7%) High (50-100%) Bypass fluid temperature Low(Sat) High (850-1000F) Boiler outlet steam temperature at shutdown and start-up High Low (except Applicable to BFP flow initiated >50% load) (OTSG) Yes Yes Applicable to steam generation flow initiated (drum boiler) Yes Limited Need for IP turbine bypass No Yes Relative cost Low High Value for turbine runback Limited Difficult to implement Turbine bypass and start-up systems have been absorption with the higher steam flow does not widely used in Europe. They have been applied necessarily result in increased superheater temp- mainly to once-through boilers with spiral wound erature. Differences between steam output required furnaces operating at variable pressure and with from the unit and the steam generation in the fur- single valve turbines. These three components, nace resulting from firing rate cause a pressure each of discrete design, have shown good operating change within the unit. experience and serve identified operating objectives. Drum Boiler Start-up System The desirable features of the turbine bypass For a drum boiler, Babcock & Wilcox advocates its system, as applied to the European design boilers drum boiler bypass and start-up system shown in and turbines, are not directly applicable to drum Figure 23. boilers. A fundamental reason is that for a once- This system has a wide range of flexibility to through steam generator the flow is initiated by generate and deliver steam to the turbine for roll- the boiler feed pump, thereby requiring either a ing through initial loading over a wide range of large motor or a large source of steam flow for the controlled throttle pressures and temperatures. turbine drive during start-up. For a drum boiler, With this start-up and bypass system, thermo- however, the boiler flow is initiated by firing, and couple probes are installed in the gas stream be- only after steam is generated at the desired tween the secondary superheater and the reheater pressure. to measure furnace outlet gas temperature. This The fact that the drum boilers physically sepa- data allows automatic control of fuel input during rate the boiling surface from the superheating sur- start-up and protects the superheater and reheater face results in a final steam temperature which is against overheating when steam flow is limited. It not a simple function of feed flow and firing rate. also provides a means of steam temperature con- The steam generated is dependent on the heat ab- trol when high steam temperatures are desired for sorbed in the furnace and economizer while the turbine rolling. superheating of that steam results from super- Superheater bypass valves are located at the heater surface absorption at low boiler loads. In- drum or primary superheater inlet discharging into creases in firing rate will result in increased fur- the condenser. Higher-than-normal steam tempera- nace absorption and consequently a higher steam ture can be maintained while retaining control of flow generation requiring an increase in feed flow throttle and/or drum pressure. By using this by- to maintain drum level. The increased superheater pass valve, high temperature, up to the gas temp- 542 Generator Primary superheater seco 501 OX 500 oma Condenser Secondary superheater Superheater by-pass Reheater Figure 23 Superheater bypass system for drum boiler 20 erature limit or full load steam temperature, can be achieved for hot restarts with a limited amount of steam bypassed to the condenser. Steam attemperators mix nearly saturated steam with superheated steam at the secondary super- heater and at the reheater outlet to provide low steam temperature to high pressure (HP) and intermediate-pressure (IP) turbines for cold and weekend starts. They are also available for positive temperature control for occasions when the desired rolling temperature is exceeded by the normal rise of boiler outlet temperatures. Steam temperatures as low as 50°F above saturation can be maintained for rolling the turbine while maintaining normal start-up firing rate. Superheater division valves located between the primary and secondary super- heater provide for dual pressure operation where the drum may be operated at a pressure other than throttle pressure. This dual pressure opera- tion permits taking the unit off line for an over- night shutdown with optimum conditions of the maximum drum pressure, steam temperatures, tur- bine metal temperature and valve position, and a minimum throttle pressure. This system also per- mits storage of excess energy from overfiring dur- ing start-up by allowing the drum pressure to rise rather than dumping the excess steam generated to the condenser. The superheater division valves also have an unique value during on-line operation for variable throttle pressure operation. Adequate drain lines (2 percent steam flow at 500 psi) at the HP turbine inlet should be provided to warm the line to the HP turbine. Adequate drains are also necessary at the IP turbine inlet. A small line from the superheater bypass line to the reheater inlet may be provided for warming steam flow from the reheater to the IP turbine before rolling the turbine. Variable Pressure Operating Modes The simplest arrangement used on many older units that are cycled on and off daily is to roll and synchronize the turbine at low pressure. After the initial load is applied with the turbine valves in manual, the boiler is ramped up to full pressure. The turbine valves are then released to automatic or manual load control and the boiler is released to a conventional boiler following system. Such a system provides no load control during ramping. The principles of an integrated boiler-turbine- generator control system to operate variable pressure with load control are shown in Figure 24. For such a hybrid system, the turbine valves 21 ™ Pressure program Pressure limits for load response Pressure Start 4+ ——— Integrated control — —_—- up |.~______ Modulate turbine valves — —+ Variable pressure —-}- Modulate We = — Load —— = Fast Slow Figure 24 Integrated control modulate to control load until the under-or-over- firing can restore throttle pressure to the set point. Figure 25 shows the results of operating with such a system at a load change rate of percent/minute. 3.3 100 Turbine valve position Steam flow % full load values Attemperator flow Superheater Reheater Z 5 10 15 20 Time, minutes 25 Figure 25 Variable pressure load change (3.3%/min.) The B&W drum boiler bypass and start-up system previously described (Figure 23) provides another dimension to variable pressure operation. The superheater division valves between the primary and secondary superheater are similar in location, but of larger capacity than those pro- vided on all North American B&W once-through boilers. Drum or once-through boilers with super- heater division valves can be operated in a dual pressure mode. By throttling through the super- heater division valves, the secondary superheater and turbine inlet piping can be operated by vari- able pressure over a wide range. This mode of pressure control is now being used on many boilers and is expected to be used extensively in the future for both drum and once-through boilers. Figure 26 describes three types of throttle 2800 a Drum press. SHDV Drum press. CTP Throttle press. CTP Drum press. Pressure, psi Variable throttle press. 800 VDP & SHDV 400 0 20 40 60 80 100 Load, % Figure 26 Modes of pressure control pressure control, called boiler operating modes. The constant throttle pressure (CTP) is well understood. The pure variable pressure VDP (variable drum pressure) with the turbine valves at or near the wide-open position is also a recognized mode. The dual pressure or superheater division valve mode (SHDV), however, is not as familiar. 22 Division valves allow optimum temperature con- ditions for both the boiler and turbine (Figure 27). For optimum boiler temperature control as well as load response, the drum pressure should be maintained at nearly full pressure. For uniform Throttle temperature 100% turbine valve position 800 worn Impulse chamber temp. - --" -- Temperature, F a x Ss 8s go 8 Saturation temperatures Py S 300 0 20 40 60 80 100 Load, % Figure 27 Turbine and boiler temperatures control of turbine metal temperatures, the turbine valves should be maintained near to the wide-open position. Each operating mode described has a critical boiler or turbine temperature. For constant throttle pressure operation, especially with sequenced valve operation, the critical temperature is at the turbine impulse chamber. For variable drum pressure, the limit is the change in satura- tion temperature of the drum and boiler. For the superheater division valve operation, the critical temperature is the steam temperature at the secon- dary superheater inlet header. Figure 28 compares the rate of temperatures change limits for each of the three critical temperatures. The critical temperature limit for superheater 1600 1400 Sec. SH inlet steam 1200 using Division Valve (501) 2 vA 3 Be 5 = 1000 aw 52 = 3 800 Turbine rotor metal of . o? Constant Pressure Operation 2 © <x 600 4 400 Sat. temperature Variable Pressure J Operation 200 0 100 200 300 400 500 600 Temperature change division valve mode, the requirements are com- parable to those for constant throttle pressure and are significantly less than the pure variable pressure. Figure 29 shows the effect of turbine admission and of boiler operating modes on cycle heat rate for a typical 2400 psi 1000/1000°F turbine. Figure 28 Critical temperature limits division valves is well above the level for operating transients. The result is no critical temperature limit within the boiler or turbine when operating in the dual pressure mode. Another advantage of the superheater division valve mode of control is reduced need for under- and-overfiring during load changes when compared to pure variable pressure operation. Operating data shown in Table 11 compares the maximum and integrated values of underfiring and overfiring required for the three modes. Table 11 Operating data for variable pressure operating modes Load Maximum Decrease Rate Underfiring Area MW MW/min. % % Min. Variable Pressure 210 12 11% 131 Constant Pressure 200 9 7% 24 501 Valve 100 12 4% 22 Load Maximum Increase Rate Overfiring Area MW MW/min. % % Min. Variable Pressure 250 9 18% 247 Constant Pressure 225 9 8% 118 501 Valve 100 10 5% 28 These values can be related directly to both the limits for rate of load change and the difficulties to be encountered in steam temperature control. This tabulation indicates that for the superheater 2400 psig 1000/1000 F turbine turbine-driven bfp Constant pressure (CTP) Single Admission Variable pressure (VDP) 2 + r 3 (Full Throttling) Variable throttle pressure s (SHDV) oy __ = °?) (sHov) ~ 2 £ 4 Admissions , = 0 5 “ee Base line is locus of Valve Loop oy 4 yf “valve-best-points” 7 oo (partial arc admission) > ¢* (voP) 20 40 60 80 100 % VWO Load Figure 29 Effect of turbine admission and boiler operating modes on heat rate The loss of cycle efficiency at reduced loads is especially evident for a single valve machine operating at constant throttle pressure. The super- heater division valve mode operating over the en- tire load range shows a significant improvement. Neither arrangement is as efficient as a sequence valve system for four admissions operating at con- stant pressure. The most efficient arrangement is to operate in a hybrid mode with variable drum pressure over 60 to 75 percent of the load range and sequence valve operation at full pressure above that point as described in Figure 24. Air Pollution Control Equipment Flue gases leaving the airheater of the PC-fired lignite unit should be treated with a combination of electrostatic precipitators for particulate control and a wet lime or limestone flue gas desulfuriza- tion (FGD) system for SO2 removal. Electrostatic Precipitator Flyash obtained from the burning of high sulfur, high ash content Teruel lignite results in a low resistivity ash that is easily collected with an electrostatic precipitator located after the air- heater. The main problems to be considered in designing this system have to do with the high dust loading leaving the airheater and its effect on inlet flue and precipitator component erosion. Con- tinuous ash removal should be provided for the first field hoppers since up to 75% of the total ash will be removed from the hoppers of the first field. The high SO2 and SO3 content of the flue gas re- quires fabricating flues, casing and hoppers from Corten A material and utilizing a double layer of insulation to reduce corrosion that will occur in low temperature areas due to the low acid dew point of the flue gas. Very high collective efficiencies, greater than 99.9%, can be achieved in the precipitator with a moderately sized rigid discharge frame design with 300 mm lane spacing. A three-field precipitator with an SCA of about 60m?/am*/sec would be re- quired to obtain the 99.9% efficiency. Flue Gas Desulfurization System The extremely high sulfur dioxide concentration of the flue gas leaving the precipitator is treated by passing the flue gas through a wet scrubber sys- tem that utilizes either lime or limestone as the reactant. The choice between the two will depend on the availability and delivered cost of each. We have assumed that high calcium limestone would be selected in the Teruel area. This system is shown in Figure 30. For maximum SO: removal, a system consisting of three 33% capacity absorbers, each similar to the one shown in Figure 31, would treat all of the flue gas leaving the precipitator and remove 90% of the sulfur dioxide. Flue gas is treated by pass- ing it countercurrent through multiple spray levels of recycled limestone slurry. Mist eliminators <> Stack Flue Gas From ~C} Limestone Scrubber Precipitator ~ Module Silo Fan Oxidation Tank Classifier Spent \ c W Slurry Tank — Thickener | { oH Ball Mill | > o To Scrubber 4 } Recycle y H—6) Slurry Storage Clarified Recycle | Oo Sludge Stabilization t+————> To Ball Mill System Air Compressor To Moisture Separator [> Wash System Blend Station | Fresh Water Figure 30 SO> scrubber flow diagram 24 ¢ Absorber Recycle Pump ¢ Mixers Figure 31 FGD absorber remove entrained slurry from the flue gas in the top section of the absorber from where the cleaned gases pass to the stack. Typical reactant and aux- iliary power values for this system are given in Table 12. Waste Disposal - Stabilization System The reacted limestone-sulfur dioxide reaction products consist of about 60% calcium sulfite (CaSO3°2H20), 26% calcium sulfate (CaSO4°2H20) and 14% inerts and unreacted limestone. Effluent from the absorbers is first pumped to a thickener or hydroclone system which concentrates the re- action products from 15% suspended solids to 35% solids. Final dewatering then takes place in 25 Table 12 Limestone FGD System Reactant - and Auxiliary Power 350 MW lignite unit 1,750,000 ACFM @ 300F (826 AM3/SEC @ 149C) 4800 Btu/Ib 6.7% S SO. removed, Ib/hr 96,200 Limestone, 90% CaCO3, Ib/hr 183,400 System pressure loss, inwg 8.0 Auxiliary power (ID fan not included) Installed, hp 10,840 Operating, kw 7,200 vacuum filters which dewater the effluent to 55% cake solids. Dewatered cake is then mixed with fly- ash in pug mill mixers from where it is conveyed to a stack-out pile for final disposal, either back at the mine or in a land disposal area. Recovered water from the dewatering processes is pumped back to the FGD system for use in the limestone milling system and as make-up to the absorbers. These systems are shown in Figure 32. If local or other authorities require a fully oxi- dized effluent from the FGD system, then this can be accomplished by the addition of an atmospheric oxidation system to the absorbers. Complete oxida- tion is achieved by directing a small portion, about 5% of the recycle slurry leaving the absorber, to a separate oxidation tank consisting of a mixer and air sparge ring in the tank which promotes mixing of compressed air with the unoxidized slurry. This sytem can achieve over 99% oxidation of the scrubber effluent resulting in a commercial gypsum product. FGDS Thickener| OS Lime Truck Lime Storage Silo Lime Screw Conveyer [_] Rotary Feed Valve Fly Ash Screw Conveyer Pug Mill Pug Mill Mixer Mixer Thickener Collection To Thickener Trucked To Land Disposal Area Figure 32 Secondary dewatering and waste stabilization Summary Utility pulverized coal-fired boilers can be designed to efficiently burn lignite, with high availability, while generating low levels of NOx, SOg, and par- ticulate emissions. Proper design of the systems and components requires technology specifically developed to accommodate the properties of lignite. The high quantity of inerts intrinsic to lignite pose several challenges to the boiler designer. The fuel handling system must be designed to overcome plugging tendencies of the coal due to surface moisture. Pulverizer capacity determination is crucial due to the high fuel throughput requirements. However, conventional methods do not adequately compensate for dif- ferences in grindability between the softer portions of the coal and hard to grind minerals which preferentially collect in the grinding zone. Small scale pulverizer grinding tests most accurately correlate grindability and coal abrasiveness to full scale mill performance. Hazards of mill fires or explosions require the use of vertical spindle mills equipped with systems to inert and clear the mill when warranted. The B&W Enhanced Ignition burner has demonstrated its ability to burn low grade lignites over a wide range of conditions, with little dependence on furnace conditions. The flow charac- teristics of the burner compensate for the low heat release and high quantity of inerts characteristic of the low-grade lignites. High iron and sulfur lignites, which are abundant in Spain, require precise control of secondary air to maintain oxi- dizing conditions throughout the furnace to minimize corrosion and slagging. Compartmented windboxes regulate secondary air on a mill basis, and individual burner air flow measurement is recommended to fine tune air distribution. Coal ash corrosion of the high temperature superheater and reheater surface can be prevented by tube metal selection and surface arrangement. Ash deposition tendencies depend on ash chem- istry, mineral formation, particle size, thermal environment, flow patterns, and conditions of the impacted surface. Slagging and fouling indices have been developed empirically to relate ash pro- perties to field experience. Computer generated solutions of numerical models are becoming the most accurate means of predicting ash deposition and means of control. A superheater bypass system for drum boilers provides a wide range of controlled throttle pressures and temperatures for rolling and loading 26 the turbine. Superheater division valves permit dual pressure operation during start-ups and cycl- ing, and consequently improve temperature control while reducing auxiliary fuel consumption. The division valves also provide for variable throttle pressure during on-line operation to provide opti- mum temperatures for both the boiler and turbine. An electrostatic precipitator downstream of the air heater can readily exceed 99.9% collection effi- ciency with Teruel lignite, but the high ash loading and SO2/SO3 content of the flue gas require measures to reduce erosion and corrosion. The extremely high SO» concentration can be reduced by 90% by a wet scrubber using either lime or limestone as the reactant. The solid waste can be mixed with flyash and disposed of, or oxidized to produce a commercial gypsum product. References 1. Steam/Its Generation and Use, Babcock & Wilcox, 38th Edition, 1972. 2. H. H. Lowry, Chemistry of Coal Utilization - Volume 1, Copyright 1945 New York - John Wiley & Sons, Inc. 3. A. L. Bennett, ‘“‘Coal Flow and Design Con- siderations Bunker to Pulverizer,”’ Stock Equipment Company, July 1979. 4. “B&W Guide Specifications MPS Pulverizer - Recommended Procedure for Pulverizer Isola- tion, Inerting, and Clearing’”’ August, 1986. 5. R. R. Piepho, D. R. Dougan, ‘‘Grindability Measurements on Low Rank Coals,”’ Coal Technology Conference, November, 1981. “Coal Pulverizer Materials Development: Effect of Coal Characteristics on Wear Perfor- mance and Reliability,’’ EPRI Project 1883-2, October, 1985. 7. D. R. Dougan, ‘Coal Tests,’’ B&W In-house Report, July 22, 1986. 8. C. F. Eckhart, ‘‘Enhanced Ignition Dual Register Coal Burner Tests,’ B&W In-house Report, March 22, 1985. A. D. LaRue, M. A. Acree, P. L. Cioffi, “Utility Steam Generator NOx Control Update - 1985,’’ 1985 Joint Symposium on Stationary NOx Control, May 6-9, 1985. Vecci, S. J., Wagoner, C. L. and Olson, G. B. - “Fuel and Ash Characterization and Its Effect on the Design of Industrial Boiler’ - American Power Conference, Chicago, Illinois, April 24-26, 1978 (Babcock & Wilcox BR-1117). 10. 11. 12. 13. 14, 15. Heil, T. C. and Durrant, O. W. - ‘Designing Boilers for Western Coal,” Joint Power Generation Conference, Dallas, Texas, September 10-13, 1978 (Babcock & Wilcox BR-1121R). Attig, R. C. and Duzy, A. F. - ‘‘Coal Ash Deposition Studies and Application to Boiler Design’’ - American Power Conference, April 22, 1969. Durrant, O. W. - “Design, Operation, Control and Modeling of Pulverized Coal Fired Boiler” - Boiler-Turbine Modeling and Control Seminar, University of New South Wales, Sydney, Australia, February 14-18, 1977 (Bab- cock & Wilcox BR-1082). Wagoner, C. L. - “Generic Variables Affecting Deposition and Performance” - Sixth Annual Workshop on Coal-Liquid and Alternate Fuels Technology, Halifax, Nova Scotia, September 29 - October 3, 1986 (Babcock & Wilcox). Wagoner, C. L., and Wessel, R. A. - “Initiating the Change from Empiricism to Generic Engineering, Part 2: Initial Results” - Paper to 27 16. 17. 18. 19. be presented at the 1986 Joint Power Genera- tion Conference, Portland, Oregon (October 19-23, 1976 (Babcock & Wilcox). Wagoner, C. L., ‘Influence of Fireside Deposits on Performance of Heat Exchangers” - American Institute of Chemical Engineers, Houston, Texas, November 24-28, 1985 (Bab- cock & Wilcox). M. F. R. Mulcahy, J. Boow, and P. R. C. Goard, ‘‘Fireside Deposits and Their Effect on Heat Transfer in Pulverized Fuel Fired Boilers, Part 1: The Radiant Emittance and Effective Thermal Conductance of the Deposits; Part 2: The Effect of the Deposit on Heat Transfer From the Combustion Chamber Considered as a Continuous Well-Stirred Reactor,” Jour. Inst. Fuel, Part 1, p. 385, Part 2, p. 394; 1966. T. F. Wall, A. Lowe, L. J. Wibberly, and M. C. Steware, ‘‘Mineral in Coal and the Ther- mal Performance of Large Boilers,’’ Progress in Energy and Combustion Science, 5:1; 1979. W. J. Peet, “‘Bypass Systems,” Babcock & Wilcox Canada, Cambridge, Ontario, Canada (BR-1264). | Technical Paper Continuously monitoring furnace temperatures in refuse boilers using acoustic pyrometry P. S. Larsen, P.E. Refuse Marketing Domestic Fossil Operations Babcock & Wilcox Barberton, OH Presented to Waste Technology ‘86 Chicago, IL October 20-22, 1986 Babcock & Wilcox BR-1292 a McDermott company Continuously monitoring furnace temperatures in refuse boilers using acoustic pyrometry P.S. Larsen, P.E. Refuse Marketing Domestic Fossil Operations Babcock & Wilcox Barberton, OH Presented to Waste Technology ‘86 Chicago, IL October 20-22, 1986 PGTP-86-38 Abstract Twenty years ago there was very little concern with flue gas emissions, other than those visible to the eye. Eventually, mechanical dust collectors were replaced by highly efficient precipitators and bag- houses. Now, sulfuric and hydrochloric acids are controlled by scrubbers, and NOx is controlled by staged combustion. Recently, the fluid bed boiler has become popular for control of these unwanted pol- lutants. Thus, boiler design philosophies have evolved from ignoring pollutants, to trapping them in the back end of the flue gas train, to now controlling the combustion process and preventing their initial formation. In many large metropolitan areas of this country, the accumulation of the nation’s refuse is at a crisis level. Landfills are closing and the difficulty of siting new ones has spurred communities to look at alternative methods of disposal. A very logical alternative used over the past few years has been to incinerate the refuse and transform the valuable Btu’s into usable energy in the form of steam and/or electricity. The combustion of refuse, however, has led to a new series of pollution problems in the form of toxic gases, hydrocarbons and heavy metals. Many of these detrimental pollutants exist as a direct result of incomplete combustion and can be greatly minimized with a properly designed combustion control system. Introduction Over the years, there has been an unfulfilled need to continuously monitor the high temperature zones of the furnace and superheater of steam- producing power generation equipment. During boiler startup, flue gas temperatures entering the superheater must be continuously monitored to prevent the overheating of the superheater tubes until steam flow has been established. At normal operating loads, flue gas temperatures can be monitored at various points within the furnace and across the superheater, economizer, and boiler tube banks to provide indications of surface cleanliness. A decrease in flue gas temperature differential, for example across a tube bank, would indicate decreased heat transfer or an increase in surface fouling. Slagging in the furnace and fouling of heat transfer surface in the tube bank zones causes decreased cycle efficiency by increasing the flue gas temperature leaving the boiler. Today’s poorer quality fuels such as municipal solid waste (MSW), lignite, sub-bituminous and culm coals, petroleum coke, and others have a higher tendency toward slagging and fouling because of their high ash contents. The combustion efficiency of refuse-fired boilers has been determined to be an important parame- ter in preventing or minimizing the formation of certain unwanted combustion products. With the continuously changing constituents of refuse fuels, controlling this combustion process is diffi- cult without monitoring the results of the process, the furnace temperature. Knowing furnace temperature, the operator can take corrective action, such as: e Firing auxiliary burners e Changing excess air e Redistributing air flow e Sootblowing Recently, construction and operating permits for several refuse-to-energy plants have been con- tingent upon being able to continuously monitor the combustion zone temperature. On one recent project, the San Diego Air Pollution Control Board required a gas retention time of two seconds at a minimum of 1800 F any time refuse was being fired. They further required that this temperature be continuously monitored and inter- faced with their computer to determine com- pliance with this condition at all times. The intent of the requirement was to achieve a degree of cer- tainty that harmful and unwanted combustion products would be destroyed as a result of time and temperature. These recent requirements accelerated Babcock & Wilcox’s efforts toward development of a reliable, cost-effective, continu- ous high temperature monitor. This instrument was named Pyrosonic 2000™. Alternative methods High temperatures, together with the corrosive and erosive nature of refuse combustion gases, make it nearly impossible to continuously mea- sure furnace temperatures with conventional thermocouples any place but beyond the economizer. Water-cooled, high velocity thermocouples (HVT) are currently used to measure gas tempera- tures above 1000 F. They are an accurate, well- accepted method for measuring spot tempera- tures, but become impractical for measuring temperatures in furnaces wider than 40 feet since the maximum practical working length has been limited to 20 feet. For this 20-foot penetration, a 27-foot probe is needed and will require at least 30 feet of clearance outside the boiler access port. Many times this accessibility is not available. To measure the average temperature of a flue gas stream with HVT’s, a series of temperature mea- surements is needed. Traversing a furnace to measure this temperature grid is not only time and manpower consuming, but results in inaccu- racies since it may take hours to complete. Gas stream temperatures are likely to change over this required time period due to fuel changes, load swings, furnace slagging, etc. HVT probes are not only labor intensive, but can be dangerous to per- sonnel because of burns from hot probes, flue gas escaping the access port, and the possibility of slag falling against the probe. HVT probes are suitable only for periodic temperature measurements. Optical pyrometers have been used for measur- ing spot furnace gas temperature but are not effective in zones below 1600 F. This type of mea- surement requires sensitivity to a number of opti- cal wave lengths, and also requires knowledge of the thermal gradients through the gas medium to the spot being measured. In addition, glowing or burning flyash particles entrained in the furnace gas stream introduce considerable uncertainty in the optical temperature readings. These burning flyash particles are normal for high ash fuels such as refuse. Although acoustic pyrometry is not a new con- cept, it is one method of temperature detection that has not been extensively explored. It was first suggested by A. M. Mayer! in 1873 and then in amore recent study by Green and Woodham? where they acoustically tested furnace tempera- tures in an operating boiler, but only compared them with a calculated temperature method. Theory of operation Acoustic waves propagate at a sonic velocity through a given gas medium, primarily as a func- tion of its absolute temperature, and to a lesser extent, as a function of the specific heat ratio, molecular weight, and a universal gas constant. By measuring the flight time of an acoustic wave and dividing it by the distance traveled, the speed of sound can be determined by the following relationships: C =(kRT/M)0-5 (1) where: C =Speed of sound - ft/sec k =Specific heat ratio- dimensionless R =Universal gas constant = 1545 ft-lbf Ibm-mole-R T =Absolute temperature - R (Rankine) M =Molecular weight - lbm lbm-mole which can be stated as: C =BT05 (2) where: B =(kR/M)9-5 (3) The speed of sound can be determined by divid- ing the distance over which acoustic waves travel by the time of flight or the time it takes to get from the transmitter to the receiver. Expressing the above equations in terms of Fahrenheit temperature, we have: F =_d2. 460 (4) (tB)? where: B =Acoustic constant - ft/sec - R9-9 F =Temperature - F d =Distance - ft t =Time - sec Figure 1 Transmitter/receiver attached to boiler observation port. The specific heat ratio (k) is the only tempera- ture dependent variable in equation (1). Specific heat ratios for a typical flue gas with 12.0 percent carbon dioxide, 6.0 percent oxygen and 88.0 per- cent nitrogen by dry volume with 5.0 percent moisture content by weight (lb moisture/lb dry gas) vary from 1.37 at 70 F to 1.26 at 3000 F. By using an average acoustic constant (B) for the above temperature range, the maximum expected error is only 0.5 percent®. System description The design criteria established by B&W in the development of an acoustic pyrometer system were: e its survival in a hostile operating environ- ment from elements both inside and outside the boiler reliability accuracy cost-effectiveness ease of installation applicability to field programming applicability to a variety of fuels and boilers The approach taken to measure furnace temperatures, as shown in Figure 1, is to mount both the transmitter and the receiver in place of the observation doors wherever possible. This allows for easy installation and removal, and does not require boiler modifications if the obser- vation port locations are acceptable. With this method of installation, setup can be conducted in some cases while the boiler is operating and in less than one day. Tests conducted and described below have made use of existing observation ports or openings for future sootblowers. New boilers Control and Display Assembly Analog Output 110 VAC 60 Hz 0-10 VDC Acoustic 4-20 ma Processor —_| | Display Output IEEE 422 Terminal Output Power = +— Receiver Amplifier Amplifier bi NEMA Enclosure Control | Signal Transmitter Receiver Plant Air bo Plant Air Purge Purge Control Control Valve Valve Waveguide/ Horns Figure 2 System configuration?. can be designed with ports in their optimum locations. As shown in Figure 2, the transmitter, an elec- tromagnetic transducer, sends out an acoustic pulse. The receiver picks up the transmitted signal and delivers it to the acoustic processor. Through a cross-correlation technique, the received signal is sampled at 30,000 samples/ second and compared to the transmitted signal. When a match is determined, the flight time and the quality of the readings are measured and stored. After a number of readings are made, an average is determined and a corresponding temperature is calculated. This averaging smooths out the instantaneous fluctuations of the dynamic temperatures existing within a furnace. Readings below a predetermined quality are not included in the average. Background noise is sampled periodically to direct the amount of transmitted power in order to receive a high qual- ity signal. A running average is provided that can be used for display, storage and/or combustion control purposes. As many as eight sets of trans- mitters/receivers can be used with each signal processor. It is possible, by using a grid of temper- atures monitored within the same plane, to com- pute average temperatures across the area of the furnace at one elevation. Field testing Field testing began in July 1985 on a 250 MW pulverized coal-fired utility boiler at Sierra Pacific’s North Valmy Station. Fourteen months later, B&W had demonstrated its acoustic temper- ature measurement device on six different boilers, with numerous tests performed on each. It was the intent to prove and refine the Pyrosonic 2000 in many different boilers, firing as wide a variety of difficult fuels as possible. At this point, the product was identified as commercial. During this shakedown period, the system was tested in three utility boilers, two process recovery boilers, and a refuse derived fuel (RDF) fired boiler. Special situations in which temperatures were measured were as follows: e at several locations within the furnace of a refuse-fired boiler e through narrow superheater cavities ¢ across a 70-foot-wide furnace e through an operating black liquor oscillator port in a recovery boiler ¢ across the black liquor bed in a recovery boiler e through a pressurized furnace For most of the Pyrosonic 2000 testing that was done, corresponding HVT traverse data was taken and calculated in terms of multiple high thermocouple (MHVT) temperatures in order to account for radiation effects. Testing results/system refinements Throughout the course of the above testing pro- gram, the Pyrosonic 2000 evolved into today’s commercial product. Some of the major problems encountered and the resulting solutions are de- scribed below. e Initial testing revealed that although a narrow band acoustic signal could be detected across a furnace, utilizing filtering to minimize the effect of background noise, a high-powered sound source would be required to achieve the high signal-to-noise ratios in order to pick up a rea- sonable number of processable signals. These high powered transmitters were found to be very expensive. As a result, the effort shifted toward utilization of a sound pulse sig- nal. The received sound pulse signal can then be cross-correlated or compared with the trans- mitted signal to accurately distinguish it from background noise. It was found that a signal could be detected even if the background noise was two times higher than the received acoustic pulse, and that the low power transmitter could still be used. A “qualifier” was also introduced into the signal processing that determined the quality of the signal received. Only measure- ments with high qualifiers are accepted. A sequential series of acceptable measurements are collected, averaged and displayed. The cross-correlation method requires a few more seconds for data processing because of the number of computations involved, but has proven to be the cost-effective solution without compromising accuracy or response. When initially comparing the acoustically mea- sured furnace temperatures with the HVT temperatures, larger deviations occurred than expected. These deviations were minimized when it was discovered that: 1) large tempera- ture gradients occurred near the furnace walls. These were found to be up to 550 F/foot within 1.5 feet of the walls. HVT measurements did not initially account for these lower temperature zones; 2) the temperatures within the horns themselves tend to be cooler and would bias the average acoustically generated temperature on the low side. (A self-compensating procedure is being developed to account for these cooler horn temperatures); 3) several of the furnaces tested were either too wide to reach across with HVT probes or only one side was accessible. This created an unknown error in the HVT method; 4) it was found that the temperatures within an operating furnace were very dynamic. Tempera- ture swings as large as 100 F were found in the acoustic data and as large as 200 F in the HVT data. Hundreds of acoustic temperature mea- surements are averaged for each set of HVT spot temperatures, and therefore, it is probable that error is introduced when the averaged values of the two methods are compared. Accounting for the above, we have found that temperatures routinely compare within one percent. When measuring temperatures through narrow cavities and along furnace tube walls, it has been found that secondary or reflected sound waves are generated that can be stronger than the original sound pulse. These secondary sound waves seem to be transmitter-position depen- dent. We should be able to account for these false signals through changes in the signal pro- cessing software. It was found that the transducers and receivers could be adequately protected from the conduc- tion of heat coming from the furnace walls by using an insulator that separates the transducer from the horn. The furnace radiation, however, required that the horn be turned at an angle and that a waveguide be added to extend the dis- tance. No form of air or water cooling was then required. The waveguide also provided the transducers and receivers with protection from the corrosive gases of refuse and process recov- ery boilers. e For temperature detection within boilers firing high ash fuels, an air puff system connected to the horn was required to not only protect the acoustic drivers from fine flyash particles, but to keep the waveguides from plugging. This air puff system has now been added as a standard part of the horn assembly. Conclusions e Testing of the acoustic pyrometer has shown that it is a practical and accurate, non-intrusive method for continuous temperature measure- ment in hostile environments. From the testing conducted, it is believed that the Pyrosonic 2000 methods discussed can be used to measure temperatures across boilers as wide as 100 feet. e When compared to currently used HVT methods, the Pyrosonic 2000 device can be used not only to continuously monitor furnace temperatures, but also as a control method for the combustion process. Control of the combus- tion process will likely lead to better control of the unwanted emissions, particularly when dealing with the constantly changing character- istics of MSW fuels. e Low power transducers can be utilized if sound pulse tones are used in conjunction with cross- correlation signal processing techniques. e It was found that the frequency within the audible range appeared to be optimum. The lower limit would be 500 Hz and an upper limit would be 2500-3000 Hz. Combustion and back- ground noise predominates below 500 Hz and attenuation losses set the upper limit. The higher end of this frequency range is best from the standpoint of signal detection, timing and temperature resolution. e Flyash, black liquor spray, and other entrained particles in the furnace do not seem to affect the use of acoustic waves for temperature detection. With the success we have seen using acoustic pyrometry techniques for quick and accurate fur- nace temperature measurements, it is not difficult to determine additional applications in industry where the Pyrosonic 2000 will prove to be a valu- able instrument. References 1. A. M. Mayer, (1873), Phil. Mag., pp. 45, 18. 2. S. G. Green and A. U. Woodhan,, “Rapid Fur- nace Temperature Distribution Measurement by Sonic Pyrometry”’, Central Electricity Generating Board, Marchwood Engineering Laboratories, Marchwood, Southampton, England, 1983. 3. S. P. Nuspl, et.al., “Acoustic Pyrometry Applied to Utility Boilers”, ASME/JPGC, Portland, Oregon, October, 1986. Technical Paper BR-1293 CWF firing of an industrial boiler on a commercial basis Richard V. Carlson John M. Wilkinson Coal-Water Fuel Business Unit Babcock & Wilcox Barberton, Ohio Robert A. Cartwright Nuclear Equipment Division Babcock & Wilcox Barberton, Ohio James J. Muckley Research & Development Division Babcock & Wilcox Alliance, Ohio Donald P. Malone Ashland Development, Inc. Subsidiary of Ashland Oil, Inc. Ashland, Kentucky Sponsored by South Point CWF Partnership, a partnership of The Babcock & Wilcox Company and Ashland Oil, Inc. Babcock & Wilcox a McDermott company CWF firing of an industrial boiler on a commercial basis Richard V. Carlson, Manager John M. Wilkinson, Sr. Design Engineer Coal-Water Fuel Business Unit Babcock & Wilcox Barberton, Ohio Robert A. Cartwright Nuclear Equipment Division Babcock & Wilcox Barberton, Ohio James J. Muckley Research & Development Division Babcock & Wilcox Alliance, Ohio Donald P. Malone Ashland Development, Inc. Subsidiary of Ashland Oil, Inc. Ashland, Kentucky Sponsored by PGTP-86-21 South Point CWF Partnership, a partnership of The Babcock & Wilcox Company and Ashland Oil, Inc. Abstract In late 1984, a 65,000 lb/hr heating boiler at B&W’s Barberton Works was converted to coal-water fuel (CWF) firing. Based on the experience gained during the 1984-85 heating season, the CWF supply sys- tem was fully automated to provide a high degree of reliability with minimal operator attention; and the CWF atomizer was redesigned to permit optimum performance with the available low-pressure (90-110 psig) steam. The boiler was fired with CWF on a commercial basis during the 1985-86 heating season, demonstrating an availability factor for the CWF firing system of 95.9 percent. One operator was used per shift as with natural gas firing, and a total of 810,000 gallons (i.e., 4150 tons) of CWF was delivered in rail tankers. The CWF was supplied by the South Point CWF Partnership, a partner- ship of The Babcock & Wilcox Company and Ashland Oil, Inc. This paper includes a summary of the retrofit design considerations and a description of the modi- fied boiler system. In addition, data are presented on the boiler performance and on the operation of the overall system. The experience from operating the industrial boiler over two heating seasons clearly demonstrates that the technology currently exists to permit routine, commercial operation of boilers with unsup- ported CWF firing. The primary key to successful operation is the installation of burners and fuel supply systems which are carefully designed to meet the handling requirements resulting from the unique rheology of CWFs. Introduction Babcock & Wilcox has successfully concluded a program initiated in 1980 to take a bench-scale technology for formulating a CWF and develop it into a commercially-acceptable product (1-4). Dur- ing this period, CWF preparation was scaled up to a 2 ton/hr continuous pilot plant and, finally, to a 20 ton/hr plant (5); CWF burners were scaled up from 5- to 20- to 40- and, finally, to 100-million Btu/hr (6-8). Moreover, B&W was a major partici- pant in the EPRI-sponsored CWF demonstration in an industrial boiler at a DuPont production plant in Memphis, Tennessee (9-11). By mid-1984, all of the elements were in place to support a commercial conversion of a boiler to CWF firing—the technological requirements had been developed and demonstrated on a commer- cial scale, and an adequate fuel supply was avail- able. In October 1984, the South Point CWF Part- nership contracted with B&W’s Nuclear Equipment Division (NED) to convert a 65,000 lb/hr saturated steam heating boiler in Barber- ton, Ohio, to CWF firing. Since this boiler is used only during the heating season, from approxi- mately mid-November to late-March, it was necessary to compromise on some of the boiler modifications in order to obtain some operating experience during the 1984-85 heating season. These results were reported at several recent con- ferences (12-15). Based on this experience, the remaining boiler modifications were completed in preparation for the 1985-86 heating season. These consisted pri- marily of: e Fully automating the CWF supply system to reduce required operator actions to those which are normal for an oil-fired boiler. e Installing CWF atomizers designed to provide the maximum achievable carbon conversion, given the limitations of the furnace due to water-cooled walls, with the low-pressure (90-110 psig) steam available. © Correcting uneven air distribution patterns to the two burners and to the mechanical dust col- lector so their design performance could be realized. This paper presents information on the CWF supplied, the retrofitted boiler, the CWF supply system, and the overall performance achieved. The results clearly demonstrate that the technol- ogy currently exists to permit routine, commercial operation of boilers with unsupported CWF firing. Coal-Water Fuel Supply In October 1983, B&W and Ashland Oil, Inc. formed a partnership to own and operate a CWF production plant at South Point, Ohio, for the purposes of: e Having an adequate supply of CWF to support full-scale demonstrations in various applications. e Generating technical and economic data that could be applied to the design and operation of future plants which were anticipated to be three- to-twentyfold larger in capacity. e Providing an economic supply of quality CWF for sale in the local marketplace to offset the costs of the facility. By late summer 1984, Ashland’s idle coal-oil mix- ture plant was converted to CWF production with several flow-sheet alternatives (5). Some features of the South Point CWF plant are: e Maximum design capacity of 20 tons coal per hour (i.e., about 180,000 tons or 850,000 barrels or 36 million gallons CWF per year). ¢ Coal grinding by wet and/or dry techniques. ¢ Coal receipts by rail or truck. ¢ CWF shipments by rail, truck, or river barge. © On-site storage capacity of 1.5 million gallons. CWF for shipment to the NED boiler in Barber- ton was produced with the plant configured in its maximum-output mode (Figure 1). The feed coal was a nominal 2:1 blend of Pond Creek and Pocahontas-3 seams with characteristics as shown in Table 1. The characteristics of the CWF are shown in Tables 2, 3, and 4. Shipment of CWF from South Point to Barber- ton was accomplished using a fleet of 15 rail tankers. These tankers have a weight capacity of about 17,000 gallons, are insulated and capable of being heated with steam. Each made three-to-four round trips to deliver the 810,000 gallons of CWF. A 250 gpm progressive cavity pump in a heated pump house was used to transfer CWF through a large-capacity basket strainer to the 15,000 gal- lon/day tank (Figure 2 is a photograph of a rail tanker in position next to the pump house). The CWF unloading system proved to be highly reliable--the pump seals required occasional greas- ing, the strainer was cleaned twice, and no freez- ing occurred in the pump or heat-traced, insulated piping. The uninsulated rail tanker discharge valves were routinely thawed using steam; and, Coal Bunkers W-Feeders Exhaust CWF Storage (1,500,000 gal.) Figure 1 South Point CWF plant. except for one tanker in which the CWF had cooled to below 35°F, no difficulties were encoun- tered. In this one case, the CWF viscosity was so high that unloading took several hours. NED Boiler Retrofit The retrofitted boiler is a 65,000 lb/hr Babcock & Wilcox designed FH-type unit built in 1948 to fire pulverized coal and operate at 125 psi saturated steam. The furnace is constructed of water-cooled tangential tubes on the side and rear walls, with the lower side walls forming a 55-degree sloped hopper bottom. The front wall has wide-spaced tubes with refractory backing. Two burners are located in the front wall. The furnace is 15-feet deep and 12-feet wide. Gas leaving the furnace is directed to a three- pass generating bank from where it exits into a two-pass tubular air heater. The gas leaving the air heater goes through a dust collector, ID fan and is finally discharged through a 6-foot- diameter, 61-foot-tall metal stack. Combustion air enters the forced draft fan from inside of the boiler building, passes over the air heater tubes, and enters the 4-foot-deep burner windbox from the upper right-hand corner. The boiler tube banks are cleaned by six Diamond G9B steam soot blowers, and the air heater surface is cleaned by a similar soot blower. There are no soot blow- ers in the furnace proper. In the early 1970’s the boiler was converted to natural gas firing, and all coal handling and dust collecting equipment was removed. When the fur- nace ash hopper was removed, the hopper bot- toms were sealed with plate and covered over with refractory brick. During the conversion to CWF firing, the furnace hopper seal plate and refrac- tory were removed, and a transition chute with a submerged chain deasher was installed (Figure 3). The latter serves as a water seal and provides for continuous removal of bottom ash. In order to obtain a permit to burn CWF from the Regional Environmental Protection Agency, an 80 percent- efficient mechanical dust collector was installed. For the 1985/86 heating season, splitter vanes were installed upstream and downstream of the Table 1 Typical Parent Coal Analyses (as received basis) Proximate Analysis, % Moisture 3.70 Volatile Matter 32.24 Fixed Carbon 57.56 Ash 6.50 Gross Heating Value, Btu per Ib 13,808 Ultimate Analysis, % Moisture 3.70 Carbon 76.45 Hydrogen 4.79 Nitrogen 1.46 Sulfur 0.59 Ash 6.50 Oxygen (Difference) 6.29 Chloride 0.22 Sulfur Forms, % as S Pyritic 0.12 Sulfate 0.00 Organic 0.47 Free Swelling Index 8.5 Grindability (ASTM D-409) 51 Ash Analysis (Spectrographic), % _ Silicon as SiO, 54.99 Aluminum as Al203 30.05 Iron as Fe203 5.43 Titanium as Ti02 1.59 Calcium as CaO 0.78 Magnesium as MgO 0.93 Sodium as Na20 0.68 Potassium as K20 2.57 Sulfur as S03 0.96 Phosphorus as P20s, 0.06 Ash Fusion Temperature, °F Reducing Atmosphere A(1.D.) 2700+ B(S.T., Sp) 2700+ C(S.T., HSp) 2700+ D(F.T., 1/16") 2700+ E (F.T., Flat) 2700+ Oxidizing Atmosphere A(I.D.) 2700+ B(S.T., Sp) 2700+ C(S.T., HSp) 2700+ D(F.T., 1/16") 2700+ E (F.T., Flat) 2700+ Pond Creek Seam Total Moisture, % 3.70 Equilibrium Moisture, % 2.03 Calculated Pocahontas for 70/30 Seam Blend 5.00 4.09 1.30 1.81 5.00 4.09 18.10 28.00 67.70 60.60 9.20 7.31 13,598 13,745 5.00 4.09 77.25 76.69 4.07 4.57 1.35 1.43 0.70 0.62 9.20 7.31 2.43 5.14 0.00 0.15 0.11 0.12 0.01 0.00 0.58 0.50 9.0 - 91 - 56.82 55.54 20.48 27.18 11.94 7.38 1.01 1.42 4.37 1.86 1.16 1.00 1.00 0.78 0.96 2.09 0.80 0.91 0.86 0.30 2450 - 2515 - 2535 - 2700+ - 2700+ - 2600 - 2698 - 2700+ : 2700+ - 2700+ - tubular air heater to correct gas and dust distribu- tions to the mechanical dust collector. Flyash that collected in the hoppers was emptied through rotary seal valves into metal collection containers which are manually changed by the operator when they become filled (Figure 4). Figures 5 and 6 are schematic side and plan views of the general arrangement of the retrofitted boiler. Performance of the deasher and mechanical dust collector was essentially trouble free-the only incidents were two occasions when pluggage of the dust collector hoppers required modest striking of the hopper anvils. Except for periodic greasing of the deasher drive system and the dust collector rotary valves, no other maintenance was required. Overall boiler reliability during the 1985-86 heating season was good with only two Table 2 Typical Analysis of CWF for NED Wet Dry Basis Basis Solids Content, wt. % 71.78 - Volatile Matter, wt. % 22.20 30.93 Ash Content, wt. % 5.34 7.44 Gross Heating Value, Btu/Ib 10,190 14,196 Sulfur, wt. % 0.59 0.82 Slurry Density, g/cc 1.2324 Viscosity, sec’ @ 25°C 900 pH 8.9 Grind, % 75 microns 76.2 Table 3 Typical CWF Proximate and Ultimate Analysis for NED As Received Dry Total Moisture, % 28.22 Proximate Analysis, % Moisture 28.22 - Volatile Matter 22.20 30.93 Fixed Carbon 44.24 61.63 Ash 5.34 7.44 Gross Heating Value, 10,190 14,196 Btu per Ib Ultimate Analysis, % Moisture 28.22 - Carbon 57.42 80.00 Hydrogen 3.52 4.90 Nitrogen 1.10 1.53 Sulfur 0.59 0.82 Ash 5.34 7.44 Oxygen (Difference) 3.81 5.31 unplanned outages--one due to an unexplained malfunction of the ID fan damper and one due to a low drum water level trip. Additionally, there were three brief boiler outages to repair the ID fan housing, a feedwater control valve, and two leaks in the plant steam piping. Outages of this type are to be expected for an old boiler, especially consid- ering that no preventive maintenance program was employed due to the non-critical need for this steam supply. There was no apparent increase in outages due to CWF firing. CWF Supply System CWF Supply to Burner Front A schematic of the CWF supply system is shown in Figure 7. The 15,000-gallon/day tank (Figure 8) was trouble-free. The mixer was operated for 20 minutes each day following fuel loading to Table 4 Typical Ash Analysis Fly CWF Ash Ash Analysis (I.C.P.), % Silicon as Si02 53.39 53.65 Aluminum as A1,03 26.16 26.55 Iron as Fe203 9.03 9.21 Titanium as TiO2 1.39 1.36 Calcium as CaO 1L7E 1.65 Mgnesium as MgO 0.86 0.82 Sodium as Na20 0.86 0.80 Potassium as KO 2.04 1.99 Sulfur as SO3 0.73 0.14 Phosphorus as P2305 0.24 0.21 Red Oxid. CWF Ash Fusion Temperature, °F A(I.D.) 2540 2670 B(S.T., Sp) 2660 2750+ C(S.T., HSp) 2680 2750+ D (F.T., 1/16") 2750 2750+ E (F.T., Flat) 2750+ 2750+ achieve temperature uniformity and to prevent the formation of a significant amount of crust at the CWF/air interface. The small quantities of crust from the liquid surface and scale from the tank side walls were adequately removed by a basket strainer that required cleaning every two- to-three days. There was not one atomizer hole pluggage incident. CWF was supplied to the burner with a progres- sive cavity pump equipped with a variable-speed drive. The flow of CWF was regulated from the Figure 2 CWF rail tanker being unloaded. Figure 3 Submerged chain deasher for removal of furnace bottom ash. Figure 4 Fly-ash collection system. burner front control station by changing the pump speed. During the 1985-86 heating season, the pump ran continuously for 2,160 hours with only one incident--the loosening of a set screw on the variable speed coupling mechanism. The only maintenance required was occasional greasing. The CWF manufactured by the South Point CWF Partnership can be heated to at least 240°F without degradation. Heating lowers the CWF viscosity, improves atomization and, thereby, substantially improves combustion quality and carbon utilization. For the NED boiler, a shell- and-tube heat exchanger was used to heat the CWF to about 220°F. A steam control valve per- mitted automatic and remote control of the CWF exit temperature to a set-point temperature. The heat exchanger operated without failure during the 1985-86 heating season. Figure 9 is a photo- graph showing the burner feed pump in the fore- ground and heat exchanger in the background. After the heat exchanger, CWF is piped to the burner front (Figure 10) where flow is directed “ - Hoppers ee [ ] Dust Collector Figure 5 NED boiler general arrangement (side view). Figure 6 NED boiler general arrangement (plan view). Burners Unloading Pump Atomizing Steam Progressive Cavity Pump metre! Gas Heat Exchanger Figure 7 CWF fuel supply system. into the burners or into a recirculation line to the day tank. The 1985-86 heating period had only one pipe pluggage incident. This occurred when a Network 90® control module failed during a boiler start-up, and several steam and CWF valves were locked into a position that resulted in slurry dry- ing out in a line at the burner front. This situation was resolved in a few hours. Instrumentation of CWF Supply System and Boiler A major improvement in the CWF supply system for the 1985-86 heating season was installation of instrumentation to detect all parameters asso- ciated with start-up, shutdown, routine operation, and indications of problems developing in the fuel Figure 8 15,000-gallon day tank. supply or boiler systems. A Bailey Controls Net- work 90® system used this instrumentation to automatically monitor and manage the overall system. Interlocks were provided to shut down operation upon indication of a potential malfunc- tion which could lead to an unsafe condition or to an extended shutdown for major repairs. With the Network 90®, boiler operation required only the use of the following eight pushbuttons: e Two for natural gas firing start and stop. ¢ Two for CWF firing start and stop. ¢ Two for heat exchanger start and stop. ¢ Two for heat exchanger valve positioning. After initial start-up, the only problems with the Network 90® system were failure of two controller modules and one master logic module which together led to 24 hours of downtime. Figure 11 is a photograph of the boiler control panel with the Network 90 ® system located in the foreground. Figure 9 CWF burner feed pump and heat exchanger. Previous combustion applications requiring detection of natural gas and CWF flames for safety purposes had experienced unnecessary shutdowns due to inadequate flame scanners. This was solved at NED by using one Bailey Con- trols type UV Flamon scanner per burner for gas light-off, main gas flame, combination gas/CWF firing, or CWF-only firing. During the 1985-86 heating season, only one boiler trip was asso- ciated with the flame scanner and that was caused by oil on a scanner lens after the separa- tors and filters in the scanner air line became overloaded. Figure 10 NED boiler burner front equipped for automated firing of CWF and/or natural gas CWF Burner and Atomizer The final components of the CWF supply system were two B&W CWF burners. These burners are offered commercially and have the demonstrated performance characteristics shown in Table 5 (8). The existing burner throats had to be enlarged from a 21-inch- to 24-inch-diameter to accommo- date the new burners. This did not require any boiler pressure part changes. For the 1985-86 heat- ing season, an adjustable shroud of 50 percent porosity was installed around the upper burner register to improve the combustion air distribu- tion. As a result, there was no observable differ- ence in the performance of the two burners except at very low loads. The CWF atomizers were a modified B&W T-jet design with tungsten carbide inserts to protect high wear areas. All previous BkW CWF atomiz- ers were designed for 150 psig steam or air. Neither atomizing medium is available at this high a pressure from plant systems at NED. For the duration of the 1984-85 heating season, an air compressor was obtained to provide 150 psig air for good atomization (12). Prior to this season, the CWF atomizer was redesigned to provide good atomization with the 100 psig steam that was readily available at NED. These new atomizers were installed and produced results equivalent to those obtained the previous year using 150 psig air. During operation of the NED boiler, no signif- icant wear of the tungsten carbide was measured. Operating Performance Achieved The boiler start-up sequence involved a normal gas light-off, consisting of a boiler purge, lighter ignition, and then main flame ignition. Once the main flame was established, drum pressure was raised until there was steam flow to the main Figure 11 NED boiler control panel. heating system header. During the warm-up and pressure raising period, the CWF supply system was activated with the CWF circulating up to the boiler front and back to the storage tank. Once steam flow had been established on natural gas firing and CWF was recirculating, the CWF burners were put into service by opening the burner valves and closing the recirculation valve. All of the CWF then flowed to the burners and established a minimum CWF input of approxi- mately 25 percent. CWF flow was increased to approximately 50 percent input while natural gas flow was cut back to the minimum stop. The CWF heat exchanger was put into service, and when heated slurry became available, the natural gas Table 5 Performance Characteristics of BaW's 100 Million Btu/Hr CWF Burner“) CWF Turndown 3:1 (steam atomization) 4:1 (air atomization) 12:1 (with CWF atomizer) 15% maximum 300°F minimum desired 5.4 inches (H20) at full load No. 6 Oil Turndown Excess Air Air Preheat Burner Pressure Drop Atomizing Medium Air or steam Atomizing Medium/Fuel Ratio 0.15 maximum Carbon Conversion Efficiency) 99.5% at full load ©) Data from EPRI-sponsored public demonstration. ®) It is important to note that carbon conversion is only partially dependent on burner design. The other major factors are combustion residence time and combustion zone temperature which depend upon furnace design characteristics. was shut off and the boiler was on 100 percent CWF firing. The boiler was operated continuously 24-hours- per-day, seven-days-per-week at loads up to 55,000 lbs/hr of steam with no natural gas firing during the 1985-86 heating season. This corresponded to a maximum firing rate of 10.5 gpm CWF. Higher loads were not possible due to the I.D. fan capac- ity limitations. The boiler availability was 87.3%, with the downtime being as follows: Start-up problems 2.4% Due to CWF firing system 4.1% Due to boiler maintenance 6.2% During both heating seasons, testing was con- ducted to obtain boiler operating performance data. Representative results obtained at 75 per- cent and 100 percent load are shown in Table 6 and Figures 12 and 13. Some of the points to note from the data are as follows: 1. The success of the atomizer redesign used in 1985-86 is evident from the reduction in CWF and atomizing medium pressures. 2. The success of correcting the air distribution within the windbox for 1985-86 is indicated by the reduced burner pressure drop. (This parameter is also influenced by the difference in boiler load for the data shown). 3. The infiltration of air into the furnace and boiler bank through leaks in this 38-year-old equipment results in a higher amount of excess air than might be otherwise expected at the air heater inlet. Most of the leaks were repaired prior to the 1985-86 heating season. 4. One of the effects of atomizer wear is evident from the high air/CWF ratio in 1984-85 when wear-protection inserts were not available at the time these data were taken. 5. One of the advantages of steam atomization is evident from the reduced NOx levels measured in 1985-86. 6. The boiler efficiency in 1984-85 was lower than predicted due to relatively high unburned car- bon loss which, in turn, was related to atomizer wear as noted above. In 1985-86, the boiler effi- ciency was lower than predicted primarily due to higher dry gas losses as a result of the air heater exit gas temperature being about 100°F higher than expected. Since the 1985-86 data shown were taken near the end of the heating season, it is presumed the high gas temperature resulted from dirty heat transfer surfaces. Table 6 Representative Boiler Performance Data 1985/86 Boiler Load, % 75.4 Steam Flow, Ib/hr 49,000 Drum Pressure, psig 135 Feedwater Temperature, °F 220 CWF Flow, Ib/hr 5,915 CWF Temperature, °F 180 CWF Pressure, psig 110 CWF Input, 10° Btu/hr 62.3 Natural Gas Input, 10° Btu/hr 0 Combustion Air Temperature, °F 475 Air Heater Exit Gas Temperature, °F 380 Burner Pressure Drop, inches H20 3.2 Air Heater Inlet Excess Air, % 23.1 Air Heater Inlet CO, ppm 130°) Air Heater Inlet NOx, ppm 390) Air Heater Inlet S02, ppm 580°) Atomizing Medium Pressure, psig 105) Atomizing Medium Temperature, °F 3702) Atomizing Medium Flow, Ib/hr 1,000) Atomizing Medium/CWF Ratio 0.18 SO, at Stack, Ib/10° Btu 11 NOx at Stack, Ib/10® Btu 0.53 Furnace Exit Gas Temperature, °F 2,080 Boiler Efficiency, % 81.7 Carbon Conversion, % 97.4 (1) — Converted to 3% 02 (2) — Steam Atomizing (3) — Air Atomizing (4) — Based on Heat Loss Method (5) — Full Load Prediction Predicted CWF 1984/85 Performance 98.9 100 64,300 65,000 119 125 220 225 8,080 7,520 23 = 170 - 79.3 72k 0 0 431 410 283 295 5.7 40 31.2 24.0 585 500 1873) 169°) - 1,968 1,128 0.26 0.15 0.95 1.08 0.80 - 1,880 1,800°) 819 84.2 95.3 98.2 7. The predicted carbon conversion is relatively low because of the water walls which result in the furnace being significantly cooler (approx- imately 400°F) than is typical, for example, in utility applications. The data in Figures 12 and 13 show the furnace is cold, especially adjacent to the wall surfaces. The lower-than-predicted carbon conversion value shown in Table 6 for 1984-85 is the result of atomizer wear and poor air distribution within the windbox. As the 1985-86 data show, these situations were large- ly corrected. . The furnace exit gas temperature in 1985-86 was significantly higher than in 1984-85 and than that predicted for a circular burner even though the 1985-86 data are at 75% load. This apparent anomaly is believed to be due to dif- fering flame shapes — burner adjustments were made in 1985-86 to obtain a longer, narrower flame than experienced in 1984-85 or than nor- mally obtained from a circular burner. This 10 explanation is supported by the data shown in Figures 12 and 13. Conclusions The results from the CWF-firing of the 65,000 lb/hr boiler in B&W’s Barberton Works over the past two heating seasons led to the following conclusions: 1. CWF can be prepared with a consistent quality on a continuous basis using commercial-sized equipment. . Properly formulated CWF can be readily trans- ported and stored in conventional truck or rail tankers with no significant degradation of quality over a period of at least one month. Cold weather poses no handling problems that cannot be resolved with conventional precautions. . The technology exists to design automated CWF supply systems having a high degree of reliability. Door Screen C0O00000000 Tubes Furnace Exit Corner Door Left Burners Corner Door Right Figure 12 Boiler temperature profile, 1984/85 season, 100% load, °F. 4. Commercial-scale CWF burners and atomizers are available for operation without support fuel except for light-off of the boiler. They meet pul- verized coal performance levels in a given fur- nace and have a commercially-acceptable life- time. Good atomization can be achieved either with compressed air or with steam at a pressure as low as 100 psig. 5. With the proper selection of parent coals to avoid any conventional slagging or fouling problems, boiler reliability is not degraded by CWF firing. Additionally, with automated CWF supply and ash-handling systems, CWF-fired boiler operation requires no more operating 11 personnel than an oil-fired boiler. In summary, the results from operating the NED boiler over two heating seasons clearly demon- strate that the technology currently exists to per- mit routine, commercial operation of boilers with unsupported CWF firing. Acknowledgment The authors gratefully acknowledge the efforts and interest shown by the boiler operators, W. Board, A. Maxim, R. Raber, and K. Saunders, in making this commercial operation on CWF firing an outstanding success. Corner Door Right n o a a -E c : o o 3 x 6 R < 5 3 oO o 3 © eo \ 8 © 3 x & = g q g 2 g o Qa € 2 ~ 2 S a ” a 2g 5 oe iz load, °F. 12 References . Sommer, T. M., Funk, J. E., “Development of a High-Solids, Coal-Water Mixture for Application as a Boiler Fuel,” ASME/IEEE Power Generation Conference, St. Louis, Missouri, October 4-8, 1981, BR-1203 (B&W paper). . Carlson, R. V., Dunlop, D. D., “Status of Co- AL Commercialization,” Proceedings Third Annual Coal-Liquid Mixtures Workshop, Halifax, Nova Scotia, October 1983, p. III-3. . Carlson, R. V., Dunlop, D. D., “Commercialization of Coal-Water Fuel,” Proceedings Coal-Liquid Mixtures: The Pathway to Commercialization, Tampa, Florida, February 7-8, 1985, p. 15. . Warchol, J. J., DeVault, R. F., Shiao, S., Vecci, S. J., “Effect of Coal Properties on Coal-Water Fuel Quality,” Proceedings Seventh International Symposium on Coal Slurry Fuels Preparation and Utilization, New Orleans, Louisiana, May 21-24, 1985, p. 186. . Carlson, R. V., Daley, R. D., Wilkinson, J. M., Emmons, P. C., “South Point Coal-Water Fuel Production Plant,” Proceedings Sixth Inter- national Symposium on Coal Slurry Combustion and Technology, Orlando, Florida, June 25-27, 1984, p. 933. . Eckhart, C. F., Lindstrom, G. D., Farthing, G. A., “Design of a Coal-Water Fuel (CWF) Burner for Low Air-Side Pressure Drop,” Proceedings Sixth International Symposium on Coal Slurry Combustion and Technology, Orlando, Florida, June 25-27, 1985, p. 1012. . Winters, P. J., Olen, K. R., Bailey, R. T., “The Effect of Fuel Formulation on the Atomization Characteristics of Coal-Water Mixtures,” Proceedings Seventh International Symposium on Coal Slurry Fuels Preparation and Utilization, New Orleans, Louisiana, May 21-24, 1985, p. 430. . Warchol, J. J., Markert, D. H., Farthing, G. A., “Commercial Coal-Water-Fuel Burner 13 10. 11. 12. 13. 14. 15. Demonstration,” Proceedings Eighth International Symposium on Coal Slurry Fuels Preparation and Utilization, Orlando, Florida, May 27-30, 1986, p. 706. . Perkins, R. P., Manfred, R. K., Taylor, B. E., “EPRI Industrial Coal-Water Slurry Demonstration,” Proceedings Sixth International Symposium on Coal Slurry Combustion and Technology, Orlando, Florida, June 25-27, 1984, p. 481. Perkins, R. P., Houghtaling, R. E., Manfred, R. K., Carlson, R. V., “DuPont Coal-Water Slurry Tests,” ASME/IEEE Joint Power Generation Conference, October 1984. BR-1261 (B&W paper). Furman, R. C., “EPRI Coal-Water Slurry Demonstration Using Co-AL,” Proceedings Sixth International Symposium on Coal Slurry Combustion and Technology, Orlando, Florida, June 25-27, 1984, p. 548. Batyko, R. J., Cartwright, R. A., Schueler, P. H., Zahirsky, R. W., “Coal-Water Fuel (CWF) Firing in an Industrial Size Boiler,” Proceedings Seventh International Symposium on Coal Slurry Fuels Preparation and Utilization, New Orleans, Louisiana, May 21-24, 1985, p. 733. Zagrodnik, A. A., Schueler, P. H., Markert, D. H., Grams, R. A., “The Economic Incentive for Conversion of an Industrial Boiler to Coal- Water Fuels,” Proceedings Seventh International Symposium on Coal Slurry Fuels Preparation and Utilization, New Orleans, Louisiana, May 21-24, 1985, p. 1109. Zahirsky, R. W., “Coal Water Conversion Consideration,” Proceedings Fifth International Workshop on Coal-Liquid Fuels Technology, Halifax, Nova Scotia, October 15- 18, 1985, p. 116. Carlson, R. V., Zagrodnik, A. A., Batyko, R. J., “Considerations for Converting an Industrial Boiler to Coal-Water Fuel Firing,” ASHRAE Transactions 1986, Vol. 92, Part 1, p. SF-86-14, No. 2. Technical Paper From Hamilton to Palm Beach: the evolution of dedicated RDF plants D. R. Gibbs Manager, Refuse Marketing Power Generation Group Babcock & Wilcox Barberton, Ohio L. A. Kreidler RDF Marketing Manager Refuse Marketing Power Generation Group Babcock & Wilcox Barberton, Ohio Presented to Processed Fuels & Materials Recovery from Municipal Solid Waste Symposium Sponsored by Resource Recovery Report in association with Gershman, Brickner & Bratton, Inc. Washington, D.C. December 1-2, 1986 Babcock & Wilcox BR-1294 a McDermott company From Hamilton to Palm Beach: the evolution of dedicated RDF plants D. R. Gibbs Manager, Refuse Marketing Power Generation Group Babcock & Wilcox Barberton, Ohio L. A. Kreidler RDF Marketing Manager Refuse Marketing Power Generation Group Babcock & Wilcox Barberton, Ohio Presented to PGTP-86-46 Processed Fuels & Materials Recovery from Municipal Solid Waste Symposium Sponsored by Resource Recovery Report in association with Gershman, Brickner & Bratton, Inc. Washington, D.C. December 1-2, 1986 Overview Refuse Derived Fuel (RDF) technology was devel- oped in the United States as an alternative to the mass burning method of refuse combustion com- mon throughout Europe. RDF firing is not only a more efficient means of converting municipal solid waste (MSW) into useable energy, steam and/or electricity, but RDF is also very compati- ble with the current trend to conserve resources via recycling. This paper reviews the evolution of refuse-to-energy facilities that were designed as dedicated RDF facilities. The first generation of these dedicated RDF facilities (such plants as Akron, Albany, Hamil- ton, Niagara Falls) used as their design basis: a) previous work done at facilities that only pro- cessed refuse for landfill with no combustion, b) units that were designed to burn coal or wood but which burned RDF on a test basis only, c) units that burned RDF as a supplementary fuel, and d) laboratory work. The first generation plants provided extensive experience to the refuse industry about the processing, handling and stor- age characteristics of RDF, as well as the inter- face requirements between the fuel feed system and the boiler. The second generation of dedicated RDF facili- ties were those that were designed initially with state-of-the-art fuel feed systems and those ear- lier facilities which underwent major modification to the fuel preparation, handling and feeding sys- tem (such plants as Akron, Columbus, Lawrence). With the ability to consistently process and feed RDF, boiler operating characteristics unique to RDF combustion became evident, and areas for boiler design improvements were recognized. The third generation of dedicated RDF facilities are essentially today’s state-of-the-art design. These facilities are based not only on the lessons learned at the first generation plants, but also on normal improvements that come only with daily operating experience. Webster defines evolution as a process of change in a certain direction. The direction that RDF facilities have taken is towards simplicity, both in the fuel processing logic and in the equipment used to process and handle the fuel. RDF preparation systems Introduction RDF is the combustible or organic fraction of MSW which has been prepared for use as fuel in a boiler, or other energy recovery system, by a mechanical processing method. Included with RDF is an inorganic fraction or ash, which is a variable percentage of that which was in the incoming MSW, depending upon the processing system design used. The RDF preparation system is a subset of the total plant material handling and processing sys- tem which transports MSW from the refuse truck tipping area through the ash, residue and mate- rial recovery loadout points for removal from the plant. The RDF preparation system employs con- veyance, size separation, shredding, material re- covery, storage and, in some cases, density sepa- ration to produce fuel of a desired characteristic which must be compatible with the systems using the RDF, such as the boiler distribution, fuel feed- ers, boiler grates and ash removal systems. The need for an integrated system, from MSW inlet to combustion stack outlet, has been realized after years of experience at these early design RDF plants. The refuse boiler, the heart of the entire refuse facility, functions properly because the other components of the plant now function properly. History Although mass burning of refuse has a long European operating history, RDF does not share a similar European heritage. RDF is largely a USA developed technology driven by goals of providing more efficient, more load responsive and more economical boiler facilities, coupled with a capability to recycle recovered materials. RDF has been burned in various ways since the late 1960's, both alone and in combination with other fuels. First generation plants adapted pro- cessing equipment from other industries and many lessons were learned. However, RDF has now come of age. The learning process typical of new technology has been met, and state-of-the-art RDF refuse-to-energy facilities are being designed and built. These have benefited from the first and second generation plants, which have provided the technological experience necessary to develop today’s RDF system designs. Installations with successful RDF technology include co-firing at Ames, Iowa; Baltimore, Mary- land; Lakeland, Florida and Madison, Wisconsin, as well as dedicated RDF firing at Lawrence/ Haverhill, Massachusetts; Akron, Ohio; Colum- bus, Ohio; Albany, New York; Hamilton, Ontario; Niagara Falls, New York and Dade County, Florida. Industry confidence in RDF as a state-of-the- art technology is evidenced by dedicated RDF facilities currently in startup or under construc- tion at fifteen sites across the nation (Table 1). Table 1 Dedicated RDF Refuse-To-Energy Facilities Plant Initial Location Capacity Operation Status Hamilton, Ontario 600 TPD 1972 Operating Akron, OH 1000 TPD 1979 Operating Hempstead, NY 2000 TPD 1980 Closed Niagara Falls, NY 2000 TPD 1981 Operating Albany, NY 700 TPD 1981 Operating Dade County, FL 3000 TPD 1982 Operating Columbus, OH 3000 TPD 1983 Operating Lawrence, MA 1350 TPD 1984 Operating Biddeford, ME 700 TPD 1986 Startup Red Wing, MN 940 TPD* 1987 Under Construction Mankato, MN 940 TPD* 1987 Under Construction Portsmouth, VA 2000 TPD 1987 Under Construction Hartford, CT 2000 TPD 1987 Under Construction Wilmington, DE 600 TPD 1987 Under ** Construction La Cross, WI 320 TPD 1987 Under Construction Erie, PA 850 TPD 1988 Under Construction Orrington, ME 800 TPD 1988 Under Construction Rochester, MA 1800 TPD 1988 Under Construction Detroit, MI 4000 TPD 1988 Under Construction Honolulu, HI 2160 TPD 1988 Under Construction San Marcos, CA 1650 TPD 1988 Under Construction Palm Beach, FL 2000 TPD 1989 Under Construction Naples, FL 600 TPD 1989 Under Construction 35,010 TPD * At these two plants, existing boilers (two at each plant) are being modified to operate as dedicated RDF refuse-to- energy facilities. ** RDF facility began commercial operation in 1986. Plant will begin operation as a refuse-to-energy facility in 1987. Evolution Hamilton, Ontario (1972) The Babcock & Wilcox Company’s first expe- rience with firing 100% RDF in a boiler designed for that fuel was at Hamilton, Ontario in 1972 when two 300 TPD stirling power boilers became operational. The original system was a “crunch and burn” type (Figure 1) consisting of a moving- bottom MSW storage bin feeding vertical ham- mermill shredders, an early desigr magnetic sep- Horizontal or Vertical Hammermill Ferrous Recovery 6 Shredder c = oath —» RDF to Boiler Figure 1 Process Flow Schematic (Hamilton, Akron & Columbus) aration system, an Atlas fuel storage bin, a metering screw feeding system, and interconnect- ing conveyors. Knowledge about the RDF storage characteris- tics and the RDF feed characteristics to the boiler was gained during early plant operations at Hamilton, with subsequent plant modifications. In a “crunch and burn” system, all of the incoming MSW is included in the boiler fuel, sim- ilar to a mass-burn system, except for a small amount of the non-processibles that can be picked from the incoming loads and a varying amount of removable ferrous, depending on the efficiency of the magnetic separator. However, by subjecting the incoming MSW to size reduction, a range of fuel sizes is produced which provides advantages to the combustion system. A suspension burn fuel fraction is produced and the ability to mix oxygen with the reduced size fuel particles is enhanced over that possible with a mass-burn system, which combusts MSW in the as-received size. In addition, by removing ferrous materials from the MSW, an abrasive and high impact component is removed from contact with the fuel handling sys- tem, the grate, and the ash removal system. The combustible fraction of the fuel is maintained while a deleterious fraction, which contributes to maintenance, is removed. Akron, Ohio (1979) The original RDF processing system in the Akron plant consisted of a moving-pit storage floor, a horizontal hammermill shredder, an early genera- tion magnetic separator, an air classifier, an enclosed screw discharge RDF storage bin, and pneumatic transport to the boiler. The system installed in Akron reflected a recognition of the abrasive characteristics of the dense fine fraction in RDF - glass, sand, grit, etc. An air classifier was provided to remove the dense fraction which contributes to maintenance of the equipment from fuel preparation through ash removal processes. Air classifiers, pneumatic RDF transport/feed to the boiler and enclosed bin RDF storage were applications of successful equipment configurations which had been employed previously for utility applications. For the utility applications, however, a primarily sus- pension burn RDF fuel was produced, which was small, 2” x 2” maximum top size. For the Akron application with a 100% RDF capability, the max- imum top size is larger, 6” x 6”, and the equip- ment performance was found to be much different than with the utility application for smaller fuel. Modifications were made to the Akron system to provide a process similar to that at Hamilton, i.e., “crunch and burn” type with no further attempts to remove fines. Columbus, Ohio (1983) The RDF processing system consists of vertical hammermill shredders, an early generation mag- netic separator, RDF screw discharge surge bins, RDF storage pit and belt conveyor discharge to the boiler front feed chutes. RDF is also produced at transfer stations using horizontal hammermill shredders. The Columbus RDF processing system was designed as a “crunch and burn” type both at the plant site using vertical hammermill shred- ders and at transfer stations using horizontal hammermill shredders. In addition, a coal handling system was installed, and the boilers were designed for com- bined firing of RDF and coal, as well as either fuel individually. Similar RDF processing systems and RDF feed characteristics were exhibited at Columbus as at Hamilton and Akron. RDF size characteristics from the different shredder configurations became apparent and modifications are in progress to provide top size control to reduce the oversized material produced from the vertical hammermill. RDF feed characteristics for the fuel produced are understood, and enhancements are underway to provide a more uniform, consistent feed to the boiler. Haverhill/ Lawrence, Massachusetts (1984) The Refuse Fuels Associates’ (RFA), Haverhill, Massachusetts RDF processing facility includes initial shredding with a vertical hammermill shredder, ferrous separation, a two stage trommel with unders diverted to landfill, sized fuel passing through the larger trommel holes, overs diverted to landfill, and a state-of-the-art feed system. A later addition was the installation of a horiz- ontal hammermill to shred the trommel overs and convey it to the RDF storage area (Figure 2). RDF Shredder ieee Fines to Landfill @ ———__—_ Ferrous Removal Shredder - 6" oe D LS RDF to. Boiler Figure 2 Process Flow Schematic (Haverhill) storage at both Haverhill, the RDF preparation site, and at Lawrence, the power plant site, is on an open floor in an enclosed building. Materials handling of the RDF from storage is by front- loader at both sites. With use of an overfed RDF supply, distribution and return system, consistent fuel flow is provided to the RDF boiler feeders. This is the first new installation to incorporate an overfeed/return of RDF. Their use has allowed uninterrupted supply of RDF to the feeders and has allowed the feeders to supply a continuous and evenly distributed supply of RDF to the boiler. The RDF feeders at Lawrence are a combi- nation surge bin/feeder designed to accept minor upstream interruptions in flow without having an impact on flow of RDF to the boiler. Each feeder has a surge hopper and a lower RDF hopper which houses a steeply inclined discharge con- veyor. A hydraulic ram operates in conjunction with the monitored level in the lower RDF hopper. The inclined discharge conveyor is driven by a variable speed DC motor, whose speed is con- trolled by the combustion control fuel demand signal. The lower RDF hopper configuration and the steeply inclined flighted conveyor impart a tumbling action to the RDF, causing a mixing and fluffing action within the hopper and result- ing in uniform feed to the chute to each airswept spout-boiler-fuel distributor (Figure 3). Uniform feed to the boiler and rapid response to demand fluctuations have been achieved, thus consistent stoker ash beds can be maintained. As a result, the fluctuation in boiler operating parameters such as steam flow, pressure and temperature, draft, etc. as seen at the earlier RDF plants are not present at RFA. The grate speed is slower, and clinker formation associated with slug feed- ing at other plants is not experienced at RFA. The trommel provided in the Haverhill process- ing system provides a successful means of remov- ing a portion of the harmful small-size fraction of the fuel, at the expense of also diverting a small- size combustible fraction to landfill. RDF Inlets Distributors of cn Figure 3 RDF Feed System Palm Beach County, Florida (1989) The RDF Processing System’ being supplied to Palm Beach County consists of a flail mill, mag- netic separator, two stage trommel, secondary shredder, disc screen, and combustible recovery Flail Mill - 1-1/2" —<— Heavies to Landfill Air Density Separator Trommel Ferrous Removal Shredder Lights Disc Screen —_S) Aluminum RDF to Boiler Recovery SED Figure 4 Process Flow Schematic (Palm Beach) from the trommel undersize rejects (Figure 4). The function of the flail mill is to open bags, break glass and provide a very coarse size reduc- tion. The advantage of using a flail mill before the trommel instead of a high speed, primary ham- mermill shredder is that glass is not pulverized and imbedded in paper and cardboard, nor is material shredded that is already the proper size for use as RDF boiler fuel. This material is removed by the trommel and conveyed to storage. Trommel undersized material is further processed by an air density separator to reclaim the light combustible fraction from the heavies. Maximum Btu utilization is expected with use of this equip- ment. The fuel-sized trommel discharge, -6”, may be air classified to densify a heavies stream so that aluminum cans can be hand picked effec- tively. This operation not only provides a poten- tial revenue stream, but is also a means of reduc- ing stoker maintenance. By positioning a disc screen and a recycle loop downstream of the secondary shredder, positive control of fuel top size is anticipated. In addition, a coarser grind fuel can be provided from the secondary shredder, thus reducing the smaller size fraction produced, and reducing shredder horsepower and mainte- nance. An air classifier is used at Palm Beach, even though one was removed from the Akron plant, the reason being the difference in the design and application. The air classifier supplied at Akron was a high air volume, high air velocity device, which was applied to the entire size range of RDF, from -1/4” size through 6” + size. Lighter mate- rials were entrained in the air stream and trans- ported up a vertical path to a cyclonic device for removal of the separated material from the trans- port air before discharging to a baghouse. The heavy fraction materials were dropped out of the air stream and landfilled. The air density separa- tor to be used on the Palm Beach project is a low volume, low velocity device, which is to be applied to a small size separated fraction, about -1-1/2” size. With this device, lighter materials are fluid- ized and gravity-tumbled down an incline while the heavier materials are friction conveyed on an upwardly sloped vibrating conveyor. A compari- son of design parameters of the two concepts of air classifier designs is provided in Table 2. The Palm Beach County facility will use the proven floor storage of RDF to avoid the plugging and bridging characteristics associated with the storage bins at earlier RDF plants. RDF supply, distribution and return of material, as well as the proven RDF feeders, will be utilized to achieve uniform operation of the refuse boiler. Metals removal characteristics Aluminum No attempt is being made to remove aluminum from the RDF at the Hamilton, Akron, Columbus or Haverhill plants. With uniform fuel feed and Table 2 Air Classifier Comparison * Early Design Stoner Type Air Air Density Classifiers Separator Throughput (TPH) 60 60 Air Volume (CFM) 60,000 12,000 Horsepower 220 50 Exhaust Air Particulate Loading 2,000 100 (grains/ft3) Exhaust Air Velocity (FPM) > 5,000 < 3,500 * Courtesy of Triple/S Dynamics Systems, Inc. distribution on the grate, as experienced at the Lawrence facility, a consistent ash bed depth is maintained which allows most of the aluminum that melts in the high temperature boiler zone to become cooled and trapped in the ash bed before reaching the grate and becoming a maintenance item. The Palm Beach facility is designed to include an aluminum beverage can hand-picking station, to be installed if the economics of aluminum can resale is favorable. Its use has the added advan- tage of removing a potential maintenance item from the fuel stream. Ferrous Removal of ferrous materials from RDF has expe- rienced steady improvement since the first appli- cations. Both drum and belt type magnetic separ- ators are now available with a supplier guarantee of 90 percent removal efficiency. The advances made with ferrous removal technology provide not only more material resale potential but the additional boiler related benefit of removing a non-combustible abrasive constituent, thus improving the maintainability of the unit. Lessons learned 1. Shred last, not first. Shredding first imbeds glass, grit, etc. in the RDF, which causes increased maintenance on the stoker and boiler. 2. Size separate and remove fines. Do not shred what is already the right size. 3. Use simple RDF transfer via conveyor rather than pneumatic systems. 4. Use simple RDF floor storage, not bins that are subject to pluggage. 5. Maintain a uniform flow of RDF to the boiler fuel feeders. Avoid slug feeding that results in unstable boiler control. 6. Use a proven RDF feeder, which maintains even grate distribution and is responsive to load change. RDF boiler evolution As the RDF processing systems have evolved, so have the RDF boilers. The RDF boiler related improvements can all be included in the two broad categories of Combustion Systems and Cor- rosion. The combustion system includes the grate, the overfire air system, furnace designed for proper furnace exit temperature, and furnace arches-all geared to achieve the best combustion efficiency. The modifications to reduce corrosion are mostly in the lower furnace area, although virtually all areas of any type of refuse-fired boiler, mass- or RDF-fired, are subject to some degree of corrosion. Combustion systems The composition of municipal refuse is very sim- ilar to that of wood (Table 3). With the removal of Table 3 Comparison of Ultimate Analysis of Wood and Municipal Refuse Hardwood Municipal Pine Bark Bark Refuse Constituent Percent Percent Percent Ash 15 5.3 14.4 Ss 0.1 0.1 0.2 Ho 5.5 5.4 5.7 c 55.3 49.7 425 O02 & No 37.6 39.5 37.2 100.0 100.0 100.0 HHV on ash 9300 8830 8600 and moisture Btu/Ib Btu/Ib Btu/Ib free basis ferrous metal and other non-combustibles in the typical RDF processing system, it was felt that RDF could be easily combusted in a boiler of the same design as that used for waste wood firing. In the United States, there is a wealth of experience with waste wood-fired boilers, particularly for the Pulp & Paper Industry. This waste wood combus- tion technology was adopted as the technology for RDF firing. In refuse-to-energy facilities, designed to burn RDF as a dedicated fuel, the RDF is introduced by overbed feeders to a traveling grate stoker (Figure 5). For both wood- and RDF-fired boilers, the grate is sized on a Btu input per square foot of grate area. Wood-fired units are in operation with fans and air supply systems must be sized to accommodate these higher excess air levels. The standard procedure for good combustion has always been the requirement for the “three T’s” - time, temperature and turbulence. This rule, of course, holds true for refuse firing. What was missing from early design RDF units was the optimum turbulence. The increase in the quantity of overfire air previously mentioned was one mod- ification to get better turbulence in the overfire air combustion zone. The original design one-inch overfire air ports have been replaced with larger overfire air ports to provide more penetration, and hence, more turbulence. The addition of lower furnace arches put the upper overfire air nozzles closer to the center of the furnace. The result of these three changes is an overfire air system that provides the optimum turbulence in the lower fur- nace combustion zone (Figure 6). Figure 5 RDF Traveling Grate Stoker with Overbed Feed grate release rates in the range of 1,000,000 to 1,200,000 Btu/sq ft/hr. From the very first dedi- cated RDF boiler to today’s state-of-the-art designs, the RDF stokers have been designed for a grate release rate of 750,000 Btu/sq ft/hr. This has proven to be an acceptable design parameter, and while very conservative when compared to wood-fired units, there is currently no considera- tion being given to increase this grate release rate design parameter. In addition to providing a platform on which the refuse can burn and convey away the ash residue, the grate also serves as a means to intro- duce part of the combustion air to the furnace. This undergrate air helps to dry the fuel and agi- tate the fuel bed. RDF boilers were initially designed with approximately 30 percent of the total air going through the overfire air ports and 70 percent through the undergrate air system. It has been found through operating experience that for good combustion, the overfire air system must be designed for broad flexibility to accommodate changes in fuel moisture, ash content, Btu value, etc. This results in designs that are capable of providing more of the total air as overfire air and less as undergrate air. Initially, RDF boilers were designed to operate at 35 percent excess air at the grate. While this is still the design parameter, wear and tear on stoker seals, plus variations in the RDF fuel supply, result in normal operating conditions that require higher excess air levels; consequently, the Overfire Air Undergrate Air Figure 6 Controlled Combustion Zone (ccz™ ) Lower Furnace This lower furnace double arch design is known as the Controlled Combustion Zone (CCZ™) design. It was developed by Babcock & Wilcox in the early 1970’s for the combustion of high mois- ture waste wood. The first CCZ™ boiler was a wood-fired boiler for Crown Zellerbach’s Camp- bell River, British Columbia mill, and it was started up in late 1980. Initial performance tests run in early 1981 indicated that the CCZ™ design was superior to previous conventional straight wall boiler designs. Prior to the CCZ™ design, wood-fired boilers and RDF-fired boilers used a reinjection system to return unburned carbon back into the furnace to be more completely combusted. Such a reinjection system was very successful in improving boiler efficiency by lowering the boiler’s unburned car- bon loss (UCL), but has also increased the ash loading in the furnace and throughout the boiler. At several RDF facilities, the operators felt that the reinjection system was the cause of some lower furnace erosion. One goal of the CCZ™ design was to achieve low UCL without the necessity of a reinjection system. On wood firing, field testing has verified that the UCL of a CCZ™ unit without a reinjection system can be held to values considerably lower than could be attained with straight wall units with reinjection systems. The CCZ™ boiler design is currently the state-of-the-art boiler design for both wood and RDF boilers. Table 4 lists the Table 4 cCZ™ Design Boiler Experience List Plant Location Fuel Startup Date Campbell River, B.C. Wood December 1980 St. Johns, New Brunswick Wood July 1985 Crosset, Arkansas Wood August 1985 Bathhurst, New Brunswick Wood September 1986 Selma, Alabama Wood September 1986 Greenville, Maine Wood November 1986 Biddeford, Maine RDF January 1987 Florence, South Carolina Wood February 1987 Springfield, New Hampshire Wood May 1987 Athens, Maine Wood May 1987 Whitefield, New Hampshire Wood September 1987 San Marcos, California RDF 1988 Palm Beach, Florida RDF 1989 CCZ™ design units in operation or under con- struction, including the first RDF-fired CCZ™ boiler which is currently in startup. In addition to being a corrosive and hard-to- burn fuel, refuse is a fuel that can cause slagging in the furnace and fouling in the convection sec- tions (superheater, boiler bank, etc.) if not designed and operated properly. The potential for slagging in the lower furnace can be minimized by a well-designed combustion system and is aided by the removal of glass and metal frag- ments in the RDF processing system. Fouling in the superheater and boiler bank areas can be controlled by cooling the furnace gases to 1600- 1700 F. The early RDF boilers were designed for this furnace exit gas temperature (FEGT) design parameter, however the furnace heat absorption rates were not as high as initially predicted and, therefore, the FEGT was higher than predicted. New absorption rates were verified by operating unit data and are used in the design of today’s units. These lower absorption rates result in more furnace heating surface being required to obtain the desired temperature and, therefore, today’s furnaces are physically bigger than first and second generation RDF boilers for the same refuse capacity. Corrosion MSW is a corrosive fuel. Refuse-fired boilers, whether mass or RDF, will be subject to some amount of corrosion. A normal part of the opera- tion and maintenance of a refuse-fired boiler is the constant awareness of corrosion rates and the repair and protection of areas that have exceeded design allowances. Corrosion in refuse boilers, mass or RDF, is related to the relatively high amount of chlorides in MSW. While an RDF processing system may remove some of the materials containing chlo- rides, it is not realistic at present to prevent boiler corrosion by the removal of materials in the RDF processing system. The rate of tube metal corrosion in a refuse boiler is a function of the metal temperature, the corrosion rate increasing with higher metal temperatures (Figure 7). This is a very similar Bare Carbon Steel Tube Corrosion Rate Bare Tube Metal Temperature - °F Figure 7 Theoretical Chloride Corrosion Rate in Refuse-Fired Boiler phenomenon to the sulfide corrosion that occurs in process recovery boilers, which burn the black liquor residue from the pulping of wood (Figure 8). This sulfide corrosion rate is also a function of tube metal temperature. Se 20 ee ge 42 10 Sa a5 COG ee o So aé 400 500 600 700 Bare Tube Metal Temperature - °F Figure 8 Sulfide Corrosion Rate in Black Liquor Fired Pulp and Paper Process Recovery Boiler The furnace tube metal temperature will be slightly higher than the saturation temperature of the steam/water mixture in the furnace tubes. The superheater metal temperature also will be slightly higher than the steam temperature in the superheater tubes. One way to minimize furnace corrosion would be to operate at very low steam pressures and temperatures. The RDF boilers at Albany, New York and Hamilton, Ontario gener- ate low pressure, saturated steam and are not known to have experienced any significant corro- sion problems. However, most of today’s refuse-to- energy plants depend on the revenue from the production and sale of electricity. The higher the steam pressure and temperature to the turbine, the more kilowatts can be produced from the same ton of refuse input. The typical refuse boiler today operates at 600 psi, 750 F steam conditions. B&W refuse boilers at two U.S. refuse facilities are being successfully operated at 900 psi, 830 F, and several more are currently under construction. For the same refuse input, the 900 psi, 830 F unit will produce more kilowatt output than the more typi- cal 600 psi, 750 F unit. Therefore, it is economi- cally advantageous to find ways to operate at these higher pressures and temperatures while minimizing the effects of corrosion. Since the corrosion rate is tube-metal depend- ent, it is critical to control water chemistry of the boiler feedwater. Poor water quality or water chemistry upsets can result in the buildup of de- posits on the inside of the furnace tubes. This buildup insulates the tube and results in higher furnace side tube metal temperatures which will result in higher-than-normal corrosion rates. Another factor on the acceleration of lower fur- nace corrosion is fuel related. Corrosion rates have been shown to accelerate in the presence of low melting point metal chlorides, particularly zinc chloride. Zinc compounds are present in tex- tiles, leather and rubber. Therefore, refuse that contains a higher-than-normal percentage of these materials could result in higher refuse boiler corrosion rates. The environment of the lower furnace of both mass and RDF units is generally one that is con- stantly changing between an oxidizing atmo- sphere (an excess of 02 beyond that needed for combustion) and a reducing atmosphere (a defi- ciency of Og below that needed for combustion). If it were possible to perfectly mix the fuel with the air, a refuse boiler could be operated at very low excess air levels and in a constant oxidizing atmosphere, thus minimizing the effect of a reduc- ing atmosphere on the corrosion rate. Unfortu- nately, perfect mixing is not feasible. Originally it was thought that RDF units would have much better mixing of air and fuel, and that the result- ing corrosion effect would be much less than that of the typical mass-fired units. While RDF units did have better mixing and operated at lower excess air levels, roughly half that of mass units, the lower furnace corrosion was still apparent. The state-of-the-art RDF unit (Figure 9) with the ~ Burner Overfire (One Each Air Ram Feed N Inclined __ Feeder —~Y/ Figure 9 State-of-the-Art CCZ™” RDF Boiler Design CCZ™ furnace design will provide for better mix- ing of the fuel and air than the first and second generation units, and will reduce oxidizing/reduc- ing corrosion. A protective coating on the lower furnace tubes is an effective means of preventing corrosion. Some coatings occur naturally, such as the nor- mal oxide coating that is inherent in refuse boil- ers, as well as light coatings of slag that develop during operation. While effective, these coatings are not uniform, may be porous, and are easily removed. With first generation RDF processing systems, the fuel is shredded first, with the glass, ceramic and grit becoming imbedded in the RDF. This imbedded abrasive material helps remove light protective coatings from the tubes, consis- tent with the trajectory of the incoming RDF. These areas of the lower furnace have shown some corrosion. The lower furnace sections of mass-fired refuse boilers are protected with pin studs and refrac- tory, otherwise they would experience lower furnace corrosion. Combustion in an RDF boiler results in higher furnace temperatures than a mass-fired boiler. These higher furnace temper- atures coupled with a refractory coated furnace could result in slagging problems in the lower furnace. One RDF boiler was originally installed with a refractory coating in the lower furnace but the refractory was soon removed when furnace slagging became a problem. On another RDF unit, an alloy weld overlay has been applied in the lower furnace, the alloy mate- rial being resistant to chloride corrosion. While this is a positive means of preventing corrosion in the lower furnace, it is also an expensive method. 10 Research is ongoing to find more cost effective materials for lower furnace protection. Excessive lower furnace corrosion has actually only been experienced on one of the eighteen RDF boilers currently in operation. In the upper furnace, superheater, boiler bank, etc., there is a normal oxidizing atmosphere. In these areas the normal oxide coating is typically enough to prevent excessive corrosion. In fact, the only areas where corrosion has occurred is where this coating is periodically removed, such as the tubes immediately adjacent to the sootblowers. Where such corrosion has been found, tube shields have been installed. The state-of-the-art RDF system Today’s design RDF refuse-to-energy plant is a state-of-the-art facility. As a result of experience with the operating RDF preparation systems, the operating RDF fuel feeding systems, and the operating RDF boilers, the state-of-the-art RDF units will burn refuse in a cleaner, more efficient way than any other process. RDF may be used in a retrofit of an existing boiler to burn refuse and is compatible with mate- rial recovery for recycling prior to incineration. RDF also provides flexibility to collect and pro- cess refuse at one location and to incinerate the beneficiated refuse at another location. Further- more, RDF can be used as a co-compost material with sewage sludge to provide a useful product. In short, RDF is the technology of the present and future, providing an ecological balance among recycling, incineration, landfilling, and composting. Bibliography . H. Franco, “The Hamilton SWARU Retrofit,” ASME June 1986 National Waste Processing Conference. . T. Barr, D. Moats, K. O’Brien, “Working Out The Kinks: An Update on Columbus’ Refuse and Coal-Fired Municipal Electric Plant,” ASME June 1986 National Waste Processing Conference. 3. F. Hasselriis, “Thermal Systems for Conversion of Municipal Solid Waste, Vol. 4 Burning Refuse-Derived Fuels in Boilers - A Technology Status Report,” Argonne National Laboratory for US DOE, March 1983, ANL/CNSV-TM-120, Vol. 4. . D. Kaminski, “Performance of the RDF Delivery and Boiler Feed System at the Lawrence, MA Facility” ASME June 1986 National Waste Processing Conference. . J. D. Blue, J. R. Strempek, “Considerations for the Design of Refuse-Fired Water Wall Inciner- ators,” Energy From Municipal Waste Conference, October 1985. . P. L. Daniel, J. D. Blue, J. L. Barna, “Furnace-Wall Corrosion in Refuse-Fired Boilers,” ASME June 1986 National Waste Processing Conference. . Institute for Local Self-Reliance, “Resource Recovery State-of-the-Art: A Data Pool for Local Decision-Makers,” June 1986. 11 | Technical Paper Corrosion of composite | BR-1296 port opening tubes in recovery boilers: appearance and occurrence J.L. Barna Domestic Fossil Operations Babcock & Wilcox Barberton, OH J. B. Rogan Domestic Fossil Operations Babcock & Wilcox Atlanta, GA Presented to 1986 TAPPI Engineering Conference Seattle, Washington September 21-25, 1986 Babcock & Wilcox a McDermott company Corrosion of composite port opening tubes in recovery boilers: appearance and occurrence J. L. Barna, Materials Engineer Babcock & Wilcox Barberton, OH J. B. Rogan, Regional Service Mgr. Babcock & Wilcox Atlanta, GA Presented to 1986 TAPPI Engineering Conference Seattle, Washington September 21-25, 1986 Abstract PGTP 86-17 Corrosion of the stainless layer of composite tubing around port openings in recovery boilers has been investigated through field inspections and laboratory evaluations of corroded tubing. Corrosion has been observed in port openings in the primary and secondary zones of both high and low pressure boilers, and affects AISI 304 and 304L composite tubing. The corrosion appears to terminate once the stainless steel has been consumed and the carbon steel exposed. The importance of field inspection techniques and instrumentation is emphasized and documentation from field and laboratory studies is provided to permit recognition of corrosion locations and appearance. Instrumental to the corrosion process are crevices that have the capability of accumulating deposits and that form a non-gastight seal between the furnace and casing side of the tubing. Twelve months of operating experience with a port design that eliminates crevices suggests that stainless layer corrosion may be eliminated through design modification. It has yet to be determined whether design changes alone will be sufficient. Elimination of corrosion through material changes also remains to be investigated. Introduction In early 1985, the composite tubing in several B&W recovery boilers was found to have expe- rienced localized corrosion of the stainless clad around primary and secondary air port openings. Subsequently, a field engineering study and related corrosion research program were initiated. The objectives of these efforts were to identify the affected B&W units, characterize all aspects of corrosion locations and morphology, and relate this information to unit design, operation and chemistry. This paper presents the results of the field study, which details the appearance and occurrence of corrosion from a survey of eleven units. The purpose of this paper is to emphasize the need for thorough consideration of construc- tion details and inspection techniques when ex- amining composite tube recovery boilers for cor- rosion. Through a comprehensive understanding of the corrosion mechanism and its contributing factors, some corrective and preventive measures can be defined. Recovery boiler furnace The furnace wall construction of all B&W recovery boilers built since 1963 consists of 3-inch OD (7.62 cm) tubing on 4-inch (10.16 cm) centers with 1- inch-wide (2.54 cm) welded membranes. The wall construction is the same for boilers built with car- bon steel, pin studded tubing and composite tub- ing. The composite tubing is comprised of a SA- 210 Al pressure bearing tube and either an AISI 304 or 304L stainless clad layer. The nominal wall thickness of the composite tube is 0.259 inches (6.58 mm) with the nominal base tube and clad thicknesses being 0.194 inches (4.938 mm) and 0.065 inches (1.65 mm), respectively. All attach- ments to the composite tubing, e.g., membrane bars and stud plates, utilize 309 stainless welds. The lower furnace incorporates three levels of air admission with liquor being sprayed through either oscillating or stationary nozzles. Illustrated in Fig. 1 is the general furnace arrangement of the B&W recovery boiler. Composite tubing is typ- ically installed up to the tertiary air level, with the tertiary air ports being either carbon steel or composite tubing. The bed height of the boilers under study ranges from no bed at all to a level above the secondary air ports, with the average being at an elevation between the primary and secondary air zones. Composite tube corrosion has been noted almost exclusively at the primary and secondary air elevations. Depicted in Fig. 2 are the pri- mary and secondary windbox and air port arrangements. Total air flow to the primary and secondary zones is controlled by duct dampers. Separate port dampers bias flow to individual ports. While usu- ally open, the primary port dampers may be slightly throttled by operators, depending on bed conditions near the port. Secondary port dampers are either completely open, maintaining high velocity air flow, or entirely closed, remaining out of service with minimal air flow. Until 1979, B&W supplied only carbon steel tubes with pin studs for recovery boiler furnaces. When the initial composite tube furnaces were constructed in 1979, the air port designs were Tertiary Air Oscillator —. WK oogoo00000 | J, Furnace Tube | Secondary Air Duct Secondary J. Rodding Port Air —, L Port iy I Sight Glass Primary Air Duct ir Primary Air Port Damper Damper Ay sient Glass bp Primary P| - Rodding Port ir - Port be Figure 1 Furnace arrangement of B&W recovery boiler. Figure 2 Primary and secondary windbox arrangement. replicas of those utilized in earlier carbon steel furnaces. Figs. 3 and 4 show primary air ports in carbon steel and composite tube furnaces, respec- tively. The primary air port design shown repres- ents that used in carbon steel furnaces since the adoption of membraned wall construction in 1963. Case Studies Presently, there are six U.S. and five Canadian units built by B&W that are constructed with composite tubing and that have been inspected for corrosion of the stainless layer. Operating information and survey data for the eleven units are provided in Table 1. Of the eleven units, seven operate at superheater outlet pressures of 1000 psig (6890 kPa) or lower, and four operate at pres- sures in excess of 1000 psig (6890 kPa). Corrosion of the composite tubing has been found in all of the higher pressure (> 1000 psig or 6890 kPa) units and has affected tubes made with Figure 3 Primary air port design of carbon steel furnaces since 1963. both AISI 304 and 304L stainless. Both primary and secondary air ports have been affected with corrosion which, experience has shown, usually occurs at both elevations if the port design is similar. Several of the high and low pressure units have also experienced a similar preferential corrosion of the stainless layer on clad stud plates (1). The survey data of Table 1 does not account for these incidents and deals exclusively with corrosion of the tubing. The units under study represent from six months to seven years of operation since installa- tion of the composite tubing. Reflecting the ten- dency of the corrosion to predominate toward the casing side and to occur in inaccessible locations, the corrosion has in only one case been detected in the initial stages. Inspection of Unit K (Table 1) revealed corrosion during the first six-month outage following installation of the composite tub- ing. The corrosion had not progressed to the point of exposing the carbon steel and was detected only because of prior experience in inspecting other boilers. In the other affected higher pressure units, the corrosion was not discovered until measurable wastage of the stainless clad had occurred, i.e., to the point of exposing carbon steel. Delays in recognizing the problem until long after considerable wastage had occurred has further complicated the understanding of an already complex corrosion process. Lower pressure units are not immune to corro- sion of the stainless layer of the composite tubing. Units C and D (Table 1) are identical in size, design and operating time and differ only with Tih LB J Figure 4 Primary air port design of composite tube furnaces through 1984. Table 1 Operating and Corrosion Survey Data For B&W Composite Tube Furnaces SH Outlet Mos. Service Operating Before Start Pressure Clad Tube Corrosion Unit Up (PSIG) Material Corrosion Noted A 7/79 870 304 No - B_ 10/80 900 304 No — Cc 1/81 900 304 Yes 42* 0 2/81 1300 304 Yes 32* E 11/82 1500 304 Yes 30* F 1/83 850 304 No _ G 1/83 915 304L No a H 5/83 875 304L No = 1 11/83 625 304L No = J 7/84 1500 304L Yes 12* K 4/85 1500 304L Yes 6 Corrosion was first detected after carbon steel was already exposed respect to operating pressure. After five years of operation, both were inspected. In the higher pressure (1300 psig or 8957 kPa) Unit D, over 250 individual corrosion sites were found in the pri- mary and secondary zones where carbon steel was exposed. Comparatively, the lower pressure (900 psig or 6201 kPa) Unit C displayed less than 10 sites of exposed carbon steel. However, clad thickness measurements of the lower pressure unit revealed many additional sites where the cor- rosion had advanced through at least half of the original stainless layer thickness. Thus, stainless wastage does occur in lower pressure units, but at rates significantly lower than in comparably designed and operated higher pressure units. The absence of exposed carbon steel and the lack of clad thickness measuring devices during inspection probably account for the reported absence of corrosion in other lower pressure units. It is possible that most lower pressure units have experienced wastage which will be discovered with continued operating time. Inspection of a 1300 psig (6957 kPa) unit 12 months after the wastage was initially detected revealed a substan- tial increase in the number of sites of exposed carbon steel. However, comparison of inspection data from other boilers indicates that additional factors such as design specifics and boiler chemis- try are also contributory. Corrosion locations relative to opening design Primary Air Ports Design (1979-1984). Fig. 5 shows typical corrosion locations at primary air ports designed through 1984. Featured are a double layer of flat stud plates at the top and bottom that form a rectangu- lar opening. A gastight barrier between the fur- nace and windbox is formed where the solid rear stud plate is welded to both opening tubes. Con- trasts and comparisons between corrosion loca- tions, relative to the top and bottom of the open- ing and between identically designed carbon steel and composite tube furnaces, are made in the fol- lowing ways: ¢ At the top of the opening, the corrosion is typi- cally positioned at the centerline of the furnace wall and in the crevice formed between the upper tongue of the port casting and tube. The corrosion does not generally extend below the upper tongue of the casting but extends upward for a distance of ~ 1-2 inches (2.54-5.08 cm) behind the rear stud plate. ¢ In contrast to the top of the opening, the corro- sion at the bottom does not envelope the rear stud plate and is not confined solely to the cre- viced area between the lower tongue of the port casting and tube. The most advanced corrosion site at the bottom is typically positioned within one inch above the tongue in regions where de- posits accumulate. ¢ Corrosion at the bottom of the opening has been observed to extend significantly toward the cas- ing side between the port casting and tube near the bolted joint. The absence of corrosion on the sides of the opening at the contact between tube and port casting possibly reflects a sensitivity to crevice size or crevice position relative to the furnace. ¢ No corrosion has been observed on the furnace side of the tube adjacent to either the stud plates or port casting. Stud Plates ™ Port Casting “7 Wastage of Stainless Figure 5 Corrosion locations in primary air ports designed through 1984. 056 Stainless Port Casting SLATAT ALYY LNT co Figure 6 Corrosion locations in primary air ports designed in 1985. e Inspection of an identically designed 1500 psig (1035 kPa) carbon steel unit after 7 years of ser- vice revealed no corrosion of the carbon steel at locations where composite tube wastage has been observed. Many years of experience with carbon steel tubing shows that the locations where composite tubes corrode are not suscepti- ble to sulfidation corrosion. Design (1985). The primary air ports were rede- signed to eliminate stud plates. Fig. 6 illustrates | — Wastage of Stainless J Figure 7 Corrosion locations in secondary air ports designed through 1982. the locations of corrosion on the 1985 primary air port design. To date, only one unit (Unit K) of this design has been inspected for corrosion. e After 12 months of operation, corrosion was noted adjacent to the upper and lower tongues of the port casting in the crevice between tongue and tube. Secondary air ports Design (1979-1982). Fig. 7 shows the secondary air port design utilized through 1982. Featured are two rows of flat stud plates backed by refractory. The stud plates are not welded together or to the adjacent membrane bar and thus do not form a gastight construction between furnace and windbox. The corrosion in this design does not differ between the top and bottom of the port. Wastage at both locations occurs behind the rear stud plate, extends toward the casing side of the tube, and spans a distance equal to the length of the stud plate. Design (1983-1984). The redesign of 1983 was made to reduce the size of the secondary air port opening. Featured are a single set of double stud plates backed by refractory with no port casting. As shown in Fig. 8, the stud plates are welded to only one opening tube and thus do not form a gas- tight seal between furnace and windbox. i j Stud Plates Ir: JJ Stainless — Figure 8 Corrosion locations in 1983-1984 secondary air port design. ¢ Corrosion occurs behind the rear stud plate at the tube centerline. ¢ Corrosion occurs on tube surfaces positioned at the wall centerline in the crevice formed between the small end of the stud plate and membrane. Design (1985). The secondary air ports were redesigned in 1985 to eliminate all stud plates. As shown in Fig. 9, a gastight seal is formed between furnace and windbox. This design does not utilize a port casting and eliminates crevices. Inspection of the 1985 design revealed no corro- sion in port openings after 12 months of operation. Observation Ports Observation ports are at the secondary air eleva- tion of all boilers discussed in this paper. The design of the observation port is identical to that shown in Figure 8 for the secondary air ports. Corrosion has been observed in observation ports, thus demonstrating that air flow is not a prereq- uisite for corrosion to occur. Summary of Corrosion Locations ¢ Opening designs that feature a more gastight construction generally exhibit less overall cor- rosion than those that are not gastight and con- tain crevices. Membrane rarara ra LLP PLS a Figure 9 Secondary air port design of 1985. ¢ The occurrence and location of corrosion gener- ally varies between the top and bottom of the openings and among differently designed openings. ¢ Variations in the occurrence and location of cor- rosion relate to the geometry of the opening and corresponding windbox construction, locations of crevices, and the capability of crevices to accumulate deposits. ¢ Contrary to earlier indications, the top of the openings do not display a universal tendency toward greater corrosion susceptibility than the bottom. ¢ Continual or high velocity air flow is not a necessary condition for corrosion to occur at air port openings. ¢ Corrosion is not unique to weld locations and may affect large areas of tubes that contain no welds but that are adjacent to attachments on bordering tubes. Corrosion appearance Visual Field inspections of corroded tubes from different units demonstrate a variety of corrosion morphol- -—15"—4 Figure 10 Dimpled or pitted topography of stainless layer. Arrow indicates exposed carbon steel. ogies. In the initial stages where wastage typi- cally does not exceed 0.010 inch (.254 mm) in depth, the corroded surface often displays rough- ening. The superficial nature of the roughening and associated tube deposits often make positive identification difficult without thorough surface cleaning and thickness measurement of the stain- less layer. Complicated by the position of corroded surfaces in confined crevices where visual inspec- tion and thickness measurement are difficult, if not impossible, inspectors commonly employ small diameter brushes and the use of methanol and compressed air to sufficiently clean the corro- sion site. In other cases, the initially corroded surface appears dimpled or pitted as shown in Fig. 10. Factors that influence the dimpled versus smooth morphology of the corrosion, such as location in the unit and length of service, have yet to be established. In the advanced stage, when the stainless layer thickness is significantly diminished or entirely consumed, the surface often appears smooth and featureless. The photograph of Fig. 11 is represen- tative of a corrosion site that displays exposed carbon steel. To the touch, the affected surface may feel flat. A smooth transition from stainless steel to carbon steel is characteristic and a dis- tinct ridge between the two materials is not often Figure 11 Stainless layer corrosion and exposed carbon steel ina primary air port opening. apparent either visually or by feel. Often, the exposed carbon steel will be rusted following a water wash. Inspectors often rely on swabbing suspect sites with copper sulfate solution to detect exposed car- bon steel. Because of misleading discolorations associated with this procedure, it is recommended that supplemental thickness measurement be per- formed to assure positive identification. To be most effective, areas to be checked with copper sulfate should be completely cleaned to a bright surface finish prior to testing. Clad thickness rather than total wall thickness measurement, such as by conventional ultrason- ics, is the preferred technique. Devices using small diameter or right angle probes are best suited for clad thickness measurement because of the inaccessible nature of the corrosion locations. Instruments utilizing magnetic induction princi- ple for clad thickness measurement should be used with caution to avoid erroneous readings due to interference from stainless welds. Metallography Tube samples exhibiting corrosion of the stainless clad were removed from two units and metallo- graphically examined. One sample represented an AISI 304-clad tube after 56 months of operation and the other represented an AISI 304L-clad tube after 12 months of operation. Metallographically, the corrosion morphologies of both samples were Corroded Surface (a) t—0.4"—4 Stud Corroded Plate Weld Surface (b) -—.01" —4 ———_ Figure 12 Photomicrographs showing smooth surface of stainless layer corrosion. consistent with those observed visually. The pho- tomicrographs of Figs. 12 & 13 illustrate corrosion sites that evidence both a smooth featureless sur- face and a scalloped profile indicative of a dimpled topography. No corrosion products are microscopically discernible. Visually and metallographically, there are no differences in the corrosion mode between the two types of stainless steel, i.e. AISI 304 and 304L. Both steels evidence transgranular wastage and no apparent selective attack to grain boundaries. Additionally, no alteration of the corrosion mode is associated with the fusion boundary (b) }— 04" —+ Figure 13 (a) Photomicrograph showing scalloped profile of 304-clad tube after 56 months (unetched). (b) Photomicrograph showing scalloped profile of 304L-clad tube after 12 months (chromic etch). between the stainless clad and carbon steel com- ponents of the tube. Fig. 14 illustrates that the corrosion path is uninterrupted and not affected by the higher carbon and twinned structure of the stainless at the boundary. In the tubing examined, the corrosion progres- sion appears to terminate once the stainless layer has been consumed, leaving the round profile of the carbon steel tube intact. The photomicro- graphs of Fig. 15 are representative of AISI 304 and 304L clad tubes after service exposures of 56 H— 0.1" —4 Figure 14 Photomicrograph showing corrosion. and 12 months, respectively. The photomicrograph of Fig. 15a further shows that the corrosion appearance is the same at sur- faces distant from welds and at weld heat-affected zones. Fig. 15a also demonstrates that 309 stain- less welds are relatively immune to attack at loca- tions where the AISI 304 stainless clad is susceptible. Corrosion mechanism and rates Evidence suggests that the evolution of hydroxide vapors from the smelt bed and subsequent con- densation at creviced areas in the furnace walls is responsible for corrosion of the stainless layer of composite tubing around port openings. Applica- ble creviced areas are those formed between tubes and attachments in regions where furnace con- struction permits access of furnace gases toward the casing side of the tubes. A study of the hydroxide corrosion mechanism was made in the 1960s when corrosion occurred on the cold side of non-membraned, carbon steel furnace tubes at the primary and secondary air elevations (2,3,4). The literature suggested that operating temperature and pressure, non- membraned wall construction, and the use of high chromium refractories were prime contributors to the corrosion. Corrosion rates in composite tubing should not be generalized. The field study indicates corrosion sensitivity to furnace construction, operation, and chemistry; all of which vary among the units and between different locations within a specific unit. Clad thickness measurements and corrosion dis- tribution data clearly demonstrate the variability Casing Side yea Stud Plate (a) -— 0.2” Furnace Side 02" — Stud Plate Figure 15 (a) Photomicrograph showing round profile of carbon steel in a 304-clad tube after 56 months. (b) Photomicrograph showing round profile of carbon steel in a 304L-clad tube after 12 months. in corrosion rate. For example, in a typical high pressure boiler after one year, there may be only a few isolated sites of relatively small size where the corrosion rate has resulted in exposed carbon steel. These may not be detected without extensive inspection. In three years, dozens of small areas of exposed carbon steel may be present. Hundreds of small areas of exposed carbon steel may be present after five years. The entire corrosion pro- gression may be slow or fast, depending on boiler operating pressure, lower furnace operating con- ditions, or liquor chemistry. Corrosion of the AISI 304 and 304L stainless layers of composite tubing generally occurs at rates that are more rapid than those reported in the 1960s for carbon steel tubing. The 309 and 309L stainless welds in the composite tube units appear relatively immune to attack. In molten salts, mild steel is more resistant to attack than some of the chromium-containing austenitic steels, such as AISI 304 or 304L (5). Since nickel is resistant to molten salts, higher levels of nickel in an alloy, such as AISI 309, will reduce the detrimental effect of chromium on cor- rosion resistance (6). No literature is available to suggest that grain boundary carbides are influen- tial to the corrosion of austenitic stainless steels in molten salts. In regions where corrosion has penetrated the stainless clad layer and subsequently exposed the underlying carbon steel tube, industry practice has been to overlay the affected area with 309 stainless weld. Inspections of boilers 12 months after weld repairs were made have revealed no measurable corrosion or other distress to the over- lay areas. Information obtained through the present field study and reports of hydroxide corrosion of car- bon steel furnaces in the 1960s suggest that no appreciable corrosion of the carbon steel portion of the composite tube will occur in areas where the stainless clad has been completely consumed. Supportive of this claim are the following obser- vations and rationalizations: ¢ When first detected, none of the corrosion sites in composite tube furnaces displayed significant wastage of the exposed carbon steel. In many cases, it is not known how long the carbon steel was exposed before the discovery was made. e The design of primary air port openings in BkW units is identical in both carbon steel furnaces built since 1963 and composite tube furnaces built through 1984. No hydroxide corrosion of B&W boilers made of carbon steel tubing has been observed after 1963, at which time the use of lower chromium refractories and membraned wall construction was implemented. ¢ If the carbon steel tube units are not affected, then exposed areas in the composite tube units should also be unaffected, primarily because these areas locally resemble the former carbon steel furnace. Evidence obtained from this study suggests that the occurrence of corrosion is further influenced by the presence of deposits. Regardless of opening design, all corrosion occurs at sites where a gas- tight construction between the furnace and casing side of the tube is not maintained and where de- posits accumulate and adhere; the latter of which are not necessarily entirely within crevices. It is speculated that hydroxide vapors condense at deposits and form a molten phase at the deposit/- substrate interface. Future studies Continued field study and corrosion monitoring of composite tube recovery boilers will supplement laboratory testing and evaluations of boiler chem- istry. Smelt, liquor and tube deposits from cor- roded and non-corroded units are being analyzed to further understand chemical influences on the incidence and rates of corrosion. References 1. Wensley, D. A., Proc. TAPPI Kraft Recovery Operations Seminar, Orlando, Fla., 1986, p. 237. 10 . Clement, J. L. and Chassell, J.R., Proc. 22nd TAPPI Engineering Conference, Atlanta, GA., 1967, p. 28. . Proc. of Finnish Recovery Furnace Corrosion Research Committee, Helsinki, Finland, 1968, p. 70. . Bruno, F., “Pulp and Paper Industry Corrosion Problems” (4):68 (1983); Swedish Corrosion Institute, Stockholm, Sweden, 1983. . Brasunas, B., “Alloy Behavior at High Temperature,’ NACE, 1975, pp. 13-31. . Pickner, D., and Bernstein, I.M., Handbook of Stainless Steels, McGraw-Hill, New York, 1977, pp. 12-30. Technical Paper The importance of proper loading of refuse-fired boilers J.F. Clunie and A. Leidner R.W. Beck and Associates Denver, Colorado James T. Hestle, Jr. and Garry R. Weaver Nashville Thermal Transfer Corporation Nashville, Tennessee Babcock & Wilcox BR-1300 a McDermott company The importance of proper loading of refuse-fired boilers J.F. Clunie and A. Leidner R.W. Beck and Associates Denver, Colorado James T. Hestle, Jr. and Garry R. Weaver Nashville Thermal Transfer Corporation Nashville, Tennessee Introduction The operator of a resource recovery project may find himself in a position of having been too enthusiastic and aggressive in his attempt to maximize energy production and solid waste disposal at his plant. Encouraged by customers to produce increased amounts of energy during peak periods of demand and urged by city officials to dispose of increased quantities of solid waste because of problems at the landfill, the plant operator may find himself a short range hero, but unwittingly, a long range villain. This situation actually occurred at the Nashville Thermal Transfer Corporation (Thermal) during the late 1970's and early 1980's and it wasn’t until the plant had actually been in operation for approximately six years before the actual effect of overloading the boil- ers became apparent and could be identified and evaluated. It is the purpose of this paper to explain the method of operation which had been employed, the changes which were implemented and the com- parative operating and financial results of the “before” and “after” operation. A Brief History of the Nashville Thermal Transfer Corporation The story of Thermal has been told and retold more often than the tale of James Watt’s invention of the steam engine and there is little to be gained by dredging up the past, beyond stating that it is extremely important for all parties to undertake proper and adequate planning in the early stages of project development even when it appears that one is making a direct application of prior experience. Briefly, the basic concept of Thermal was formulated in May 1970 by a concerned City of Nashville which created Thermal as a not-for-profit corporation for the purpose of providing low cost district heating and cooling to the Metropolitan Government of Nashville and Davidson County ("METRO") by constructing a solid waste-fired central heating and chilling plant and distribution system. The plant was financed in June 1972 through the issuance of $16,500,000 of revenue bonds which were to be paid off completely from energy revenues with the exception of a nomi- nal annual payment of $150,000 by METRO. The plant was to include two 360 TPD (327 tpd) municipal/commercial solid waste incinerator boil- ers capable of generating 218,000 Ib/hr (99,000 kg/h) of steam with 14,000 refrigeration tons of water chilling capacity, and a standby auxiliary gas/oil-fired package boiler rated at 135,000 Ib/hr (61,000 kg/h) of steam. Thermal began commercial operation in 1974 but problems with the original air pollution control equipment resulted in the United States Environmental Protection Agency issuing a compliance order requiring Thermal to undertake steps to install alternative air pollution control equipment. After Thermal obtained additional financing of approximately $8,000,000 in 1976, two electrostatic precipitators were installed in 1976 and 1977 which effectively solved the air pollution con- trol problem. Since the completion of the installation of the elec- trostatic precipitators in 1977, the two largest main- tenance problems at Thermal have been: (a) the unexpectedly high rate of boiler tube wastage; (b) frequent breakages of the reciprocating grate com- ponents, both of which resulted in increased parts and supplies expense as well as unusually high maintenance labor expense. Boiler tubes which were expected to have a useful life of anywhere from six to ten years had to be replaced after only two years. What was causing these problems? Was there some- thing in the solid waste of Nashville which resulted in unusually high corrosion? Was the design of the boil- ers inadequate in some respect, particularly with regard to an adequate mixture of over-fire and under-fire air? Was the method of operation faulty in some manner? Starting in the late 1970's and culmi- nating in 1981, Thermal management and its con- sultants began an active search to try to find an answer to this problem. Level of Operation With the advantage of hindsight, and reviewing the entire matter from the vantage point of 1984, it is relatively easy to look back and say the answer should have been fairly obvious. However, it is important to keep in mind the nature of Thermal’s operation for the five year period between 1977 and 1982. In 1976, Thermal was suffering from an acute public relations problem. Thermal had been repre- sented to the public as an entity which could provide an inexpensive and environmentally sound means of waste disposal and energy production. In fact, there was initially no tipping fee being charged at Thermal; instead a supplementary payment to Thermal was being made by METRO which was equivalent to approximately $1.00/ton. The air pollution problems experienced in 1974 and 1975 had been highly pub- licized both in the press and on television. Because the initial projections of operating and maintenance expenses had been greatly underestimated, it became necessary to institute rate increases for heating and cooling services which more than doubled the rates in less than one year. Thermal was being sued by an irate energy customer who charged that the increases were due to mismanagement and it became obvious that if Thermal lost this lawsuit, additional energy customers were likely to follow the same course of action. Furthermore, when it became evident that Thermal would require an additional $8,000,000 of financing to pay for the precipitators, replenish the debt service reserve fund and repay approximately $3,500,000 of short term financing, the need for an additional source of revenue to help pay for this additional financing became imperative. Since it was impractical to further increase the rates for heating and cooling services, the only alternative was to look to METRO to increase its annual payment to support Thermal. On December 31, 1975, METRO and Thermal exe- cuted a contract which provided for an additional payment to be made to Thermal by METRO. Under this contract, METRO agreed to pay an additional $1,350,000 annually, or a total of up to $1,500,000, which was equivalent to a tipping fee of approxi- mately $10.00/ton. The new commitment to Ther- mal by METRO was politically controversial and once again thrust Thermal into the public spotlight. In 1976, Thermal was placed in a position of: (a) trying to increase its operating revenues by increas- ing its sales of heating and cooling services; (b) decreasing its fossil fuel expense by utilizing a greater amount of solid waste as the primary fuel; and (c) increasing the amount of solid waste dis- posed of in order to decrease the unit cost of dispo- sal as well as extend the useful life of the METRO owned Bordeaux Landfill which was rapidly being filled. Thermal management's reaction to these three pressure points was understandable: run as much solid waste through the plant as possible, even if it meant exceeding the boiler manufacturer’s recom- mended rating as to loading levels of the incinerator- boilers. This particular decision coincided with another decision concerning the method of opera- tion, and when the two decisions were combined, the final results would prove to be a significant problem. The second decision dealt with the determination to operate only one solid waste boiler at a time at or above maximum rated capacity, with the second solid waste boiler to be utilized as back-up. The genesis for this decision to operate just one solid waste boiler at a time had its origin in two areas: (a) during Thermal’s initial attempts at com- mercial operation, the operating problems were so severe that it seemed as though one boiler was con- stantly down for repair while the other boiler was being operated; and (b) certain of Thermal’s energy customers had no alternative means of heating and cooling services, so there was a requirement that at least one boiler must be available at all times. As a result of these two points, an operating philosophy developed which, in somewhat simplified terms, was as follows: operate one boiler at its maximum capac- ity for as long as possible while repairing the second boiler, and meet peak energy demands by utilizing the fossil-fuel-fired package boiler. During this same time period, a question also arose as to what the maximum capacity of the solid waste boilers actually was. The boilers, when using solid waste as a fuel, were initially rated in 1972 at 360 TPD (327 tpd) with solid waste at 6000 Btu/Ib (3333 Cal/kg). In 1980, Thermal management announced that it had increased the boiler rating to 530 TPD (481 tpd) based on as-fired solid waste with a heating value of 4500 Btu/Ib (2500 Cal/kg). The original steam rating of 109,000 Ib/hr (49,400 kg/h) on solid waste with 6000 Btu/Ib (3333 Cal/kg) was reported to have been increased to 125,000 Ib/hr (56,700 kg/h). Such rerating of the system was reported to have been made possible due to a generous boiler furnace design and a series of stoker undercarriage modifications and improvements. Thermal reported that on January 9, 1979, approx- imately 590 tons (535 t) of solid waste were burned in one boiler, a rate which was 64 percent over the original design stoker loading capacity on a mass basis. However, Thermal management reported that the uprating of the boilers was not the result of an overload situation, but rather demonstrated that the capacity of the boilers when burning solid waste was equal to the design capacity of the boilers when burning either gas or oil. While utilizing just one boiler Thermal reported in 1980 that, “In operation we normally steam at 100,000 Ib/hr (45,400 kg/h) steady state. Steam not used for customer load and chilled water production is dumped into the Surplus Steam Condenser.” Confronted with a history of significant operating problems and trying to fulfill the demands of maxi- mum waste disposal, reliable service and minimum energy rates, Thermal management determined that the best way to meet its multiple objectives was to operate one boiler on solid waste at an average steaming rate of 100,000 Ib/hr (45,400 kg/h), util- ize the auxiliary fossil-fuel-fired boiler for peak demands, and undertake repairs to the second solid waste boiler with the hope that such repairs could be completed before the first solid waste boiler was in need of repair. Reaction of the Manufacturers Concern with such methods of operation was expressed by representatives of Thermal’s boiler manufacturer who, in 1980, wrote the following: “The two boilers at this plant are operated in the same way they are operated all over the world. Plants that are required to dispose of waste and produce a continuous and reliable supply of steam must have a standby refuse boiler upon which to rely. This is what is done at Nashville but, unfortunately, in doing this one boiler must carry the total load and, in doing so, is frequently forced to operate at greater than design throughputs and higher than design steam flows on refuse. We believe this is one of the main causes of the high maintenance experienced at this facility . . . . the stoker’s burning rate is higher than intended, the fuel’s residence time in the furnace is reduced, the design furnace exit gas temperature and gas velocities throughout the unit are exceeded. If (the boiler manufacturer) had known the units were going to be required to operate under these conditions, we would have increased considerably the size of the stoker, furnace and boiler.”[1] A representative of a testing laboratory also noted: “Another protection against superheater corrosion that has been learned by many European operators is: don’t overload the boiler-furnace, don’t push it too hard. The point here is that municipal solid waste is often a distressingly variable fuel. If the boiler is pushed to the limit continually, there will be many (excursions) where excess burning will cause flame to extend into the first pass. During such excursions, if the superheaters are in the furnace, or in the first pass, corrosion can be rapid.” [2] Finally, a representative of Thermal’s stoker manu- facturer stated: "In order to avoid burning auxiliary fuel, Nashville Thermal . . . rerated these units . . . to a rating of 530 TPD (481 tpd) during cold weather based on 4,500 Btu/Ib (2500 Cal/kg) refuse with peak ratings up to 590 TPD (535 tpd). "This 530 TPD (481 tpd) rating results in a burn- ing rate on the stokers of 107.5 pounds of refuse per square foot of grate (525 kg/m?) and a heat of 16.6 million Btu per foot of stoker width (13.7 million Cal/m). These burning rates are far beyond design ratings on any stoker fired unit. "These overload operating conditions have obviously had a severe and substantial effect on the maintenance costs and reliability of the equipment in this plant... . These overload conditions with a pressurized furnace and low excess air greatly increased maintenance costs and shortened the life of the stoker and contributed to the high mainte- nance costs and relatively low availability of the firing equipment. "Experience at Nashville points out the importance of matching the capacity of the boiler and the number of units to provide sufficient capacity for the maximum load requirements and allow for mainte- nance of the units, without requiring overload condi- tions or auxiliary fuel firing.” [3] It should be noted that such comments were pro- vided as an attempt to offer an explanation for the unusually high maintenance costs, not as a criticism of Thermal’s management. All parties understood and appreciated the severe operating demands under which Thermal was required to perform. Historical Operating Results Prior to describing what changes were actually made to the system, and in order to better appreciate the improvement in the level of operation which was experienced as a result of the remedial actions, it is helpful to review certain historical operating results prior to the time such changes were made. Table 1 shows an annual breakdown of operating and main- tenance expenses actually incurred during the period June 1, 1976 through May 31, 1982. Asa review of Table 1 illustrates, Thermal’s operating problems began to increase in 1980, the same year that Thermal management uprated the boilers. Annual operating and maintenance expenses increased from approximately $1,969,000 in 1979 to $3,824,000 in 1982, an increase of approximately 94 percent in three years. It is important to remember that for a portion of this time period the nation’s overall economy was plagued with double-digit inflation and a significant increase in the price of oil in 1979. However, infla- tion alone was not the sole reason as the two major contributing causes of the problem were: (a) extraordinarily high levels of maintenance parts and supplies, which consisted primarily of the cost asso- ciated with boiler tube and stoker repairs; and (b) the need to burn fossil fuel during periods when the solid waste boilers were not available. These significant increases in operating and main- tenance expenses were taking place during a period when Thermal’s energy customers were undertaking a series of conservation measures designed to decrease purchases of steam and cooling commodi- ties (See Table 2). For example, between 1979 and 1982, steam commodity sales decreased from 253,731,000 Ib (115,092,000 kg) to 183,656,000 Ib (83,306,000 kg), a decrease of approximately 28 percent. While Thermal had managed to avoid a rate increase from May 1975 through December 1979, it became necessary to implement annual rate increases of between 9 and 10 percent each year between 1980 and 1982 just to help pay for the increases in operating expenses. The annual pay- ment from METRO was also increased to its maxi- mum level of $1,500,000. In spite of these increases, Thermal’s sources of revenues were not sufficient in 1982 to meet operating and maintenance expenses and the annual debt service payment. Confronted with a continually escalating problem (maintenance parts and supplies expense had increased by approximately 250 percent over four years), Thermal management looked even more closely at the problem of the unexpectedly high level of boiler tube wastage and stoker problems and, in cooperation with its consulting engineer and the plant’s design engineer, embarked on a program to alleviate these problems through: (a) the establish- ment of optimum boiler loading and mode of opera- tion; and (b) expansion of the overfire air system. The Search For A Solution The Overfire Air System Beginning in late 1980, the year the boilers were re- rated upward to 530 TPD and the year operating and maintenance expenses increased by 41 percent, Thermal and its consultants began trying to find a solution to the severe operating problems which were being experienced at the plant. Thermal man- agement focused its attention on the question of the adequacy of the existing overfire air system. Visual inspection of the fire within the incinerators indi- cated that there was flame impingement on the refractory-covered lower sidewalls and such flame impingement was causing excessive slag buildup. Furthermore, the flame was observed to extend through the superheater and the convection bank. There was also a problem with the formation of car- bon monoxide in the combustion gases. CO causes corrosion by continually removing or preventing the formation of a protective oxide layer on the surface of exposed ferrous metal and its presence indicates either the use of insufficient combustion air or that combustion air is not optimally distributed to and within the combustion zones. Representatives of Thermal visited other resource recovery facilities which reported to have expe- rienced problems similar to those that Thermal was encountering. Following their discussions with the operators of these other facilities, Thermal manage- ment determined to undertake a program to increase the quantities of overfire air in the incinera- tors, thereby providing for operator adjustment of the distribution of overfire air around the perimeter of the stoker. Construction of the new overfire air sys- tem began in the second half of 1981 and was com- pleted in April, 1982. During the consultations, it had been stipulated by the boiler supplier that an alternating oxidizing and reducing atmosphere in the Thermal boilers together with excessively high temperatures and high flames were contributing factors to tube corrosion from chlorine (and sulfur) compounds. The original stoker design provided for 85 percent underfire air and 15 percent overfire air at an operation with 80 to 85 percent excess air. Subsequent experience by the stoker supplier has changed this design philosophy to increase the overfire air to 30-40 percent of total air. It has also been determined from operating records that operating limitations on the amount of underfire air that could be admitted to the solid waste bed from under the grate resulted in operation with excess air far below the recommended 80 to 85 percent. It has been felt therefore by the stoker supplier and the design engineer that an expansion of the overfire air capacity would be beneficial in promoting better and complete combustion in the furnace, and eliminating the reducing atmosphere in the boiler. It was also expected that the new system might reduce the height of the flame in the furnace. This plan was consistent with similar expansions of original overfire systems in boilers of European design. Consequently, in the second half of 1981 an expansion of the overfire air system, as developed by the design engineer, was implemented on the Ther- mal solid waste boilers. The expansion consisted primarily of the installation of 16 new nozzles, 1-1/16 in. (27 mm) wide by 3-1/2 in. (83 mm) high, on each side of the boiler, about 5 to 6 ft (1.5 to 1.8 m) above the grates, and the installation of new overfire air fans. The main air headers are square ducts 36 x 36 in. (91.4 x 91.4 cm) while the branches are of a round cross section 20 in. (51 cm) and 12 in. (30.5 cm) in diameter. The flow through individual duct branches is controlled by pneumat- ically and manually operated dampers. The fan header pressure is 30 in. of water (762 mm). It should be noted that the overfire air system, as origi- nally supplied, consisted of a single row of nozzles across the furnace rear wall and two rows of nozzles across the front wall. In 1975 an additional 14 noz- zles were added to the lower front wall row. The total overfire air capacity is now 30,000 acfm (849 acmm) as compared to the original 7,500 acfm (212 acmm). Observations and tests were conducted by the boiler supplier on the No. 2 boiler during the week of March 1, 1982 and further observations and record- ings were made on March 26, 1982 to evaluate the effect of the expanded overfire air system. During the tests, flue gas samples from a sampling grid between the boiler and economizer were analyzed and recorded and observations were made of the furnace and fuel bed conditions. Also recorded were air, steam and water data and other control room instrumentation readings. Flue gas data included continuous 0, and CO recordings. Periodically, Orsat analyses were run and flue gas temperatures and steam and economizer outlet water temperatures were recorded. Although not of direct importance, NO, data was also gathered during all but the first test. Six test runs were executed during two days - one at 100,000 Ib/hr (45,000 kg/h) steam flow under the old overfire air mode and the other five under the new mode. The general conclusion of the boilers supplier in its report (B&W No. S-10240) was: "The new OFA system has produced a major change in furnace appearance and reduction in flame height. The new OFA system, in conjunction with a good distribution of fuel bed burning on the two main grates (#2 and #3 sections) and burnout on #4 grate will usually keep the flame below the top of the refractory covering on the furnace walls.” Proper Boiler Loading and Mode of Operation Simultaneously with the review by Thermal man- agement of the need for modifications to the overfire air system, Thermal’s consulting engineer undertook a review of the then current loading of the boilers and method of operation of the system. The review began with extensive research of all existing files, literature, specifications and design criteria deve- loped ten years earlier when the facility was originally being designed. Such research was followed by cor- respondence and conferences with representatives of the boiler and stoker manufacturers. Following these consultations, it was determined that the com- bustion air systems in place were adequate to pro- vide the total amount of combustion air recom- mended by the boiler manufacturer. It was also determined, however, that the total combustion air supply capability had not been usable. The stoker design provided for most of the combustion air to enter the furnace up through the stoker grate, and Thermal personnel determined that uniform com- bustion could be achieved only by reducing under- grate air flow to some value below the maximum capability. The provisions for supplying combustion air above the grate (overfire air) were not adequate to compensate for the reduction of undergrate air, nor were they adequate to establish a fully effective secondary combustion zone above the grate. Upon completion of his review, the consulting engineer recommended the following: "After reviewing available information on the origi- nal specifications and history of operation of the boilers, it is the opinion of the consulting engineer that the boilers have been operated at waste burning rates greater than the original design specifications. Inquiries made of the boiler and stoker manufactur- ers corroborated this opinion and the manufacturers made specific recommendations as to maximum waste burning rate. The burning rates recommended by the manufacturers are lower than the burning rates which have been followed by the Company since the plant was put into operation. The recom- mended burning rates will require that both waste- burning boilers be operated together, a method of operation which has not been followed by the Com- pany since only one waste-burning boiler has been operated at one time with the other boiler either out of service for maintenance or on standby.” Specifically, the consulting engineer recom- mended to Thermal’s management that the maxi- mum continuous rating (MCR) of the boilers should be re-established at 80,000 Ib/hr (36,000 kg/h) of steam when burning 360 TPD (327 tpd) of solid waste with a heating value of 4500 Btu/Ib (2500 Cal/kg which will amount to a grate loading of approximately 73 Ib/ft? of grate (356 kg/m?). Con- sequently, the consulting engineer further recom- mended that in order to meet the steam and chilled water loads, the normal mode of operation should be to operate both solid waste boilers simultaneously. The consulting engineer estimated that under his regimen, the operating availability of each boiler would be in the range of 80 to 85 percent and the probability of having either one or two boilers availa- ble would be in the range of 96 to 97.5 percent. During periods of unavailability of one of the boil- ers, it was stipulated that either the gas/oil-fired aux- iliary boiler would satisfy the steam demand in excess of 80,000 Ib/hr (36,000 kg/h) or the remain- ing boiler in operation would temporarily be over- loaded. It was anticipated by the consulting engineer that under this mode of operations together with the improvement in the combustion because of the expansion of the overfire air system, the mainte- nance expenses due to tube wastage and stoker fail- ure would be substantially reduced. The reduced level of operation and the simultane- ous operation of the boilers began in the summer of 1982 and the boilers have been operated in that manner ever since. Simultaneous Operation of the Boilers Thermal began the simultaneous operation of the two solid waste boilers in August, 1982. The plan was to operate each of the two boilers simultane- ously at an average load of approximately 50,000 lb/hr (23,000 kg/h) with the loading to be increased to approximately 100,00 Ib/hr (45,000 kg/h) during periods of peak demand. Such levels of steam generation were achieved by incinerating with both boilers an average of approximately 400 TPD (363 tpd) of solid waste. This represented an increase of 68 TPD (62 tpd), or 20.5 percent, for the system as a whole over its previous method of operation. How- ever, it also represented a decrease of 106 TPD (96 tpd), or 32 percent, for each individual incinerator boiler. Boiler availability for the first ten months of two-boiler operation was 81 percent. While it is still too soon to determine the long range effect of these two changes the results thus far have been most encouraging. Total fossil fuel expense decreased from approximately $438,000 in 1982 to $246,000 in 1983, a decrease of approxi- mately 44 percent; maintenance parts and supplies decreased from approximately $1,640,000 in 1982 to $475,000 in 1983, a decrease of approximately 71 percent; and total operating and maintenance expenses decreased from $3,824,000 to $2,562,000, an overall decrease of 33 percent. Dur- ing the first three months of the fiscal year ending May 31, 1984, the trend has continued. By project- ing the first three months for the remainder of the year, an admittedly dangerous thing to do at times, the total operating and maintenance expenses for the year would be approximately $2,272,000. Furth- ermore, during fiscal year 1983, solid waste repre- sented approximately 96.2 percent of Thermal’s annual fuel requirements, the highest level ever achieved. During the first three months of fiscal year 1984, solid waste accounted for 98.6 percent of the fuel requirements. Boiler availability for the first quarter of this fiscal year has been 94 percent. Several questions come immediately to mind. How much of this improvement is due to the fact that a significant retubing program was just com- pleted in 1983? How much of the improvement is due to the installation of the expanded over-fire air system and would the same results have been achieved regardless of the level of loading? Unfortunately, the answers to these questions are not presently available. Since the retubing of the boilers, the installation of the overfire air sys- tem and the reduced level of operation all hap- pened simultaneously, it is not possible to point at any one of the three items and say, "There is the reason for improvement.” With the passage of time and the gathering of more information, an answer should eventually be available. If the present level of improvement continues beyond two years, the retubing of the boilers can be dismissed as the answer. Additional tests are being made of the overfire air system and if the results of the next set of tests are not more conclusive than the first set of tests, the overfire air system may also eventually be dismissed. However, in the meantime, we will have to wait until such results are actually available. Summary There are several valuable lessons to be learned from Thermal’s experience. The most obvious deals with the importance of maintaining a proper level of loading on the boilers. It is possible that this particu- lar lesson could have been learned sooner by paying more attention to both the operating experience which had been gained in Europe and the ratings of the equipment established by the boiler and stoker manufacturers. However, a somewhat less obvious lesson deals with a determination by policy makers and plant superintendents concerning the primary function of a solid waste resource recovery facility. Is the princi- pal goal to dispose of the maximum quantity of solid waste possible, or is it to generate energy in the most cost effective method possible? By attempting in the short term to help solve the serious waste disposal problem in Nashville by operating the plant at or above its maximum capacity, Thermal defeated its second objective of generating energy at the lowest possible cost. Ironically, in the long run, Thermal may defeat its initial objective of maximum waste disposal. Policy makers should answer these ques- tions prior to the design of a facility so that the engi- neers can take such matters into account in the design of the project. References [1] Rochford, Robert S., "Discussion of Paper- Update on Nashville Thermal,” Proceedings of the 1980 National Waste Processing Conference, ASME, New York. [2] Engdahl, R.B., "Discussion of Paper-Update on Nashville Thermal,” Proceedings of the 1980 National Waste Processing Conference, ASME, New York. [3] Reschly, D.C., “Discussion of Paper-Update on Nashville Thermal,” Proceedings of the 1980 National Waste Processing Conference, ASME, New York. Table 1 NASHVILLE THERMAL TRANSFER CORPORATION HISTORICAL OPERATING EXPENSES Fiscal Year Ended May 31 Operating Expenses 1977 1978 1979 1980 1981 1982 1983 Production Operation Electricity $158,643 $174,625 $218,343 $246,910 $325,491 $373,860 $474,610 Natural Gas 176,645 88,644 153,595 249,638 407,941 354,328 229,238 Fuel Oil 120,554 78,668 127,304 24,519 65,809 83,538 17,004 Water 126,926 100,581 72,279 50,907 50,276 45,777 55,013 Chemicals 54,254 33,209 23,937 36,942 21,789 24,146 24,206 Payroll 470,398 521,306 533,295 581,609 715,289 826,620 829,348 Other Expenses 43,697 42,004 23,774 20,987 35,506 35,712 25,050 Maintenance Parts & Supplies 344,961 508,926 467,875 1,184,419 1,225,686 1,639,867 475,424 Total Production Operation 1,496,078 1,547,963 1,620,402 2,395,931 2,847,787 3,380,848 2,129,893 Distribution 14,767 69,895 42,750 38,922 50,880 50,846 86,779 General and Administrative 419,373 308,892 305,484 346,824 350,957 392,183 345,262 Total Operating Expense $1,930,218 $1,926,750 $1,968,636 $2,781,677 $3,249,624 $3,823,877 $2,561,934 Percent Increase (Decrease over Previous Year) (17.8%) (0.2%) 2.2% 41.3% 16,8% 17.7% (33.0%) Tons of Solid Waste Processed 105,972 123,947 134,370 118,730 131,664 120.889 145,641 O&M Expense per Ton of Solid Waste Processed $18.21 $15.54 $14.65 $23.43 $24.68 $31.63 $17.59 Table 2 NASHVILLE THERMAL TRANSFER CORPORATION HISTORICAL OPERATING REVENUES & COMMODITY SALES Fiscal Year Ended May 31 1977 1978 1979 1980 1981 1982 1983 Operating Revenues Steam $1,448,093 $1,418,361 $1,515,987 $1,555,400 $1,529,846 $1,665,064 $1,754,761 Cooling $1,433,864 1,473,017 1,383,196 1,413,872 1,903,779 2,104,990 2,263,558 Total $2,881,957 $2,891,378 $2,899,183 $2,969,272 $3,433,625 $3,770,054 $4,018,319 Rate Increase None None None 10% 9% 10% None Payment from METRO(1) $1,425,000 $1,300,000 $1,300,000 $1,300,000 $1,300,000 $1,500,000 — $1,500,000 Sales Volume Steam Demand (Pounds per hour)(2) 176,805 178,490 190,805 190,805 190,305 186,350 196,015 Commodity (Thousand Pounds) 236,204 230,218 253,731 230,014 198,671 183,656 190,872 Cooling Demand (Tons)(2) 8,625 8,665 8,715 10,000 10,390 10,340 10,960 Commodity (Thousand Ton Hours) 18,234 19,109 17,410 16,293 21,595 20,135 21,027 Solid Waste as Percent of Annual Fuel Requirement 87.8% 93.8% 91.0% 91.2% 88.6% 89.9% 96.2% Cost to METRO per ton to Incinerate(3) $13.45 $10.49 $9.68 $10.95 $9.87 $12.41 $10.30 (1) Thermal receives an annual payment from METRO in lieu of a tipping fee (2) Peak Demand during 12 month period (3) Payment by Metro divided by Solid Waste Processed THE AMERICAN SOCIETY OF MECHANICAL ENGINEERS 345 E. 47 St., New York, N.Y. 10017 15 months after the meeting. Printed in USA. 86-JPGC-Pwr-A The Society shall not be responsible for statements of opinions advanced in papers or in discussion at meetings of the Society or of its Divisions or Sections, or printed in its publications. Discussion is printed only if the paper is published in an ASME Journal. Papers are available from ASME for Montana- Dakota Utilities 80 MW AFBC Retrofit B. Imsdahl Generation Manager Power Production Montana-Dakota Utilities Co 400 North Fourth Street Bismarck, ND 58501 ABSTRA' Montana~Dakota Utilities Co.'s Unit 2, at R. M. Heskett Station, is being retrofitted to atmospheric fluidized bed combustion (AFBC). e unit will burn North Dakota lignite using the existing spreader feed system. Improved performance and increased boiler capacity will be realized by retrofitting. Included in this project is the installation of an all water-cooled AFB combustor beneath the existing wreader fed boiler, a new tubular air heater, forced aft fan, AFBC circulation pumps, and other auxiliaries. The new AFB combustor and new/or replaced auxiliaries are being designed and installed by Babcock & Wilcox. Discussed are the major design parameters, arrangement, and expected performance of this utility AFRC retrofit project. INTRODUCTION To increase unit capacity and improve overall unit performance, Montana-Dakota Utilities Co. has undertaken an extensive program to upgrade and convert Unit 2, at its R. M. Heskett Station, Mandan, North Dakota, to fluidized bed comb ion. This unit is currently a spreader stoker-fired boiler and is believed to be the largest of its type in the country. After retrofitting to fluidized bed combustion, the unit will fire Beulah North Dakota lignite, reusing the existing spreader feeder system. In fact, the entire boiler convection pass and furnace enclosure walls will be reused. Tew modifications to existing pressure parts will be required. A new tubular type air heater, new forced draft fan control system modifications and various auxiliary fluid bed stems will be installed. This project exemplifies how fluidized bed combustion can satisfy the utility market's increasing need for power plant upgrades, utilization of BR-1301 R. L. Gorrell Manager, Fluidized Bed Boiler Development Babcock & Wilcox 20 S. Van Buren Ave Barberton, OH 44203 H. L. Johnson Product Development Manager Babcock & Wilcox 20 S. Van Buren Ave. Barberton, OH 44203 available low cost fuel, and emission reductions. Also worth noting is the fact that this is currently the largest utility AFBC contract funded solely by the utility itself. OPERATIONAL HISTO! The existing Unit 2 at Montana-Dakota's Heskett Station consists of a Riley lignite-fired stoker boiler and a General Electric turbine. The capability rating of the boiler is 650,000 lbs/hr (294,835 kg/hr) at 1300 psig (8963 kPa), 950°F (513°C), and the nominal rating of the turbine is 81.2 MW with a steam low of 682,700 1bs/hr (309,668 kg/hr). The unit was placed in commercial operation November 1, 1963. Originally, the grate moved slowly away from the feeders - the combustion being completed as the grate reached the rear of the furnace. The ash was then discharged into hoppers below. Five years after the unit went commercial, the direction of travel of the grate was reversed. With the reversal of the grate, the feeders would now throw the fuel to the rear of the boiler, with the grate discharging the ash to hoppers located underneath the feeders. As a result, better combustion efficiency has been experienced. During the modification of the grate reversal, the three vertical platens, which had been fed from the front of the boiler, were rearranged, lengthened, and are being fed from the rear to collect more radiant and convection heat from the furnace. In order to combat the slagging and fouling which reduced the effective load carrying capacity of the boiler, many different fuel additives were tried throughout the years. However, none of these fuel additives were successful in reducing the slagging or fouling for long-term periods. The installation of water lances near the high temperature superheater tubes was successful in Presented at the Jt. ASME/IEEE Power Generation Conference Portland, Oregon — October 19-23, 1986 removing some of the slag on the superheater, but had limited overall success in improving continuous steam output. Because of the slagging and fouling experienced when the unit was loaded to near original rating, low combustion efficiency was experienced, not only because of the unburned carbons that went to the bottom ash hoppers, but also because of carbon carryover past the reinjection hoppers and into the dust collectors. Another problem experienced in burning the Beulah lignite was the build-up of a porcelain-type coating on the generating tubes between the two boiler drums. This deposit on the generating tubes was directly related to the amount of sodium found in the lignite ash. To remove this deposit, the unit must be shut down for water washing, since the installation of sootblowers is impractical due to the close tube spacing. DESCRIPTION OF UNIT Furnace Arrangement The overall general arrangement of the unit is typical of most spreader stoker-type boilers. A sectional side view of the unit as it exists is shown in Figure 1. The furnace is approximately 40 ft wide (12.1m) by 21 ft (6.4m) deep and contains three water-cooled wing walls. The furnace wall construction is of a water-cooled tube and tile construction with a cold gas-tight casing. The wing walls, which are fed by downcomers from the lower drum, penetrate the lower rear furnace wall, rise through the furnace, and connect directly to the upper drum. Convection Pass Arrangement The convection pass contains superheater, steam generating, and economizer surfaces. The superheater is an all pendent type with spray attemperation for steam temperature between the primary and secondary banks. The generation surface has a 60 in. (1.52m) diameter upper drum and a 36 in. (0.9im) lower drum with all long-flow heating surface. The economizer surface is a bare-tube-counterflow type. The superheater and generating bank enclosure is a water-cooled tube and tile construction while the economizer enclosure is refractory and insulation lined. Both enclosures have a cold gas-tight casing. Fuel Feed System The coal feed system consists of three bunkers, three conical distributors, and ten stoker spreader feeders which are evenly spaced along the front wall of the furnace. Each unit has a separate cup-type rotary volumetric feeder with a drum-type rotary flipper. Back-End Equipment Flue gas and air handling equipment consists of a multiclone-type dust collector, one regenerative-type air heater, one FD and ID fan, and an electrostatic precipitator. The multiclone dust collector is used for the first stage of particulate control only. A from the collector is removed by the plant ash handling system, and is not reinjected to the furnace. Currently, ash is reinjected from the boiling bank hopper only. Reinjection is done pneumatically through the lower rear wall with injection air provided by a separate cinder return fan. Main ‘Steam C Outlet Economizer heater Division Wall Dust Collector Spreader/ Feeder | | (| Overtire Air Fa ~ Lr” Figure 1 Side view of existing boiler at Montana-Dakota’s Heskett Station, Unit 2. RETROFIT FLUID BED PERFORMANCE PARAMETERS Performance ~ The existing turbine capacity is greater than the existing boiler capacity. The retrofit design will increase the boiler capacity to meet the existing turbine capacity. The new steam conditions for the unit are given in Table 1. Other selected performance parameters are summarized in Table 2. Table 1 Steam Conditions Superheat Flow* 700,000 1b/hr (317,515 kg/hr) SH Outlet Pressure 1300 psig (8963 kPa) SH Outlet Temperature 955°F (513°C) Feedwater Temperature 443°F (228°C) *Existing Unit steam flow is 650,000 lb/hr (294,835 kg/hr) Table 2 Selected Performance Parameters Fluidization Velocity 12 ft/sec (3.7m/s) Normal Bed Temperature 1500°F (816°C) Bed Depth 51 in. (1.3m) Overall Excess Air 25% Air Heater Gas Exit Temperature 275°F (135°C) Bed Material Sand As mentioned before, the primary factor setting the fluidization velocity is the available maximum bed plan area, The bed depth is set by the height of the in-bed tube surface. The bed material to be used is a local sand. It was chosen for both functional and economical reasons. Because of the low sulfur content of the fuel, sulfur capture is not required. Since limestone is not locally available, it would be much more costly than the local sand. To minimize the overall costs of this project, including those of demolition and erection, modifica- tions to the existing unit have been kept to a minimum. These limitations, however, impose several major constraints on the design and arrangement of the new fluidized bed retrofit. Reuse of Coal Feed System One of the first limitations imposed is due to 2 reuse of the existing coal handling and feed stem. The front and rear wall locations of the asiuid bed and windbox are all set by the type and location of the spreader feeders. The front wall has to be in line with the feeder while the windbox compartmentalization has to match up with feeder spacing; the rear wall of the bed is limited by the maximum throw of the feeder. Reuse of Lower Side Walls It is desirable not to alter any part of the existing side wall pressure parts including the lower headers. This limits the maximum width dimension of the fluid bed and, with the restriction set by the coal feed system, establishes overall bed plan area. This, in turn, sets the nominal operating fluidization velocity on the unit. Structural Steel Modifications Because the weight of the new tubular air heater and new fluid bed, including pressure parts and bed material, is substantially more than the existing stoker and air heater, both have to be bottom support- ed. This is done to minimize modification to structural steel and foundations. In addition, the new air heater has to be arranged to fit with the existing air heater column rows and the resulting available space. Precipitator and ID Fan The existing electrostatic precipitator and existing ID fan are also to be reused. The retrofit design has to incorporate the functional capacities of w both of these systems. The major limitation, set by the ID fan, is the maximum-allowable gas side pressure drop across the new tubular air heater. REQUIRED CHANGES FOR RETROFIT Equipment Removal To accommodate the new equipment required for the retrofit design, several existing systems will have to be removed. For instance, the existing stoker grate, stoker ash hoppers, and grate cooling fan will have to be removed to allow for installation of the fluid bed proper. The existing overfire air fans, cinder rein- jection fan, and associated duct work also will be removed to allow for installation of boiler circula- tion pumps. In addition, the existing regenerative air heater, along with associated flue and duct work, will be taken out to allow for installation of the new tubular air heater. Furthermore, the existing FD fan will have to be removed. Pressure Parts Removal As discussed, existing pressure part modifica- tions have been kept to a minimum, but a few modifications will be required. The only major water-side pressure part change will be the removal of the lower drum-end downcomers and the existing wing-wall inlet headers. These downcomers will be re-routed to feed the inlet of the boiler circulation pump. The wing wall tubing will also be removed to a point just inside the furnace. The existing main steam piping will be modified to connect the existing convective superheater section to the new in-bed superheater,. New Pressure Parts New downcomers will be connected to the existing lower drum and routed to new boiler circulation pumps located at the rear of the unit. From the pumps, supply tubes will be routed to the fluidized bed floor, in-bed boiling surface, and to the bed enclosure walls. The entire bed enclosure and distributor plate is a water-cooled membrane type construction. All new water circuits, including the bed enclosure tubes, floor tubes, and in-bed boiling bank will be routed and connected directly to the existing furnace wing walls. A complete sectional side view of the unit after retrofitting is shown in Figure 2. The steam side pressure part modification will also be minimal. The entire existing convective superheater will be used. The main steam line leading from the convective superheater will be re-routed to connect to the new in-bed superheater. From the new in-bed superheater outlet, a new steam line will be added and routed back to the existing main steam line to the turbine. A new superheat attemperator will also be added in the new interconnecting piping between the convection pass and in-bed surface. Aur Heater Main Steam Line Forced Bratt fan Windbox Pumips Ash. Removal System Figure 2 View with fluidized bed and new air heater added. AUXILIARY SYSTEMS Fuel Feed System On this retrofit the entire existing fuel handling and feed system will be reused. The only change will be in the feeder drive controls area. Here, the individual feeder drives will be grouped and con-trolled on a bed compartment basis. Air Heater and Fans The existing regenerative air heater is being re- placed with a new tubular-type air heater. This replacement is required due to: 1) the higher air side pressure requirements, which would have significantly increased air heater leakage and required air heater reconstruction, and 2) the desire to reduce the flue gas exist temperature from the air heater to improve boiler efficiency. The new air heater being installed is a three-gas pass, one-air pass arrangement. In addition to the new air side pressure and gas exit temperature requirement, the new air heater arrangement has to fit in the existing space available, and the gas side pressure drop has to be such that the existing ID fan static capacity will not be exceeded, since it has to be reused. Because of the higher air side pressure requirements, the existing FD fan and drive will have to be replaced. A new single centrifugal type fan will be installed. The existing two overfire-air fans and cinder return air fan will be eliminated. The existing overfire air ductwork and ports and boiler-hopper-cinder return system will, however, be reused. The air for these systems will be taken directly from the new secondary air system, with a air being provided by the new FD fan. Boiler Circulation Pumps Since the new fluidized bed combustor enclosure walls, floor, and in-bed boiling surfaces are mostly horizontal, water circulation through these circuits must be pump-assisted. Three, 50% capacity, wet motor-type pumps will be installed to pump these circuits. Only the new fluidized bed combustor water circuits, all of which connect directly to the existing furnace wing walls, will be pumped. All the remaining furnace enclosure walls and boiler bank will stay in natural circulation. Bed Ash Removal System To remove bed material, seven letdown systems consisting of individual drain points, downspouts, valves, and ash coolers will be installed. Since sulfur capture is not a requirement, a relatively small amount of sand will be used to maintain a bed instead of the larger quantities of limestone normally required for most fluidized bed boilers. Because of this, the bed drain rates will be relatively low. However, of major concern with this retrofit, is the removal of oversized bed material, particularly the "ege" type clusters. These "eggs" were discovered during the test burn of the Beulah lignite. The number, sizing, and location of the bed drain systems to be installed are set by oversized material removal requirements. FLUIDIZED BED DESCRIPTION The fluid bed plan area is approximately 40 ft (12.1m) wide by 25 ft (7.6m) deep. As described earlier, the plan area is limited by the existing unit arrangement. The fluid bed contains both boiling surface and superheater surface. The boiling surface is located in the front of the bed, horizontally positioned, and spans the entire 40-foot (12.lm) width of the unit. The superheater is also horizontal. It too spans the entire 40-foot (12.1m) width of the unit and is located in the rear portion of the bed. Because of the abrasiveness of the selected bed material, all in-bed surface is provided with additional wall thickness. Furthermore, erosion-type shields will be installed in those areas where higher erosion rates are expected. The entire in-bed tube bundle design and tube spacing will be set with adequate clearance to prevent bridging of potential oversized bed material. The entire distributor plate is a water-cooled membrane type. The fluidizing and combustion air is introduced to the bed from the windbox by means of bubble caps through the distributor plate. The windbox located below the water-cooled distributor plate is divided into four main compartments for load control, in which additional compartmentalization is provided to facilitate start-up operations and to allow for partial compartment fluidization. ADDITIONAL CHANGES AND MODIFICATIONS Many other changes and modifications to the existing plant and its system are obviously necessary for this project. Control Systems Modifications A new Network 908 system will be used to control the major portion of the boiler. This system alone will be a great improvement over the present pneumatic control. Sand Handling System A new handling system is being installed for the sand bed material. This system consists of a storage area, a 10-ton (9072 kg) feed hopper, from which the material will be fed through a lump breaker into a 10-ton per hour (9072 kg/hr) natural gas-fired drier. From this drier, the sand will be conveyed into a 2-ton (1814 kg) surge hopper and transported from that point by dense phase system to two 30-ton (27,216 kg) storage silos. The sand will then be fed from the yrage silos through gravimetric feeders which will itrol the flow of the material to the fluidized bed. Bottom Ash System The bottom ash will be conveyed through seven bed drain water-cooled screw conveyors. These continuously running screw conveyors will discharge the bed ash into two bed ash collection screw conveyors which will discharge into a bed ash transport flight conveyor which will, in turn, dis- charge to a 32-1/2-ton (29,480 kg) storage hopper. From the storage hopper, the ash will be conveyed into the existing ash handling system. New Coil Air Preheater With the addition of a new FD fan, a new coil air preheater will be installed. This new coil air preheater will be connected to the existing glycol heating system. Electrical The electrical system modification necessary for the retrofit includes a 15 kV switchgear with circuit breaker and protective relays for the power supply to the new FD fan motor. Other modifications include: a new motor control center for the boiler circulating pumps, the ash cooler drives, the ash conveyor drives, the sand transport system and the motor operated dampers for the air supply ducts. An uninterruptible power supply system will also be installed for the new control system. Additional lighting around new equipment will be required. Structural Steel and Foundations When possible, the present foundations and structural steel will be used, although some new foundations for the new tubular air heater and for the fluid bed will have to be added. The new F.D. fan will be located on the original stack foundation. PROJECT SCHEDULE The following are the milestone dates of the project schedule: Contract Award January, 1986 Begin Required Demolition of Existing Unit October, 1986 Start of Material Shipment November, 1986 Hydrostatically Test Unit Early February, 1987 First Coal Fire in Retro- fitted Bubbling Fluid Bed Combustor Late February, 1987 BENEFITS The fluidized bed retrofit will greatly improve boiler efficiency and operation. The lower temperature at which combustion will occur will greatly reduce the fouling and slagging. The unit originally was load limited due to furnace slagging and fouling of the convection pass when lignite was fired on the grate. With the retrofit, the capacity of the unit will be increased from 50 MW to 80 MW in continuous generator output. Another benefit of elimination of fouling and slagging will be increased boiler availability. Decreased availability resulted from a unit shutdown twice a year to remove slag by water washing the convection pass, Due to the effects of fouling and slagging, the existing air heater, exit gas temperature is approximately 70°F (21°C) greater than design. The gas temperature will be reduced to the original design values as a result of elimination of the fouling and slagging. This drop in exit gas temperature will increase the efficiency of the unit by slightly less than two percentage points. The fluid bed retrofit will allow the continuous use of the local Beulah North Dakota lignite to 700,000 1b/hr (317,515 kg/hr). In summary, Montana-Dakota expects the unit's thermal efficiency, availability and capacity to be increased by retrofitting the steam generator with a bubbling fluidized bed combustor while continuing to burn a locally available fuel. REFERENCES l. G. M. Goblirsch, S. A. Bensor, D.R. Hajicek, and S. L. Cooper, "Sulfur Control and Bed Material Agglomeration Experience in Low-Rank Coal AFBC Testing." Proceedings of The 7th International Conference on Fluidized Bed Combustion, Philadelphia, Pennsylvania, October 25-27, 1982. 2. J. N. Duqum, J. L. Esakov, and W. C. Howe, "AFBC Performance Comparison for Underbed Feed Systems." Proceedings of The 8th International Conference on Fluidized Bed Combustion, Houston, Texas, March 18-21, 1985. THE AMERICAN SOCIETY OF MECHANICAL ENGINEERS 86-JPGC-FACT-G 345 E. 47 St., New York, N.Y. 10017 The Society shall not be responsible for statements of opinions advanced in papers or in discussion at meetings of the Society or of its Divisions or Sections, or printed in its publications. Discussion is printed only if the paper is published in an ASME Journal. Papers are available from ASME for 15 months after the meeting. Printed in USA. nnn neu ee ee | Development of a Retrofit Low Nox Cell Burner M. J. Clark A.D. LaRue Fossil Operations Division Babcock & Wilcox Barberton, Ohio A. D. Liang Research & Development Division Babcock & Wilcox Alliance, Ohio D. Eskinazi Electric Power Research Institute Palo Alto, California ABSTRACT develop a low-cost combustion system specifically for coal-fired, cell burner-equipped boilers that Pulverized coal-fired cell burner units which will accomplish these goals. comprise over 26,000 MW of generating capacity are located primarily in the Northeast and Midwest United States. These units generate almost 15% of the pre-NSPS utility NOx emissions, and may be likely targets for eventual retrofit NOx control. The Babcock & Wilcox Company (B&W), with support from the Electric Power Research Institute (EPRI), has designed and tested, at two pilot scales, an advanced low NOx burner applicable to existing pre-NSPS wall-fired boilers equipped with B&W's standard two-nozzle cell burners. CELL BURNERS Economic considerations, which dominated boiler design in the 1960s, led to the development of the cell burner, Two-and three-nozzle cell burners were designed to burn coal efficiently in relatively tight (high input per unit volume) utility boilers. The tight spacing, plus rapid mixing of the cell, minimized the flame zone while maximizing the heat release rate. This paper addresses two-nozzle cell burners Results of both pilot test programs proved that since they account for 90% of the the new design could achieve a NOx emission cell-burner-equipped boilers. The concepts reduction of greater than 50% as compared to the developed here for low-NOx combustion would, in standard cell burner. The new design achieved this principle, apply to three-nozzle cells. For reduction with little change in boiler performance two-nozzle cells, shown in Figure 1, two and in a configuration which would require no circular-type burners were stacked vertically to pressure part modifications. In addition to pilot comprise one cell. As is evident in the figure, scale tests, a full-scale low NOx cell burner at the nozzles (throats) of a cell are nearly on Dayton Power & Light's (DP&L) Stuart Station No. 3 tangent with each other. has been firing successfully for over one year, and continues to be evaluated. Future plans include a Approximately 35% of total pre-NSPS pulverized complete retrofit of a full-scale unit to coal wall-fired capacity is generated by two-nozzle demonstrate the NOx reduction capability. cell burner units representing over 26,000 megawatts of capacity. All of the 38 units are opposed INTRODUCTION wall-fired with two rows of burners, and have an average size of 699 MW. A typical unit is shown in In the past several years, much attention has Figure 2. been devoted to the issue of acid rain. Oxides of nitrogen emissions are believed to be contributors to the acid rain problem. Boilers sold after 1971 are currently required to achieve lower NOx emissions due to New Source Performance Standards (NSPS). However, boilers sold before 1971 comprise the majority of pulverized coal-fired generating capacity (1). These pre-NSPS boilers may be the targets of legislation designed to reduce NOx The unique configuration of the cell burners and the tight furnaces directly results in high NOx emissions. Typically, these units operate in the range of 1.0 to 1.8 lbs per million Btu (430-770 ng/J). Regulations are expected to eventually mandate reduction of NOx from these older units and have motivated development of retrofittable combustion technology hardware. emissions and, therefore, are the focus of programs PROGRAM OBJECTIVE AND OVERVIEW aimed at developing combustion systems to alleviate these emissions as effectively and inexpensively as The primary objective of the program was to possible. EPRI and B&W contracted in 1983 to develop a retrofit low NOx technology for pulverized Presented at the Jt. ASME/IEEE Power Generation Conference Portland, Oregon — October 19-23, 1986 BR-1302 Figure 1 Standard two nozzle cell burner. [ Headers —.| i Risers. Cell =) Cell Burners al Te Burners Windbox Figure 2 Typical cell burner arrangement in utility boiler. coal-fired cell burners. Specifically, the goal was to obtain a 50% reduction in NOx by employing a retrofit which would have minimal impact on the boiler and its operation. To accomplish this goal, several tasks were performed over a three-year period: (1) pilot combustion tests were conducted at two scales to develop the best equipment for NOx reductions, (2) mechanical verification tests were conducted at Dayton Power & Light's Stuart Station on a single full-scale low NOx cell burner, (3) furnace corrosion potential was evaluated based on laboratory-scale corrosion experiments and in-furnace gaseous species measurements taken duri the combustion tests (4) numerical analysis to predict full-scale performance, and (5) a feasibility study to assess economic and technical risks. This paper summarizes the pilot combustion tests and full-scale, single low-NOx cell evaluation. Furnace corrosion evaluation and numerical analysis are reported elsewhere (2,3). PILOT COMBUSTION DEVELOPMFNT TESTS The objectives of the overall combustion development task was to develop a low NOx cell burner and characterize the burner performance with respect to NOx emissions, combustion efficiency, fuel adaptability and impact on boiler performance. This task was performed in three steps: ° Small scale burner screening test: Two low NOx cell burner designs were tested and compared with the standard cell burner in the six-million Btu/hr combustor located at B&W's Alliance Research Center. The best low NOx cell burner design was selected for further development (1). ° Small scale burner characterization test: The selecged low NOx cell burner was further tested at 6 x 10° Btu/hr. Burner adjustments and parametric variations were conducted to investigate the sensitivity of the burner design to NOx emissions and combustion efficiency under normal an4 staged combustion conditions (1). ° Large scale burner characterization test: The low NOx cell burner performance was characterized in a 100 x 10° Btu/hr test furnace to provide scale-up information and further verification of the burner design. Small Scale Screening and Characterization Tests Test Facility. The Combustion and Fuel Preparation Facility (CFPF) at the B&W Alliance Research Center was used for the small-scale combustion tests (Figure 3). The furnace involved was designed for six-million Btu per hour heat input with coal, oil or gas firing. The vertically-designed, wall-fired furnace measures 4.5 feet wide x 6.0 feet deep x 14.0 feet (1.4 x 1.8 x 4.3 m) high and models the geometry and residence times of B&W commercial boilers. The inside surface of the furnace is insulated to yield a furnace exit gas temperature of 2300°F (1, 260°C) at the design input. The location of the burners, sampling ports, air staging (OFA) ports, and slag panel is shown in Figure 4. The ports were designed to accept inserts to vary momentum ratio and were located vertically at two, three, or four burner spacings above the top burners. One set of front and/or rear wall ports could be used simultaneously. A Beckman gas analysis system provided continuous monitoring of the stack 0,, NOx, CO and CO, concentrations. The analyzers were calibrated datly with standard gases, and gas sampling was Elevator ~— for Crushed Coal System Bag Filter Iverizer oar Crushed Coal Bin and Feed System Pulverized Coal Combustion Bin and Feed Furnace Figure 3 Combustion fuel preparation facility. YLLLLLIL MLE Le hy A Gas Sampling ® rt ay x Insulation XN 4; 3rd OFA Level rt Observation Door ® ee @¢ 6 AO SSS t I et +] 2nd OFA Level F £ = ba ey a f Ist OFA Level a Com ae oA si Ll | || Panel Cell Burners SASS Hopper Figure 4 CFPF test furnace. performed on a continuous basis. Detailed gas temperature, gas species, and unburned carbon measurements were taken during selected combustion tests. The CFPF furnace was directly fired with an MPS-32 pulverizer, while secondary air was metered and controlled separately to each burner. A slag panel was designed and positioned on the CFPF rear wall to provide information relative to changes in deposition rate and deposit chemistry for the different burner and test conditions. The panel was constructed of materials normally used in boiler tube walls, e.g. Croloy 1/2 tubing with membrane wall construction. The panel was cooled by a low Pressure coolant system which used Dow-Therm as the working fluid. The slag panel and circulation system were instrumented such that temperatures at the tube wall were maintained in the 700-800 F (370-430°C) range, similar to field performance. Enclosure for Fuel Systems Secondary air Air disk Adjustable vanes 54— air only Spin vane and coal assembly Coal deflector Conical diffuser Adjustable air zone sleeve Adjustable inner zone spin vanes (gear operated) Coal impeller Figure 5 Low NOx cell burner No. 1. Adjustable outer air zone shroud Adjustable inner Primary air \ e air zone disk and coal \ Outer Pe zone register assembly i fe Primary air and ‘i Coal deflector Conical diffuser Adjustable inner zone spin vanes Secondary air Figure 6 Low NOx cell burner No. 2. Burner Designs. Burner screening tests were performed with three different burners: standard cell burners for baseline comparison, and two candidate low NOx cell burners (Low NOx No. 1 and Low NOx No. 2). One low NOx cell design criterion stipulated that the new burner be retrofittable without furnace waterwall pressure part modifications, Low NOx No. 1, as shown in Figure 5, was designed to fire all the fuel through an enlarged coal pipe in the lower throat with a portion of the secondary air through the upper throat. The enlarged coal pipe was sized to maintain the original cell firing rate. In principle, this provided each cell with an integral NOx port. The elements of the lower throat assembly closely resemble the B&W "S" burner, a commercial design used for mechanical enhancements without NOx reduction. Low NOx No. 2, as shown in Figure 6, consisted of two modified Low NOx Dual Register Burners (DRB) sized to fit the cell location. This sizing forced higher-than-normal velocities for the Dual Register Burner. Each DRB consisted of an axial coal nozzle with a conical diffuser used to disperse the coal prior to injection. Burner Configuration. In the small scale combustor, the CFPF, the burner configuration included two cells, one vertically above the other. Test Coal. A fuels study was conducted to select fuels for the combustion tests that would bracket the key characteristics of coals in use in the field. Key factors were identified as volatile matter and the fixed carbon/volatile matter (FC/VM) ratio. The two coals selected for testing, based on these parameters and field usage, were Ohio No. 6 and Lower Kittaning. The Ohio No. 6 represented the high range of volatiles in the population and a lower FC/VM ratio. Lower Kittaning is representative of lower volatility with higher FC/VM ratio. Fuel nitrogen on a Btu basis was similar. Table 1 presents the average fuel analyses from the CFPF screening and characterization tests. Screening Test Results. Screening tests provided performance comparisons for the standard cell, Low NOx Cell No. 1 (LN1) and Low NOx Cell No. 2 (LN2) while firing Ohio No. 6 coal in the CFPF at several loads and excess air levels. The performance was judged relative to the standard cell, as well as to field data and previous low NOx burner (DRB) tests in the CFPF. The standard cell and DRB produced emissions in the CFPF which were in the range of their respective field results. In these tests, coal impellers were used in the standard cell but were not used in LN1 or LN2. NOx results for the standard cell at full load (5.6 million Btu per hour) and 3% 0, were 1.4 lbs NO Table 1 Coal analysis Ohio #6 Coal Lower Kittaning Raw Coal Basis As Received As Received Total Moisture, % 10.03 3.66 Proximate Analysis, % Moisture 10.03 3.66 Volatile Matter 38.83 22.28 Fixed Carbon 44.15 63.31 Ash 6.99 10.75 Gross Heating Value Btu per Ib 12015 13330 Btu per Ib (M&A Free) were Ultimate Analysis, % Moisture 10.03 3.66 Carbon 66.74 75.58 Hydrogen 465 4.42 Nitrogen 1.24 1.20 Sulfur 2.16 1.82 Ash 6.99 10.75 Oxygen (difference) 8.19 2.57 Total 100.00 100.00 Fixed Carbon/Volatile Matter Ratio 1.14 2.84 Nitrogen Ib/10® Btu 1.03 0.90 per million Btu (600 ng/J), while LN2 produced 0.7 lbs NO per million Btu (300 ng/J) and LN1 produced 0.35 lbs NO per million Btu (150 ng/J) (Figure 7). NOx sensitivity to excess air was relatively low for all three burners. CO emissions were very low for all three burners (less than 50 ppm) and comparable to cell burner field performance. Unburned carbon loss as shown in Figure 8 was low (less than 0.2%) for the standard cell and LN1. LN2 exhibited an increase in unburned carbon emissions. Several other performance criteria were also evaluated. No changes in furnace exit gas temperature were measured for LN1 and LN2 in comparison with the standard cells. Flame stability at lower loads was best with the LNI burner. Flame length for the standard cell was about 3 feet (0.9 m) at full load, 3% 0,, with Ohio No. 6 coal. Both low NOx burner flames’ impinged on the back wall in the 6-foot (1.8 m)deep furnace. This raised concerns about slagging and corrosion. Burner adjustments were somewhat effective in shortening flame length but with some sacrifice in NOx reductions. Based on these screening tests, LNl, the most promising design, was selected for further development. NOx reductions of greater than 50% were achieved. The characterization tests included a wider range of conditions than the screening tests, Low NO, No. 2 Burner —_-—- -_—- _——_—-—- * 5° Impeller No. 1 Burner with azo \mpe' NO,, [b/108 Btu 3 3.5 4 45 Excess Oxygen, % Figure 7 Comparison of NOx levels for all burners. Unburned Carbon, % Excess Oxygen, % Figure 8 Comparison of carbon loss for all burners. evaluation of a second coal, and reproducibility tests. A major objective was to shorten flame length for this burner to produce compatibility with furnace depths in the field. This was achieved by iddition of a 25° coal impeller at the burner tozzle, The impeller shortened the flame to about 5 feet (1.5 m) at the same conditions, at full load with 3% 0. NOx correspondingly increased to 0.5 lbs per million Btu (215 ng/J) which was an acceptable trade-off. Figure 7 shows the comparison with other burners. CO and unburned carbon did not change with the impeller. Burner adjustment effects on NOx and flame stability were examined and indicated that the burner was stable under nearly all combinations of settings. NOx was insensitive to upper and lower slide damper positions. Emissions increased 50 ppm as swirl was increased by spin vane settings. Adjusting air vanes in the upper throat were directed from 20 degrees up, to parallel with the flow, resulted in a 150 ppm increase. Numerous tests were conducted under staging conditions to evaluate low NOx cel] performance with overfire air. NOx reductions of up to 90% (relative to the standard cell) were achieved with source flame impingement. Since the original project goal of 50% reduction was achieved with burners alone, staged combustion was not pursued. Large Scale Low NOx Cell Burner Characterization Test Test Facility. The EPA's Large Watertube Simulator (LWS) located at the Energy and Environmental Research Corporation's (EER) Irvine, California facility was used in the large scale burner test. It was a 100-million Btu per hour combustor test facility. The furnace was designed to match the size and geometry of a large industrial or small utility single wall-fired furnace. Figure 9 shows its general arrangement. 1. Coal Storage 2. Main Coal Hopper 3. Crusher 4. Coal Feeder 5. Pulverizer 6. P.A. Exhauster 7. Main Combustion Air Fan 8. Tubular Air Heater 9. P.A. Trim Heater 10. P.A. Venturi 11. Oil/Gas Fired Preheater 12. Air Plenum The LWS furnace was 22 feet (6.7 m) deep and 16 feet (4.9 m) wide. The overall height was 50.5 feet (15.4 m) from the hopper to the top. The furnace was front-wall-fired with the nose directly above the rear (or target) wall. It was externally spray-cooled with water to absorb the heat of combustion and control furnace temperature. The vertical walls of the furnace in the vicinity of the burner were insulated with refractory so that furnace internal temperatures were similar to those of field operating boilers. The furnace exhaust system was designed to permit a wide range of emission measurements. The sampling location was near the end of a long straight duct meeting EPA specifications for the minimum number of sampling points. Most of the measurements were made continuously and results were processed into engineering units in real-time by a microcomputer. Commercial gas analyzers were used for measurements of 0,, co, CO, NO/NOx and S0,. Unburned carbon emisstons were measured during selected tests, Measurements of gas phase species were made at several locations in the lower furnace and along the wall near the burners to determine if reducing conditions existed for selected test conditions. The probing locations inside the furnace are shown in Figure 10. Slag panels were also installed during the detailed furnace probing tests. The panels were located on a sidewall and one on the rear wall. Burner Design and Configuration. The low NOx cell burner No. ] was scaled to test in the LWS furnace and compared with the standard cell burner. The cell burners were arranged in a ? x 2 array in the LWS to account for side by side interactions between the burners. Each cell was designed for 25-million Btu per hour of heat input. The vertical and horizontal spacing was scaled directly from a cell-equipped utility boiler. 13. Combustion Air Venturis 14. Coal Splitter 15. Cell Burners/Windbox 16. LWS 17. Exhaust Duct 18. Spray Tower Scrubber Figure 9 Large watertube boiler simulator facility. 5 Test Coals. Ohio No. 6 and Lower Kittaning coal were tested in the cell burners. Table 2 presents the average coal analyses from the LWS test. An extensive test series was conducted in the LWS to characterize the low NOx cell burner performance. The test series demonstrated the sensitivity of the burner to a number of its adjustable parameters: coal impeller design, spin vane position, lower damper position, directional vane position and upper damper position. Similar effects were obtained for each configuration tested in the LWS, which are summarized below: ° Impeller Design: The design of the coal impeller was a critical parameter in determining flame shape and operating characteristics. The impeller with a 50 degree included angle tended to produce shorter flames with correspondingly higher NOx emissions than the 25 degree impeller. The steeper angle was more effective in dispersing the coal into the swirling combustion air stream. Impeller position had no significant effect over the range investigated. ° Spin Vane Position: The spin vane position was the dominant parameter used during this test program to control the air flow distribution between the upper and lower throat of each cell. Closing the spin vanes and increasing the angle to the flow not only increased the degree of swirl but also increased the pressure drop across the lower throat. This forced more air to be diverted to the upper throat, the integral staged air port. This reduced the burner zone stoichiometry and thus decreased NOx emissions. ° Lower Damper Position: The lower damper adjustments were limited during these tests. However, this device did demonstrate that it could provide a similar degree of control on NOx as the spin vanes, but at higher burner pressure drop conditions. ° Directional Vane Position: The upper throat air vanes proved to be an effective tool for flame shaping and NOx control. Diverting the staging air away from the flames tended to decrease NOx emissions and increase flame length. In particular, the upper cell burner directional vanes were most effective. ° Upper Damper Position: Upper damper adjustments were limited during these tests due to excessive windbox-to-furnace differential when closed 50% or more. Closing the upper damper, in any case, was not desirable from an emission standpoint. By increasing the restriction of air through the upper throat, the burner zone stoichiometry increased as did NOx emissions. The burner variables evaluated during these tests produced a wide range of NOx emissions and flame lengths. A close relationship of NOx to flame length was developed for full load operation at an overall stoichiometry of 1.17. The NOx varied from about 820 ppm with a flame length of 10 feet (3 m) with the standard cell down to a minimum of 205 ppm with a flame length of 22 feet (6.7 m) with the low NOx cell. The flexibility of the low NOx burner design, however, allows adjusting burner performance Port “A” ——+ Port “B” Furnace | | Burner | 1-Plate Oil Burner Figure 10 General locations for in-furnace probing. Basis Proximate (wt %) Moisture Ash Volatile Fixed C Heating Value Btu/Ib MMF Btu/Ib MAF Btu/Ib Ultimate (wt %) Moisture Carbon Hydrogen Nitrogen Sulfur Ash Oxygen* Forms of Sulfur (wt %) Sulfate Pyritic Organic Table 2 Composition of test coals Ohio #6 Coal As Received 2.37 10.48 42.35 44.82 12,860 2.37 69.12 5.09 1.18 4.82 10.46 6.96 0.17 2.49 2.16 *Oxygen determined by difference Lower Kittanin; Raw Coal —_— As Received ——— 1.04 15.12 23.73 60.11 12,959 1.04 72.34 4.37 1.12 3.80 15.12 2.21 0.43 2.42 0.95 1000 7 800 —. . E a = 600 oO & m~m % 400 BF x So 2 200 Coal: Ohio No. 6 Excess O, (%) r > = Oo N S p ° iy NO, (Ib/10® Btu) 100 @ 48 Standard 0 40 Optimum Low NO, ©@ 100 MKB x 10° Btu/hr 80 44 75 MKBx 10° Btu/hr E O® 60 MKBx 10° Btu/hr a = 60 Qo o 7 ° 5 A m~m = 40 Oo g 9 g 20 a Excess O, (%) Figure 11 Comparison of emissions from the standard cell burner and optimum low-NO x design with Ohio No. 6 coal. 1000 i on = a = 600 S x ~m 400 oa y o 2 200 Coal: Lower Kittaning Excess 0, (%) 1.4 = ny tant o S © NO (Ib/10® Btu) o a 100 @ B Standard © O Optimum Low NO, O @ 100 MKB 80 O @ 60 MKB 60 40 CO at 3% O, (ppm) 20 Excess O, (%) Figure 12 Comparison of emissions from the standard cell burner and optimum low-NOx design with Lower Kittaning coal. to the application. Lower NOx emissions can be achieved if long flames can be accommodated in the retrofit application. An up to 75% reduction in NOx emissions was achieved by the low NOx cell burners with the 25 degree impeller. The flame length for this minimum NOx condition was about twice that for the standard cell burner. The optimum configuration from this test series was with the 50 degree coal impeller. NOx reduction of 55% was achieved with a 60% increase in flame length. Flame stability was very good over expected load and excess air ranges. The emissions performance characteristics were compared for the optimum low NOx cell burner and the standard cell burner as shown in Figures 11 and 12 for Ohio No. 6 coal and Lower Kittaning coal, respectively. The emission data are presented as a function of excess air for several firing rates. Although the NOx emissions were about 55% lower with the low NOx cell burner, the effect of excess air was similar for the two burners with each coal. The effect of firing rates, shown explicitly in the two figures, was substantially different. The NOx emissions from the standard cell burner were much more sensitive to firing rate than the optimum low NOx cell burner. Combustion efficiency for the two burners was compared, At nominal full load, 100-million Btu per hour, flyash carbon content was considered exceptionally good with both coals for the two burners, compared to previous LWS burner tests. Comparison of field and LWS test results in previous programs has verified the similarity of performance in the key areas of flame shape, stability, and operating range. Therefore, scale-up of the low NOx cell burner should result in 3AIUT&B Hine p-------- L-_ Figure 13 Location of low NOx cell burner on front wall. (Viewed from outside furnace) comparable performance in the field. However, NOx emissions are generally lower in a research furnace like the LWS than with operating boilers, primarily due to thermal environmental difference. The percent emissions reduction with the low NOx cell burners in an operating boiler and in the LWS, having comparable conditions, are anticipated to be similar, FULL SCALE EVALUATION In March of 1985, a full-scale two-nozzle cell burner was replaced with a low NOx cell at Dayton Power & Light's Stuart Station No. 3. This burner has now been in operation for over one year. Evaluations of the burner include mechanical reliability, corrosion checks of furnace tubes surrounding the cell opening, visual observations of flame, air flow measurements of secondary air, and temperature measurements of burner components. All the mechanical components of the burner have operated properly throughout the past year and are staying within material temperature limits. In a low NOx scheme, where combustion is staged, reducing atmospheres will be present, possibly accelerating corrosion rates on furnace wall tubes. Because of the concerns about corrosion, ultrasonic (UT) measurements of furnace tubes were taken to document metal thickness before installation and after one year of operation. This unit has had a history of tube wastage in the furnace due to reducing atmospheres, especially along the side walls. The location of the burner with respect to the furnace (Figure 13) coupled with the design of the windbox, causes this burner to receive less than its share of combustion air. The air flow measurements (30 point impact-suction pitot grid, Figure 14) indicate that it is operating at a stoichiometry of 0.81, a worst case scenario for corrosion. The UT measurements performed after one year by DP&L's own nondestructive testing team showed that virtually no tube wastage had taken place, thus DP&L felt no steps would need to be taken to prevent corrosion due to the burner. Visual observations of the burner array at the Stuart Station are difficult since the wraparound Thirty Point Sampling Grid Air Sleeve (6" Into Air Sleeve - Typical) Coal Nozzle Figure 14 Air measuring device. windbox does not allow conventional observation ports on the side walls in the combustion zone. However, observations from the upper furnace and through the burner barrel indicate that the low-NOx cell flame is stable and originates at the throat; consequently, a specially designed side wall port is being installed through the windbox at a location close to the low NOx cell. Utilizing a fiber optic probe through this port, better observations can be made of the flame. ENGINEERING FEASIBILITY Included in the program was a study of the technical feasibility of retrofitting low NOx cells on an entire unit equipped with cell burners. This study looked specificelly at DP&L's Stuart Station as the candidate unit for a full retrofit since it was very representative of cell burner units. The average cell-equipped unit is 699 MWe and has 24 burners. (Stuart Station is 610 MWe and has 24 burners.) The study considered performance, operations and maintenance with the proposed trofit of the low NOx cell. Of utmost importance to any utility is the ability of its boilers to convert heat available in the fuel to steam, Since carbon contributes the great majority of heat in coal, its utilization must be maximized, From two-scale combustion tests performed on this burner, high carbon utilization is anticipated in field applications. The main goal of the new burner is to reduce NOx by at least 50%. At both pilot scales, this level of reduction was achieved with acceptable increases in flame lengths. A field unit could expect a reduction in NOx emissions of the same magnitude with this new burner. Combustion tests also showed no appreciable change in furnace exit gas temperature. Regarding increased corrosion potential, based on UT measurements taken around the low NOx cell at Stuart Station and gaseous species measurements in the pilot combustion tests, no corrosion problem would exist with a full retrofit. Therefore, maintenance costs should not increase. If the flame of a low NOx burner is too long, it may impinge on the furnace walls, causing slagging and possible corrosion. Based on the combustion tests, flame lengths on a full scale boiler are expected to be easily accommodated within the confines of the furnace. Another consideration when retrofitting new irners is the added resistance of the burners. The «eserve fan capacity of a a unit must be able to Overcome the added pressure drop and maintain the quantity of air flow desired. The low NOx cell does have higher resistance than the standard cell, requiring a 2-3 inch W.G (51-76 kg/m). additional Pressure, When considering a total drop for the secondary air system of a 20-inch W.G. (510 kg/m) or more, the additional resistance should be well within the range of capacity for the fan. In addition, these pressurized furnaces have low windbox-to-furnace differentials leading to poor air distribution, The higher resistance burners will serve to improve distribution, which may aid corrosion prevention, and improve overall combustion, The technical feasibility of retrofitting a cell burner unit with low NOx cells is very high. Further, considering the design and age of pre-NSPS burners in general, retrofitted low NOx burners could represent a reduction in maintenance costs. CONCLUSION Boilers equipped with cell burners have a unique problem. The closed-space design of the highly turbulent burners does not lend itself to the conventional low NOx technology available today. Technically, this problem has been solved with the burner developed in this program, the low NOx cell. The result of this program is a burner capable of significant NOx reduction without requiring pressure part modifications. Tests at two scales indicate that over 50% NOx reductions have easily been achieved with the low NOx cell relative to the standard cell burner, In addition, flame length increases have been qualified - i.e. for a 55% reduction in NOx, a non-impinging flame is easily anticipated. A full scale demonstration of the burner's capabilities is now required. EPRI is currently seeking host sites for this demonstration. Upon successful completion of an entire full scale retrofit, a commercial product will exist to reduce oxides of nitrogen emissions from power plants equipped with cell burners. ACKNOWLEDGMENTS The authors would like to acknowledge the assistance and effort given by the Energy and Environmental Research Corporation in the large scale combustion test. Specifically, we thank Messrs. A. Abele and Y. Kwan for the conduction and supervision of the test. The authors also acknowledge Dayton Power & Light Company for participating in the full scale burner evaluation. In particular, M. Linsberg deserves a special note of thanks for his assistance, REFERENCES 1. A, D. LaRue, and L. W. Rodgers, "Development of Low NOx Cell Burners for Retrofit Applications", 1985 Symposium on Stationary Combustion NOx Control, EPRI Report CS - 4360, Volume 1, January 1986, page 17-1. 2 S. F. Chou, P. L. Daniels, L. W. Rodgers, G. T. Theus, and D. Eskinazi, "Fireside Corrosion in Low NOx Combustion Systems", 1985 Symposium on Stationary Combustion NOx Control, EPRI Report CS - 4360, Volume 1, January 1986, page 19-1. Se "Development of a Retrofit Low NOx Cell Burner", Final report being prepared, EPRI Research Project 2154-7. | Technical Paper Design and BR-1304 performance of 80 MW bubbling bed retrofit at Montana- Dakota Utilities Co., R. M. Heskett Station D. H. Roy R. L. Gorrell Babcock & Wilcox Barberton, Ohio B. Imsdahl Montana-Dakota Utilities Co. Bismarck, North Dakota Presented to American Power Conference Chicago, Illinois April 27-29 1987 Babcock & Wilcox a McDermott company Design and performance of 80 MW bubbling bed retrofit at Montana- Dakota Utilities Co., R. M. Heskett Station D. H. Roy R. L. Gorrell Babcock & Wilcox Barberton, Ohio B. Imsdahl Montana-Dakota Utilities Co. Bismarck, North Dakota Presented to American Power Conference Chicago, Illinois April 27-29, 1987 Introduction Montana-Dakota Utilities Co. (MDU) has retro- fitted Unit 2 at its R. M. Heskett Station near Mandan, North Dakota (Figure 1) to fire North Dakota lignite in a bubbling fluidized bed com- bustor (FBC). When placed into service in 1963, the unit was the world’s largest spreader stoker with a moving grate area of 969 ft?. The boiler was rated at 650,000 lb/hr, 1300 psig, and 950°F main steam temperature. The nominal rating of the General Electric turbine at the unit is 81.2 MWe at a steam flow of 682,700 lb/hr. Lignite for the station comes from the South Beulah, ND, mine of the Knife River Coal Mining Co. The fuel is typically high moisture (37.4%), Figure 1 R. M. Heskett Station, Mandan, ND. PGTP-87-18 low sulfur (0.9%), high sodium (9% of ash) fuel that resulted in excessive slagging and fouling of heat transfer surfaces during its years of opera- tion as a stoker- fired unit. As a consequence, combustion efficiency was low when the unit operated at or near original rating (1). Several types of fuel additives were tried over a period of many years but were not successful in reducing the slagging and fouling problem for long periods. Since installation of sootblowers was impractical due to close tube spacings, water washing of the unit was required on the average of twice a year. In addition to these concerns, exit gas tempera- ture from the original regenerative air heater was approximately 70°F greater than design, and the original unit was limited to approximately 50 MW in order to sustain operation at acceptable limits over long periods of time. The primary purpose of the bubbling bed retro- fit is to upgrade boiler capacity to take full advan- tage of turbine capability and to increase boiler efficiency. With the bubbling bed combustor, the unit will deliver 700,000 lb/hr steam at 1300 psig, 955°F steam temperature at the superheater outlet. Feed water temperature at the economizer inlet is 443°F. Power output will increase to 80 MW and design boiler efficiency to 80.3%. The boiler retrofit project consisted of remov- ing the stoker and fitting the boiler with a lower enclosure, air distributor floor, windbox, and inbed heat transfer surface. To minimize overall costs of the project, including those of demolition and erection, modifications to the original unit were kept to a minimum. This objective, however, imposed several major constraints on the design and arrangement of the fluidized bed retrofit. The existing coal feed system, most of the existing pressure parts, and the existing structural steel and foundations were to be used with little or no modification. In addition to the boiler retrofit, the plant’s pneumatic boiler control system was upgraded to a modern distributed control system; a new FD fan and tubular air heater were installed; and pumps were installed to provide forced circulation cooling for the new enclosure and inbed surfaces. The contract for the retrofit was awarded in January 1986, and demolition began October 1986. Hydrostatic testing of the modified unit was completed on February 18, 1987, with start-up activities currently in progress Atmospheric bubbling bed combustion An atmospheric bubbling bed is an ideal solution for the problems experienced at R. M. Heskett Station Unit 2. A fluid bed combustor uses air to fluidize a bed consisting of a mixture of fuel and either inert material or reactive sorbent material such as limestone or dolomite. The amount of fuel in the bed at rated power is very small, generally less than 2% of the total bed solids; air velocities are typically 6-12 ft/second. In the fluidized state at MCR, bed density will be approximately 40-50 lb/ft3. When fluidized, the bed looks much like a pot of boiling fluid with substantial intermixing of solids and gases in the bed. Heat transfer sections are immersed in the bed to control com- bustion temperature and to provide for steam generation and superheating. Combustion temperatures are normally main- tained in the range of 1500°F-1600°F. In this range, thermal NOx production is insignificant, and NOx formation is dependent upon the nitro- gen content of the fuel and the staging and mixing of the air in the furnace. For most ap- plications, NOx production is well below Federal emission standards. The bubbling bed operating temperature range is also near optimum for removing sulfur con- tained in the fuel with sorbents such as limestone or dolomite. The bubbling bed boiler permits excellent intermixing of solids and gases which facilitates efficient use of the sorbent and results in high levels of sulfur removal (90%), thus, strin- gent limits on NOx and SOx production can be met without the use of backend flue gas desulfuri- zation equipment (2). The bed material to be used in the MDU retrofit is sand. A sorbent bed material is not required since sulfur content in the lignite fuel is very low (0.9%/wt). This means that concerns associated with the efficiency of the sorbent-sulfur reaction are not an issue for this project. SOx emission, however, will be reduced since approximately 40% of the sulfur will be removed due to the scrubbing action of alkalines in the ash. This sulfur-scrub- bing ability was confirmed in a test burn of the fuel conducted in B&W’s 7 MWt pilot unit at the Alliance Research Center in Ohio (3). Flue gases from the combustion process pass out of the furnace through convective heat transfer surfaces that provide for additional steam superheat and for raising the temperature of entering feedwater. An existing multiclone dust collector will continue to be used for first-stage particulate control. Since the carbon content of the collected ash is expected to be very low, this ash will be removed by the plant ash handling system and not reinjected to the furnace. The combustion temperature chosen for the R. M. Heskett Station retrofit is 1500°F. This temperature was selected to minimize bed mate- rial agglomeration caused by the formation of low melting temperature eulectics which can form when burning North Dakota lignite in a sand bed. The retrofit planned for the R. M. Heskett Station is ideal from another vantage point. Fuel for the plant contains a very low fraction of fines (particles ranging in size from 75 microns to 600 microns). This means that overbed fuel feed can be used with a relatively low fraction of combus- tion occurring above the bed (free board burning). As a result, combustion efficiency will not be significantly affected and the existing coal feed system, up to and including the stoker spreaders, can be used as well. In addition, overbed feed is a simpler, more reliable form of fuel entry than an underbed feed system, all of which reduces cost of the upgrade. In summary, the slagging and fouling problem at the R. M. Heskett Station Unit 2 will be signifi- cantly reduced, improving both boiler efficiency and operation. With the addition of inbed surfaces and reuse of existing heat transfer surfaces, power output capability will be increased to approximately 80 MW. Boiler emissions will be reduced due to the low temperature combustion, high residence time, and excellent solids and gas mixing occurring in the bubbling bed. Use of test and demonstration facilities The design of the retrofit for the Montana-Dakota Utilities Co. unit has been directly influenced by operating experience from two pilot units placed in service at B&W’s Alliance Research Center and from the 20 MW demonstration plant which is owned and operated by the Tennessee Valley Authority (TVA) in Paducah, KY. These facilities resulted from a long-term development project jointly sponsored by EPRI, TVA, and B&W and initiated in 1975. The TVA unit was placed into operation in 1982. The fluidized bed combustor used with this unit was designed and built under a turnkey con- tract by B&W. The company also provided boiler controls and other accessory equipment such as fuel and ash handling systems and the baghouse. Experience obtained with this unit significantly affected the design of inbed surfaces, inbed sur- face supports, distributor plate and bubble caps, windbox, and bed drain systems for the MDU re- trofit. Operating experience from this facility also influenced many functional design parameters such as fluidization velocity and bed depth. To help further establish functional and mechanical design parameters, a test burn of the retrofit fuel was conducted at B&W’s 7 MWt pilot unit. This unit is a 6 ft x 6 ft atmospheric fluid bed combustor. Results of these tests were used to set the bed temperature to minimize bed agglomera- tion and to confirm adequacy of overbed feed (3). In addition, extensive testing of lignite from the Beulah Mine was conducted in a fluidized bed combustor by the Energy Research Center of the University of North Dakota (4). Data from this testing was also used to establish design parame- ters for the retrofit. Description of original unit As shown in Figure 2, the general arrangement of the original unit at the R. M. Heskett Station was typical of most spreader-stoker type boilers. The furnace was approximately 40 ft wide and 21 ft deep and contained three water-cooled wingwalls. The furnace wall used water-cooled tube and tile construction with a gas-tight casing. The wing- walls, which were fed by downcomers from the lower drum, penetrated the lower rear furnace wall, rose through the furnace, and connected directly to the upper drum. The convection pass contained superheater, steam generating, and economizer surfaces. The superheater is an all pendant type with spray attemperation between the primary (long flow) and secondary banks (counter flow). The genera- ting surface has a 60-in.-diameter upper drum and a 36-in.-diameter lower drum with all long flow heating surface. The economizer surface is a bare tube-counter flow type heat exchanger. The superheater and generating bank enclosure is of water-cooled tube and tile construction while the economizer enclosure is refractory and insulation lined. Both enclosures have a gas-tight casing. Flue gas and air handling equipment consisted of a multiclone dust collector, one regenerative- ce | _ Main Steam ix Outlet Economizer Division [) Dust Wall F Collector Spreader/ Feeder Air Heater NNNMININNHTINTELL Im Stoker — Ll LT Force d Dratt Fan Overfire Air Fans — Figure 2 Sectional side view of original unit. type air heater, one FD and ID fan, and an electrostatic precipitator. The multiclone dust collector was used for first-stage particulate con- trol only. The ash from the collector was removed by the plant ash handling system and was not reinjected into the furnace; ash was reinjected from the boiler bank hopper only. Reinjection was done pneumatically through the lower rearwall with injection air provided by a separate cinder return fan. The coal feed system consists of three bunkers, three conical distributors, and ten stoker spreader feeders. The feeders are evenly spaced along the frontwall of the furnace (Figure 3). Each unit has a separate cup-type rotary volumetric feeder with a drum-type rotary flipper. Retrofit unit Fluidized bed description One of the first limitations imposed by the retrofit arose as a result of the reuse of existing coal han- dling and feed systems (Figure 4). The front and rearwall locations of the fluid bed and windbox were set by the type and location of the spreader feeders. The frontwall has to be in line with the feeder while windbox compartmentalization has to align with feeder spacing; the rearwall of the bed is, in turn, limited by the maximum throw of the feeder. The maximum width of the fluid bed was set by the constraint that existing sidewall pressure parts, including the lower headers, were not to be altered. Thus, the overall bed plan area Figure 3 Furnace frontwall. was set which, in turn, established the nominal operating fluidization velocity on the unit. The fluid bed plan area is approximately 40 ft wide and 25 ft deep; fluidization velocity is 12 ft/ second. i—— Primary Economizer | Superheater ti i Ay | Division Wall Vn =~ Dust Air Heater Ce val Boiling Bank Forced fp Draft | iL foo Fan NL | Boile! Circulation Ash Se —__ Pumps Removal Secondary Main Steam Line System Superheater Table 1 Design Parameters Fuel - North Dakota Lignite (HHV = 6680 Btu/Ib) Fuel input @ MCR - 136,500 Ib/hr Bed material - Sand Steam @ SH outlet - 700,000 Ib/hr, 1300 psig, 955°F Bed temperature @ MCR - 1500°F Fluidization velocity -12 ft/sec Bed depth @ MCR -51in. Excess air - 25% Gas temperature @ air heater exit - 275°F Entering feedwater temperature - 443°F As previously mentioned, the entire lower fur- nace enclosure including the lower sidewall and rearwall headers and spreader firing equipment were reused. All of the original furnace pressure parts are top supported and have a downward thermal expansion growth of approximately three inches at full pressure. The firing equipment on the frontwall, however, is supported locally from the main steel and is stationary relative to the furnace. The new fluidized bed combustor is bottom supported and expands upwards approxi- mately 1/2 inch. To accommodate this expansion differential between the FBC unit and the original furnace and to seal the furnace from the atmo- sphere, a special bed-to-furnace expansion joint- seal was designed and installed. In addition to vertical expansion, the seal also accommodates Figure 4 Unit 2 with fluid bed retrofit. lateral expansion differentials up to 1-1/4 inches. The design of the expansion seal was kept as simple as possible, consisting of two insulation- filled pockets protected on the furnace side by stainless steel shields and skirts, and sealed on the external side by a nonmetallic expansion joint. The fluid bed contains both boiling surface and superheater surface.The boiling surface consists of horizontal tubing located in the front of the bed, spanning the entire 40 ft width of the unit. The superheater is also horizontal, spans the entire 40 ft width of the unit, and is located in the rear portion of the bed. Since the new fluidized bed combustor enclosure walls, floor, and inbed boiling surfaces consist of primarily horizontal tube runs, water circulation through these circuits must be pump assisted. Three, 50% capacity, wet stator-type pumps were installed to pump these circuits. Only the new fluidized bed combustor water circuits, all of which connect directly to the existing furnace wingwall, will be cooled by forced circulation. All remaining furnace enclosure walls and boiler bank are cooled by natural circulation. Water from the lower drum is routed to the pump inlets by two new downcomers connected to form the pump inlet manifold. Two 12 in. supply pipes are connected to the pump outlet manifold and run parallel to the sidewalls up to the front of the unit. At this point, two circuits are created, called the inbed boiler circuit and the enclosure wall circuit. As shown in Figure 5, the floor tubes, inbed generating tubes and wingwall tubes (new and existing) are connected in series to form the inbed boiler circuitry. A total of 40 floor tubes on 8 in. centerline-to-centerline spacing forms the floor. Division Wall Connection Tubes In-Bed Boiling Surface Floor Intermediate Header Floor Tubes \ (Distributor Plate) Floor Inlet Header Figure 5 Circuitry for inbed boiling surface. The floor tubes run across the width of the unit and flow into an outlet header. The inbed generat- ing tubes (88 total, 2-1/2 in. OD, internally ribbed) are taken off the floor tube outlet header. The last row of tubes on each bank is bent upwards to form the wingwalls. These wingwall tubes are bifurcated to attach to the original wingwall tubes approximately 34 ft from the furnace floor; the original wingwall headers and tubing were removed from an area just inside the furnace (Figure 6). The enclosure wall circuit, as shown in Figure 7, is formed from 16 enclosure tubes, 2 in. OD on 4 in. centerline spacing, running horizontally along the perimeter of the fluid bed floor. These ribbed tubes exit the enclosure walls, run along the outside of the unit, and tie into the existing wingwall tubes 32 ft above the furnace floor. The inbed superheater is the final stage of superheat with the original convection pass superheater acting as the primary pass. Spray attemperation is used for steam temperature con- trol and is located in the piping connecting the convection pass primary bank to the inbed sec- ondary bank. The inbed superheater is comprised of five rows of stainless steel tubes, 15 rows deep, and is comprised of 2.25 in. OD tubes. Because of the abrasiveness of the bed material, all inbed surfaces were provided with additional wall thickness, and shields were installed in areas where high erosion rates were expected. All inbed tube spacings were set to assure that adequate clearance was provided to prevent bridging of potentially oversized bed material. The entire distributor plate is membraned and water-cooled. The fluidizing(primary) combustion air is introduced to the bed from the windbox by means of bubble caps welded to the membrane between floor tubes. The windbox located below the distributor plate is divided into four main compartments for load control; additional com- Figure 6 Wingwall connections. Division Wall Connection Tubes Inlet Connection a == Enclosure Wall A <E Tubes Bed Enclosure Wall Pump Outlet Manifold Floor Inlet Header uu Figure 7 Circuitry for enclosure wall. partmentalization is provided to facilitate start-up operations and to allow for individual compartment fluidization. Air heater and fans The original regenerative air heater was replaced with a tubular-type heater. This replacement was required to meet higher air side pressure require- ments and to reduce flue gas exit temperature to improve boiler efficiency. The new air heater arrangement had to fit into the existing space with little or no structural steel additions or modi- fications. The gas side pressure drop was not allowed to exceed existing ID fan static capacity since the ID fan was to be reused. Because of the higher air side pressure require- ments, the existing FD fan and drive were replaced. A single centrifugal-type fan was in- stalled, and the existing over-fire air ductwork ports and the boiler-hopper cinder return system were reused. The air for these systems is taken from the secondary air system with all air being provided by the new FD fan. The fan is rated at 870,704 lb/hr with a net static pressure increase of 74.3 in. w.g. Bed ash removal system To remove bed material, seven letdown systems consisting of individual drain points, downspouts, valves, ash conveyors, and hoppers were installed. The bed drain rate is relatively low since sulfur capture is not a requirement and sand makeup is only that required to maintain the bed and adequate bed drain purge. The number, siz- ing, and location of bed drains were set by over- sized material removal requirements. To verify and evaluate the design parameters for the fluid bed retrofit, a test burn of the fuel was necessary since little or no fluidized bed com- bustion data was available for lignite with over- bed feed. A total of 250 hours of testing was con- ducted using the 6 ft x 6 ft pilot test facility. This testing revealed the presence of a small “egg”’- sized agglomerate or clusters consisting of low melting temperature glass and eutectic. It is believed these clusters were formed during the burning process of large, individual pieces of coal. Throughout the testing, the agglomerate material caused no operational problems nor was there an accumulation within the bed. A continuous bed drain allowed for removal of bed material while a small amount of sand was added to maintain bed level. Control system A Bailey Controls Co. Network 908 Distributed System was installed to control the fluid bed. The Network 908 system will provide bed level, fur- nace draft, tubular air heater cold-end metal temperature, steam dump valve, firing rate, com- partment air flow, overfire air, steam temperature, turbine bypass, and coal feeder interlock control. Feedwater and turbine controls remain part of the existing pneumatic system. The bed ash disposal system that operates continuously during service is on manual on/off control. The new system has a single CRT operator interface unit. Critical control loops within the distributed control system are backed up by hardwired, board-mounted, manual/automatic stations. A 60-point annunciator and two 30-point multipoint recorders have been installed for alarm and temperature monitoring of new equipment for the fluidized bed operation. Turbine bypass system During start-up of the fluidized bed (prior to admitting steam to the turbine), steam flow must be maintained through the inbed superheater to prevent overheating of the tubes. Since the amount of steam flow required is significant (25% of MCR flow) and could be on demand for several hours, a turbine bypass system was installed. This system takes steam from the new main steam piping downstream of the inbed super- heater, passes the steam through a combination pressure reducing/desuperheating station, and discharges it into the condenser through a new internal distribution header above the condenser tube bundle. The turbine bypass system is designed for a flow rate of 175,000 lb/hr with steam supply conditions of 650 psig at 955°F. Dur- ing start-up, the steam flow rate is manually con- trolled and modulated to keep inbed superheater outlet steam temperature within limits. Sand handling system The sand specification for the retrofit requires washed, screened, and dried sand. Because of local supply limitations, on-site provisions for drying as well as storing and transporting sand to the fluidized bed were included in the retrofit project. Washed and screened sand is transported to the site, dried in a vibrating bed-type dryer, and conveyed by means of a screw conveyor and bucket elevator to an elevated sand surge hopper with a capacity of two tons. Dried sand from the sand system is transported to two 30-ton sand storage silos by means of a dense-phase sand transport system with a capacity of 10 tons/hr. The sand storage silos are located in the plant just below the coal bunkers, so that sand is fed to the fluidized bed by gravity. The total sand stor- age capacity of 60 tons is sufficient for 24 hours of operation at full load. Two belt-type feeders, one for each storage silo, convey sand from the bottom of the storage silos through 6-inch-diameter sand feed pipes to feed ports located approximately 15 ft above the bed on the front furnace wall. Under normal operat- ing conditions, feeder speeds vary as required to maintain the desired bed level. The maximum sand requirement under normal operating condi- tions is 2-1/2 tons/hr. Operations In a bubbling atmospheric fluid bed boiler, a large portion of the heat transfer occurs in the fluidized bed. This heat transfer, however, must be properly split between the boiling surface and the super- heater surface. Steam flow rate and pressure at the turbine are established by the amount of heat absorbed by the boiling surface while main steam temperature is a strong function of the heat absorbed by the inbed superheat surface. The arrangement of the inbed surface and compart- mentalization of the fluidized bed was set so that a proper split between boiling and superheat absorption can be maintained over the operating range of the unit. The primary method used to change load on the unit is a combination of fuel and air turndown and compartment slumping; i.e., either removing bed compartments from service (slumping) or placing them into service (fluidizing) to decrease or increase load, respectively. The total number of compartments in service is a function of boiler load. One compartment in the boiling bank portion of the bed was designed specifically for start-up operations. While each bed/windbox compart- ment has its own air supply duct with flow moni- tor and control damper, the air supply duct to the starter compartment also contains a natural gas- fired duct burner. During start-up, this burner is fired to heat the fluidizing air to the start-up compartment and the bed material within that compartment. When bed material reaches approx- imately 900°F with the compartment in a flui- dized state, coal feed is initiated. When coal-fired and stable, the compartment temperature is controlled by adjusting coal feed rate, and com- partment bed level is controlled by the sand feed rate. When desired bed temperature and level has been established, an adjacent compartment totally within the boiling surface portion of the bed is brought into service. To do this, the adja- cent compartment is fluidized and mixed with the active start-up compartment. Fuel input to the start-up compartment serves as the heat input to warm up the newly fluidized portion of the bed. When the temperature of the adjacent com- partment reaches 900°F, coal feed to that compartment is initiated. This procedure is repeated as additional compartments are brought into service. Before a superheat compartment is placed into service, a steam flow of approximately 25% must be established and maintained through the inbed superheater to prevent overheating. The turbine bypass system is used until minimum steam flow has been established. Turbine load is then in- creased while the turbine bypass valve is throttled until the entire steam flow passes through the turbine. The load is then increased by adding sand until the full bed inventories are reached and all compartments are placed into service. Operator training During final construction on-site, training ses- sions were held with plant operators and other personnel. In this case, special emphasis was placed on familiarizing plant personnel with fluidized bed operation. In addition to covering basic operating procedures relative to start-up, shutdown, load control, system trips and inter- locks, information on basic fluid bed combustion, specific to this unit, was presented. Because of the characteristics of the fuel being fired, special attention was placed on avoiding bed agglomera- tion and how to monitor for, and manage, the boiler should it occur. References 1. Imsdahl, B., Gorrell, R. L., Johnson, H. L., “Montana-Dakota Utilities 80-MW AFBC Retrofit”, Presented at the Jt. ASME/IEEE Power Generation Conference, Portland, OR., October 1986. 2. American Boiler Manufacturers Association, Fluidized Bed Combustion Guidelines, First Edition, Washington, D.C., 1987. 3. Gorrell, R. L., Strong, D. N., “Description of 80-MW Fluidized Bed Retrofit at Montana- Dakota Utilities Co.,” Presented at EPRI Seminar on Atmospheric Fluidized Bed Tech- nology for Utility Application, Palo Alto, CA., April 8-10, 1986. 4. Goblirsch, G. M., Benson, S. A., Karner, F. R., Rindt, D. K., and Hajicek, D. R., “AFBC Bed Material Performance with Low-Rank Coals,” University of North Dakota, Energy Research Center, Presented at Twelfth Biennial Lignite Symposium, Grand Forks, N.D., May 18-19, 1983. Technical Paper BR-1312 Design and construction of a wood-fired circulating fluidized bed boiler R. F. Johns Manager, Fluidized Bed Program Babcock & Wilcox 20 S. Van Buren Ave. Barberton, OH 44203 R. E. Wascher Vice President, Babcock & Wilcox West Enfield Power, Inc. 20 S. Van Buren Ave. Barberton, OH 44203 Presented to The Ninth International Conference on Fluidized Bed Combustion Boston, MA May 3-7, 1987 Babcock & Wilcox a McDermott company Design and construction of a wood-fired circulating fluidized bed boiler R. F. Johns Manager, Fluidized Bed Program Babcock & Wilcox 20S. Van Buren Ave. Barberton, OH 44203 Presented to R. E. Wascher Vice President, Babcock & Wilcox West Enfield Power, Inc. 20S. Van Buren Ave. Barberton, OH 44203 The Ninth International Conference on Fluidized Bed Combustion Boston, MA May 3-7, 1987 ABSTRACT Babcock Ultrapower has completed the design and construction of two 25 megawatt wood-fired power plants. The plants are located in West Enfield and Jonesboro, Maine and are designed to burn wood waste generated by nearby forest operations. These projects include Babcock & Wilcox circulating fluidized bed (CFB) boilers, turbine-generator sets, fuel preparation and handling systems and all other necessary auxiliary equipment. The 25 megawatts generated from each plant will be sold to Bangor Hydro Electric under a long-term power sales agreement. The project was initially developed by Ultra- systems, Inc. of Irvine, California as a biomass-fired, small power producer. The B&W CFB boilers were selected in a competitive bidding process. Subsequent to the boiler contract award, Babcock & Wilcox was offered and accepted an equity position in the project. The owners then formed Babcock-Ultrapower West Enfield and Babcock-Ultrapower Jonesboro to design, construct and operate the two power plants. This paper provides a description of the design, manufacturing, construction, and performance parameters of the circulating fluidized bed boilers and balance of plant equipment. BOILER SPECIFICATION AND SELECTION Ultrasystems, Inc. as project developers, conceived the wood-fired power plant project in early 1984. The key elements of the project included a power sales agreement with Bangor Hydro, a fuel supply agreement with local forestry operations, and the requirement to achieve low emissions. This combination led to the specification of a wood-fired fluidized bed boiler to provide the solution. B&W was informed of the intent of Ultrasystems to proceed with the project in August of 1984. By October of 1984, B&W was awarded the contract to deliver and erect the CFB boilers at the West Enfield and Jonesboro sites. The following performance criteria were specified: Maximum Continuous Steam Flow ...... 218,640 lb/hr Main Steam Pressure ..........++.+++- 1250 psig Main Steam Temperature .. » 955 F Feedwater Temperature .........++-+- 269 F Air Heater Outlet Gas Temperature .. 275 F so, EmissionS .......eeeeeeeeeeeeees 0.15 1b/MKB NO, Emissions ...+++eeeeeeeee rere eee 0.15 1b/MKB CO Emissions .. eee + 0.15 1b/MKB Hydrocarbon Emissions ..... see eeceee 0.15 1b/MKB Particulate Emissions ...........+4- 0.5 grains/DSCF @ 32 0, (entering precipitator) In addition, the normal, commercial boiler guaran- tees, such as boiler efficiency, power consumption, draft loss, air resistance, and steam and water pressure drop were required. These performance guarantees were based on firing wood waste with the following analysis (as fired, 2 by weight): © ccccccccvccevercccccece 31.40 HY a ulb we etlee sive tlee cites cle 3.62 S cecscccceserccccccccsese 0.02 Ny oso ala 0.30 H,0 see es ale 40.00 o, tee ececccccrecsecceees 21.20 Ash ......--.. eee eee eeeee 3.46 Btu/1b ...eeeeeeeeeeeeeee 5040 In addition, the performance is guaranteed for fuel with up to 50% moisture. SCHEDULE The following milestone schedule was achieved on these projects: Award ..eeeeeeeeeeeeeeees 10/84 Shipment ... ++ 10/85 Hydrostatic Testing . 06/86 Boil Out ... ++ 10/86 First Power Production .. 11/86 DESCRIPTION OF UNIT The CFB boiler (Figure 1) consists of a water cooled, membraned furnace which is 11 ft-10 in. deep by 17 ft-10 in. wide by 66 ft-0 in. high from the furnace floor to the roof, a hot particle separator and solids recycle system, and a convection pass enclosed by membraned, water cooled tubes containing a two-stage superheater and an economizer. Primary air enters the furnace via a bubble cap air distributor located in the floor. Along with primary air, fuel and recycled bed material are introduced into the lower or primary section of the furnace. The expected primary air is about 50% of the total air required for combustion. The balance of combustion air is supplied through secondary air nozzles located about five feet above the floor and tertiary air nozzles located about half way up the furnace shaft. The turbulent mixing action of the bed material in the primary zone provides both combustion kinetics and fuel distribution. The combus- tion process continues and is completed in the secondary zone of the furnace. Entrained bed material (sand) and flue gas enter the U-beam hot particle separators located at the furnace exit. The highly efficient Figure 1 Side view of the West Enfield and Jonesboro CFB boilers. U-beams separate the solids for recycle back to the furnace and allow the cleaned flue gas to flow into the convection pass. A horizontal, drainable, two-stage superheater arranged for counter-flow is followed by a horizontal, bare-tube economizer. At the bottom of the convection pass, the gas flows through a multiclone dust collector before entering into a two-pass tubular air heater where it is cooled to 275°F. The multiclone dust collectors remove small quantities of sand and unburned carbon which were not collected by the U-beams. The primary purpose of the multiclone is to collect and recirculate bed material back to the furnace in order to reduce the quantity of sand make-up. Incoming air from the forced draft fan is preheated by the tubular air heater before flowing to the primary and secondary air systems. Following the air heater, the flue gas passes through a B&W Rothemuhle electrostatic precipitator (ESP) before discharging to the atmosphere via a 135 ft high stack. The ESP has three electrical fields and is designed to attain specified performance with only two of the three fields in operation. BED MATERIAL A circulating fluidized bed boiler acts as a classifier of bed material. Oversized bed material will sink to the lower portion of the furnace and be purged through floor drains along with rocks and other inert materials. Very fine bed material (< 20 microns) has a higher probability of escaping collection in the hot particle separator and will be purged from the system at either the duct collector or electrostatic precipitator. Since SO, reduction is not a requirement for the wood fuel, thé CFB boilers at West Enfield and Jonesboro utilize sand as bed material. Design specifications call for this sand to be sized at minus 16 mesh. U-BEAM HOT PARTICLE SEPARATOR The boilers at West Enfield and Jonesboro utilize the patented U-beam hot particle separator to disengage the solids from the flue gas at the furnace outlet (Figure 2). The U-beam separator is a labyrinth-type Ultrasystems U-Beam Configuration WH ss 5 —— UUUYUYHUYHUUUUUUUY / UYUUUUUYUUUUUUDUUUUY UVUUUUUUYHUUUUUUYU re UUUUUUHUUUUUUY buns UUUUUUHUUHUUUUY UUUUUYUHNUHUUUYUUUUUU UUUUUUUUUUUUUUY UUUUUYNYUUNUHNUYNUUUUYU UUUUYUUHUYUUUUUUUUHY 1 cy Detail U-Beam UYUUUUUUUUUUUUUY (1 eee Se ee te | i SRS UNL ES RR ATLL EE Se Figure 2 Plan view of the U-beam particle separators installed at West Enfield and Jonesboro. 6.17/32" 14.0" 63/4 mechanical separator formed from a staggered array of high strength stainless steel channels. These channels collect the bed material along with a high percentage of the ash and char present in the flue gas stream. The cleaned flue gas continues to the convection pass and the solids are discharged into a storage hopper located directly beneath the U-beam cavity The U-beams are formed from 1/4 in. thick TP-310 stainless steel plate and are 6-1/4 in. wide by 6-3/4 in. deep. The clear spacing between adjacent U-beams is 5-3/4 in. The West Enfield and Jonesboro boilers have eleven rows (deep) of U-beams. In addition to separating solids from the flue gas stream for recirculation, the U-beams impart a turbu- lent, mixing action on the flue gas. This mixing action coupled with a one second residence time is designed to complete the conversion of CO to co,. U-BEAM FUNCTIONAL REQUIREMENTS The purpose of the U-beam material separator is to disengage the circulating bed material from the flue gas stream. This component is responsible, therefore, for insuring that the furnace solids material balance is maintained (through recycle of the collected solids) and also for reducing solids carryover to the convec- tion pass to minimize the potential for tube erosion. Although previous work indicated that efficiencies of 99+Z% were achieved with the U-beam separator, a conservative value of 97% was used as the basis of design (Figure 3). Studsvik Pilot Plant 4 Rows Gas Temperature 850 °C 100 99 = 8 oe 5 97 2 96 = 95 3 94 Gas Velocity: 1-4m/s 5 93 2-8m/s 92 91 —— 4 6 810 20 40 60 80 100 Solids Flow into U-Beams, 10? kg/hr Figure 3 Collection efficiency of U-beams vs. solids flow and velocity. U-BEAM DURABILITY In addition to the functional requirements, the U-beam separators must be durable. The design basis of this durability requirement is two-fold. First, the gas velocity entering the U-beam separator is set at one-half the furnace shaft superficial gas velocity, or 13 fps. At this low velocity, even with the high solids loading of CFB's, erosion is negligible. Second, the collector channels are filled with solids (to varying depths) during operation. The incoming solids, therefore, impact other solids rather than base metal. B&W selected TP-310 stainless steel for the U-beams. The expected operating temperature for the U-beams at West Enfield and Jonesboro is 1550°F and the beams are designed for 1700°F. U-BEAM ENCLOSURE The U-beam enclosure is formed by steam-cooled, membraned tubes (Figure 4). A thin layer of insulating refractory, attached with refractory anchors to the tubes, is applied over the enclosure tubes. An insulat- ing refractory was selected to reduce heat transfer to the boiler tubes, thereby enhancing CO burn-out and keeping flue gas temperature high for better convective superheater heat transfer. we 41 N / i Figure 4 Installation of the material storage hopper and U-beam enclosure. The enclosure roof is formed from water-cooled, membraned tubes which form the furnace front wall The U-beams are hung from the enclosure roof with a double-bolted attachment through the roof membrane bar. At the bottom of each beam, a guide pan is attached which forms a separation between the collector channels and the storage hopper. These guide pans are designed to maintain U-beam alignment and also to prevent re-entrainment from the storage hopper. PARTICLE STORAGE HOPPER Located directly below the U-beams is the hot particle storage hopper. This hopper is formed with membraned tube construction and is cooled with satu- rated steam from the drum. The storage hopper is designed to hold 100% of the circulating solids inven- tory. By functioning as an accumulator, the storage hopper decouples the collection process from the recycle flow process thereby providing an independent control of recycle. The particle storage hopper tubes are covered with a layer of refractory material. The refractory is attached with anchors and was installed by a "gunning" process. The function of this refractory coating is to prevent hopper tube erosion. L-VALVE The L-Valve is a solids flow control device that has been ytilized for many years in the process industry.“ B&W applied this knowledge to the CFB product in order to have a precise, non-mechanical method of controlling the solids recirculation rate (Figure 5). 160 120 a o i oO Mass Flow Rate 1000 Ibs/hr 8 12 16 20 24 L Valve Aeration Flow SCFM Figure 5 L-valve characteristic curve showing solids flow vs. aeration air flow. The boilers in West Enfield and Jonesboro include four L-Valves for each unit. These L-Valves discharge hot bed material at a controlled rate into the primary zone of the furnace. The L-Valves are located opposite the fuel feed points. The L-Valve device actually starts at the bottom of the material storage hopper. Hot solids flow at low velocities (<.2 meters/second) through the stand pipe. In addition to conveying the solids to the primary zone of the furnace, the vertical section of the L-Valve provides a pressure seal between the primary furnace (operating at approximately 30 in. H,0) and the material storage hopper (operating at approximately - 2 in. H,0). The lower section of the L-Valve is where the flow control occurs. By injecting a small quantity of compressed air (20-50 ACFM @ 75 psig) above the 90° bend in the L-Valve, the horizontal leg is aerated. By increasing or decreasing the aeration air flow, the resistance in the L-Valve is increased or decreased and, consequently, the solids flow is increased or decreased. The key to using solids flow control to control furnace density and, therefore, heat transfer is to design a system independent of boiler load. The particle storage hopper/L-Valve system is designed to decouple solids recycle from boiler load, thereby providing independent control of furnace density. The construction of the stand pipe/L-Valve compo- nent consists of a high temperature stainless steel liner, surrounded by several inches of insulation and sheathed in a carbon steel outer casing. The West Enfield and Jonesboro L-Valves are 12 in. ID, and are equipped with a high temperature expansion joint at an intermediate elevation to account for differential expansion. MECHANICAL DUST COLLECTOR During the design stage, B&W decided to install a mechanical "multi-clone" dust collector at the econo- mizer flue gas outlet. This decision was derived due to the low ash content in the wood fuel and by the desire to minimize the amount of bed "make-up" sand needed. The design efficiency of the multiclone is 92%, which, added to the conservative U-beam efficiency of 97%, provides an overall solids collection efficiency in excess of 99.52. The solids collected in the mechanical dust collector are either returned to the furnace via a pneumatic transport system or are purged to the ash disposal system. The amount recycled to the furnace is based on maintaining the solids material inventory. WOOD FUEL SYSTEM The plant will burn about 260,000 green tons of fuel per year. The fuel is supplied primarily from wood waste material taken from forests within 30 to 50 miles of the plant. Mill residues are also a source of fuel to the plant. Waste wood chipped in the forest and the mill residue is transported to the plant in trucks. At the plant, the trucks are unloaded in a truck dumper. The wood is carried by conveyor to a disk screen which passes two inch minus wood. Oversize fuel is reduced in size in a hog before being conveyed with the screened fuel to a traveling tripper in the fuel yard where it is stored until it is burned in the boiler. A front end loader transports the wood from the storage piles and piles it over a reclaimer. Two traveling screws under the reclaim pile load the wood on a conveyor which carries it to a metering bin located inside the boiler building (Figure 6). The live-bottom wood metering bin transports the fuel through rotary seal feeders into variable speed screw conveyors. The fuel is conveyed into the lower furnace front wall through four screw feeders which are evenly spaced along the front wall. The boiler feed system is designed to achieve full boiler load with one of the four feed screws out of service. Figure 6 Layout of wood yard equipment at West Enfield EMISSIONS The air emissions permit specifies the following allowable limits: Total Suspended Particulates .. 43.6 tons/year co . se eeeeeee . 232.5 tons/year NO. .. sees . 239.4 tons/year x SO, ... . 44.8 tons/year Volatile Organic Compounds .... 145.3 tons/year To achieve these levels of emissions, the following equipment design philosophy was applied: PRECIPITATOR The important decision regarding final particulate clean-up came early in the design stages. Of the two options (ESP or Baghouse), B&W/Ultrasystems elected to use an electrostatic precipitator. This decision was derived by standard commercial practices which suggest that using baghouses on biomass fired jobs causes a high risk of baghouse fires due to "sparkler" carryover. The B&W precipitator (rigid frame) is designed to achieve the particulate emissions with only two of the three electrical fields in operation. co To ensure that CO is reduced to the specified permit level, the B&W CFB boiler takes advantage of several design features. First, the furnace is equipped with both secondary and tertiary air ports. The forced draft fan was designed to provide enough flexibility to bias the primary/ secondary air split from 70/30 to 30/70. Although the expected split is 50/50, this biasing flexibility enhances CO burnout performance. Second, the gas residence time coupled with the mixing action of the bed provides the necessary time and turbulence to compensate for lower combustion tempera- ture. Third, the U-beam hot particle separator along with the CO-burnout cavity provides a mixing chamber where CO can continue to reduce to CO,. HC The key to controlling Volatile Organic Compounds (VOC) or hydrocarbons is to control CO levels. Previous field test data taken at several B&W boiler installa- tions indicates that the HC emission level is about 10-15% of the CO emission level. Therefore, the basic combustion control process designed to control carbon monoxide will, in turn, control hydrocarbon emission. NO. x A 7 110 7 Controlling NO. to .15 1b/10 Btu input was the most critical requirement of the project. This level, or less, was the cut-off point for which a construction permit was obtainable without performing extensive monitoring of the existing atmospheric conditions. The B&W CFB utilizes a combination of low combus— tion temperature (1550°F) and staged combustion (50/50 air split) to achieve the NO. guarantee. x SO. x There are no provisions to the boiler design to control SO. Referring to the fuel analysis, the SO emission guarantee will be obtained with 100% converSion of fuel bound sulfur to SO,. The unique capability of a fluidized bed to capture SO, was, therefore, not significant to the purchasing decision for this project. MANUFACTURING The manufacturing process followed a standard B&W project (Figures 7,8,9). Pressure parts, including the steam drum, furnace walls with headers, U-beam and convection pass enclosures, and the convective super- heater were fabricated in the B&W manufacturing facili- ties. Boiler accessories, such as fans, air heaters, steel, and feeders were purchased from qualified suppliers using B&W specifications. The components of the boiler that were new to the manufacturing process included the U-beams, the steam- cooled material storage hopper, and the L-Valves. Although new, the relative simplicity of the design resulted in no unusual manufacturing problems. Figure 7 Panel fabrication of the U-beam enclosure and material storage hopper in B&W’s West Point, Mississippi manufacturing facility. Figure 8 Shop assembly of the boiler floor, water-cooled windbox and lower rear wall. The bubble caps were shop installed along with the dense stud pattern for refractory application oad Ay ula ANS Ni - ak ui Na aS i AON Figure 9 Shop assembly of the water-cooled convection pass enclosure showing the superheater support attachments. CONSTRUCTION Construction of the West Enfield unit began in October 1985. At that time work commenced to pour the foundation and erect buildings. Site clearing and rough grading had been done a year earlier in anticipa- tion of starting construction after the spring thaw in early 1986. However, financing problems arose because of the impact of the Seabrook nuclear project on Bangor Hydro's financial health. This delay in financing delayed the start of construction by about six months. Even with this late start it was possible to get the turbine island and the administrative spaces under roof to permit work inside throughout the winter. Figure 10 Construction progress at West Enfield Figure 11 Construction at West Enfield showing boiler, precipi- tator, and stack arrangement. The U-beams are shown in the lower-right foreground Figure 12 Aerial photo of the completed West Enfield job site However, it was not possible to get the boiler island under roof. Instead the boiler support steel was erected during the winter. This led to low produc- tivity and extra cost in the boiler construction. But, by working through the winter and by getting materials delivered to the site at that time, it was possible to work during the spring thaw season which would other- wise have been lost. As a result, the boiler was hydrotested early in June of 1986. This effort by the B&W manufacturing facilities and B&W Construction Co. permitted the project to maintain the original schedule for start-up in December of 1986 (Figures 10,11,12). SUMMARY The West Enfield and Jonesboro units are industrial sized power plants for production of electricity. They use B&W circulating fluidized bed boilers to burn waste biomass while meeting the low emissions specifications required to keep the environment clean. The design, manufacturing, and construction followed standard B&W practices. The projects were built and started up on a two-year schedule in spite of significant delays associated with unusual financing problems and with winter construction in Maine. REFERENCES l. Whitney, S. A., Johnson, H. L. and Hemmingsson, L. "Wood Fired Fast Fluidized Bed Boiler". Proceedings of Energy Technology Conference and Exposition, Washington, D.C., March 25-27, 1985. 2. Knowlton, T. M. and Hirsan, I. "Solids Flow Control Using a Nonmechanical L-Valve", Proceedings of the Ninth Synthetic Pipeline Gas Symposium, Chicago, Illinois, October 31 - November 2, 1977. Technical Paper Development Status of B&W’s Second Generation Low NOx Burner - the XCL Burner A. D. LaRue M. A. Acree Domestic Fossil Operations Babcock & Wilcox Barberton, OH C. C. Masser U. S. Environmental Protection Agency Research Triangle Park, NC Presented to Joint Symposium on Stationary Combustion NOx Control New Orleans, LA March 23-27, 1987 Babcock & Wilcox a McDermott company BR-1315 Development Status of B&W’s Second Generation Low NOx Burner - The XCL Burner A. D. LaRue C. C. Masser M. A. Acree US. Environmental Domestic Fossil Operations Protection Agency Babcock & Wilcox Research Triangle Park, NC Barberton, OH Presented to PGTP-87-12 Joint Symposium on Stationary Combustion NOx Control New Orleans, LA March 23-27, 1987 ABSTRACT Due to the national concern with acid rain, the U.S. NO,| emission standards may become more stringent for new sources, and uncontrolled*sources may face NO emission limits. B&W has consequently proceeded to develop a second generation low NO, pulverized coal (PC) burner suitable for new or retrofit applications. A primary objective is to minimize NO, by burner design to avoid slagging and corrosion concerns associated with Staged furnace operation. Development of the XCL burner stemmed from the technology of Babcock Hitachi's HT-NR burner design. The development program was cosponsored by the EPA, and consisted of full scale (80 million Btu/hr)* tests of a standard Dual Register burner, an HT-NR burner, and an XCL burner. Air flow tests were conducted to characterize flow patterns and improve swirl efficiency, and combustion tests were performed to evaluate and minimize emissions. The XCL burner proved capable of NO, emissions (unstaged) of 0.3 to 0.5 1b/million Btu with high efficiency and adjustable flame shape. Consequently, a full complement of XCL burners was retrofitted to Ohio Edison's Edgewater Unit 4 for the EPA Limestone Injection-Multistage Burner (LIMB) demonstration. The paper describes development and field results with the XCL burner. INTRODUCTION In the U.S., several factors have caused a need for combustion equipment capable of lower NO, emissions. Federal New Source Performance Standards * Readers more familiar with metric units may use the conversion factors listed at the end of this paper. (NSPS) for No, emissions from pulverized-coal-fired boilers are under review and may become move stringent in the years ahead. Since their enactment in 1971, the emissions limits have been reduced once. Also, areas of the country designated as Non-Attainment (i.e., not meeting National Ambient Air Quality Standards) require emission offsets for new sources. This requires the new source to be equipped with combustion systems capable of extremely low NO, emission levels, usually backed up by expensive Selective Catalytic Reduction (SCR) gas cleanup systems. Even in attainment areas, Prevention of Significant Deterioration (PSD) reviews can result in NO, emission limitations well below federal standards for a specific new source, Lastly, NO, emissions from existing sources have been regulated in some states, and the possibility exists of regulation on a federal basis. It will often be more difficult to achieve a specified level of emission (e.g., NSPS limits) in an existing source compared to a new source. Many existing sources have compact combustion zones that contribute to thermal NO, generation and limit flexibility in arrangement of low NO. equipment. Consequently, the combustion system requires enhancements to satisfy emission limits. In any case, the combustion system is expected to control No, emissions while efficiently burning the coal; i.e., with low levels of CO aid unburned combustibles. One of the most effective means of augmenting present day low No, burners is the use of two stage combustion. Reducing combustion zone stoichiométry to 1.0 and below is a proven means to sharply lower NO, emissions. However, many boilers have experienced extensive furnace tubewal1* wastage as a result of operating at reduced stoichiometry, particularly when burning eastern U.S. coals with 2 to 3% sulfur or more, The tube damage is attributed to H,S and FeS attacking carbon steel tubes in a reducing environment (1). Further} the higher iron contents in the ash of eastern coals contribute to ash fusion temperatures hundreds of degrees lower for reducing conditions. This leads to wet or plastic slag deposits forming in the combustion zone during two stage operation and associated operating problems with uncontrolled slagging. Consequently, two stage operation is not well suited for many eastern coals. A better solution is a burner capable of very low NO, emission levels in normal conditions. Such a burner should be capable of reduéing NO. emissions well below present designs without resorting to two stage combustion with NO ports. Eliminating NO, ports would avoid the combustion zone slagging and corrosion problems, whilé reducing costs and operating complexities. In addition, the problems of locating, feeding, and controlling the ports in retrofit applications would be avoided. These factors led Babcock & Wilcox to initiate a program to develop a second generation low NO, burner. BACKGROUND The most promising combustion technologies to further reduce NO, emissions were involved with controlling fuel nitrogen conversion to No. during devolatilization. Previous work (2,3) had shown the majority of NO. formed*in PC combustion, 75 to 90%, was contributed by oxidizing nitrogen bound*to the fuel molecules. Further, the fuel nitrogen evolved during devolatilization had the greatest contribution to overall NO, . The effectiveness of present day low NO, burners, such as B&W's Dual Register burner (Figure 1), depends on this principle The Dual Register burner (DRB) introduces fuel to the combustion zone as an axial fuel jet, with which secondary air gradually mixes as combustion proceeds into the furnace. Although all of the combustion air is introduced through the burner, the stoichiometry apparent to the fuel drops rapidly as 0, from the primary air (transporting the coal) is consumed and 0, from secondary air is gradually introduced. The low apparent stoichiometry Occurs during devolatilization and produces competing reactions for volatile nitrogen which result in formation of Ny rather than NO or NO,. CFA Lighter Conical Phase 5 Diffuser Register Assembly Burner BY Inner Air Outer Air A - Axial Fuel Jet: Oxygen deficient Elbow Zone Zone devolatilization Weldment B - Hot Gas Recirculation Zone: Stabilizes flame base C - Mixing Region: Gradual mixing of fuel and air to complete combustion Figure 1 Phase 5 Dual Register Burner. Recent developments by Babcock Hitachi K.K. (BHK) in the Hitachi NO. Reduction (HT-NR) burner have advanced the potential of ultralow NO, emissions from combustion of pulverized coal. The HT-NR burner technology is available to B&W through the In-Furnace NO. Reduction license from BHK to B&W. The HT-NR burner (Figure 2) was reported if the Joint NO Symposium in Boston (4) and subsequently has achieved NO reductions of 30 to 50% relative to BHK's Dual Register burner performance, and up to 80% reduction relative to uncontrolled emission levels from circular burners. The HT-NR burner accelerates the combustion rate during devolatilization. Pulverized coal leaving the axial fuel nozzle is rapidly ignited by virtue of recirculation patterns set up by the Flame Stabilizing Ring at the nozzle exit, yet done so at low stoichiometry. Hydrocarbon radicals and intermediate nitrogenous species, formed immediately downstream of the ignition zone in a fuel rich area, reduce NO formed during devolatilization. As the flame continues forward, highly swirled secondary air re-enters the flame core Devolatilization Zone Oxidizing Zone P Secondary Hydrocarbon NOx Pulverized Air Radical Reduction Coal & Tertiary Air Generation Zone Zone Primary Air Figure 2 Hitachi-NOx Reduction Burner. downstream to complete combustion. The combination of the Air Separation Sleeve with highly swirled secondary air throws the air away from the flame during the critical devolatilization and NO_ reduction stages, and subsequently returns the air rapidly to the flame downstréam allowing char reactions to proceed rapidly. This rapid reintroduction of secondary air downstream limits flame length, as set by final char reactions. METHODOLOGY The proven NO reduction capability of the HT-NR burner (reported upon further in this symposiui) became a fundamental part of the program described here. In addition, several other features appeared desirable for inclusion in a second generation burner for its use in the U.S. First, many older wall-fired boilers, which may eventually need such a burner, experience problems with secondary air flow distribution. On modern units, B&W uses the compartmented windbox (Figure 3) to measure and control air flow to the burners served by each pulverizer. Older units, often not arranged to permit compartmentation, experience wide variations in air flow horizontally and vertically in the windbox. Consequently, a need exists for individual burner air flow measurement to facilitate secondary air distribution in these units. Second, mechanical performance has suffered on many burner designs due to the hot, dusty environment in which they operate. Burners experience high thermal gradients and cycling as they cycle in and out of service and this can result in components overheating or locking up. B&W's S-type burner, developed a few years ago to alleviate these problems, incorporates air measurements capability. The second generation low NO. burner would be configured to include these features. The importance of secondar}-air-flow peripheral distribution within the burner, swirl generation, and mixing patterns were anticipated from the design of the HT-NR design. Testing appeared prudent to evaluate flow performance of the HT-NR burner and to check the influence of streamlining on pressure drop, since available forced draft fan static pressure limits burner pressure drop in retrofit applications. Finally, the fuel jet was seen as a crucial factor in determining overall NO performance and flame length. Alteration to the fuel jet momentum and dispersion*was expected to provide a Furnace Observation Doors Compartmented Windbox Typical Burner Burner Secondary Air Control Dampers Burner Secondary Air Foils Figure 3 Compartmented Windbox. Fixed Vanes Air Control Disk Tube Wall Opening Conical Diffuser Adjustable Vanes Spin Vanes Figure 4 Babcock & Wilcox XCL Burner. measure of control over NO, and flame dimensions. Flame width and length should be adjustable to tailor flame shape to a given furnace geometry so as to avoid impingement. The B&W XCL burner (Figure 4) addresses these concerns. Functionally, the XCL burner uses criteria for the HT-NR burner, and its predecessor, the DRB. The prototype XCL uses slide dampers to regulate air flow to the inner and outer air zones, with adjustable vanes in both zones to impart swirl. The air separation vane was retained from the HT-NR burner. Several hardware arrangements were evaluated for the burner nozzle, including the flame stabilizing ring, conical diffuser, and impellers. PROGRAM DESCRIPTION AND OBJECTIVES The U.S. Environmental Protection Agency had expressed interest to B&W in the performance of the HT-NR burner. The EPA was interested in conducting combustion tests with this burner in their Large Watertube Simulator (LWS). Further, it appeared to be a promising candidate for demonstration in the Limestone Injection Multistage Burner (LIMB) project for which B&W was prime contractor to the EPA. It was subsequently agreed that the EPA and B&W would share the cost of a development program to evaluate the HT-NR burner relative to B&W's Dual Register burner and XCL burners. The LIMB demonstration project called for retrofitting low NO, burners in Ohio Edison's Edgewater Station Unit 4, in Lorain, Ohio. This unit (Figure 5) is a B&W natural circulation drum boiler rated at 690,000 1b/hr mainsteam flow at 1,480 psig at 1,000°F. The unit is 30 ft. wide, 22 ft. deep, and 114 ft. high from lower headers to the roof. The unit is equipped with four E64 pulverizers, each supplying three burners in an individual burner level (burner matrix is four levels high with three burners per level). The LIMB demonstration project brought several operating constraints and burner sizing criteria to the program, specifically burner input rating, pressure drop, OO0800 OOOO OOs00 Figure 5 Unit 4 of Ohio Edison's Edgewater Station. and flame length (Table 1). Burner input would be 78 million Btu/hr (MKB) and the burner would have to fit the existing walltube opening (no pressure part alterations) and windbox. Taken individually each criterion does not appear too restrictive; however, as a group they multiply the degree of difficulty to achieve all simultaneously. To accomplish these goals, a development program was formulated that consisted of three phases: ° Air flow testing. ° Single burner combustion tests. ° Retrofit demonstration at Edgewater 4. Table 1. BURNER OBJECTIVES AND CRITERIA FOR LIMB DEMONSTRATION Burner Objective Criteria Burner input, 10®Btu/hr 78 NO, emissions, Ib/10® Btu <0.5 Unburned carbon (in ash), % <5 Flame length, ft <22 Burner air resistance, in. H20 <5 Burner throat, in. 35 (or less) Windbox depth, in. 63 Burner depth, in. 49 TEST APPARATUS AND FACILITIES The air flow and combustion tests involved evaluation of the DRB, HT-NR, and XCL burners. The burner capable of meeting all the objectives was intended for use in the retrofit demonstration. One full scale version of each burner was fabricated to meet the size constraints of Table 1 at the design input. The HT-NR burner was designed according to BHK standards. These criteria were met with a burner throat diameter of 35 in. The burner nozzle diameter and air zone design criteria of the HT-NR were matched for the XCL design. The DRB was sized to meet the same throat diameter as the HT-NR and XCL, which resulted in lower than standard design velocities for the DRB. This was done in the interests of simplifying burner changeover during testing, with the expectation of limited impact on results. During subsequent testing, the actual impact on DRB performance due to the oversized throat was significant; consequently, a true standard DRB was built and tested to establish baseline performance. Air flow tests were conducted at B&W's Alliance Research Center in Ohio. The test facility is shown in Figure 6. Ambient air was separately measured and controlled to the burner nozzle and to the windbox. Local velocity and swirl measurements were taken in a plane 7 in. downstream of the inlet of the throat cone using a Fecheimer two-dimensional velocity probe. The radial component of velocity in the secondary air region of the burner exit was found by a wool tuft to be such that the flow is parallel to the throat cone. Fecheimer traverses in this plane were taken with the probe perpendicular to the throat cone. Velocity and swirl angle measurements (from which axial and tangential velocity components can be calculated) were obtained. Overall, isothermal burner exit flow profiles were taken downstream of the burner using wool tuft grids. These patterns were used to develop the flow profiles of the burners. Windbox (Width = 9") Secondary Air Flow Perforated Plate Primary Air Venturi Section 40,000 scfm Fan Figure 6 Full Scale Burner Air Flow Facility. Combustion tests were performed in the EPA's LWS facility operated by the Energy and Environmental Research Corporation, El Toro, California. The LWS furnace (Figure 7) is 16 ft. wide and 22 ft. deep, conveniently the same furnace depth as the Edgewater 4 unit. The furnace is 50.5 ft. high, refractory lined to a height of 16 ft. (in addition to the hopper), and spray water cooled on the outer casing. Secondary air is preheated by an indirect fired airheater and metered and controlled to the windbox attached to the furnace frontwall. A bowl pulverizer directly provides pulverized coal to the burner. Gas is sampled at a multiple point grid in the exhaust duct at about l, 000°F. The sampling system and train for continuous analysis of emissions conforms with EPA Performance Specifications 2 and 3. A data acquisition system provides continuous and time-averaged data collection. The coal used in the combustion test program matched criteria established in the LIMB program. Table 2 shows the coal to be a high volatile bituminous A with high sulfur and a fuel ratio (FC/VM) of 1.5 indicating a reactive coal. B&W empirical indices predict medium slagging and high fouling tendencies. The combustion test matrix consisted of parametric testing of the DRB, HT-NR, and XCL burners in sequence to screen burner performance. Several hardware variations were performed with each burner to evaluate the effectiveness of certain components, primarily as they affected NO, » unburned carbon loss (UCL), flame length, and burner pressure drop. All coifigurations were tested at full load (78 MKB) over a range of burner adjustments and excess air. Key arrangements were further tested over a range of loads, but facility limitations prevented operation below about 45 MKB. The burner of choice from the screening tests was further characterized by tests with a variety of coals, including low volatile bituminous and subbituminous. Major findings are covered here, and complete documentation will be found in the final report to the EPA for this project, to be issued later this year. The burner selected from the combustion test program was approved by Ohio Edison and EPA for use in the LIMB demonstration. Twelve low NO. burners were fabricated and installed in Edgewater Unit 4 during August 1986. (NSte: performance tests were conducted on the unit prior to this outage to establish baseline conditions with the circular burners.) The low NO, burners have been undergoing field testing since that time for purposes of*shakedown and final tuning to the Cal cones os Feeder ZEA Sy | Hammer Ta Crusher P.A. Venturi Flowmeter Duct Heater Tempering P.A./Coal Air Inlet Combustion | aw I~ Panels UTube & muttiple VN Sampling ot ¢ Venturi Location Scrubber xchanger 4 Flowmeters Furnace Flue Gas Main Combustion Air Fan Recirculation Venturi Flowmeter Preheater Oil/Gas Fired Air Fan Combustion Recirculation Chamber Fan Additional Insulation (Speckled) Additional Insulation Additional Insulation Insulated Mounting Plate 52° Existing Refractory Existing 22' Firebrick Front Wall South Side Wall Rear Wall North Side Wall Figure 7 LWS Test Facility Arrangement and Furnace Layout. Table 2. COAL AND ASH CHARACTERIZATION Coal Source: Mine: Pittsburg No. 8 County: Belmont State: Ohio Raw Coal and Ash Data (As Received Basis) Coal Chemical Analysis: Ash Chemical Analysis: Proximate Analysis: SiO, 47.44 Alz03 20.55 Total Moisture: 2.20 TiO, 0.87 Volatile Matter: 34.10 Fes03 25.05 Fixed Carbon: 51.01 CaO 1.76 Ash: 12.69 MgO 0.52 NERO) 2.00 Btu per Ib: 12344 K20: 0.32 P20s: 1.11 S03: 0.00 Ultimate Analysis: Ash Fusion Temperatures (°F): Moisture: 2.20 Reducing Oxidizing Carbon: 69.18 Hydrogen: 4.67 IDT 2118 2331 Nitrogen: 1.15 AST 2248 2558 Sulfur: 3.33 AHT 2457 2584 Ash: 12.69 AFT 2487 2607 Oxygen: 6.78 Calculated Results of Characterization Analysis Rank: High Volatile Bituminous A FC/VM Ratio: 1.50 C/H Ratio: 2.49 Ibs Nitrogen per MKB: 0.93 Results Interpretation: Ash Type: Eastern Bituminous Slag Propensity: Medium Foul Propensity: High Edgewater unit. These tests are in progress at this writing, but some data will be presented to indicate results thus far. RESULTS AND DISCUSSION AIR FLOW TESTS Air flow tests performed on the Dual Register burner verified that the axial primary jet penetrated the swirl-induced secondary recirculation zone downstream. In practice, the reverse flow along the boundary of the fuel jet returns hot gases to the ignition zone to stabilize the flame. The HT-NR flow pattern, while similar, has several important differences. Primary jet momentum is disrupted at the nozzle exit by obstructions into the flow stream and the bluff body effect of the collar. This acts to cause localized recirculation patterns at the outer boundary of the primary jet as it leaves the nozzle and reduces primary jet momentum. Swirled air in the inner secondary air zone follows the contour of the Air Separation Vane and attaches to the highly swirled air exiting the outer (tertiary) zone along the burner throat. This highly swirled air stream, physically separated from the primary jet, develops a low pressure zone downstream, which in turn draws the air rapidly back in, developing a recirculation zone. In subsequent combustion tests, with sufficient outer air swirl, it was possible for the secondary induced recirculation pattern to overcome 10 the primary jet and sharply reduce flame length. Pressure drop was measured during the air flow tests and related to velocity head to predict pressure drop with preheated air. This led to a decision to alter the Air Separation Vane during the XCL air flow tests to reduce resistance. Air flow patterns of the XCL generally followed those of the HT-NR, although maximum swirl generation was less. Pressure drop readings across the XCL were reduced 32% by streamlining flow and other modifications. Figure 8 compares velocity profiles measured at a plane in the burner throat for the three burners. The break in the curves is due to low velocity recirculation in the area downstream of the inner air zone, which could be evidenced by wool tufts but not accurately measured. The HT-NR displayed the highest velocity profile peak due to highly swirled air moving in a narrow band along the wall of the throat. The XCL showed a similar pattern but at lower velocity. Concerns about high pressure drop discouraged raising the swirl generating capability of the XCL to match the HT-NR. V Local/V Outer Zone lh pach 1 1 6 8 10 12 14 16 18 20 Distance from Burner Centerline (in.) Figure 8 Velocity Profile Comparison. DRB, HT-NR, & XCL Burner. COMBUSTION TESTS Dual Register Burner The Dual Register burner (with a 35 in. throat) was the first burner tested in the LWS in this program. NO, appeared unusually low for the DRB firing bituminous coal, with results averaging 210 ppm at 3% oO, (all concentrations are reported at 3% 0 » or 0.29 1b/MKB, at full load and 18% éxcess air. Flames appeared somewhat lazy? and flame length in every case greatly exceeded the 22 ft criterion. Unburned carbon loss (UCL) was in the 3 to 6% range (reported as percent combustible in fly ash). The facility operators advised that UCL levels of less than 10% should be considered very good in this furnace and that field values would be lower. An impeller, of the type used in B&W circular burners, was installed in the DRB burner nozzle to simulate circular burner performance. Previous operating experience with DRB burners and impellers, circular burners, and the LWS test 11 Table 3. BURNER PERFORMANCE COMPARISON Field LWS. Data@ Data? Circular Burner (with impeller) 700 (700) DRB with Impeller 640 660 DRB without Impeller 330 210 - NOx @ 3% O. 5 > Approximate. results are compared in Table 3. The field results (opposed fired utility boiler) indicated the DRB with this type impeller produced No, emission similar to a standard circular burner (640 and 700 ppm respectively). This is not unexpected: this impeller rapidly disperses the coal into the secondary air, eliminating the delayed diffusion/fuel rich zone normally produced by the DRB. The field data for the DRB burners without an impeller were 330 ppm, representing a 48% reduction relative to the DRB plus impeller burner. By comparison, the LWS tests resulted in 660 ppm No, for the DRB plus impeller (close to the 640 ppm field value) but produced 210 ppm without the impeller (versus 330 for the field). The LWS test resulted in a 68% NO, reduction by removing the impeller. The conclusion was drawn that the DRB tésted in the LWS was producing uncharacteristically low NO, . This was attributed to the oversized burner throat (refer to Test Apparatus for a description) and lower velocities with this DRB compared to standards. While the low NO, performance was virtuous, the long flames exceeded the project objectives. Since the large throat DRB results were significantly different than expected, a second DRB designed to B&W standards was later tested to establish baseline emissions for the program. The standard DRB with a venturi diffuser in the burner nozzle averaged 278 ppm NO. (0.38 1b/MKB) at full load, 18% excess air, and NO, ranged from 230 to 319 ppm depending on burner adjustments. NO, emissions weré also sensitive to load and excess air. UCL results were good, Fanging from 2 to 6%. Flame length was controllable to about 20 ft with high swirl and high burner 400; 400, NOx Load Carbon in Ash 14 © 80MKB @ 4 60MKB & o 350) 45MKB @& 350) 12 -~ 300 10 -~ 300 3 9 3 & g & ® 8) 2 S = 250 5 E 250 s : 6S 45 MKB < 200 oN. lS 200 2 sSaMKs 4 oe \. 150 a Q 150 Full Load (80 MKB) Conditions - 80 MKB 2 © Conical Diffuser fat o 4 Venturi 100 J 100 115 120 125 130 135 1.40 115 120 125 130 135 1.40 Burner Stoichiometry Burner Stoichiometry Figure 9 Dual Register Burner Performance with Figure 10 Dual Register Burner Venturi Conical Diffuser. vs. Conical Diffuser. 12 differential, 10 in. H,O. Reducing swirl brought windbox pressure in-line with objectives, but flame fength exceeded 22 ft Flames were stable at all test conditions, regardless of load or register adjustments. CO emissions were less than 50 ppm except at low excess air test conditions. The standard DRB was also tested with the B&W conical diffuser in the burner nozzle. NO typically averaged 268 ppm (0.37 1b/MKB) with the conical diffuser and ranged from 202 to 320 ppm, depending on burner settings at full load with 18% excess air (Figure 9). Point-to-point comparisons with the venturi (Figure 10) showed approximately a 12% reduction with the conical diffuser. UCL was satisfactory, and other performance was like the venturi diffuser nozzle. In summary, the DRB satisfied the NO» UCL, and mechanical requirements, but required excessive burner resistance to meet the flame length criterion. The standard impeller reduced flame length but with high NO. emissions. HT-NR Burner The HT-NR burner was tested in several configurations, but only the optimum configuration is detailed here. Shakedown tests indicated that use of the conical diffuser ahead of the flame stabilizing ring produced lower NO. results (about 40 ppm) than the swirler at the best conditions of both. The conical diffuser was used for all subsequent tests. The air register assembly was modified with BHK's assistance during the tests to generate more swirl. Prior to register modifications, NO emissions were typically 240 ppm (full load, 18% excess air) with flame lengths of 20 to 21 ft., UCL about 3%, and burner resistance of 7.5 in. H,0. 2 Following register modifications for increased swirl, NO. emissions at full load (18% excess air) averaged 195 ppm (0.27 1b/MKB) and rangéd from 190 to 310 ppm, depending on register adjustments. Full load NO. emissions varied from 147 to 297 ppm as excess air increased from 7 to 37%. Unbutned carbon was low (Figure 11), amounting to about 3% at full load with 18% excess air. Flame length increased to 400; 400; NOx load Carbon in Ash 14 380 80 3 GOMKE 360 350 oO 5SOMKB @ 340 12 320 S 300 S 300 10 x % 280 x . g © 260 c 8 8 & 240 a 250) Ss x & > $ 220 5 6a 200 = S 180] 200 S 160} - 140] 150 120 2 100! 114 1.18 1.22 1.26 1.30 1.34 1.38 Burner Stoichiometry 100 115 120 125 130 135 140 Burner Stoichiometry Figure 12 XCL Burner Performance Compared to Figure 11 HT-NR Burner Performance. HT-NR Burner at 80 MKB. 13 over 22 ft. with a corresponding burner resistance of 7 in. H,O. At maximum swirl conditions, flame length could be reduced to about 19 ft., bu€ burner resistance increased to 13 in. H,0. Flame stability was excellent over the load range regardless of burner adjustments. CO emissions were typically less than 25 ppm at full load. The effect of coal fineness on No, was briefly investigated with the HT-NR burner. Fineness (expressed as percent paSsing a 200 mesh sieve - minus 74 micron) was increased from 70 to 82% without any change in NO.. However, when primary air flow was increased 12% in addition to the higher fineness, NO. decreased from 190 to 170 ppm, about 10%. Corresponding flame length exceeded 23 ft. In summary, the HT-NR proved capable of very low NO, emissions with low levels of UCL and excellent flame stability, and satisfied me¢hanical size limitations. NO, emissions were typically 30% less than the DRB with venturi. Pressure drop and flame length exceeded the program objectives. XCL_ Burner The XCL burner tested in the LWS was equipped with the Air Separation Vane, and the secondary air zones were sized in accordance with HT-NR standards for the flow rates and velocities. The XCL was first tested with a conical diffuser in the burner nozzle. At full load with 18% excess air, NO, emissions were 237 ppm (0.32 1b/MKB) with burner resistance of 5 in. HO, and UCL*was 7%. Flame lengths exceeded 22 ft. except at very high swirl’ and accompanying high burner resistance (10 in. H,O). Flame stability was excellent over the range of load and burner adjustments. A flame stabilizing ring was then installed in combination with the conical diffuser. This resulted (Figure 12) in No, emissions (full load, 18% excess air) which averaged 204 ppm (0.28 1b/MKB) and ranged from 190 to 276 ppm depending on burner adjustments. Burner resistance was typically 3.5 in. H,O and flame lengths exceeded 22 ft. except at very high swirls and high resistance’ (12 in. H,0). Screening Test Evaluation Following these tests, the results of the DRB, HT-NR, and XCL burners were evaluated relative to the program objectives, summarized in Table 4. The NO, emission limit of 0.5 1b/MKB was easily met by the DRB, HT-NR, and XCL burnets. The DRB was typically less than 0.4 1b/MKB, and the HT-NR and XCL achieved less than 0.3 1b/MKB. In comparison to uncontrolled emissions, the DRB achieved 60 to 62% NO. reductions, while the HT-NR and XCL burners achieved over 70% NO reductions. Combustion efficiency was simultaneously very good, with low CO levels and unburned combustible in the fly ash of 2 to 7%. (Note: the facility operators' experience indicated that UCL levels in the field would be lower, while NO, emission was predicted to increase about 10% in the application at Ohio Edison's Edgewater 4 Unit.) The primary concern was the strong tendency of the flames to exceed the 22 ft limit at full load conditions. In most cases, high secondary air swirl would reduce flame length to just under 22 ft. But the accompanying burner air side resistance would increase to 8 to 13 in. H,O and greatly exceeded the program objective in this regard. It was concluded that use of secondary air swirl to control flame length was not satisfactory. The alternative proposed was to reduce 14 fuel jet momentum to reduce flame length while maintaining moderate burner resistance. The XCL had demonstrated very low NO, emissions with low burner resistance, so the decision was made to pursue flame length optimization using the XCL burner. Table 4. BURNER PERFORMANCE SUMMARY - FULL LOAD, 18% EXCESS AIR (LWS TEST FURNACE) NOx Flame Burner Unburned NO, Reduction ppm Length AP Combustible Relative to Condition (@ 3% Op) Ib/MKB (ft) (in. w.c.) (% in fly ash) Uncontrolled (%) Program 367 0.50 S22 <5.0 5 Objectives Uncontrolled 700 0.95 13 25 2-3 0 (Circular Burner) a DRB & Wide 660 0.90 16 - 1-2 6 Impeller DRB & Venturi 278 0.38 >22> 5.5 2-6 60 DRB & Con. 268 0.37 >22> 45 2-6 62 Diff.© HT-NR & Swirler 280 0.38 >22P 5.0 3 60 HT-NR & Con. Diff. 240 0.32 21 7 3 66 HT-NR & Con. Diff. 195 0.27 >22> 7.0 3 72 Plus High Swirl XCL & Con. Diff. 237 0.32 >22> 5.0 7 66 XCL & Con. Diff. & FSR 204 0.28 >22> 3.5 NA 71 a - Estimated results b - Flame could be shortened to less than 22 ft with 8-13 in. burner AP c - Conical diffuser XCL BURNER CHARACTERIZATION TESTS The XCL burner was subsequently tested with several devices designed to reduce fuel jet momentum, including burner nozzle modifications to reduce exit velocity and a device which preferentially reduced fuel velocity while accelerating primary air velocity. In addition, several types of impellers were used to impart radial deflection to the fuel jet and thereby reduce the axial component of velocity. Four of the configurations proved to be capable of simultaneously satisfying program objectives for NO, flame length, and burner resistance. The most versatile of these was an* impeller used with the standard burner nozzle, as shown in Figure 13. This modified impeller imparts less deflection on the fuel stream compared to the standard impeller reported previously with the DRB. Figure 14 shows the relationship between flame length and NO_ at full load, 18% excess air conditions. The flame length could be varied by burner adjustments much more effectively when the XCL was equipped with the impeller. The data in Figure 14 represent moderate burner resistance; i.e., 3.2 to 5.5 in. H,0. The lowest NO levels were achieved by partially retracting the impeller info the coal nozzle*to lessen its ability to flare the fuel jet. A shorter flame (with higher NO_) was repeatedly produced with the impeller at the normal location at the nozzle*exit. A longer flame (with lower NO.) was repeatedly produced with the impeller retracted to the preferred alfernate setting. However, the impeller in positions intermediate to these produced a long or short flame depending on how the condition was approached. (Note: Figure 15 represents repeatable conditions that were outside of the hysteresis band.) 15 Air Control Pitot fixed. Disk Manifold / Tube Wall Opening Adjustable Vanes Impeller Spin Vanes Figure 13 Babcock & Wilcox XCL Burner with Impeller. ~ s a Staged Fixed Vanes o Adjustable Vanes © Staged Fixed Vanes © Adjustable Vanes tO 80 MKB 80 MKB 18% Excess Air 18% Excess Air = a 2 10 o So = = mo ow ® a E £ s § 6 8 = x 4 2 0 12 14 #16 18 20 22 = 24 1.10 1.20 1.30 1.40 Flame Length (ft) Burner Stoichiometry Figure 14 NOx vs. Flame Length for XCL Burner. Figure 15 Combustion Efficiency for XCL Burner. XCL Burner Plus Modified Impeller. XCL Plus Modified Impeller. Unburned carbon loss was in the 4 to 7% range during these tests (Figure 15), and was expected to be lower in field applications. However, B&W decided to modify the air swirl vanes in the outer (tertiary) zone to improve peripheral air distribution in order to reduce UCL. Dual stage swirl vanes were installed in the XCL outer zone for this purpose. Figure 14 shows that NO, and flame length relationships were nearly the same. Unburned carbon was excellent (Figure 15), in the 2 to 3% range (full load, 18% excess air). Burner resistance was typically 4.1 in. H,0 with this arrangement. Therefore, with a modified impeller and dual stage air vanes, the XCL burner satisfied all program criteria simultaneously. NO. was variable from about 0.3 to 0.6 1b/MKB (0.5 1b/MKB goal) with: flame lengths ¥rom over 22 ft to 12 ft (22 ft goal), burner resistance near 4 in. H,O (5 in. H,0 goal), and UCL of 2 to 3% (5% goal). The mechanical size limitations were also satisfied in this design. 16 Table 5. XCL TEST COALS Fuel Lower Analysis (as received) Kittaning Utah Moisture 3.66 5.46 Ash 10.75 10.00 Volatile Matter 22.28 38.26 Fixed Carbon 63.31 46.28 Higher Heating Value Btu/Ib 13330 11930 600: 16 800 NOx Fuel Carbon in Ash / Coal Analysis (as rec'd) O Pittsburgh 8 7 SS ——— & Utah VM 33.46 550 © Lower Kittaning . ‘ 14 750 FC 47.84 Load: MCR (104 MWe) / ‘ Moisture 9.48 ° i Lower Kittaning Pitt 8g Ash 9.50 é ‘ HHV (Btu/Ib) 11772 ’ 500 12 700 Sulfur 172 f U ’ _ & g 450 10 Z 650 9 5 g 8 g mo ea 3 ® 400 8 § ® 600 5 E 5 7 z 3 z s 2 x s x 9 So 350 6 S$ 550 < - Lower Kittaning —= 300 4 500 250 2 450 O NOx @ Carbon in Ash 200! 400 1.10 1.20 1.30 1.40 1.05 1.10 1.15 1.20 1.25 Burner Stoichiometry Burner Stoichiometry Figure 16 Fuel Effects: XCL Burner Plus Figure 17 Baseline Performance of Ohio Edison’s Modified Impeller. Edgewater Station - Unit 4 with Circular Burners. To more fully characterize the XCL burner in this configuration, additional tests were conducted on two alternate coals--a western bituminous Utah coal and a lower volatile eastern bituminous (Lower Kittaning). Key fuel parameters are summarized in Table 5, and full load performance results are shown in Figure 16. The Utah coal resulted in No, emissions and unburned carbon levels very similar to the base Pittsburgh 8 coal. *The Lower Kittaning coal produced higher NO, emissions and higher unburned carbon levels as well. The previous discussions on the significance of volatile matter on NO, control support this trend; i.e., the lower volatile coal released less fuel nitrogen and generated fewer hydrocarbon radicals in the fuel rich portion of the flame and consequently increased overall NO, emissions. LIMB DEMONSTRATION With concurrence from Ohio Edison and the EPA, B&W manufactured 12 full-scale XCL burners for installation at Ohio Edison's Edgewater Station Unit 4. Figure 5 shows the furnace and burner arrangement. (Note: investigation outside of this 17 program had determined that the optimum sorbent injection locations were in the upper furnace, meaning that no sorbent injection equipment was installed in the burners.) Prior to the XCL burner installation, baseline tests were performed on the boiler with the existing circular burners. B&W inspected and tuned the burners prior to these tests. Figure 17 summarizes NO. and UCL results from the tests and also lists pertinent information about the*coal. The excess air values were determined from a multipoint gas sampling grid installed at the airheater inlet. These readings were compared to 0, measurements taken in a vertical plane at the entrance to the boiler convéction pass (at the furnace exit). The difference in excess air measurements between the furnace exit and the airheater is due to air infiltration to the setting. These tests were performed at three loads to account for infiltration variances with load. The results were used to correct excess air values measured at the airheater inlet in all subsequent tests. In essence, NO emissions were typically 650 ppm (0.89 1b/MKB) at full load with 3% excess 0, and 2.0% carbon in the fly ash (from isokinetic dust sampling). Flames were about 12 ft long. The XCL burners were installed during a 6-week outage which concluded in September 1986. Early test results, listed as Condition 1 in Table 6, indicated that NO emissions were slightly above the program goal of 0.5 1b/MKB and that flames were relatively short. This indicated a need to tune the burners to the Edgewater furnace. The bottom burner elevation had longer flames than the other rows as a probable consequence of their proximity to the furnace hopper cooling surface. Consequently, these burners were not adjusted further. The impellers in the upper three burner rows were retracted and produced results listed as Condition 2 (Table 6). NO, emissions were reduced to 0.38 1b/MKB but CO emissions unexpectedly increased to high levels. Flame lengths were acceptable. B&W sampled furnace gas near the walls at several locations in the combustion zone to determine if H,S was present at levels which would cause tube wastage. Gas analysis indicated H g levels less than 10 ppm, even under conditions with elongated flames, and these were considered to be innocuous (1). The high CO emissions encountered with retracted impellers had not been observed during earlier single burner tests in the LWS. This led to tests to check air/fuel balance between burners, which Table 6. LIMB - XCL BURNER CHARACTERIZATION/OPTIMIZATION PERFORMANCE DATA co Burner Flame NOx ppm @ % Carbon AP Lengths Ib/106 Btu 3% Oz in ash in. H20 ft <0.50 Program Goals (0.40) <5.0% <5.0 <22 Burner Configuration 1. Initial - 30 deg impeller 0.53 115 ~28 4.5-5.5 15-18 45 deg outer vanes (0.49-0.58) (40-140) 2. Retract impellers 12” on top 3 rows to simulate 0.38 2500 conical diffuser (0.34-0.40) (830-3000) N/A 45-55 18-22 3. Bias air to lower west burner zone (1 test point) 0.40 75 N/A 5.3 18-22 4. Replace retracted impellers on top 3 rows with standard coal distribution equipment 0.35 700 11.2 (conical diffuser/deflector) (0.31-0.41) (100-2000) (6.5-14.6) 4.8-5.0 18-22 All data is full load, normal O2 (3.0-3.5% O2 furnace outlet, 4.5-5.0% Oz at air heater inlet) 18 indicated air deficiency to the left-hand burners. Biasing air to these burners reduced CO to background values while maintaining low NO, (Condition 3 in Table 6). Conical diffusers were installed in the top three burner elevations to permanently provide the mixing function of the retracted impellers. High CO and unburned combustibles returned sporadically (Condition 4 in Table 6), and furnace probing revealed major air/fuel imbalances. (Note: the conical diffusers reduce pulverizer-to-furnace pressure drop and could have changed fuel distribution among burners.) The commissioning process has indicated a tendency in this unit for significant air/fuel maldistribution. This is probably the consequence of furnace size, coal piping and burner arrangement, and pulverizer controls. NO, emissions below present NSPS were readily achieved with good combustion performance. NO, was further reduced to 0.35 1b/MKB, by addition of conical diffusers, but futther air/fuel imbalance resulted. Steps are presently being taken to confirm stoichiometries of the furnace and to balance these mixtures. However, this program has shown the difficulty in achieving very low NO, emissions by burner retrofits to existing boilers, which place a variety of constraints upon adaptation. These difficulties appear resolvable, but require more tuning of the combustion system to the boiler. CONCLUS IONS e The HT-NR and XCL burners produced 25 to 30% lower No. emissions than the Dual Register burner, while providing efficient combuétion. e A strong dependence was observed between No, emissions and flame length. Lowest NO, emissions were usually accompaniéd by relatively long flames, and measures Which reduced flame length increased NO. e Flame length could be reduced by high recirculation induced by high secondary air swirl and accompanied by high burner resistance. Flame length was also reduced by several devices which reduced fuel jet momentum, but NO, emissions increased somewhat. e Retrofit application of B&W XCL burners to a 104 MW utility boiler resulted in No, emissions of about 0.5 1b/million Btu while achieving less than 3% carbon in the ash, and with moderate burner resistance and flame length. NO ports were not required. Lower No, emissions can be attained but require burner/boiler tuning. 19 CONVERSION FACTORS Readers more familiar with metric units may use the following to convert the non-metric units used in this paper. Non-Metric Times Yields Metric Btu 1.055 kJ oF 5/9 (CF-32) °c in. 2.54 cm ft 0.30 m 1b 0.45 kg 1b/in.? 703.1 kg/m? 1b/10° Btu 0.43 kg/10° kJ REFERENCES 1. Chou, S. F., et al., "Fire-side Corrosion in Low No, Combustion Systems," In Proceedings: 1985 Symposium on Stationary Combustiod NO, Control, Volume 1, EPA-600/9-86-02la (NTIS PB86-225042), July 1986. Pershing, S. W. and Wendt, J. O. L., "Pulverized Coal Combustion: The Influence of Flame Temperature and Coal Combustion on Thermal and Fuel NO, " 16th Symposium on Combustion, p.389, Combustion Institute, Pittsburgh, Pa* (1977). Pohl, J. H. and Sarofim, A. F., "Devolatilization and Oxidation of Coal Nitrogen," 16th Symposium on Combustion, Combustion Institute, Pittsburgh, PA (1977) pg. 491. Narita, T. et al., "Development of the Low-NO_. Burner for Pulverized-Coal- Fired In-Furnace NO, Reduction System," In Préceedings: 1985 Symposium on Stationary Combustion No, Control, Volume 1, EPA-600/9-86-02la (NTIS PB-86-225042), July 1986" 20 Technical Paper Montana-Dakota Utilities 830MW AFBC retrofit approach to design and erection B. Imsdahl Generation Manager Power Production Montana- Dakota Utilities Co. Bismarck, North Dakota 58501 H. L. Johnson Product Development Manager Babcock & Wilcox Barberton, Ohio 44203 R. E. Spada Product Development Manager Babcock & Wilcox Construction Co., Inc. Americon Inc. Copley, Ohio 44321 Presented to Energy Technology Conference Washington, D.C. April 14-16, 1987 Babcock & Wilcox BR-1308 a McDermott company —— Montana- Dakota Utilities 830MW AFBC retrofit approach to design and erection B. Imsdahl Generation Manager Power Production Montana-Dakota Utilities Co. Bismarck, North Dakota 58501 H. L. Johnson Product Development Manager Babcock & Wilcox Barberton, Ohio 44203 R. E. Spada Product Development Manager Babcock & Wilcox Construction Co., Inc. Americon Inc. Copley, Ohio 44321 Presented to Energy Technology Conference Washington, D.C. April 14-16, 1987 PGTP-87-22 Abstract Montana-Dakota Utilities Co.’s R. M. Heskett Unit 2 has been retrofitted to atmospheric fluidized bed combustion (AFBC). The unit originally burned North Dakota lignite using the existing spreader feed system. Included in this retrofit project was the installation of an all water-cooled AFB combustor beneath the existing spreader stoker boiler, a new tubular air heater, forced draft fan, AFBC circulation pumps, and other auxiliaries. The new AFB combustor and auxiliaries were designed and installed by Babcock & Wilcox. The integrated approach taken to the design and erection, the major design parame- ters, arrangement, and expected performance of this utility AFBC retrofit project will be discussed in this paper. Introduction Montana-Dakota Utilities Co. has undertaken an extensive program to upgrade and convert Unit 2, at its R. M. Heskett Station in Mandan, North Dakota, to fluidized bed combustion. Objectives of forced draft fan controls and auxiliary fluid bed systems. Description of original unit the retrofit program were to complete demolition and erection within a five-month span, increase the unit capacity, and improve overall perfor- mance. This original unit, a spreader stoker-fired boiler, is believed to have been the largest of its type in the country. After retrofitting, the unit is still firing Beulah North Dakota lignite, reusing the existing spreader feeder system. In fact, the entire boiler convection pass and furnace enclosure walls have been reused. Few modifications to existing pres- sure parts were required and only some new com- ponents were added: a tubular type air heater, The original Unit 2 (Figure 1) at Montana- Dakota’s Heskett Station, consisting of a Riley lignite-fired stoker boiler and a General Electric turbine, started commercial operation November 1, 1963. Selected design data of the original boiler and turbine are: Original Boiler Capacity .......... 650,000 lb/hr Steam Temperature ... Steam Pressure ..............0.002000e Feedwater Temperature .................. 443°F Nominal Turbine Rating .. 81.2 MW Turbine Throttle Flow ............ 682,700 lb/hr Economizer Collector Overfire Air Fans ( ———— Main Steam Outlet Division Wall Spreader/ Feeder Figure 1 Original Unit 2 - R.M. Heskett Station. Convection Pass Arrangement The convection pass contains superheater, steam generating and economizer surfaces. The super- heater is an all pendant type with spray attem- peration for steam temperature between the primary and secondary banks. The generating surface has a 60-in.-diameter upper drum and a 36-in. lower drum with all long-flow heating surface. The economizer surface is a bare-tube- counterflow type. The superheater and generating bank enclosure is a water-cooled tube and tile construction while the economizer enclosure is refractory and insulation lined. Both enclosures have a cold gas-tight casing. Fuel Feed System The coal feed system consists of three bunkers, three conical distributors, and ten stoker spreader feeders which are evenly spaced along the front wall of the furnace. Each unit has a separate cup- type rotary volumetric feeder with a drum-type rotary flipper. Ash Reinjection System The stoker ash reinjection system was comprised of a knock-out hopper located beneath the lower boiler drum, which removed large pieces of unburned char from the flue gas stream. The col- lected ash was then pneumatically reinjected into the furnace through the lower rear wall. The air for the system was provided by a separate cinder return fan. Backend Equipment The flue gas leaving the economizer was passed to a multiclone mechanical dust collector, an induced draft fan, and electrostatic precipitator. Ash was not reinjected from the mechanical dust collector due to the low fusion temperature of the lignitic ash. If ash were reinjected, grate slagging would have occurred. Operational history The steam generator on Unit 2 had fouling and slagging problems throughout its operational life. The first modification involved the rearrange- ment of the three furnace wing walls. The heating surface of the wing walls was increased in an attempt to reduce the flue gas temperature enter- ing the superheater thus eliminating the tube fouling problem. This modification did not achieve its desired purpose. Further attempts to combat the slagging and fouling included many different fuel additives, but none of these additives were successful in reducing the slagging or fouling for long-term periods. The installation of water lances near the high temperature superheater tubes was successful in removing some of the slag on the superheater, but had limited overall success in improving continu- ous steam output. Because of the slagging and fouling experienced when the unit was loaded to near original rating, low combustion efficiency was experienced. This caused an increase in unburned carbon in the pit ash and carbon carryover by the flue gas into the mechanical dust collector. Another problem experienced in burning the Beulah lignite was the buildup of a porcelain-type coating on the generating tubes between the two boiler drums. This deposit on the generating tubes was directly related to the amount of sodium found in the lignite ash. To remove this deposit, the unit had to be shut down twice a year for water washing since the installation of sootblow- ers was impractical due to the close tube spacing. Options available to improve operation The options investigated by Montana-Dakota Utilities Co. to improve overall unit operations and return the capacity of the unit to its design level were: ¢ Change fuels from lignite to higher cost Montana subbituminous, having a lower ash content. When tested, the grate ran hotter with this coal leading to increased maintenance problems. e Add pulverizers or cyclones. This major modification was not feasible as the boiler was not large enough. e Adapt the unit to fluid bed operation. After an evaluation of the options, the utility decided to convert the unit to atmospheric fluidized bed combustion (AFBC). This system offered greater fuel flexibility and lower operating temperatures. Babcock & Wilcox was selected to redesign, modify and erect the boiler. The retrofit specifications required the steam flow of the unit to be increased to 700,000 lb/hr. With this increased steam flow, the turbine gen- erator would be able to reach its rated capacity. Retrofit design approach At the inception of this project, it was recognized that close coordination between the customer, B&W, and those involved within B&W, had to be established to achieve the goals in the project schedule. Discussions began immediately to co- ordinate the engineering and shop fabrication schedule with the erection span. Based on its sys- tem load demand, the demolition and erection span was determined by Montana-Dakota to be a maximum of five months between November 1986 and March 1987. Given time for demolition and initial operation of the retrofit, the time span for erection was determined to be approximately 3-1/2 months. An innovative approach was required to insure completion of the project. The approach taken in the preliminary design of the AFBC retrofit was to minimize the modifi- cations to the pressure parts, utilize existing aux- iliary equipment to the greatest extent possible and minimize the down-time required for demoli- tion and erection of the retrofit components. Furnace and Fluid Bed With respect to minimizing pressure part changes, R.M. Heskett Unit 2 provided an ideal retrofit candidate. Preliminary engineering indicated the plan area of the existing unit was adequate to provide the required fluidization velocity. There- fore, the furnace walls did not require modifica- tion. The fluid bed plan area is approximately 40 ft wide by 25 ft deep and is constructed of water-cooled membrane construction. The fluid bed contains both boiling surface and superheater surface. The boiling surface is located in the front two-thirds of the bed, horizontally positioned, and spans the entire 40-foot width of the unit. The superheater is located in the rear third of the bed horizontally spanning the entire 40-foot width of the unit. Because of the abrasiveness of the selected bed material, all in-bed surface is provided with addi- tional wall thickness. Furthermore, erosion-type shields will be installed in those areas where higher erosion rates are expected. The entire in- bed tube bundle design and tube spacing will be set with adequate clearance to prevent bridging of potential oversized bed material. The entire distributor plate is a refractory covered water-cooled membrane. The fluidizing and combustion air is introduced to the bed from the windbox by means of bubble caps which are welded into the membrane between the floor tubes and the distributor plate. The windbox located below the water-cooled distributor plate is divided into four main com- partments for load control, in which additional compartmentalization is provided to facilitate start-up operations and to allow for partial com- partment fluidization. These compartments run parallel with the depth of the unit. Economizer ~ | Primary Superheater Circulation ~~ Division Wall Dust Collector Air Heater Main Steam Forced Line Draft Fan Pumps Ash Removal Secondary System Superheater Figure 2 Unit 2 retrofitted with atmospheric fluid bed combustor. The enclosure wall tubes wrap horizontally around the perimeter of the bed, exit from the enclosure, and feed a portion of the existing division wall. The in-bed generating surface is also horizontal and feeds a portion of the existing division wall. The division walls were modified by removing the headers on the rear of the unit and a portion of the lower wall which was located in the furnace. The side view of the retrofitted unit is shown in Figure 2. Convection Pass The additional capacity required a review of the existing convection pass. This review revealed that the maximum convection pass gas velocity of the original unit was within design limits for a fluid bed. (Due to the increased flue gas flow at 700,000 lb/hr.) If no changes were made to the thermal performance of the air heater, modifica- tions would have to be made to the convection pass to maintain acceptable velocities. Unit efficiency was improved to maintain the same maximum flue gas flow rate as the original unit. Thus, no convection pass changes were required. Also, the performance of the mechanical dust collector and electrostatic precipitator was not affected. Selected performance parameters for the retrofitted fluid bed combustor are: Fluidization Velocity ............. 12 ft/sec Normal Bed Temperature ......... 1500°F Bed Depth .............02: cece eee eee 51 in. Overall Excess Air ...............0005 25% Air Heater Gas Exit Temperature ... 275°F Bed Material ....................20005 Sand Air Heater The structural design of the regenerative air heater was reviewed with the vendor. Two prob- lem areas became apparent; namely, the ability of the housing to withstand the higher forced draft air pressure and the ability of the sealing system to adequately seal. Both problem areas are due to the higher required static air pressure for the fluid bed in comparison to the stoker unit. The struc- tural review by the vendor determined it may not be feasible to structurally modify the casing to withstand this higher static pressure. The deter- mination of replacing the regenerative air heater with a tubular air heater was based on the air-to- gas side leakage. The tubular air heater has zero air-to-gas leakage since the air heater tubes are rolled into their respective tube sheets. In addi- tion, the tubular air heater was designed so that its gas side pressure drop would be compatible with that of the existing regenerative air heater, allowing the reuse of the induced draft fan. Structural Steel The existing stoker was supported by its own grid steel system. Thus, the weight of the stoker was not carried by the furnace walls or the top boiler support steel. If the fluid bed were designed to be supported by the top boiler support steel, the increase in load would require major modification to the steel to increase its load carrying ability. The bed enclosure was designed to be bottom sup- ported so that the existing structural steel would not be required to support any additional weight. By bottom supporting the bed enclosure and designing the bed itself to have the same physical dimensions of the existing furnace, the lower fur- nace wall headers also could be reused. A seal was provided between the lower furnace wall headers and the bed enclosure to prevent leakage of air into the furnace and provide for expansion between the two components. Boiler Circulation Pumps Since the new fluidized bed combustor enclosure walls, floor, and in-bed boiler surfaces are mostly horizontal, water circulation through these cir- cuits must be pump-assisted. Three, 50% capacity, wet motor-type pumps will be installed to pump these circuits. Only the new fluidized bed com- bustor water circuits, all of which connect directly to the existing furnace wing walls, are being pumped. All the remaining furnace enclosure walls and boiler bank will remain in natural circulation. Bed Ash Removal System To remove bed material, seven letdown systems consisting of individual drain points, downspouts, valves, and water-cooled screw conveyors have been installed. Since sulfur capture is not a requirement, a small amount of sand is used to maintain a bed. A major concern with this retrofit is the removal of oversized bed material, particu- larly the “egg” type clusters. These “eggs” were discovered during the test burn of the Beulah lignite. The number, sizing, and location of the bed drain systems to be installed are set by over- sized material removal requirements. The water-cooled screw conveyors will operate continuously and discharge to two bed ash collec- tion screw conveyors. These conveyors will dis- charge to a flight conveyor which will convey the ash to a 39-1/2 ton ash storage silo. The dis- charge from this storage silo will go to the exist- ing ash handling system. Coal Feed System The stoker coal spreaders were modified with vari- able speed drives in order that the coal flow could be controlled by each feeder. Fabrication and erection As mentioned previously, Montana-Dakota sys- tem’s electric demand dictated that the unit could only be out of service for a period of five months between November 1986 and late March 1987. In order to erect the modifications in this period of time, it was recognized that major components, the bed enclosure and in-bed surface, would need to be shop modularized. The following limitations were reviewed before arrangement of the modules could be finalized: e Shipping e Shop Fabrication ¢ Erection (including limitations imposed by the existing building steel) Truck shipment was determined to be the pre- ferred method of transportation. This decision was based on the shorter transit time of truck vs. rail shipment and the overall impact on the fabri- cation and erection schedule. Special permits were required for truck transit due to the size and weight of the modules. Module width was not a transportation issue due to building limitations at the plant site. The actual transit time for one of the modules was one week, the other two modules took somewhat longer due to inclement weather. The width of the module was limited by the building structural steel to approximately 12 feet . This limitation dictated that the bed be modu- larized into three subsections, each module being approximately 10 feet deep. Two of the modules included the generating surface while the third included the in-bed superheat surface (Figures 3, 4, 5). The heaviest module, the superheater module, weighed approximately 55 tons. Before truck shipment, the modules were trial assembled in the shop to check tolerances between headers. The as-built tolerances were found to be acceptable. With the module being completely assembled, tolerances between headers were critical, since these headers could not be eas- ily aligned in the field for welding. Figure 3 Shop assembly of bed enclosure floor with bubble caps installed. Figure 5 Generating module being lifted from truck at plant site. In order to move the modules into position beneath the existing furnace, a rolling track (Figure 6) was designed on which the modules would be rolled into the building. This rolling Figure 6 Rolling track in place beneath existing furnace. track was perpendicular to the sidewall of the boiler. Once the modules were positioned under the building, they would then be rolled on addi- tional rolling track to their proper position. The critical area of coordination between the fabrication shops and the erector was the on-line computerization of the fabrication, shipping and erection schedules. Any change in one schedule could be evaluated by all those involved who would then determine how their particular sched- ules would be affected. This would especially aid the erector since he was constantly aware of shipping schedules and could staff the job site as needed to insure the smooth flow of erection work to complete the job on schedule. The erection of the bed modules is a good example of the ability to coordinate work based on on-line scheduling. After the first generating module reached the plant site and was positioned on the rolling track, it arrived at its final position under the furnace within a period of one hour. The second module to be positioned was the one containing the in-bed superheater, and the third module positioned was the center one containing the remainder of the in-bed generating surface. Once the modules were positioned beneath the furnace, the headers were fitted with strong backs and the assembly was raised into its final posi- tion by a hydraulical jacking system (Figure 7). When positioned beneath the existing furnace, all pressure part and seal welding was done. The length of time required, from the arrival of the first module to completion of welding activity, was fourteen working days. Other components that were modularized included the air heater, modifications to the fur- Figure 7 View showing modules and cribbing after modules were jacked into final position beneath existing furnace. nace division walls, and bed enclosure walls. Again, the on-line scheduling system greatly enhanced the ability to complete the erection of these components on schedule. Project schedule The following are the milestone dates of the pro- ject schedule: Contract Award .............. January 1986 Begin Demolition of Existing Unit ............. October 14, 1986 Demolition Complete ...... October 31, 1986 Start In-Bed Module Material Shipment ..... November 19, 1986 Start Erection of In-Bed Modules ................. November 26, 1986 Complete Erection of In-Bed .................5 December 17, 1986 Hydrostatically Test Unit .............. 22. e eee February 18, 1987 First Coal Fire in Retrofitted Fluid Bed Combustor ...... Week of April 1, 1987 Conclusion The Montana-Dakota Utilities Co., R.M. Heskett, Unit 2 project illustrates how a fluid bed combus- tor retrofit can be a technically and economically attractive choice for unit upgrades and plant improvements. Also illustrated is the importance of integrating the requirements of the design, fabrication and erection schedule with the cus- tomer’s outage span to insure that erection can be completed within that time. Furthermore, this project exemplifies how flu- idized bed combustion can satisfy the utility market’s increasing need for power plant upgrades, utilization of available low cost fuel, and emission reductions in a cost effective manner. Also worth noting is the fact that this project is currently the largest utility AFBC con- tract funded solely by the utility itself. References Gorrell, R. L., and Strong, D. N., “Description of an 80MW Fluidized Bed Retrofit at Montana- Dakota Utilities Co.” EPRI Seminar on Atmo- spheric Fluidized Bed Technology for Utilities Application, Palo Alto, CA, April 8-10, 1986. Imsdahl, B., Gorrell, R. L., and Johnson, H. L., “Montana-Dakota Utilities 80 MW AFBC Retro- fit”, Joint ASME/IEEE Power Generation Con- ference, Portland, OR. October 19-23, 1986. (86-J PGC-Pwr-A). Goblirsch, G. M., Bensor, S. A., Hajicek, D. R., and Cooper, S. L., “Sulphur Control and Bed Material Agglomeration Experience in Low-Rank Coal AFBC Testing”. The 7th International Conference on Fluidized Bed Combustion, Phila- delphia, PA, October 25-27, 1982. Technical Paper BR-1316 80 MW fluidized bed retrofit for Montana-Dakota Utilities Co. - What changed? What didn’t? R. L. Gorrell T. J. Haynes Domestic Fossil Operations Babcock & Wilcox Barberton, OH D.L. Knighton Black & Veatch Engineers Architects Overland Park, KA Presented to The Ninth International Conference on Fluidized Bed Combustion Boston, MA May 3-7, 1987 Babcock & Wilcox a McDermott company 80 MW fluidized bed retrofit for Montana- Dakota Utilities Co. - What changed? What didn’t? R.L. Gorrell T. J. Haynes Babcock & Wilcox Barberton, OH D. L. Knighton Black & Veatch Engineers Architects Overland Park, KA Presented to The Ninth International Conference on Fluidized Bed Combustion Boston, MA May 3-7, 1987 ABSTRACT To increase unit capacity and improve overall unit performance, Montana-Dakota Utilities Co. has undertaken an extensive program to upgrade and convert its R. M. Heskett Station Unit 2 to fluidized bed combustion (FBC). The existing unit is stoker fired and burns North Dakota lignite. During the design of the conversion, emphasis was placed on reuse of as much of the existing equipment and boiler pressure parts as practical. However, certain modifications were necessary because of the arrangement and operational requirements intrinsic to the fluid bed combustor. This paper discusses many of these modifications and changes necessary to both the boiler and balance of plant equipment. INTRODUCTION Babcock & Wilcox (B&W) provided the fluidized bed unit, its auxiliary equipment, and the subsequent construction. Black & Veatch Engineers - Architects (B&V) provided balance of plant design which included connecting B&W furnished equipment with the existing system, procurement of auxiliary equipment, and construction specifications for B&V scope of work. Montana-Dakota Utilities Co. provided construction coordination of all contractors as well as performing some demolition and minor construction with plant personnel, The overall project schedule was about 18 months duration with the construction lasting from October 1986 through February 1987. DESCRIPTION OF ORIGINAL UNIT Furnace Arrangement The overall general arrangement of the unit was typical of most spreader stoker-type boilers (1). A sectional side view of the original unit is shown in Figure l. The furnace is approximately 40 ft wide by 21 ft deep and contains three water-cooled division walls. The furnace wall construction is of a water-cooled tube and tile type with a cold gas-tight casing. The division Main +—Steam | Tl | Outlet Economizer Division wall Oust Collector NK Spreader/ Feeder Air Heater Forced Oraft fan Overtire Air Fans. “I Figure 1 Side view of existing boiler at Montana-Dakota’s R.M. Heskett Station, Unit 2. walls, which are fed by downcomers from the lower drum, penetrate the rear furnace wall, rise through the furnace and connect directly to the upper drum. Convection Pass The convection pass contains superheat, steam generating, and economizer surfaces. The superheater is an all pendant type with an attemperator for steam temperature control located between the primary and secondary banks. The generation surface has a 60 in. diameter upper drum and a 42 in. lower drum with all long flow heating surface between. The economizer surface is a bare-tube-counterflow type. The superheater and generating bank enclosure is a water-cooled tube and tile construction whereas the economizer is refractory and insulation lined. Both enclosures have a cold gas-tight casing. Fuel Feed System The coal feed system consists of three bunkers, three conical distributors, and ten Stoker spreader feeders which are evenly spaced along the front wall of the furnace. Each unit has a separate cup-type rotary volumetric feeder with a drum type rotary flipper. Back-end Equipment Flue gas and air handling equipment consist of a multiclone-type dust collector, one regenerative-type air heater, one FD and ID fan, and an electrostatic precipitator. DESCRIPTION OF FBC RETROFIT The fluidized bed enclosure (Figure 2) is approximately 40 ft x 26 ft and is constructed of membraned water cooler tubes. The enclosure wall tubes wrap horizontally around the bed plan area and then exit vertically upward along the outside of the rear furnace wall. The bed side of the enclosure wall tubes are studded and covered with refractory. The floor tubes run parallel with the 40-foot length and are also refractory covered. The windbox below the floor tubes is divided into four main compartments parallel with the 26-foot bed depth. Each compartment is further subdivided for start-up and part load operations. The bubble caps which provide the air flowpath between the windbox and the fluid bed are welded into the membrane between the floor tubes. The caps extend above the refractory on the floor tubes to prevent erosion of the floor refractory. Within the fluid bed enclosure there is both boiling and superheat surface. All of the surface is horizontal and runs parallel with the 40 ft bed enclosure dimension from wall to wall. Approximately the front two thirds of the bed contains boiling surface, while the rear third is superheat surface. All of the in-bed surface has additional wall thickness due to the abrasiveness of the selected bed material, sand. And, in addition, erosion shields have been installed on the tubes in areas where the highest erosion rates are expected. WHAT CHANGED? Air Heater and Fans The existing regenerative air heater and associated duct work were removed to allow for the installation of a new tubular air heater. The new tubular air heater is a three-gas pass, single-air pass arrangement. There were two major constraints that had to be considered when designing this air heater: 1) The air heater had to fit in the limited area available, and 2) the gas side pressure drop had to be within limits for the existing ID fan which was reused. Pamary Superheater WU Econonnzer Dust Collector Aw Heater Main Steam Forced Line Dealt fan Bowler Cucutation Pumps Windbox Ash. Removal System Secondary Superheater Figure 2 View with fluidized bed and new air heater added. The existing FD fan was removed and replaced with a new TLT-Babcock centrifugal double width, double inlet fan. The new fan was sized to provide all the necessary flow for the overfire air system and the cinder reinjection system in addition to the secondary air. This made it possible to eliminate the existing overfire air and cinder reinjection air fans and duct work. With the installation of the new FD fan a new glycol coil air heater was also installed. The existing glycol heating and circulating system was, however, reused. Coal Feed System Variable frequency drives were installed on each coal feeder motors to allow independent operation of each feeder. However, within the control system, the drives will be grouped to run in sets based on the compartmentation of the fluid bed. The only other changes necessary to the feeders were minor modifications to the trajectory plates and spreader padders to readjust coal distribution. Sand Handling System For bed material, a washed, double screened, and dried sand was specified. The washing and screening requirements could be met by local sand suppliers, but none had drying equipment. Thus, on-site provisions for drying had to be included in the new sand handling and feed system. Washed and screened sand will be transported to the plant site by dump truck, stockpiled on the ground and covered by canvas. On a daily basis, or as needed, sand will be transported to the drying and handling system by a front end loader. The sand drying system consists of a feed hopper with feeder and lump breaker, a vibrating hot-air bed dryer with associated gas burner, blower, and clean-up cyclones, and a small surge hopper. From the drying system, sand is transported by a dense phase system to two new 30-ton storage silos installed between the existing three coal silos. Two new belt type feeders, one under each storage silo, will feed sand to the FBC unit. The sand will be fed by gravity through two ports on the front wall of the unit above the bed area. The control of the sand handling system will be performed from local control panels. Bottom Ash Removal System The existing bottom ash hopper and bottom ash transport system was removed and replaced with a bed material removal system to handle the hot sand and ash drained from the fluid bed. The bed material drain and cooling system consists of seven water-cooled screw-type coolers located under the windbox of the FBC unit. Under normal operation all seven screws will run continuously, removing approximately 4.2 tons of bed material and ash per hour total. Ash and spent material from the water-cooled screws is collected and conveyed by a system of a screw and drag- type conveyor to a new 32 1/2 ton capacity ash storage hopper installed near the unit. Ash from this storage hopper is transported to an existing fly ash silo by the existing vacuum conveying system. In order for the new bed material disposal system to be installed, a pit had to be excavated underneath the fluid bed enclosure. The bed drain screw coolers are at ground level. The bed ash collection conveyors, the bed ash transfer and transport conveyors are located within the pit. The difference between the ground floor elevation and the pit floor elevation is 4ft 8in. Performance By changing over to fluid bed combustion, the boiler capacity will be raised to meet the existing turbine capacity. The existing unit maximum steam flow was 650,000 lb/hr at 1300 psig and 955° F, With the retrofit the unit will produce 700,000 lb/hr steam at the same conditions. The normal operating temperature for the fluid bed will be 1500° F, At this reduced combustion temperature many of the slagging and fouling problems associated with the fuel, Beulah lignite, will be eliminated (2). This will allow continuous boiler operation at the rated steam flow. The tubular air heater installed will reduce the air heater gas outlet temperature by 70° F. This drop in gas temperature will increase overall unit efficiency by approximately two points. Test burns of the Beulah lignite in the B&W/EPRI 6 ft x 6 ft and 1 ft x 1 ft test rigs also indicate that carbon losses from the unit will be greatly reduced by the conversion to fluid bed combustion (1). Structural Steel The fluid bed enclosure, the in-bed tubes, and the sand inventory for the fluid bed combined weigh significantly more than the existing stoker grate and associated equipment. To avoid major modifications to existing pressure parts and structural steel, the FBC unit was separately bottom supported. The existing stoker steel and foundation were reused, where practical, minimizing the amount of additional foundation and steel. The tubular air heater, which replaced the existing regenerative air heater, also required additional support steel due to its increased size and weight. As with the fluid bed, the air heater had to be bottom supported. Controls and Instrumentation A Bailey Controls Network 908 distributed system was installed to control the fluid bed. To support this new control system, an uninterruptible power supply also had to be installed. Included in the new logic are controls for the following: ° Bed level Furnace draft Tubular air heater cold-end metal-temperature Steam dump valve Firing rate Compartment air flow Overfire air Steam temperature and turbine bypass Coal feeder interlocks Feedwater and turbine controls remain as part of the existing pneumatic control system. Control of the bed ash disposal system, which operates continuously during boiler operation, is performed from local panels. Only a single CRT operator interface unit was installed. In case of failure of this unit, critical control loops within the distributed control system are backed up by hard-wired, board-mounted manual-automatic stations. Existing Control Panel Modifications The existing control panel in the main control room is an L-shaped vertical-type panel. On the boiler gage board, all the old pneumatic manual/automatic control stations associated with the stoker-firing combustion control and steam temperature control were removed. In their place, a modified board was installed containing all the necessary hard-wired stations of the new control system, control switches for the new FD fan and circulation pumps, other pushbottoms, indicators and recorders, In addition, a 60-point annunciator and two 30-point multipoint recorders were installed for alarm annunciation and temperature monitoring of the new equipment for fluidized bed operation. Circulation Circulation through the entire unit was natural prior to the fluid bed retrofit. After retrofitting, the reused furnace wall of the original unit remains in natural circulation; however, the entire fluidized bed enclosure walls, in-bed steam generating bank and fluidized bed floor must be pumped. Three 50% wet-stator boiler-circulator pumps were installed to supply the required head for these pumped circuits. The downcomers which had fed the existing division walls were rerouted to the boiler circulation pumps. The lower drum was baffled to split the water flow between the natural circulation of the boiler and the pumped circuit. Cooling Water System Cooling water was required for the boiler-circulating- pump motor coolers, the pump fill and purge, and the bed ash drain cooler, all installed as part of the FBC conversion. For the new boiler circulating pump motors, and the fill and purge supply cooler, the existing closed- cooling-water system was utilized. However, the existing closed-cooling-water heat exchangers already exhibited a high pressure drop and were not capable of any additional cooling water flow. To utilize this system without increasing the existing water flow rate, the supply needed for the pumps was taken from the existing closed-cooling-water system return header which operated at an acceptable temperature. Two full capacity booster pumps were installed to provide the additional pressure head required for the pressure drop across the boiler- circulating-pump heat exchangers and fill-and-purge supply cooler. The returning water was sent to the existing cooling water head tank. Thus, the cooling water flowrate through the existing closed-cooling- water system was not increased, but the system heat load was increased by approximately 12 percent. Cooling water for the new ash coolers is supplied from the existing service water system (river water). Individual strainers and pressure reducing valves are provided for each ash cooler. The cooling water is discharged into the existing circulating water discharge piping. Pressure Parts With the FBC arrangement selected, pressure part changes were kept to a minimum. On the water side, the downcomers which fed the existing division walls were re-routed to the boiler circulation pumps, and the vertical headers which fed the division walls were removed. The division wall tubes are now fed by the bed enclosing tubes which enter the furnace through the rear wall, and the in-bed boiling bank tubes which leave the bed vertically, forming division walls in the lower furnace. On the steam side, the existing main steam line was cut at a convenient point and a new line routed to the new in-bed superheater. This line also contained a new spray water attemperator and a connection for extract steam for the plant steam eductor vacuum system. From the in-bed superheater, a new main steam line was installed back to a convenient point on the existing main steam piping to the turbine. This piping is cold sprung and required an expansion loop to meet the allowable forces and moments at its connections. Routing of this lie is such that there is only one low point in the line, and a motor-operated drain valve is installed at this point. An existing main steam stop valve and electromatic relief valve were relocated from the convection superheater outlet to the in-bed superheater outlet. A new safety relief valve was also installed downstream of the in-bed superheater, as well as a new atmospheric vent valve which is provided to maintain cooling steam flow, if required, through the in-bed superheater after a unit trip. Turbine Bypass System During start-up of the fluidized bed (prior to admitting steam to the turbine), steam flow must be established and maintained through the in-bed superheater to prevent overheating. Since the amount of steam flow required is significant, and may be sustained for a period of several hours, a turbine bypass system was installed so that the steam could be condensed and the condensate recovered. Prior to the retrofit, any steam generated before placing the turbine on line was vented to the atmosphere. The turbine bypass system consists of new steam piping from downstream of the in-bed superheater to the condenser with a combination pressure reducing/desuperheating valve. Inside the condenser an internal distribution header was installed above the condenser tube bundle. During start-up the steam flow rate is manually controlled and is modulated to keep the in-bed superheater outlet steam temperature within limits. Overfire Air System Like stoker firing, a significant amount of combustion can be expected above the bed with fluidized bed combustion. Because of this, a portion of the total combustion air can be introduced as overfire. A portion of the overfire air system was changed to insure adequate capacity and improve mixing in the freeboard, and to accommodate changes in the overfired air conditions. These changes included increasing nozzle sizes, removal of those nozzles which would have jet impingement upon pressure parts and reinforcement of ductwork where necessary because of higher air pressure, Two automatic dampers and flow monitors were also added to the system so that the total overfire air flowrate can be varied with load. WHAT DIDN'T CHANGE Convection Pass All of the existing economizer, boiling, and superheat surface in the convection pass was reused without modification. Sootblowers remain at their original locations, and the existing multiclone dust collector required only normal maintenance. Pressure Parts With the exception of tube bends added to several of the front wall tubes to allow the sand feed ports to be cleared, no modifications were made to the front, rear, or side walls, and headers. The three division walls in the furnace remain intact but now are fed from the new in-bed generating tubes and bed enclosure tubes. With the exception of the baffle added in the lower steam drum, both drums remain unchanged, and no modifications were made to the steam/water separators in the upper drum, Overfire Air The existing headers for the overfire air and stoker air were reused in their original location. No additional penetrations were made into the boiler casing to modify the air nozzles, although the upper front wall penetrations were enlarged to accommodate the larger nozzles. The existing spreader cooling air fan and ductwork remain unchanged. Controls The existing pnematic controls for the turbine and feedwater remain as they were prior to the retrofit. The sequencing of the vacuum ash removal system for the multiclone hoppers and precipitator hoppers remains independent of the distributed control system. Furnace Plan Area As mentioned, no major changes were necessary to the existing furnace walls to accommodate the FBC addition. This is due to the fact that the desired heat release rate per unit of plan areas for bubbling fluidized bed combustors is similar to that desired for stoker firing. Thus the bed was sized, based on the maximum available plan area. The active area of the bed is approximately 20 percent larger than the existing furnace plan area. With this bed size, the resulting fluidization velocity at maximum capacity will be about 12 ft/sec. Electrostatic Precipitator and ID Fan No changes were made to the electrostatic precipitator or the fly ash disposal system. No mechanical modifications were made to the ID fan except the replacement, in this area, of the existing pneumatic control loop with a new electronic control loop to improve fan response time. Ash Reinjection System The ash reinjection system remains as it was for stoker firing. There are 13 injection lines that transfer ash from the convection pass boiling bank hoppers back into the furnace. The spacing and number of ports has not changed with the retrofit. Coal Feed and Preparation System No changes were required for the coal size reduction and classifying equipment. The coal feeders have been rebuilt for maintenance reasons with no major hardware modifications made. The feeder spacing across the front wall and elevation also remains as it was for stoker firing. The spreader cooling air fan and ductwork will also remain unchanged. CONCLUSION This project illustrates how an FBC retrofit can be a viable, and economically attractive choice for unit upgrades and plant improvements. REFERENCES 1. R. L. Gorrell and D. N. Strong, "Description of 80 MW Fluidized Bed Retrofit at Montana-Dakota Utilities Co." EPRI Seminar on Atmospheric Fluidized Bed Technology for Utilities Application, Palo Alto, CA, April 8-10, 1986. 2. B. Imsdahl, R. L. Gorrell, and H. L. Johnson, "Montana-Dakota Utilities 80 MW AFBC Retrofit," Joint ASME/IEEE Power Generation Conference, Portland, OR., October 19-23, 1986 (86-JPGC-Pwr-A). Technical Paper BR-1318 Conversion of a Recovery Boiler to Bark Burning J. A. Barsin, Member ASME, TAPPI Manager, Industrial Projects Babcock & Wilcox Barberton, OH 44203 J. Proterra Sr. Utilities System Engineer Georgia Pacific Corporation Atlanta, GA 30348 G. Stewart, P.E. Power/ Staff Engineer BE&K Engineering Company Birmingham, AL 35202 Presented to: TAPPI Engineering 1987 New Orleans, LA September 1987 Babcock & Wilcox a McDermott company Conversion of a Recovery Boiler to Bark Burning J. A. Barsin J. Proterra G. Stewart, P.E. Manager, Industrial Projects Sr. Utilities System Engineer Power/Staff Engineer Babcock & Wilcox Georgia Pacific Corporation BE&K Engineering Company Barberton, OH 44203 Atlanta, GA 30348 Birmingham, AL 35202 Presented to TAPPI Engineering 1987 New Orleans, LA September 1987 Abstract Georgia-Pacific Co.’s Crossett, Arkansas mill recently completed the installation of one of the U.S. pulp and paper industry’s largest biomass-fired boilers. The project includes an extensive biomass fuel receiving and handling system, involving receipts from trailer trucks and rail cars, a stacker-reclaimer, and an overfeed spreader feeder system. A retired 900-ton recovery boiler has been converted from black liquor firing by installation of a spreader stoker, new economizer, and tubular air heater. A unique firing system combining a control combustion zone or CCZ furnace bottom with a water-cooled vibrating grate is used. Particulate emissions are controlled by a wet venturi scrubber; and a wet ash handling system is used for sluicing ash to an ash pond. The fuel handling system, boiler, and scrubber are remotely controlled by a distributed-process control system that also controls a recently installed chemical recovery complex. Operating results from start-up through the first sixteen months of opera- tion have been excellent and the project is considered a success. Introduction Georgia-Pacific (GP) operates a large integrated pulp and paper mill in Crossett, Arkansas which produces in excess of 1400 dry tons per day of various grades of bleached kraft paper. Steam generation in the mill is approximately 1.2 million lb/hr, which is supplied by a 1500-ton recovery boiler, a 400,000 lb/hr wood-waste boiler, and two power boilers. Because GP wanted to minimize use of natural gas as a boiler fuel and had an existing retired recovery boiler which could be converted, it was decided to proceed with this fuel conversion project as a means of reducing GP’s energy cost per ton of product. Biomass Fuel Handling System The biomass fuel for the boiler consists mainly of low-grade whole-tree hardwood chips. Approxi- mately 500,000 tons per year are handled by the fuel handling system. About one-third of this volume is shipped to Crossett by rail from GP’s Fordyce, Arkansas plywood mill. The other two- thirds is brought in by trucks from local sources. The fuel handling system is designed to receive and process unsized ‘“‘rough” woodwaste. The rail car unloading system includes a car puller anda rail car unloader, which is hydraulically powered and designed to unload four 70-ton rail cars per hour. Each car is individually positioned on the unloader, which tilts up and over to dump the contents of the car into the receiving hopper, alongside of the rail siding. The system did not require any extensive excavations which would have been necessary with a rotary unloader or bottom dump hopper system. The unloader dis- charges into a receiving hopper provided with five parallel drag chain conveyors which transfer the unloaded fuel to a belt conveyor. The receiving hopper/conveyor system is designed to handle up to 300 tons per hour of fuel when the truck system is idle. During periods when truck receipts are being accepted, the rail unloading system is designed to operate at one-half capacity or 150 tons per hour (Figure 1). The truck receiving system consists of a single hydraulically operated truck dumper. The truck dumper is of the back-on type, rated at 55 tons Figure 1 Stacker reclaimer. capacity and able to handle up to six semi-trailer trucks per hour. A receiving hopper with chain conveyor and spike roll is provided for the truck dumper and is rated at 145 tons per hour. This system discharges onto a belt conveyor. Eventu- ally, a second truck dumper may be added and so space has been allowed in the layout of the facility. The belt conveyors from the rail car and truck receiving operations merge on a common 60-inch- wide belt conveyor that carries fuel to the screen- ing and shredding operations area. This conveyor contains a self-cleaning metal and magnet detec- tor for removal and detection of tramp metal. The screening and shredding operations are accomplished by a disc screen and two vertical shaft biomass shredders. The screen is designed to accept biomass of 2-in. size and smaller. The rejects are sent to the shredders for size reduction to 2-in. x 0-in. A flop gate is used to direct the screen rejects from one shredder to another or to divide the flow. Each shredder is rated to reduce up to 80 percent of material flow to minus 2-in. size at 150 tons per hour unloading rate. Screened and sized biomass is conveyed to a stacker reclaimer. The stacker boom rotates in a 280° arc to build a doughnut-shaped active stor- age pile. Active storage capacity is 6600 tons of biomass fuel. Inactive storage capacity of 25,000 tons is provided. A scraper reclaimer boom rotates in a 315° arc to reclaim fuel into a buffer storage pile. Normally, half of the active storage area is used for receiving while the other half is being reclaimed. The buffer storage pile serves to insure an uninterrupted flow of fuel to the boiler while the scraper reclaimer and stacker boom are being repositioned and the pile is being dressed. The railcar receiving hopper is kept full as an auxil- iary boiler fuel supply. The screw reclaimer under the buffer storage pile feeds onto an inclined belt conveyor to the boiler which in turn feeds an over-feed drag chain distributing conveyor. The distributing conveyor feeds five screw-feeder bins which meter fuel into the airswept fuel spouts on the boiler front. The feeder bins provide only about 15 minutes of fuel storage and are intended to provide a constant head of fuel on the feeder screws. The feeder screws are arranged in pairs, each screw with its own direct current (DC) motor. Each pair of motors is controlled by a DC drive system which allows variable speed operation of the screw feed- ers. The controls are set up to vary the speed of the five sets of screws in parallel, in response to load changes with biasing of individual feeders allowed. The distributing conveyor also discharges an excess amount of fuel onto a return conveyor. This amount is normally about 35 tons per hour, but can be as high as 125 tons per hour depending on boiler load. The return conveyor is actually the same belt as the stacker infeed conveyor. There- fore, all returned fuel from the boiler is returned to the active storage pile by the stacker reclaimer. Steam Generator Objective GP planners decided to convert a 900 ton, 900 psig, 830°F, 454,000 lbs steam/hour retired recov- ml sind esd esd dined e iC Figure 2 Existing retired recovery steam generator. ery steam generator (1968 vintage) to a power boiler, fired by wood, and thereby permit the dis- placement of steam generated by oil and gas fuels in the existing pdwer boiler. Maximum possible steam generation was to be attained (Figure 2). Development Background Converting an existing steam generator from black liquor to an alternative solid fuel presents the challenge of maximizing steam generation when starting with an existing fixed-sized fur- nace. Project economics do not usually permit major furnace modifications and, therefore new combustion techniques must be applied to achieve increased steam outputs with acceptable performance. The conversion of a recovery steam generator to a wood-fired power steam generator is not novel -- many units have been successfully converted. This specific conversion, having established a clear objective of maximizing steam generation, provided a unique opportunity to apply an advanced, highly turbulent, combustion system that would permit the massive bark processing necessary to meet the steam flow objective. The wood chip fuel was specified to contain 50% average moisture (55% peak), 2.2% ash, and a Btu content of 4850 Btu/pound with natural gas as the standby back-up fuel. Preparation (sizing) was to be -2 inches. Fuel Firing System Several biomass fuel-firing systems were investi- gated for this project. They included conventional traveling grate spreader stokers, sloping grate systems, and water-cooled vibrating grates. Also, different types of overfire air systems were con- sidered such as the high volume low pressure type and the low volume, high pressure type. Many of these fuel firing systems have been installed on biomass burning boilers in the pulp and paper industry with varying degrees of suc- cess: the limitations of each are well known and documented. However, GP was looking for a unique solution to the problem of fuel-firing sys- tem selection. This project involved the conver- sion of an existing recovery boiler, which meant that several factors in design were already fixed such as furnace dimensions, and boiler drum design pressure. Also, it was desired to use the existing superheater, if possible, to hold down the project cost. The steam-side design limitation effect upon maximum boiler capacity indicated that a maximum continuous rating (MCR) of 514,000 lb/hr at 900 psig operating pressure was attainable, based on calculated superheater pres- sure drop and the 1035 psig rating of the steam drum. The specifications for the boiler conversion called for maximum steam generation from bio- mass. If the MCR of 514,000 lb/hr was to be achieved on biomass, the technology available in biomass-firing systems would be stretched beyond historically established limits. For a fuel moisture content of 50 percent average (55 percent maxi- mum), semi-suspension firing, i.e. spreader stoker (SS) has been the traditional method. However, the majority of these SS systems have been limited to approximately 400,000 lb/hr MCR on biomass because of grate depth and width limita- tions and grate heat release rate. The limit here is 1.0 x 106 Btu/ft2-hr, considering grate design temperature and unburned carbon loss resulting from excess velocity of undergrate air. Even at this nominal level of heat release, most semi- suspension systems in the past have used multi- clone dust collectors and sand classifiers to return unburned carbon to the grate. These reinjection systems are subject to high maintenance and risk of fires. Therefore, if their use could be avoided without undue loss in thermal efficiency, it would be beneficial. The maximum fuel moisture and the generous furnace volume available indicated that an alter- native to the standard traveling grate combustion system could be an attractive application. It has been demonstrated that improved air/fuel distri- bution enhances carbon utilization and that recy- cling high temperature furnace flue gases across the flame front improves combustion when using high moisture or low volatile fuels (1). Also, air side staging increases downstream turbulence, mixing and combustion efficiency (2), and higher combustion air temperatures aid in raising the low adiabatic flame temperatures inherent with high moisture wood fuel (3). B&W had already applied a novel two-zone fur- nace concept to a new power boiler for Crown Zel- lerbach’s Elk Falls, B.C. mill. The furnace was fit- ted with asymmetrical arches and multiple Over Fire Air (OFA) nozzle rows. The Elk Falls furnace had been cold-flow modeled (1/4 full size), and the OFA matrix statics and quantities were based upon the evaluation of the modeling data. Elk Falls had been in operation and had been tested, confirming predicted performance, prior to B&W’s undertaking of the Crossett retrofit. This exten- sive laboratory and field developmental program provided excellent technical back-up for the Cros- sett conversion (Figure 3). The combination of arches and flexible high- static OFA system referred to as the Control Combustion Zone (CCZ) system duplicates the approaches utilized for the successful firing of low volatile (2%), high ash (33%) pulverized Korean anthracite. The low volatility requires that the WINDS WE 4S TRIBLTOR SPOUTS] BABCOCK DE TRON / ROTOGRATE STOKER Figure 3 Elk Falls controlled combustion zone furnace hottest furnace gases be recirculated across the flame front to aid in stabilizing the flame and to heat the carbon. The arches and OFA system together establish the recirculation pattern and provide a high temperature particle residence time in the lower furnace, approximately twice as long as anon-arch/ OFA furnace. The addition of arches to a conventional retrofit unit would reduce furnace volume and therefore, the re- stricted residence time might increase unburnt carbon levels. In the Crossett retrofit case, the furnace volume is generous by B&W’s evaluation, and the arch/OFA system addition should be beneficial for sustained unsupported combustion of the high-moisture wood chips (Figure 4). To improve fuel-distribution control and mea- surement, five separate and independently bias- able twin-screw metering feeders were applied for the fuel feeds, and five refuse-type airswept spouts were utilized for fuel feeders. All the air streams were measurable and controllable to the following subsystems: The stoker, the auxiliary gas burners, the auxiliary NOx system, and the over- fire air system. A stoker that permitted higher-than-standard air supply temperatures was required for the Figure 4 Air flow patterns (modeled). upmost flexibility in fuel moisture and fuel quan- tity. The application of the improved combustion system (furnace arch and OFA) was to require an equivalent improvement in grate release rates to justify the additional capital costs invested in dis- tribution and control. A Hydrograte was attrac- tive because it did permit up to 600°F undergrate air temperatures, and the suppliers would permit up to 1,000,000 Btu/ft2 release rates for this fuel. Testing indicated that with proper fuel and air distribution, release rates higher than that limit could be attained. The initial sizing of the system utilized a value of 1,113,000 Btu/ft2 grate release rate which, after operation, proved conservative. GP decided on the Hydrograte versus a traveling grate because of reduced maintenance and asso- ciated downtime. Design Approach The basis of the design approach was to minimize capital costs and meet or better the specified per- formance. The furnace arch/OFA system com- Figure 5 Integration of hydrograte cooling with boiler circuitry. bination (CCZ) was the most cost-effective of all those reviewed as it required the removal of only 30-feet of the existing lower furnace. The removal of the original floor, primary air and secondary air zones would have been required with any approach. The primary and secondary superheat- ers were not revised, but gas temperatures enter- ing the superheater were modified by adding fur- nace surface (arches). A spray water condenser system and spray attemperator were added to control final steam temperatures. The boiler generating bank was not revised. However, the steam drum internals were completely replaced with cyclone separators and primary and secon- dary steam scrubbers to insure steam water sepa- ration at the higher predicted steaming rates. An economizer was added to reduce boiler outlet gas temperatures and improve thermal efficiency. Also, a tubular air heater was added to provide high temperature air to the stoker and OFA sys- tem (547°F), and reduce exit gas temperatures to 300°F to further improve thermal efficiency. A steam coil was provided to aid in start-ups and in protecting the cold end of the tubular air heater from potential corrosion. A natural gas Low NOx Burner System, capable of generating 525,000 pounds steam/hour was also supplied as a back- up for the wood fuel. A major innovation concerned the integration of the Hydrograte stoker cooling circuit within the boiler circuitry. This novel approach reduced the costs that would have been incurred by integrat- ing the Hydrograte cooling with the feedwater circuit. The Detroit Stoker Co.’s Hydrograte consists of four grate sections. The grate bars are bolted to the boiler water tubes (for cooling) and are drilled for high resistance to improve undergrate air dis- Figure 6 Georgia Pacific conversion side elevation. tribution. Each of the four grates are sloped towards the ash hopper and are fitted with a vibrating drive which cycles to aid in removing ash from the grate. The stoker ash discharge tank and ash sluice system were also included. Down- comers from the mud drum were utilized as sup- plies for the Hydrograte cooling circuit. The grate was to be periodically vibrated to remove ash, and so both the water supplies and water/steam risers had to be designed with sufficient flexibility to withstand this periodic movement. More than suf- ficient flexibility was provided and both riser and supply lines had to be hydraulically snubbed to reduce boiler-induced vibration by adding shock absorbers (Figure 5). The CCZ furnace system was expected to reduce particulate carryover, however, the reduction could not be quantified at the time of contract and dust loadings were predicted to not exceed 4 grams/dry standard cubic ft (DSCF) at 50% excess air and 2.22% ash in the fuel. The increase in grate release rates depended upon the successful increase in suspension firing. To insure that fuel was suspended, the overfire air portion of the CCZ system booster fan was designed to take suction from the new FD fan and sized to provide 24-inch HO static at the OFA nozzles with 38% of the total air flow required. The system is hot, takes suction from the air heater hot outlet, and also supplies distributor air to the windswept spouts to assist in distributing fuel. Pulsation dampers are provided in the air ducts to vary the distributor air pressure to assist in uniformally spreading fuel from front to rear on the grate and increasing the amount of fuel in suspension above the grate. No sootblowers were added and the existing blowers were removed in total, maximizing net steam available to the plant and eliminating those maintenance costs associated with the blowers. Operation over the past 18 months has indicated that the sootblowers were not required (Figure 6). Flue Gas Particulate Control The primary air pollutant of concern was particu- late emissions from biomass firing. These emis- sions are removed by an AirPol wet venturi scrubber. This method was preferred over other methods such as electrostatic precipitation because of the lower risk of fire, which has been known to plague precipitators on biomass appli- cations. The scrubber was designed to meet 0.1 lb/MBtu particulate at 10-in. w.g. pressure drop. The scrubber is presently operating at 13-in. to meet this required emission rating. A unique feature of the scrubber system is the location of the induced draft (ID) fan. Most biomass-fired boilers employ a multiclone dust collector ahead of the ID fan or a final particulate cleanup device, be it a scrubber or precipitator. The multiclone traditionally served a dual pur- pose: First, to capture unburned carbon or char, allowing it to be reinjected into the furnace; and secondly, to scalp the larger size particulate prior to entering the fan, thus reducing erosion. The fir- ing system was designed to limit unburned car- bon loss without reinjection, and limit particulate carryover. The dust collector would protect the ID fan from erosion if no scrubber were utilized. Decisions concerning which air pollution con- trol device to apply involved weighing some pros and cons. The potential advantages (of a scrubber) would be the elimination of the mainte- nance headaches and fire risks of the multiclone and lower flue gas system draft losses. Also, the lower temperature of the flue gas downstream of the scrubber would result in lower volume requirements and lower brake horsepower for the ID fan. The possible disadvantages would be ID fan corrosion in a saturated gas environment, fan wheel buildup and resulting imbalance from scrubber outlet residuals, and increased solids load to the scrubber system. The initial comparison demonstrated a signifi- cant energy cost savings in the I.D. fan power with a scrubber arrangement. Also, it was believed that the overall availability of the unit would be improved by the elimination of the dust collector and its associated rotary seal valves and ash handling system. The problem of corrosion of the ID fan was put before fan manufacturers. The major history was found to be in the electric util- ity industry with scrubber booster fans down- stream of flue gas desulfurization (FGD) systems. However, GP’s application is not designed to burn any sulfur-bearing fuels and, therefore, does not produce any SOx gaseous constituents in the flue gas, which could affect the corrosivity of the satu- rated flue gas. In fact, the scrubber uses a pulp mill waste water stream which has a pH ranging from 11.6 to 12.0, which is expected to neutralize any acid-forming flue gas components. GP also investigated its existing flue gas scrubbers on wood-fired boilers and found there was no serious corrosion problem in scrubber components or stacks, even when fabricated in mild steel. The confidence in the fan’s ability to survive in the wet environment was increased based upon these observations. The potential for buildup of material on fan blades was also investigated. The heat of com- pression through the ID fan, which is designed for a maximum static pressure rise of 35 inches, w.g., would increase the gas temperature about 10°F above the saturated inlet condition. Over an extended time period, residual solids in the flue gas from the scrubber could dry out or cake on the fan wheel. A water spray on the fan inlets was provided to cool the gas stream and mitigate this effect. The scrubber solids load was predicted to be a maximum of 5 grains/DSCF, which is about twice the design value normally used when downstream of a multiclone collector. This was not deemed to be a problem by the scrubber manufacturer. The bleed rate from the recycle tank is somewhat higher than normal because of the higher inlet loading, which actually is expected to help pre- vent concentration of small amounts of acidic compounds from NOx or CO9. The decision was therefore made to locate the ID fan downstream of the scrubber. This decision resulted in a decrease in installed motor capacity of 600 hp. The estimated average brake horse- power savings is about 325 hp. The fan has a radial tip-wheel design, and is provided with an inlet damper and a variable frequency electric motor drive system. The wheel is fabricated from A517 steel, the shaft is 1040 forged steel, and the housing is fabricated from Corten steel. Other Environmental Concerns NOx emissions were of concern during the opera- tion of the auxiliary natural gas burners. Proven low NOx technology was provided for natural gas firing. Two-stage combustion was utilized and actual NOx was found to be lower than predicted. Ash Handling Bottom ash discharges from the Hydrograte into a water-impounded hopper, which is periodically sluiced to an ash pond. The four sections of the Hydrograte are each provided with a vibrator drive, which is energized on a timer cycle to clean the grate surface of ash and maintain a thin fuel bed. The siftings hoppers below the stoker grate are periodically emptied by pneumatic/hydraulic conveyor system, which discharges into the bot- tom ash hopper. Air heater ash is continuously removed from the hoppers by two rotary seal valves, which dis- charge into a water impounded sand tank. This sand tank is periodically sluiced to the ash pond. Scrubber ash is removed through a bleed stream from the recycle system at about one percent sol- ids concentration. This stream continuously flows to the ash pond. The existing ash pond was reconstructed with a dividing partition to allow one side to receive sluiced ashes while the other side is being dredged. A division gate was provided to direct flow to one side or the other. Decanted water from the pond overflows into an existing system for collection and treatment of mill waste water. Instrumentation and Control The design ‘philosophy for the control system was to minimize direct operator control in the boiler house or biomass fuel system and to automate the system to the fullest extent possible. The control system had to be provided with reliable feedback and feedforward signals, especially in regard to fuel properties and quantities. All normal opera- tor functions are accomplished from a remote location known as the COMP I control room. COMP I houses the CRT operator interface units used to supply the boiler operator with on-line data from the boiler and to allow him to make Figure 7 Control Room process adjustments. This control room is located about 1/3 mile away from the boiler house and also houses the controls for a recently installed chemical recovery complex. The Fisher Provox distributed system was used for control of boiler combustion, drum level, the scrubber venturi throat, etc. All field devices, such as transmitters, control valves, and damper oper- ators are wired to an Input/Output (I/O) rack located in an air-conditioned motor control center adjacent to the boiler house. From there, all I/O data is transmitted on a data highway to COMP I (Figure 7). All discrete logic controls are accomplished by three Modicon programmable logic controllers (PLC). The discretes include areas such as the fuel handling system, auxiliary gas burner interlocks, and draft fan motor logic. The PLC’s are tied into the COMP I Provox system through a program- mable controller interface unit (PCIU) to allow for manual-automatic control of motors where neces- sary. This unit also provides system status infor- mation on all equipment controlled by the PLC’s As mentioned above, the totally automated design philosophy used on this project demanded reliable, high-quality feedback and feedforward signals. A reliable fuel flow signal was one ele- ment of the system that was investigated exten- sively. Several belt-scale designs were reviewed, however, because of the overfeed system employed, two signals would have to be obtained, flow in and flow out, with the difference being the amount burned. The in-flow signal would be rela- tively constant and easy to measure. However, the outflow signal would vary by as much as a 10 to 1 ratio, which would require accuracy of measure- ment over a wider range than that was available commercially. Table 1 Original Upgrade MCR Present MCR Actual at New Wood Firing Initial After Econ. Predicted Max. Steaming Predicted Actual Modification Performance (Before Econ. Mod.) Steam Flow Kib/hr 514,000 505,000* 514,000 650,000* 652,000 Steam Pressure psig 900 885 864 870 866 Steam Temp.°F 830 830** 837 830 837 Feedwater to Econ. °F 350 341 340 378 334 Feedwater to Drum °F 415 437 502 430 467 Boiler Gas Outlet Temp. °F 703 726 - 796 776 Economizer Gas Outlet Temp. °F 614 667 578 689 - Air Heater Gas Outlet Temp. °F 300 336 305 373 373 Air Heater Air Outlet Temp. °F 547 526 - Flue Gas 02 % Dry Volume 5.5 46 5.7*** 5.5 4.5*** Flue Gas COz % Dry Volume - 15.7 15.1*** 15*** Flue Gas CO PPM Dry Volume - 210 590 250 Flue Gas NOx PPM Dry Volume - 132 - Flue Gas NO, Ibs/KKBtu 0.5 0.192 - 0.30 Grate Release Btu/ft?-hr 1,132,000 1,110,000 1,132,000 1,466,800 1,470,000 Input to Furnace KKBtu/hr 773 760 774 1,001 1,004 Grate Temperatures (highest) °F 600 615 555 625 656 Input/Furnace ft2 28.5 28.02 28.54 38.1 38.22 Liberation Rate Btu/ft3 hr 14.2 13.95 14.22 18.4 18.43 Dustloading Grains/DSCF (Ivg. Air Htr.) 40 1.16 - 25 Carbon Utilization % UCL (Fly + Bottom Ash) 1.0 15 - * Average Over 4 Hour Test Period. ** Spray Attemperator Spraying. *** Assuming 24% Moisture in Flue Gas (actuals in initial tests). One area where it was felt that combustion con- trol could be enhanced was fuel moisture content. This parameter could have a significant effect on main steam header pressure control. A flue gas moisture analyzer was provided to give an indica- tion of fuel moisture and to adjust fuel feed and air/fuel ratio accordingly. This boiler should carry full load with biomass fuel, including the widely varying swing loads. Natural gas would be brought on only after sus- tained periods of inability to maintain header pressure with biomass. Project Schedule Work began on the project in March 1984, with the preparation of an appropriation grade capital cost estimate for GP’s board approval. Mean- while, specifications for the boiler modifications were issued at the same time. After board approv- al in May, bid evaluations commenced followed by award of the boiler contract in July. All other major equipment purchases were completed by October, 1984. Field activity commenced in July, 1984, with demolition and site preparation for the fuel han- dling facilities. Site preparation was followed directly by demolition of the boiler components that would not be re-used such as the cascade evaporator, etc. New equipment deliveries com- menced in December, 1984, and reconstruction activities began in January, 1985. Construction activities continued until the end of August, 1985, at which time start-up activities began. The boiler was placed in commercial service carrying between 350,000 and 500,000 lb/hr load in early September, 1985. Performance on Bark The objective of maximizing steam generation was achieved. Actual steam generation up to 650,000 lbs/steam/hour has been demonstrated during sustained operation-relating to an actual grate heat release of 1,470,000 Btu/ft2-hr or 35% above traditional grate release limits. The CCZ system permitted the higher throughputs. Absorp- tion in the superheater was lower than predicted and the carbon loss was higher than predicted (Table 1). Performance on Natural Gas This performance produced 537,000 lbs/steam/- hour at a NOx level of 0.13 Ib NOx/106 Btu. The final steam temperature of 759°F was lower-than- predicted due both to a lower-than-predicted fur- nace exit gas temperature and lower-than- expected superheater absorption. Boiler gas outlet temperatures were 25°F higher than predicted and these temperatures carried through to the air heater gas outlet which was 40°F higher than predicted. Modifications to Achieve Predicted Performance Air The system pressure drop in the OFA was found to be higher than predicted and this resistance limited the amount of OFA to 29% of the total air. This limitation did affect the CCZ system and contributed to the higher-than-expected carbon losses measured. Fuel Initially, the metering twin screw feeders were found to deliver different quantities of wood to the boiler even though their speeds were the same. The larger wood chips would fall into the first two feeders (at south wall) with the remaining feeders receiving progressively smaller chips. The south front of the Hydrograte received wood and ash coverage, leaving the north front half with little or no coverage. This maldistribution could be biased, utilizing the feeder speeds. The south two feeders were slowed down and the north two were accelerated with respect to the central screw feed- ers. The flexibility with separate feeder controls was useful and so was utilized. Currently, the fuel supply system is stabilized and all feeders are run at the same speed with satisfactory grate distribution. Absorption The higher-than-predicted economizer exit gas temperature (16°F high on bark) caused by the under absorption (actual vs. expected) in the superheater and in the economizer was addressed by adding economizer surface. Low absorption resulted in lower-than-predicted superheater attemperator spray flow and could have been improved by adding superheater surface. Soot- blower additions would not have been effective as the surface is clean. After evaluating the options, the most cost-effective approach appeared to be the addition of economizer surface. Additional economizer surface would be added, beyond that required to correct the under absorption, to pro- vide equivalent system thermal efficiency as compensation for the higher-than-predicted unburnt carbon losses. Fourteen rows of econo- mizers were added in total, and gas outlet temper- atures to the air heater are now 578°F. Addition- ally, an ash reinjection system was added to the two air heater hoppers to insure highest carbon utilization. The system thermal efficiency with these enhancements now exceeds 74% on bark. The steam generator currently has the capabil- ity to exceed guaranteed steam generation by 26% and process the equivalent additional bark. Balance of Pressure Parts No plans were made to replace any other pressure parts, but during the construction phase, an oper- ational nondestructive test check of the generat- ing bank indicated that many tubes were below the minimum wall established by code. The entire generating bank was replaced during the rebuild outage. During the initial hydro-test following the rebuild, numerous superheater leaks were dis- covered in the lowest part of the superheater loops. Out-of-service corrosion in the superheater due to oxygen at the wet/dry interface was the cause, and the superheater was replaced in total duplicating the original design. Summary The Crossett recovery conversion was an excel- lent project, completed on time and within budget, exceeding by 26% the original performance objec- tives of generating 514,000 lb/steam/hour firing wood. The desired flexibility in fuel selection for power generation has been achieved with an improvement in thermal efficiency and recycling of retired equipment for minimum capital invest- ment. The reliability of the repowered Process Recovery unit is high and has demonstrated the ability to consume more wood than originally planned. References 1. Barsin, J.A., Boiler Design Considerations, Coal Combustion Technology and Emission Control Conference, California Institute of Technology, Pasadena, California, February 1979. . Fiveland, W., Cornelius, D.K., and Oberjohn, W.J., Como - A Numerical Model for Predict- ing Furnace Performance in Axisymmetric Geometries, paper presented to the American Flame Research Committee, Akron, Ohio, October 1983. 3. Topley, A.W.R., The Controlled Combustion Zone (CCZ) Boiler for Efficient Burning of Waste Wood, a technical paper available from B&W (BWC81-73). we Technical Paper relocating, repowering, and reconfiguring a 125MW industrial power plant J. A. Barsin, Member ASME, TAPPI, J. E. Scheatzle Babcock & Wilcox Barberton, OH C. W. Bush Florida Crushed Stone Corporation Leesburg, FL Presented to Industrial Power Conference American Society of Mechanical Engineers Atlanta, GA Oct. 25-28, 1987 Babcock & Wilcox a McDermott company BR-1320 Relocating, repowering, and reconfiguring a 125MW industrial power plant J. A. Barsin, Member ASME, TAPPI Domestic Fossil Operations Babcock & Wilcox Barberton, OH J. E. Scheatzle ’ Domestic Fossil Operations Babcock & Wilcox Barberton, OH C. W. Bush Florida Crushed Stone Corporation Leesburg, FL Presented to Industrial Power Conference American Society of Mechanical Engineers Atlanta, GA Oct. 25-28, 1987 Abstract PGTP-87-19 Florida Crushed Stone Corporation, a major producer of Florida aggregate and limestone rock base material, generates more than a ton of waste product for each ton of useable rock. The waste product is suitable for Portland cement feedstock, but cement manufacturing also requires fuel and electrical energy. The concept of combining a power plant with a cement plant is believed to be beneficial in lowering the capital and operating costs of both plants. This paper describes the background of this novel combination and the details associated with the redesign, relocation and reconfiguration of a 125-MW pulverized-coal-fired wet-bottom retired steam generator to a 125-MW PC-fired combination limestone calciner steam generator, meeting all New Source Performance Standards (NSPS). This presentation will include initial operational performance data. Background Florida Crushed Stone Corporation, the major producer of construction aggregate and limestone rock base materials in central Florida, had a prob- lem. Its quarry, located near Brooksville, Florida, is one of the largest crushed stone plants in the United States but due to the poor quality of lime rock, the plant generates more than a ton of waste product for each ton of useable rock. This waste material is produced because the rock from the quarry has to be washed to remove soft lime- stone fines and other deleterious material. Dispo- sal of this waste material has been a costly problem. The Brooksville waste material is not suitable as an aggregate and cannot even be used as a stable fill material, but it does contain large quantities of calcium carbonate, silica and some alumina. This is about the same composition as is required in raw materials for the manufacture of Portland cement. A cement plant constructed at Brooksville would therefore benefit from having “free” raw materials. Cement, however, is a depressed indus- try, and even though the volume of cement sales has been increasing annually, the availability of low cost imported cement in Florida has kept prices and profit margins low. Even with very low raw material cost, a new cement plant would not be an attractive capital investment. Other than raw materials, the major costs in manufacturing cement are fuel and electrical energy, and the integration of a cement plant with a power plant could be beneficial in lowering the capital and operating costs of both plants. In addition to the benefit of lower power cost, fly ash from the combustion of coal in the power plant steam generator would provide an excellent source of additional iron and aluminum, two cement raw material components that are not present in adequate quantities in the Brooksville waste material. The power plant would use the hot air from cooling the cement clinker from 2700F to 350F as combustion air; this heat is presently wasted in virtually every cement plant. Also, the combined fuel consumption of the cement and power plant would be large enough to justify buying coal in unit train quantities, which would reduce fuel cost for both the cement and power plant. No cement plant in the United States receives coal by unit train at this time. Additional interconnections between the power plant and the cement plant were adopted as the concept developed, including using hot exhaust flue gas for drying raw materials, augmented with heat provided by steam extraction. New power plants are expensive, therefore, an alternative choice to reduce capital cost was a used power plant. The lowest heat rate plant on the used market in 1984 was American Electric Power Corporation’s (AEP) Mishawaka, Indiana, Twin Branch station Unit 5 that was built in the late 1940s, and “retired” in 1980. This 125MW unit of advanced technological design (for the 1940s) included a Babcock & Wilcox PC-fired steam generator rated at 930,000 pounds per hour of superheated steam at 2080 psig and 1050F, with 1000F reheat. The General Electric turbine generator was a cross-compound machine, and the plant had an actual heat rate less than 10,000 Btu per kWh. Florida’s Environmental Protection Agency viewed this project as a new source and therefore required SO2 reduction even with the low sulfur coal. Meeting New Source Performance Standards for SO2 emissions provided another challenge, but with limestone available for cement production, another opportunity for coupling the processes appeared feasible. The initial concept involved a separate calciner to produce CaO for cement, but this could also be used to capture SO» from the steam generator. Construction of the new facility began in January 1985 and the cement plant became operational in February 1987. Start-up of the power and lime plants is currently underway. While nothing in the individual plants is unique, the combination and interconnections are a novel approach that illustrates how creative thinking can improve efficiency and reduce capital and operating cost of existing proven processes. The steam generator repowering and reconfiguration provided the greatest challenge and is the focus of this paper. The design approach, commercial compromises, and predicted performance are all discussed in some detail. Gas Out Air Preheater ——>-_—»> Combustion Air Tube and Tile Wall Division Wall Furnace Exit Figure 1 OP-28 original steam generator. Design Approach Steam Generator Background The steam generator (OP-28) went into service in 1949. This unit is a single drum “open-pass” boiler. It utilized a tube and tile water-cooled fur- nace with a division wall separating the primary furnace from the “open-pass” section (Figure 1). The unit was designed to fire pulverized coal through cross tube type burners. The lower pri- mary furnace walls on all four sides were full- studded to retard heat transfer by holding slag to provide a high enough temperature for liquid-ash removal (slag-tap). The unit was originally designed to produce 930,000 lb/hr of main steam at 1050F and 2080 psig at the superheater outlet and a reheat steam flow of 835,000 lb/hr at 1000F and 390 psig. The unit design pressure remains 2300 psig. During the start-up of this unit, both the main steam and reheat temperatures were below the values originally predicted at full load (approxi- mately 200F on main steam). In an effort to raise the steam temperature, the division wall screen was lowered 21-ft. 3-in. which reduced some of the upper furnace as useful heating surface and raised the steam temperatures to design levels. The pulverizers were converted from type E-70 to EL-70 over the operational life. In 1972, the unit converted to oil-only firing. This was continued until 1980 when the boiler was shut down and put in storage. Objectives as established by Florida Crushed Stone were: Maintain the existing performance on coal; i.e., steam flow and final steam tempera- tures; generate 125MW net minimum; utilize the maximum amount of existing OP-28 material to. minimize capital requirements; integrate with the lime and cement production processes; and upgrade to meet the NSPS. The major concern was with the combustion sys- tem and how to upgrade it to meet NSPS and maintain the maximum reuse objective. The wet bottom design generates high NOx, twice that permitted by NSPS; clearly, therefore, the major upgrades would fall into the combustion system. The wet bottom had to be eliminated to permit lower flame temperatures; the tangent tube fur- nace construction had to be made gastight to permit lower air infiltration and oxygen place- ment where it was needed; and the existing inter- tube pulverized coal burners had to be updated to a low NOx design; consequently, the furnace volume had to be increased to compensate for the lower flame temperatures, as carbon conversions would be lowered if the volume was left constant. Finally, the internal and external condition of the “stored” steam generator had to be determined to permit the use/replace fitness evaluations that would impact scope and cost. The objectives set the design approach, the tech- nical innovations applied created the contract detailed engineering, and some of these high- lights are described in the next section. Final Design Combustion System Major modifications to the original pulverized- coal-fired steam generator were required (Figure 2). One obvious option that was explored was the retrofit of a bubbling fluidized bed under the exist- ing furnace. It would have been possible to do this but the original steel structure (the footprint) Intertube Division Wall Burners Figure 2 Original lower furnace steam generator. would have changed and the furnace exit gas temperature reduction (to 1700F) would have mandated extensive redesign of the existing superheater and reheaters. The maximized “reuse” objective would have been lost. Once the decision was made to stay with pulver- ized coal and stay within the existing footprint, NOx control became the prime driving force for the combustion system. The proven B&W Low NOx Dual Register Burner could not develop suf- ficient flame length if located on the front wall; i.e., duplicating the existing burner location. It was decided to remove the existing division wall (Figure 1), and locate the low NOx burners on both side walls - six on each wall directly opposed. The available flame length of 42 feet was suitable compared to the 20 feet available in the existing furnace. Removing the furnace wall that created the “open pass” provided additional volume for “burn out”, but removed a good deal of furnace- effective heat transfer surface and permitted the combustion gases to short circuit the design gas path and exit the furnace before being sufficiently cooled. The water/steam circulation design had to be redone to determine if the division wall could be removed and that fluid flow would be shifted to other circuits. In addition, the decrease in furnace absorption could increase the furnace exit gas temperature and affect reuse of the convection pass material. The low NOx burners would aid this problem, however, as peak flame tempera- tures are deliberately depressed (to reduce the thermal NOx contribution), absorption therefore hi Figure 3 Typical pulverized coal lower furnace. is spread over a greater furnace area as compared with absorption patterns created by high turbu- lence, high NOx combustors.! The burner zone was raised to the top of the furnace to increase absorption and decrease the possibility of furnace bypassing. This relocation increased the spot absorption in the roof circuits which required another redesign, namely a change in roof slope and a subsequent change in drum elevation to accommodate the new slope, thereby insuring that minimum circuit fluid velocities would be met. The tube and tile construction utilized in the exist- ing furnace, three-inch tubes on six-inch centers, was not gastight and permitted air infiltration. This air infiltration required higher total O2 to be carried to insure sufficient combustion air at the burners. The higher than required O2 would have increased NOx levels, increased gas-side draft losses, increased gas-side erosion of tube metal and increased horsepower requirements for fans. The lower furnace was covered with pin studs and refractory (wet bottom) and had a liquid slag tap outlet. The pin studding/slag tap construction was not desirable for low NOx (flame tempera- tures) or for slagging control. The slag tap was not desirable as “dry bottom” was required. It was decided to replace the entire furnace with one constructed of 2 1/2-inch OD tubes on three-inch centers with membrane walls. The flat floor wet bottom was replaced with a hopper-type slope (typical PC bottom). Available head room below Limestone Injection Ports =! a! sl Limestone Feed x: Upper_,¥ Cooling —t ES Lower Cooling Het 4 Burning Bed Esegesescooooees Figure 4 Reconfigured steam generator lower furnace. the hopper dictated the use of a low-head submerged-type chain deasher for bottom ash removal (Figure 3). Details such as new compart- mented windboxes, longer coal lines, and primary air fans (for longer runs) had to be applied but the pulverizer/bunker location was maintained to save the original steel. Overall, a cost-effective update to meet NSPS NOx was developed. DeSO x The new furnace configuration with the hopper slopes provided a potential location for a fluid bed calciner to be close coupled. The space available in the front area was maximized by offsetting the hopper throat to the rear and rearranging the front wall slope angles (Figure 4). By rearranging the hopper slope walls this way, the resulting area created was 42-ft. 6-in. wide and 20-ft. deep. A bubbling fluid bed of 40 1/2-ft. x 20-ft. could be fit into that area. The freeboard above the bed could be increased through further rearrangement of the hopper slope wall--in this case, a section of vertical slope was utilized to maximize the free- board. The extensive bubbling bed experience obtained both within B&W laboratories (1x1, 3x3, 6x6) and on the 20MW TVA demonstration indi- cated that approximately 34.5 tons of CaO per hour could be produced using crushed coal as the bed heat source. The walls and floor of the bub- bling bed calciner are membraned water-cooled tubes. The product CaO drains from the bed at 1700F, much too great a heat loss and too hot to handle. Two cool-down beds were added to pre- PC Burners Low NOx Radiation Screen Cyclone Outlet Plenum Cyclone Expansion: Joint Calciner 1 Screen ¥+— Rotary Feeder Water Cooled Chute —§s | Panasztsssse ——Calciner Bed Coal Spreader Rotary Drum Injection h—" _ fluidizing Nozzle Limestone Air System Distribution Bottle Boiler Wall u Limestone Nozzle fj Fluidizing Cooling Air | Control 7 Room Lt Console I LA L4 Air Transport Receiver Air Compressor Figure 5 Calciner outlet. heat the calciner air system to 485F. The cooler (lowest) operates at 225F lowering the product temperature to 225F by utilizing the latent heat contained within the product to preheat the air. For cold start-ups and as a supplementary heat source, a steam coil is available to provide up to 470F air temperature. The final arrangement is now a stack of three fluid beds using gravity to feed hot material from bed to bed. Coal feed is over-bed using seven mechanical spreaders, and limestone is fed underbed pneumatically. Air enters through bubble caps and provides the fluid- izing air during normal steady state operation. Fitting the start-up burners, coal, limestone, and air systems in the congested lower front wall zone proved to be one of the major engineering chal- lenges of the project. The fluid bed calciner exhausts solids and gas (Figure 5) to three hot cyclone dust collectors (in parallel) where solid CaO and elutriated lime and coal ash are sepa- rated from the gas stream, collected and cooled in a water drum cooler, and blended with the bed product. The design and location of the hot transport flue, calciner outlet to hot cyclones, and hot cyclones outlet return to the steam generator furnace, was a second engineering challenge. The flue is pres- surized and transports 1700F gases. It would have been desirable to close couple the cyclones but the location of the coal bunkers did not make that possible. The cyclones had to be located outboard of the existing bunker steel columns which added new steel and drastically increased the length of Figure 6 Limestone injection system. the hot flue runs. The calciner system’s balance pressure point (start of negative draft) is at the return flue entrance to the pulverized coal furnace. The gas exhausts from the cyclones carrying fine lime, and coal ash, into the pulverized coal fur- nace, just at the relocated low NOx burners. The unreacted lime then will mix with the furnace gases and capture the SOo.”. The combining of these two systems, the lime calciner with the pul- verized coal-fired furnace, is truly unique and we believe an excellent idea for future retrofit projects of this nature. This is not only because the cou- pling avoids the need of a wet or dry scrubber addition, but because the combined heat (from calciner and PC) reduces the need for one pulver- izer. The unit could make full load with four pul- verizers, but had one spare. Now with the calciner spare heat capacity available, the fifth pulverizer is available for pulverizing fuel for the cement kiln. Thermal Performance The need to be able to operate the steam generator with or without the calciner in operation compli- cated the functional performance design. Approx- imately 25% of the total-full load steam generator input is provided by the calciner and, when it is down, a major variation in heat input, gas flow, and SO, capture occurs. This added complexity and expense but the flexibility had to be available even though “normal” operation will always be with the calciner “on”. In the unlikely event that FGRE = Flue gas reheat exchange AH = Airheater SCAH = Steam coil air heater DC = Dust collector ~ Turbine 125 MW +—Electricity Lime ime Product 47.5 TPH Clinker |_ Cooler Figure 7 Process flow diagram. the calciner is not operating but the power pulverized-coal system is, a limestone injection system has been included and would inject finely pulverized limestone pneumatically into the fur- nace zone within a 1900F temperature window (Figure 6). The unit is to be base loaded at MCR so the 1900F window location would be a constant. The B&W Limestone Injection system was first demonstrated full size in 1971 at TVA’s Shawnee Power Plant.? The original steam generator secondary air sys- tem utilized a Ljungstrom air preheater located on top steel. The higher gas weights (calciner and power), the higher dust loadings (CaSO4, CaO and flyash), and the potential for increased leak- age as a function of seal life made us question the reuse of that type of air heater for this applica- tion. Florida Crushed Stone, in the interim, had purchased several other used power plants and it was decided to recycle two tubular air heaters from Commonwealth Edison’s Ridgeland plant to this project. Those tubular air heaters will provide air to the burners and the primary air fan at 500F and maintain a fairly constant gas volume over time to the baghouse (low leakage). The cement plant clinker cooler uses ambient air to cool the clinker down from 2700F to 350F. That cooling air at 350F will be integrated with the secondary air system and represents about 33% of the full load steam generator combustion air requirement. Cement plant upsets could divert 800F cooler air to the combustion system so it was decided as a precaution, to add a “really hot air” diversion sys- tem to manage this event if it occurred and bypass the air directly to the baghouse. Gas Stream Flue gas velocities in the convection pass are higher than desirable and will increase metal loss due to erosion with the heavy dust loadings. How- ever, the desire to reuse material prompted accep- tance of this risk with the extensive use of tube shields added to aid in protecting the existing convection pass tubes. Air heater outlet gas at 290F will be utilized to dry the limestone feed for the calciner. A flue gas reheater was included (using steam) to be available to supplement the lower flue gas temperatures at low steam genera- tor loads. To insure full flexibility of operation, a system of bypass flues with associated fans, con- trols and shut-off dampers permits varying the flow, or bypassing totally, the flue gas reheater, the limestone dryer or both subsystems and going directly to the baghouse. A second flue gas stream permits air heater bypass with its own set of fans, cyclone dust col- lector, control and shut-off damper that permits hot gases to go directly to the limestone dryer or baghouse. The total steam generator gas stream eventually arrives at the baghouse where the remaining par- ticulate is essentially removed and the SOg2 reduc- tion reaction continues.’ Exhaust gas streams from the cement kiln are also directed to this bag- house; from where the combined gas stream exits through a 320-foot high concrete stack (Figure 7). Table 1 Coal Analyses Source Mine Ryder Seam Hazard #7 State Kentucky Size (as Fed) 11/4" x 1/4" Grindability 47 Surface Moisture, % Ash Soft, Temp. F (reducing) Approximate Analysis Moisture, Total 4.38 Volatile Matter 30.71 Fixed Carbon 52.64 Ash 12.27 Total 100.00 Ultimate Analysis Ash 12.27 H,0 4.38 c 68.29 H, 4.44 N, 1.48 Ss 78 0, 8.27 CaCO; CL 09 Total 100.00 Btu/Ib 12200 The final design just described resulted in the expected performance that is discussed in the next section. Performance The final design is, of course, never really final but represents a frozen design at a point in time to permit hardware to be designed, procured, fabri- cated and installed to make the various subsys- tems whole. The design coal had to be set and the design limestone had to be specified to formulate performance predictions, performance guarantees, and emission limitation guarantees. For without the guarantees, no external financing would have been possible and the project would never have left the study stage. The design fuel, an Eastern Kentucky Bituminous (Hazard No. 7) analysis, is provided in Table 1. The Hgl of 47 and low moisture would permit near full load operations with the calciner out of service. The slagging and fouling characteristics could be met using the existing boiler “footprint”, the low NOx burners, and the four pulverizers. The same coal would be crushed for the calciner- fluid-bed feed and pulverized by a fifth pulverizer dedicated to the cement kiln firing system. The pulverized coal was one with which we had exten- sive experience and, therefore, are confident of the design and resulting performance. Calciner The design limestone feed was the local stone mined at the site having an analysis of: Chemical Percent SiO. 1.95 MgCO3 1.0 S03 0.05 H,O0 2.0 CaCO3 95.0 Sizing initially will be 3/8 in. by 1/8 in. with not greater than 15% fines. The stone was new to B&W and to evaluate it properly it was decided to utilize our in-house Research & Development Center 1-ft. by 1-ft. fluid bed steam generator. Unknowns created concerns which were evaluated at the test facility and at the BkW/TVA 20MW bubbling fluid bed. The following factors required assumptions that had to be evaluated and quantified to assure that the final performance could be not only pre- dicted, but guaranteed: e Inbed/overbed combustion splits, e degree of sulfation of the bed material, e elutriation of the stone, e rock removal with a zero-degree floor slope, e free board height to disengaging height as a func- tion of this special geometry (close coupled -cross section taper), e resulting effects upon entrainment, e inbed vs. circumferential cooling surface. Limestone Injection System When the calciner was out of service, the power boiler would still require SO2 capture. At the time of contract, we had only one commercial demonstra- tion of limestone injection technology with disap- pointing results and considerable laboratory scale work.’ Preparation, particle size, transport, injec- tion location, number of injectors, velocity of injec- tion, temperature window, and residence time required had to be fixed and were the subject of a special, separate laboratory investigation and study. Currently, this project and the joint U.S.A. EPA/State of Ohio/B&W /Ohio Edison LIMB Table 2 Unit Performance 1987 1949 After Upgrade Original Expected Performance Actual Pulverized Performance Pulverized Pulverized Coal and Pulverized Coal Firing Method Coal Coal Calciner and Calciner Steam Leaving SH, lb/hr 930,000 930,000 930,000 N Steam Leaving RH, lb/hr 835,000 835,000 835,000 oO Fuel to Calciner, Ib/hr 0 0 40,730 T Fuel to PC Boiler, Ib/hr 115,000 104,970 82,500 Limestone to Calciner, lb/hr 0 0 193,480 Y Comb. Air (To Beds/Transport), Ib/hr 0 0 375,560 E Air to Burners, lb/hr 1,058,000 1,118,770 864,270 T Flue Gas Entering Air Heaters, Ib/hr 1,245,000 1,229,260 878,950 Calciner Gas Entering PC Boilers, lb/hr 0 0 494,150 A Fluidizing Velocity, FPS 0 0 8.75 Vv Steam Pressure at SH Outlet, psig 2,080 2,080 2,080 A Steam Pressure at RH Inlet, psig 440 440 440 | Steam Leaving Superheater, °F 1,050 1,050 1,050 L Steam Leaving Reheater, °F 1,000 1,000 1,000 A Steam Entering Reheater, °F 670 670 670 B Water Entering Economizer, °F 443 443 443 L Gas Leaving Economizer, °F 625 597 610 E Gas Leaving Air Heater, °F 246 290 290 Air to Burners, °F 552 513 523 Air Entering Air Heater, °F 80 115 172 Air Leaving Air Heater, °F 552 513 523 Demonstration Project are running concurrently. Both are private demonstrations, with one R&D funded and the other a commercial venture. Inland Steel, East Chicago currently has a successful 35% SOz LIMB Program internally funded that indi- cates our assumptions and system design for this installation are on target.® Circulation Combining both combustors into an existing foot- print challenged the fluid dynamic designers. Smaller tubes, 2 1/2-inch OD vs. original 3-inch OD, were utilized to improve mass flow velocities; geometry was changed to improve sloped tubes; and, finally, judicious use of B&W’s internally ribbed tubes in various circuits permitted circula- tion freedom that was not available 30 years ago. The final circulation reflects our actual experience and is conservative, based upon more severe opera- tional experience. Concerns The hot flue and cyclone collectors, while not of high technology, or required by our design to be highly efficient, are of concern. The gas tempera- ture of 1700F is high, and experience is limited at these temperature levels. However, we have observed high temperature collectors that operate with acceptable efficiencies and lives. The operation performance and associated maintenance require- ments of these cyclones will be closely monitored over the life of the project. Expected Performance Table 2 compares OP-28 original performance, expected performance following the upgrade, and preliminary actual performance. Guarantees Subject to the conditions specified, the following guarantees are included as a portion of the design/- material supply package for this upgrade. Full Load NOx - 0.55 Ib/MMBtu hour input (tied to a fuel nitrogen). Maximum Normal Capacity (MNC) - 930,000 Ib. steam/hour at 2080 psig. Steam Temperature Main Steam - 1050°F + 10°F. Reheat Steam - 1000°F + 20°F. Lime Quality - 90% lime (CaO) by weight. TN Gas Outlet Bunkers Cyclones [ : Pulverizer ———= Lower Cooling’ Figure 8 Reconfigured steam generator. Lime Quantity - 47.5 tons/hour blend from cal- ciner and cyclone product catch. SOp2 Reduction - At Maximum Nominal Capacity (MNC) economizer outlet 760 Ibs. SO2/hour (tied to a coal sulfur of .78%). The start-up schedule is currently set for initial fire September 15, 1987. A shakedown period, initial operation, and fine tuning periods should produce operational data by October 1987, to compare against predicted performance and to measure guarantees. Summary Although other bubbling fluid bed retrofit projects (5) are underway (e.g., our Montana-Dakota Utili- ties 80-MW retrofit currently undergoing start-up), this FCS project, involving several interconnec- tions, is unique not only from an engineering standpoint but also that of economic efficiency. Florida Crushed Stone’s, use of an old boiler/- cement plant for the cogeneration of electrical energy, processing of steam, and the production of lime also generates these associated process benef- its (Fig. 8): ¢ Calciner heat is utilized for steam generation-not wasted as at present in non-integrated plants. e Waste hot air from the cement plant is utilized for boiler combustion air. e Coal handling, preparation and storage is com- mon for the three processes. e Boiler flue gas waste heat is utilized to dry lime- stone dust. e The limestone dust aids in reducing SO2 emissions. e Lime production of 47.5 tons/hour. e A common baghouse is utilized to capture particu- late and reduce SOz from both the power boiler and cement kiln. The project is within budget and the cement plant went on stream in February 1987. The power sales agreement is firm with Florida Power and Light, and the power plant start-up is scheduled for October 1, 1987. Acknowledgments A project this complex and different does not happen easily. The “visionary” who appreciated the downstream benefits and turned skeptics into sup- porters is Mr. Browne Gregg. He provided the lead- ership and drive that enabled this project to become real. Mr. Gregg is Chief Executive Officer of Florida Crushed Stone Corporation. References 1. Barsin, J.A., “Fossil Steam Generator NO xControl Update”, EPRI/EPA Joint Symposium on Stationary Combustion NOx Control, Dallas,TX, 1982. 2. Smith, J.W., Whitney, S.A., “Industrial Fluidized Bed Design and Operation of the TVA Test Facility”, International Coal and Lignite Utilization Conference, Houston, TX, November 15-19, 1983. 3. Barsin, J.A., “Reducing SO2 Emissions from Coal Firing Utility Furnaces”, University of Kentucky Coal Conference, Lexington, KY, April 23, 1986. 4. Doyle, J.B., Jankura, B.J., “Furnace Limestone Injection with Dry Scrubbing of Exhaust Gases”, Spring Technical Meeting, Central States Section Combustion Institute, Columbus, OH, 1982. 5. Dickerman, P., “Limestone Injection for SO2 Capture Inland Steel; East Chicago, Indiana”, Council of Industrial Boiler Owners, Washington, DC, February, 1987. Technical Paper Nondestructive technology to evaluate superheater condition W. E. Beak Plant Superintendent Erickson Station Board of Water and Light Lansing, Michigan D. W. Bonin Condition Assessment and Diagnostic Services Babcock & Wilcox Barberton, Ohio Presented to EPRI Conference on Boiler Tube Failures in Fossil Plants. Atlanta, Georgia November 10-12, 1987 ie Babcock & Wilcox a McDermott company Nondestructive Technology to Evaluate Superheater Condition W. E. Beak Plant Superintendent Erickson Station Board of Water and Light Lansing, Michigan D. W. Bonin Condition Assessment and Diagnostic Services Babcock & Wilcox Barberton, Ohio Presented to EPRI Conference on Boiler Tube Failures in Fossil Plants. Atlanta, Georgia November 10-12, 1987 Abstract Upon implementation of its new preventative maintenance program, the City of Lansing’s Board of Water and Light Erickson Station wanted to be able to predict remaining superheater tube life to better plan and budget for future tube replacement. Unit No. 1 at Erickson Station is a 165 MW, drum-type boiler. Although there were few previous failures, there was concern about the general condition of the secondary superheater, and a remaining life analysis of this component was desired as the unit was nearing 15 years of operation. Rather than extensive tube sampling, the Board of Water and Light chose to evaluate the secondary superheater with the Babcock & Wilcox Nondestructive Oxide Thickness Inspection Service (NOTIS™). The NOTIS system allows the ultrasonic measurement of wall and internal oxide thicknesses on large numbers of steam-carrying tubes, without tube removal. The inspection took place during a short outage in November of 1986. After scaffolding and surface cleaning, a total of 330 tubes were inspected by B&W ina single day. From the field data, a region of severe wall loss (greater than 30% of wall) was identified in the secondary superheater outlet sections on the left side of the unit. Internal oxides in the outlet tubes ranged up to 20 mils in thickness. The NOTIS data was used to perform remaining creep rupture life calculations. Remaining lives were found to be low in some of the outlet sections, primarily as a result of the extensive wall loss observed. Based upon considerations of low wall thicknesses and remaining lives, the Board of Water and Light is currently planning the replacement of several outlet sections in April of 1988. Continued monitoring of the secondary superheater with the NOTIS system is also planned, again with the idea of predicting future tube replacement to better facilitate maintenance outages and budget planning. Introduction Background Unit No. 1 at the Erickson Station is a single 165 MW, coal-fired Babcock & Wilcox drum boiler with 1005°F superheater and reheater steam temperatures. This unit has been in opera- tion since 1973. The Board of Water and Light had observed wall thinning in the secondary superheater and, although few failures had occurred, was concerned about the general condi- tion of the superheater tubes. Since the unit was nearing 15 years of operation, the Board of Water and Light (BWL) decided to perform a remaining life analysis of this component. BWL had been contemplating the removal of 10 to 20 tube sam- ples from the Erickson secondary superheater. These tube samples were to be sent to a metallur- gical laboratory for measurement of wall thick- ness and internal oxide thickness, for use in remaining life determination. A Board of Water and Light representative became aware of the NOTIS (Nondestructive Oxide Thickness Inspection Service) technique at a Babcock & Wilcox Executive Service Seminar in Akron, Ohio in October, 1986. The NOTIS system is B&W’s patented*, microprocessor-controlled ultrasonic test designed to measure the steamside oxide scale of a steam-carrying tube (wall thick- ness is also ultrasonically measured). The tech- nique has been previously documented (1). The method has the capability of accurately examin- ing a large number of tubes in a short time, with- out their physical removal from the superheater. It is a viable alternative to tube sample removal, which has been the only other method of oxide thickness measurement to date. Upon request, the Condition Assessment and Diagnostic Services group in B&W’s Barberton, Ohio office, provided BWL with a proposal to inspect 300 tubes in the secondary superheater with the NOTIS system. The proposed nondestructive inspection was priced comparably to the originally planned metallurgical tube sample analysis. The Board of Water and Light found the NOTIS technique advantageous and chose to use it during a short outage at the Erickson Station in November. The Importance of Oxide Measurement When a steam-carrying tube enters service, the tube metal in contact with the internal steam begins to form a layer of scale that is mostly magnetite (Fe304). As the tube’s service life pro- gresses, this I.D. oxide gradually grows in thick- ness at a rate dependent upon temperature. The growth kinetics of this scale have been exten- sively studied, and growth rate laws may be found in several sources (2,3). The growth of the oxide scale can have a major effect on the creep life of the tube. Creep is the process by which metal under stress at high temperatures will fail. The higher the applied temperature or stress, the more quickly failure will occur. Steam-carrying superheater and reheater tubes operating at temperatures above 900°F are typically subject to failure by creep- rupture. The expected creep life of a tube can be simply and quickly estimated from tabulated data, provided the applied (hoop) stress and the temperature of operation are known. Unfortu- nately, this simple calculation is greatly compli- cated by the very presence of the oxide scale. Magnetite is a ceramic material with a thermal conductivity many times less than that of the tube metal. This scale acts as a barrier to heat transfer from the gas side to the steam side and causes an increase in the tube metal temperature. This augmentation in temperature is a function of * USS. Patent No. 4,669,310 of January 2, 1987. scale thickness, and gradually increases with time as the scale grows (3). The rate at which the tube’s creep life depletes is also dependent upon temperature and therefore also increases with time. The temperature increase can range from 1 to 2°F per mil of scale thickness, depending upon the tube wall thickness-to-outer diameter ratio and the through-wall heat flux. This may not seem to be much, but a tube metal tempera- ture of 20°F above design (as may be caused by a scale of 10-15 mils) can have a marked effect on the life of the tube. The measurement of the inter- nal oxide is therefore crucial to the remaining life determination of a superheater tube. Another reason for the measurement of oxide thickness is to detect the presence of exfoliation (also known as spalling). This phenomenon occurs when stresses within the scale layer (caused by differences in thermal expansion prop- erties between the oxide layer and the base metal) cause bits of the scale to break loose from it. Flakes of broken oxide can become entrapped in lower tube bends, stifling flow through the tubes and leading to short-term overheat failures. Worse yet, these bits can be carried downstream into the turbine, where they can cause extensive (and expensive) erosion damage to the blades. While more common in reheaters, exfoliation can occur in secondary superheaters as well, particu- larly in cycling units. This phenomenon was known to be occurring at the Erickson Station. A tube sample, removed from the secondary super- heater outlet in early 1986, showed widespread exfoliation of the internal scale. Pre-Inspection Planning Selection and Numbering of NOTIS locations B&W selected locations within the Erickson secondary superheater for NOTIS testing. An arrangement drawing of this component, showing tube dimensions and materials, was studied in detail. This diagram is shown in Figure 1. The precise locations of NOTIS tests are chosen according to basic guidelines designed to obtain the most meaningful data possible. First, all of the tubes in the element that are desirable for inspection are marked with a crossbar, which represents a level (or plane) of inspection. This crossbar is usually set to intersect tubes of each steel grade just prior to a transition weld to the next higher material grade. In this manner, the hottest tubes of each material (the most likely to fail by creep) are examined. After the planes of inspection are approximated, the number of elements to be inspected is decided. At least 10 elements across should be checked in smaller units, 20 in larger ones. The numbers of tube rows and elements to be inspected are varied within these basic guidelines until the total number of test points equals the desired (proposed) number. For example, the inspection level (crossbar) may be selectively broken at spots to miss certain tubes or extended to reach others; the number of elements to be inspected may be changed from every third to every fifth, etc. Of course, many other factors affect the choice of NOTIS locations in any given unit: accessibility to the bank (Where are the access doors and crawlspaces?), element sidespacing (How many tube rows into the element can be reached?), and elevation (How many elevations in a single bank should be inspected?) are among them. Once the most desirable tubes have been selected, they are numbered for the purposes of identification. Within each bank (inlet or outlet), elements are numbered from the left hand side- wall to the right hand sidewall. Within each ele- ment, tube rows are numbered from front to rear, with the gas flow (this may be seen in Figure 1). This allows any tube in the bank to be identified by a set of coordinates (element number, tube row number). This standard nimbering system is convenient to use, reduces confusion during data collection, and allows for useful graphical repre- sentation of the data once it is returned from the field. Application to the Erickson Superheater The secondary superheater inlet bank in the Erickson Station consists of 23 ten-tube parallel- flow sections. Tube rows 1 through 10, the down- flow tubes leaving the inlet header, were carbon- molybdenum and 1/2 Cr-1/2 Mo steel, depending on their location in the bank. Tube rows 11 through 20, the upflow tubes entering the mani- fold headers, were 1/2 Cr-1/2 Mo steel. The upflow tubes were of greatest interest, as these tubes contained the hottest 1/2 Cr steel in the bank, and were therefore expected to exhibit the thickest oxides. The NOTIS test points were set at the 223 foot elevation, a few feet below the roof. All 10 upflow tubes were inspected in each of 12 elements (every other element) at this elevation, for a planned total of 120 test points. The secondary superheater outlet bank consists of 46 four-tube counter-flow sections, composed entirely of 2-1/4 Cr- 1 Mo tubes. Tube rows 1 through 4 carry the coolest steam in this bank; Inlet Outlet Header Header SSH Inlet Bank All Tubes 1.5” 0.D. Roof Line SA-213 T2 SA-209 Tla ssi 0.203” Wall Outlet Bank . fa 0.165" Wall SA-213 T22 All Tubes 1.75" 0.D. 0.203" Wall 0.310" Wall 0.270" Wall 0.240" Wall Tube Numbers Tube 1 Numbers | 4 5 6 7 5 9 1 SS ee SY —— YY Figure 1 Arrangement drawing of the secondary superheater at the Erickson Station. NOTIS © locations are indicated by heavy black lines crossing the inspected tubes. The diagram is not to scale. rows 5 through 8, the hottest. It was these latter rows that were of greatest interest to BWL. Origi- nally, B&W had planned to examine every tube row in the outlet bank at a single elevation, 223 feet. This level, just a few feet below the roof, was set to catch the hottest portion of the outlet legs, where the thickest oxides could be expected. Unfortunately, several sections of the reheater were being replaced during the same outage as the NOTIS inspection. This component was immediately adjacent to the secondary super- heater outlet bank, and shared a crawlspace with it. Because the activity in this area would hinder the access to the rearmost tube rows, the inspec- tion was limited to the frontmost tubes in the bank. The planned number of NOTIS test points was split between two elevations: the original 223 foot level below the roof, and 207 feet, just below the lower wraparound tube. The tube rows inspected were different at each elevation, but both elevations included the outlet leg tubes (rows 5 through 8). Seven tube rows were inspected in each of fifteen elements (every third element) at both elevations, for a planned total of 210 test points. The final distribution of NOTIS locations for the Erickson inspection is illustrated in the arrangement drawing of Figure 1. The total number of test points in the inlet and outlet banks was 330, which approximated the originally pro- posed amount. Once the precise locations and the numbers of NOTIS points were known, they were transmitted to the plant personnel. A B&W service engineer was on-site at the start of the outage to mark and label the necessary tubes, and oversee the cleaning of the NOTIS locations in prepara- tion for testing. On-Site Preparation A local contractor had been hired by BWL to pro- vide the labor and support for the replacement of the reheater, which was also occurring during the November outage. They were also charged with the responsibility of preparing the NOTIS loca- tions in the secondary superheater for testing, in accordance with B&W’s requirements. Scaffolding was erected in the superheater enclosure to allow comfortable access to the tubes, at both the 207 and 223 foot elevations. Proper lighting and a nearby 110 volt electrical source were also made ready for the advent of the NOTIS team. Then, preparation of the tube O.D. surfaces at the marked locations, for measurement of the oxide and wall thickness measurements, commenced. Flyash deposits and loose scale were first removed from each NOTIS test point by wire- bushing. More stubborn scale was ground off with a 36-grit wheel, to leave a 1/2” x 1” area of clean metal, free of ridges or pits. Each point was then smoothed with a 100-grit flapwheel, which leaves a surface suitable for accurate oxide measurement (in actuality, BWL went one better with an addi- tional 150-grit flapwheel finish; a finer surface serves to improve ultrasonic signal quality and can reduce the inspection time). Access to tubes deep into the banks had been considered during the planning stages and was not a problem in this unit. The elements in the outlet and inlet banks had sidespacings of 9” and 18”, respectively, between tube centers (in most superheater compo- nents, 6” centers is the smallest element sidespac- ing that will allow unlimited penetration into the bank; any smaller, and only the first one or two tubes can be accessed). The preparation of the superheater for NOTIS testing was secondary to the reheater replacement, so manpower was assigned to it as time permitted. Nonetheless, the entire pre-inspection work, including scaffolding and NOTIS location cleaning, was completed within a three-day period. Data Collection and Treatment B&W’s NOTIS team was at the Erickson Station the day after the completion of the pre-inspection preparation. The team consisted of a UT techni- cian and a metallurgist. The technician is skilled in the complete operation and maintenance of the NOTIS equipment, and performs the actual thickness measurements. The function of the metallurgist is to record and interpret the data as it is taken, and to provide a second set of hands if needed. Oxide and wall thickness measurements on the 330 test points were completed in a single day. Such speed is not unusual, and was attri- buted to several factors -- the quality of the finished O.D. surfaces, the closeness of the test locations within the boiler, and the immediate access to necessities such as lighting and electricity. Prior to departure from the site, Board of Water and Light was provided with a copy of the oxide and wall thickness data, and the opportunity to discuss this data with the service engineer and the metallurgist. Afterward, these results were returned to the Condition Assessment and Diag- nostics group. This information was entered into the NOTIS Data Management System (DMS), which performed the remaining life analysis and generated color plots. Remaining Life Determination The DMS model for estimating remaining creep life has been previously discussed (1). Each tube in the superheater has grown internal oxide and (perhaps) suffered wall thinning during its service life, so both operating temperature and hoop stress have slowly increased. In order to deter- mine remaining life, therefore, an integrated creep life fraction approach must be used. The tube is assumed to begin its service life operating near design temperature and stress conditions. Wall loss and oxide growth rates are estimated for the tube from its NOTIS wall and oxide thickness measurements; therefore, the temperature and stress at any time in the tube’s service life can be estimated. Then, utilizing B&W creep data (which is contained within the ASTM compilations of creep data), a function describing the fraction of creep life used up, moment by moment, is gener- ated. The point in time at which the total fraction of life used (the area under the curve) equals one is determined. This point is when the tube would be expected to fail by creep; the distance to this point from the time of the NOTIS inspection is therefore the remaining life of the tube. The DMS repeats this process for every inspected tube in the superheater, and assigns each remaining life to the coordinates of its tube. Graphic Presentation of Data Once the remaining lives of each tube have been calculated, they are organized by element number and tube number into a grid. The DMS then puts these numbers into a remaining life band accord- ing to preselected ranges and plots them, using a number-color code, each on the coordinates of its tube. The result is called a full component plot, and is effectively a map of tube remaining life across the superheater, in a single plane of inspection. It is helpful for targeting areas of the superheater in need of replacement. The ranges are usually in 50,000 hour increments (i.e., the first or critical range is 0-50,000 hours). Deter- mination of remaining life was the Board of Water and Light’s goal, but many utilities are interested only in the oxide and wall thickness measurements. The DMS can generate full com- ponent plots for tube oxide thickness (ranges in mils of oxide measured) and wall thickness (ranges in percent of specified wall thickness lost). As will become apparent, these maps are an invaluable tool for identifying regions of higher gas stream temperatures, or areas of greater cor- rosion or erosion. Linear graphs of oxide thick- ness and wall thickness across the unit, within a single tube row, are also possible. Discussion of the Erickson Inspection Results Outlet Bank The results of the inspection of the secondary superheater outlet bank at the 223 foot elevation are summarized in Figure 2. At this elevation, the oxide and wall thickness data exhibited several interesting patterns. These may be seen in Figures 2a and 2b, which are maps of the distri- bution of oxide thickness and wall thickness across the outlet. Oxide thicknesses ranged from less than 6 mils to as high as 20 mils in the exam- ined tube rows. As could be expected, the thickest oxides were mostly found in the outlet leg rows five through eight, although several of the down- flow tubes in rows 9 and 10 contained very thick oxides also (Figure 2a). Oxides also tended to be thinner in the elements nearest the sidewalls, the likely result of slower, lower temperature gas flow in these regions. In the midst of outlet leg tubes with 13 to 20 mils of oxide were several tubes with less than six mils of scale. Of particular interest is element 11, in which three of the four outlet leg tubes had oxide scales less than six mils thick. Scale exfoliation was blamed for these lower scale readings. Extensive scale exfoliation had been observed in a tube sample removed from the lower elevation in the outlet bank about eight months prior to the NOTIS examination. Wall thickness data is given in Figure 2b in the form of percent of specified wall lost. Over half of the inspected tubes at this elevation were above specified wall thickness (0% wall loss). These tubes were concentrated primarily toward the right side of the superheater. Tubes exhibiting slight (1-8% of specified wall) and moderate (9-15% of specified wall) loss made up much of the remainder, and were scattered around the center and left side of the superheater. Several tubes had lost over 15% of specified minimum wall; these are indicated by “4” and “5” symbols in the figure, and can be found at the left quarterpoint of the superheater. Interestingly, the wall loss was max- imized in element 11, the same section that ex- hibited a majority of exfoliated oxides. Oxide and wall thickness data were then util- ized to estimate the remaining creep lives of the tubes. A map of the distribution of remaining life across the outlet is shown in Figure 2c. Direct 10/3 2 3 23 1 S$ 2 1 3 33 43 3 a) 2» 913 3 535 33 4 4 2 2 S$ 2 22 Oxide Thickness z - 8 8|3 e435 435 4 1 2 5 4 5 4 Symbol Mils Tube Row 7la 1 4122 2 &§ 51 25 2 41 5 17-20 4 13-16 6/3 2 4 12 Ss S$ S 2 2 8 2 2.2 3 10-12 2 7-9 5)1 4 4 13 4 Ss 4 S58 S 124 3 1 1-6 4)e 2213 2 122 2 2 3 2 2 0 5 10 15 20 25 30 35 40 45 1 13 5 1 3 3 1 2 1 ,oaoa4 10 e . (B) g}2 1 3 83 3 1 2 12 1 .o2 41 1 Percent of Spec. Wall Loss g/1 2 3 S52 3 3 21 2 Bor.ri Symbol Percent Tube Rot 7/1 3 3 S32 1 3 1 3 1 1 1 1 1 ” 5 30-42 6/1 2 5 5853 43 2 21 yor roan 4 16-29 3 6-15 5}1 2 85 53 3 42 3 1 rororou 4 2 1-5 1 441,33 33 3 1 3 11 21 4 2 dt ° 0 5 10 15 20 25 30 35 40 45 10/1 1 3 S22 1 41 12 1 lo 2 2 | 2 (C) 9|}3 2 4 854 3 2 3 2 2 21 3 2 2 1 | Remaining Life 8 3 3 4 s4 4 4 4 4 2 3 4 4 4 3 Symbol 1000 Hours Tube Row 7/22 4 S83 2 1 4 4 2 1 4 1 3 1 5 0- 50 6}ee 5 $3154 41 2 4 11 3 4 51-100 3 101-150 5/2 3 8 S53 43 4 4 3 4 1 13 2 2 151-199 1 200- 4... 2 22 2 2 29 2 a yoaoawaoa. 0 5 10 15 20 25 30 Element Number Figure 2 Full component plots of the inspected outlet bank tubes at the 223 foot elevation, showing A) oxide thickness, mils, and B) percent of specified wall lost, both directly from the NOTIS™ inspection, and C) calculated remaining life, hours. Note the heavy wall loss in the left side of the unit (particularly in element No. 11), and its effect on remaining life. Symbols are normally in color to aid in detection of patterns. comparison of this figure with the thickness data in Figures 2a and 2b will show the effect of wall loss and oxide growth on tube remaining life. Wall loss, when severe, can be expected to be the dominating factor in creep life reduction, by virtue of rapidly increasing stresses (and acceler- ating creep) in the thinned areas. Tubes with greater wall loss will tend to have lower remain- ing lives. This trend may be verified by compari- son of Figures 2b and 2c; both figures contain crit- ical “5” symbols in the same left quarterpoint tubes. Despite the powerful domination of wall thinning on creep life reduction, the effects of oxide growth should not be underestimated. It is reasonable to expect that when wall thinning is not a problem, tubes with thicker oxides will have lower remaining lives, by virtue of having expe- rienced greater temperatures. This may be veri- 20F |= 3S 3sSB 3 3 2 1 2] Wis 2 a 2 a 2 2 a 2 18/2 2 2 2 2 2 2 2 2 17}2 2 2 2 2 a 2 aia Tube Row 16}1 1 ]=ee 1 2 2 3 15}/1 1 1 1 2 1 1 1 1 14/2 1 1 1 1 1 1 1 1 13}2 1 1 1 1 1 1 1 1 12}1 1 1 2 1 1 2] 1 2 11j2 21 1 1 1 2 2 1 1 10 15 Element Number 20 1 1 1 1 1 ou Remaining Life so Symbol 1000 Hours 1 1 a 1 1 5 0- 50 ie 4 51-100 2 oa 3 101-150 1 a 2 151-199 1 200- 1 1 25 Figure 3 Full component plot at the remaining lives of the inspected inlet bank tubes at the 223 foot elevation. Note that the tubes with the least lives were concentrated in tube rows No’s 16 and 20, toward the left side of the unit. fied by examining the three maps at the center and right quarterpoint of the unit. While wall loss was slight or nil in these regions, very heavy oxide scales were observed. These thick oxides were sufficient to reduce tube remaining lives into the sub-critical range (““4”’ symbols; 51,000-100,000 hours). The results of the outlet bank inspection at the 207 foot elevation were not nearly as interesting as those at the higher elevation. The outlet leg tube rows five through eight again exhibited variable oxides of almost 20 mils in thickness, but no pattern of wall loss was observed at this level. As a result, no tube had a calculated remaining life below 100,000 hours. Inlet Bank The inlet bank was inspected at the 223 foot ele- vation, the same level as was the outlet bank. Oxide thicknesses in the inlet bank ranged up to 10 mils. There was no tendency for oxides to be lower near the sidewalls, or higher in any single element. However, the highest oxides tended to cluster in tube rows Nos 16 and 20. This was par- ticularly gratifying, as B&W design data indicated these two tube rows were expected to be the hot- test in the bank. Wall thinning was generally slight when observed (1-8% of wall), and tended to occur at the left side of the superheater. A map of the remaining lives calculated for the inspected tubes is shown in Figure 3. Most of the tubes had remaining lives in excess of 200,000 hours (“1” symbols), although several tubes had lives down to 130,000 hours. As expected, the tubes with the lowest lives were found mainly in rows Nos 16 and 20, and tended toward the left side of the unit in these rows. The fact that the outlet bank tubes thinned more than those in the inlet bank strongly sug- gests that the mechanism of wall thinning is one that is dependent upon tube metal temperature, e.g. coal ash corrosion. The fact that both the inlet and outlet banks thinned on the left side of the superheater is not surprising; this may be the result of localized higher temperatures in the gas stream. Higher internal oxide thicknesses were not detected in the left side of the unit to verify this hypothesis, although there was substantial evidence to implicate scale exfoliation as a possi- ble explanation for this. Recommendations After examination of the NOTIS inspection data and the estimated tube remaining lives, BeW generated a set of recommendations for the exam- ined secondary superheater. These reeommenda- tions consisted of tasks that Board of Water and Light could execute at selected intervals in order to extend the life of the superheater and prevent forced outages. Inlet Bank The inlet bank did not exhibit severe wall thin- ning or thick oxide scales, and was found to have several years of creep life remaining. Therefore, no immediate action needed to be taken on this component. B&W thus recommended reinspection of the inlet bank with the NOTIS system after approximately 40,000 hours of additional service. A reinspection recommendation implies that the very same tubes be inspected the second time -- even the very same spots on the tubes, if possible. This is to facilitate the comparison of maps from both inspections, and so, best track the progress of the superheater. As with the initial NOTIS examination, the future inspection should include both oxide and wall thickness measurements, and remaining life estimation. Such advice is stan- dard for a component with no apparent problems. The reinspection interval is typically close to 50,000 hours, which is the standard band width on the remaining life maps. Outlet Bank As with the secondary superheater inlet bank, the outlet bank at the lower elevation was found to exhibit no apparent problems. However, it is the condition of these same tubes at the higher eleva- tion that was to dictate the final advice for this entire component. B&W generically recommends replacement of all tubes with remaining lives of 50,000 hours or less (‘5” symbols) in the near future. This recommendation is not meant to be imperative, but may be applied loosely by the util- ity to the particular situation. This allows flexibil- ity in the time and scope of the replacement. For example, it may be preferable for the utility to schedule the replacement during a future outage that was already planned, or it may be economi- cally feasible to replace entire elements rather than specific tubes, etc. Sometimes it may be desirable to follow this recommendation strin- gently, as in the case where the superheater is experiencing an increased frequency of creep fail- ures. B&W also standardly suggests the replace- ment of steam-carrying tubing with less than 85% of specified wall thickness. This is because thinned tubes have less capability of handling possible extended periods of above-normal temperature operation. An examination of the wall thickness and remaining life data for the outlet bank tubes at the 223 foot elevation showed eight tubes met both of these replacement criteria. These tubes were clustered around the left quar- terpoint of the superheater -- two tubes in element 8 and six in element 11 (Figures 2b and 2c). The recommendation to BWL was therefore the replacement of elements 8 and 11. Elements in between and flanking these elements could be reasonably assessed with wall thickness UT (since wall thinning was the predominating mechanism of creep life reduction in this region) to determine the full extent of a planned replace- ment. This step was also suggested. The fact that scale exfoliation was observed in the outlet bank, in both the removed tube sample and the NOTIS oxide data, was a matter of great concern. The occurrence of scale exfoliation is another criterion used to evaluate the condition of the superheater. If it is observed in conjunction with a creep-accelerating mechanism, the desire for replacement is reinforced; if alone, the problem can be eliminated by acid cleaning. The latter option was suggested for BWL’s secondary super- heater, in conjunction with the outlet bank replacement. Future Steps Planned Board of Water and Light is pleased with the results of the oxide thickness inspection, and intends to follow Babcock & Wilcox’s reeommen- dations. Eight consecutive outlet banks are sched- uled to be replaced during a planned outage in April, 1988. BWL plans to continue to track the condition of the secondary superheater with the NOTIS system at the proper time. BWL is also currently considering chemically cleaning the secondary superheater in the near future, possibly during the same outage as the outlet bank replacement. References 1. D. W. Bonin. “Nondestructive Oxide Thick- ness Measurement in Superheater and Reheater Tubing”, presented to the Electric Power Research Institute Fossil Plant Inspections Workshop, San Antonio, Texas, September, 1986. 2. S. R. Paterson, K. J. Clark. “Remaining Life Assessment of 2-1/4 Cr-1 Mo Superheater/ Reheater Tube”, EPRI RP2253-5, prepared for the Electric Power Research Institute Indus- try Advisory Group Meeting, Barberton, Ohio, May, 1987. 8. D. N. French. Metallurgical Failures in Fossil Fired Boilers, New York: John Wiley and Sons, 1983, pp. 155-163. Technical Paper Reduction of boiler tube failures via root cause corrective action in design and operation A. Banweg Engineering Technology Domestic Fossil Operations Babcock & Wilcox Barberton, OH J.M. Tanzosh Engineering Technology Domestic Fossil Operations Babcock & Wilcox Barberton, OH Presented to EPRI Conference on Boiler Tube Failures in Fossil Plants. Atlanta, GA November 10-12, 1987 Babcock & Wilcox BR-1322 a McDermott company Reduction of boiler tube failures via root cause corrective action in design and operation A. Banweg Engineering Technology Domestic Fossil Operations Babcock & Wilcox Barberton, OH J. M. Tanzosh Engineering Technology Domestic Fossil Operations Babcock & Wilcox Barberton, OH Presented to EPRI Conference on Boiler Tube Failures in Fossil Plants. Atlanta, GA November 10-12, 1987 Abstract BR-1322 Presently, boiler tube failures (BTF) account for more than five percent of the forced outage availability loss in utility industry fossil plants. EPRI has expended considerable effort to document the root causes of BTF (CS-3945). The dissemination and effective use of this information in the utility industry is essential, if appropriate corrective action is to be implemented to prevent repeated failures. Many of the root cause mechanisms of BTF are known to be aggravated by a cycling mode of operation, but relatively little is generally understood of exactly how cycling affects the response of boiler materials to the failure modes. For water-side initiated failures, pressure boundary material options for improved performance are severely limited. Therefore, control of the water chemistry environment, and an understanding of the relationship of boiler design features and failure mechanisms are essential. Recommendations and guidelines are described herein. For fireside corrosion mechanisms several resistant materials and coating options have been tested. Their relative benefits and shortcomings are discussed. Introduction Forced outages due to boiler tube failures (BTF) account for more than five percent of the availabil- ity loss in the fossil plant utility industry.) This has a significant economic impact on the genera- tion cost of electricity. Also, BTF have the greatest single impact on lost availability of the failure categories statistically tabulated and recorded by NERC.) The U.S. utility industry through EPRI and in co- operation with the major boiler manufacturers, has expended considerable effort to document the root causes of BTF (RP-1890-1). This effort identified 22 BTF mechanisms (Figure 1). Of these, 18 were doc- umented to have root causes that are well under- stood and have permanent identifiable solutions. Of the remaining four, two, fireside corrosion (coal ash & oil ash) and fly ash erosion, were identified as having root causes and solutions that have not been sufficiently demonstrated to the utility indus- try. Thus two state-of-the-art demonstration projects (RP-2711) are presently under way. Two other fail- ure mechanisms, corrosion fatigue and waterwall fireside corrosion (circumferential cracking) were identified as having neither identified root causes nor, permanent solutions. These two projects have since been funded by EPRI in RP-1890, as combina- tion laboratory and field test programs. Since 18 of the 22 BTF have identified root causes and permanent solutions (other than those whose root cause is human error or quality control) there is a very significant potential to reduce the availabil- ity impact of BTF by implementing appropriate corrective action. The product of EPRI Project RP-1890-1 was Man- ual For Investigation and Correction of Boiler Tube Failures (CS-3745). This manual is being distributed to EPRI member utilities. Additionally, EPRI has also solicited 20 host utilities to implement formal BTF reduction programs based on the root cause A. Stress Rupture © Short-Term Overheating ¢ High Temperature Creep © Dissimilar Metal Welds B. Water-Side Corrosion © Caustic Corrosion e Hydrogen Damage Pitting (Localized Corrosion) Stress Corrosion (Cracking) C. Fire-Side Corrosion e@ Low Temperature Waterwall © Coal Ash © Oil Ash D. Erosion © Fly Ash e Falling Slag © Sootblower © Coal Particle E. Fatigue ¢ Vibration e Thermal © Corrosion F. Lack of Quality Control Maintenance Cleaning Damage @ Chemical Excursion Damage © Material Defects e Welding Defects Figure 1 Failure mechanisms of boiler tubing. and corrective action information contained in the manual. This dissemination of information is essential to educate utilities in the diagnosis of the root cause of tube failures and to promote the implementation of the proper corrective actions to prevent repeat failure, and eventually to reduce the impact of BTF on unit availability. Waterside Initiated BTF Pressure boundary material options for improved resistance to waterside damage are limited by the ASME Boiler and Pressure Vessel Code Section I (hereafter referred to as the Code) (Table 1). In par- ticular, the use of austenitic stainless steel materials for water wetted surfaces of a fired boiler is prohi- bited, due to the susceptibility of these alloys to catastrophic failure from stress corrosion cracking. Though these materials might offer increased general corrosion resistance, the Code limits their application to dry steam cooled surfaces where exposure to stress-corrosion inducing agents, halides, and caustic, are unlikely. Maintenance of the protective oxide film of mag- Table | Commonly Used Boiler Tube Steel Grades (1) Tube Nominal Steel ASME Composition Type Spec Grade Percent Carbon Steel ERW?) SA-178 A 0.06-0 18C Cc 0.35C Max D 0.27 Max Seamless SA-192 - 0.06-0 18C Seamless SA-210 Al 0.27C Max c 0.35C Max Ferritic Alloy ERW SA-250 Tla C -0.5Mo Seamless SA-209 Tla C -0.5Mo Seamless SA-213 T2 0.5Cr - 0.5Mo T5 5Cr - 0.5Mo T99) 9Cr - 1Mo T1l 1.25 Cr - 0.5Mo-Si 712 1.00 Cr - 0.5Mo T22 2.25 Cr - 1.0Mo T91% 9Cr- 1Mo-V Austenitic Stainless Alloy Seamless SA-213 TP304H 18Cr - 8Ni TP316H®) 16Cr - 12Ni - 2Mo TP321°) 18Cr - 1ONi - Ti TP3479) 18Cr - 10Ni - Cb TP347H 18Cr - 10Ni - Cb “© ASME B&PV Code, Section 1, Power Boilers -Part PG 9, Pipes Tubes and Pressure-Containing Parts °) Electric Resistance Welded 3) Not commonly used in modern boilers Recently approved netite on the tube metal is critical to the protection of the steam generator tubes from waterside corro- sion. The purpose of boiler feedwater conditioning and internal boiler water treatment is to maintain an environment conducive to the formation and repair of this protective oxide film.) For most boiler pressure part materials a min- imum corrosion rate is exhibited between a pH of 9.0 - 11.0 (Figure 2). In addition to total dissolved solids (TDS) and pH control, the dissolved gas (oxygen, carbon dioxide) content of the feedwater must be controlled to very low levels. Dissolved oxygen is very corrosive to boiler pressure parts and is typically specified to be less than 7 parts per bil- lion (ppb) at the economizer inlet. Actually, this value is typically attained at the deaerator outlet, and chemical oxygen scavenging reduces it to near zero at the economizer inlet. Boiler Design Provided that these basic boiler water parameters Approximate ph value at 77°F (25°C) Steam 4 Steam quality 7 5/9 alii 10 by weight 100: 250} : 3 ffir i2 13 14 Heat input to Q 200 smooth tube 2 Film Smooth tube . boiling 3S 150 ane = z a o ~ g ‘ 4 apes = 100 —_ Transition —4 2 boiling zone J —_~ m., Nucl 7 0220] Nucleate | Tube failure 50 10,0, boiling zone | 28 0 ee 0 | | | 0 ps 1. 1 | | 365,T][f1g0400 4000 | 20,000 100,000 200,000 i i 3650 38 Sel 04 10,000 40,000 —T 675 sat 850 1050 | 193810, m NaOH —————— 4 —_ a +<-—____—__—_——- pp jal Water Tube metal temperature, F Figure 2 Effect of ph on the rate of corrosion of steel by water at Figure 3 Effect of boiling on tube metal temperature. Furnace 310°C (590°F). smooth tube failure related to DNB. (alkaline pH, TDS and deaeration) are met, the waterside corrosion rate on pressure parts is neglig- ible. Therefore, no waterside corrosion allowance is either warranted or included in the design of fossil units. Operational waterside corrosion failures in boilers are not of the general or uniform corrosion type, but rather are due to localized corrosion phenomena. These localized phenomena require a concentrating mechanism to elevate the normal boiler water chemistry to levels that are destructive to the boiler materials. The thermohydraulic design parameters of the boiler manufacturer must maintain sufficient cool- ing of the waterwall tube material so that the use limits of the material are not exceeded for once- through type boilers, and nucleate boiling condi- tions are maintained at all operating conditions in the steam generating surfaces of recirculating drum type boilers. Departure from nucleate boiling conditions (DNB) in a recirculating boiler circuit will result in a short term overheat failure (Figure 3), if the tube is vertically oriented and located in a high heat flux zone; or possibly a corrosion failure if the tube is inclined (e.g. a nose tube), allowing steam-water separation in a lower heat flux zone. In these failures the steam/water interface serves to both significantly retard heat transfer through the tube, thus raising metal temperatures, and supplies a concentrating mechanism that allows low level boiler water solids to deposit locally and build-up to corrosive levels. Once-through boilers, by design, incorporate a region in which film boiling controls the heat transfer process. Correct design locates this region in an area of relatively low heat flux, and incorpo- rates materials with suitable strength and oxida- tion resistance at the temperatures achieved. Such a design requires the strictest control of water quality. Waterside Deposits The most common condition that causes waterside initiated BTF is the accumulation of porous depos- its on waterside heat transfer surfaces. In the case of once-through boilers these deposits act as an insulating barrier to heat transfer and cause an ele- vation in tube metal temperature, which, if left uncorrected, leads to a long term overheat BTF (Figure 4). In the case of the drum type or recirculating boiler, this porous deposit, in addition to acting as an insulator, which can result in an overheat BTF, can also act as the concentrating machanism that elevates the originally low level boiler water solids to corrosive levels at the deposit/tube interface (Figure 5). The nature of the boiler water solids then dictates whether the under-deposit corrosion BTF is of the strong-alkali, caustic-gouging, or the hydro- gen damage type associated with under de- posit acidic conditions.” As seen from the previous discussion, controlling waterside BTF must then also include the control of deposit accumulation on heat transfer surfaces. Direction of flow Typical internal fluid side surface appearance of the magnetite layer in a once-through boiler furnace tube Protective oxide layer ‘t— Base meta 0.25 mm (0.01 inch) Transverse section through internal magnetite layer in a once-through boiler furnace tube Steam escaping from mouth of steam chimney by successive formation and release of steam bubbles Capillary channels drawing liquid to the base of the steam chimney Heat flow Figure 4 Transverse section through internal magnetite layer ina once-through boiler furnace tube. This control is economically accomplished through an appropriate maintenance chemical cleaning schedule. The key word here is appropriate. Many guidelines for chemical cleaning exist, from the boiler manufacturer as well as from consultants, architect/engineers and EPRI.) But they are just that; guidelines. Many plant specific parameters affect this frequency, so each individual plant must determine its own unique (appropriate) chemical cleaning frequency. This uniqueness extends further to an appropriate cleaning system and procedure suited to the type and extent of deposition and its composition. A recent situation with a supercritical once- through type utility boiler did lend itself to a mate- rial substitution to correct an overheat BTF prob- lem. A furnace division wall was discovered to be experiencing a very high heat flux that led to increased internal deposition, and resulted in over- heat BTF. It was determined that the unit required chemical cleaning at six month intervals. The orig- inal division wall material, 1/2 Cr-1/2 Mo (SA-213 T2), was redesigned with modified 9Cr-IMo material (SA-213 T91). Modifications were also made to the tube array pattern (more nesting). The improved 2 Deposit | | | | = Thickness | | | | ‘| 2 Metal Surface Figure 5 Model of “wick boiling” in magnetite deposit. 70 [Vv A213 TP 304H 60 F 3 = 50F 2 o Z 407 © 30} z o SA213 T2 DH 20b g < oF sn SA213 T91 0 1 n mn —— 4 — 800 900 1000 1100 1200 1300 Temperature, (°F) Figure 6 Comparative stress rupture characteristics. Figure 7 Pendant return bend exhibiting out of service dissolved oxygen pitting. high temperature mechanical properties of the new material could safely tolerate higher levels of depo- sition than the original material, before reaching an effective use limit temperature (See Figure 6). Through this combination of improved design and material, this unit’s present chemical cleaning fre- quency has been increased to approximately 2 years. It is now controlled by the deposition in the first pass furnace wall circuitry, which is typical for units of this design rather than by that in the division wall. The unique configuration of this supercritical boiler made material substitution part of a solution to the problem. The more common solution of this problem would have been to find a way to reduce the corrosion product transport from the pre-boiler to the unit. Presently corrosion product transport is being reviewed as the utility would like to extend the unit’s required chemical cleaning frequency even further. Out-of-Service Corrosion The control of dissolved oxygen concentration dur- ing operation has been noted previously, but the need to control dissolved oxygen in the out-of- service boiler condition is crucial and cannot be over emphasized. Out-of-service corrosion damage to utility boiler reheaters due to dissolved oxygen attack in condensate is a well recognized phenom- ena (Fig. 7). Though this problem is typically asso- ciated with reheaters, all the ferritic pressure part materials of construction exposed to oxygenated condensate are equally susceptible to this mode of Normal Horizontal Reheater RH Outlet Header RH Inlet Header Improved Drainability Reheater Design RH Outlet Header RH Inlet Header Figure 8 Normal vs. improved drainability design for horizontal reheater surface corrosion attack. The reheater tubing generally has the thinnest wall and, therefore, is the first to be perforated by the resultant pitting attack. Dry stor- age under nitrogen, or wet storage with chemically treated demineralized water, are means to prevent this type of corrosion. Plant configurations and unit start-up schedule requirements often interfere with the application of these techniques to many plants, especially in the reheaters. Even though austenitic materials, with their improved corrosion resistance, would be permitted by the Code in this steam cooled application, these alloys may only change the failure mode. If the out- of-service condensate environment is severe enough to pit through ferritic materials, then it may be cor- rosive enough to cause intergranular attack or cracking of austenitic materials. Analysis of con- densate from various units with significant out-of- service corrosion problems has identified chlorides, sulfates, and low pH in addition to saturated dis- solved oxygen concentrations. A new high chro- mium, high molybdenum ferritic alloy that prom- ises to offer better resistance to this problem is presently under test at B&W. Even drainable reheater surfaces have been sus- ceptible to dissolved-oxygen pitting corrosion dam- age, since, with time, these horizontal surfaces may sag between supports, allowing condensate to col- lect. An improved drainable reheater has recently been designed and installed, (Fig. 8). Though this improved design is appropriate for horizontal con- figurations, a material substitution for pendant arrangements will still be pursued. Cycling Operation When a unit enters a cycling mode of operation (load cycling, on/off cycling, etc.), as opposed to base loaded service, a critical review of the increased potential for water side corrosion prob- lems must be made."”) The frequent start-ups and load transients associated with a cycling unit make it inherently susceptible to out-of-service corrosion damage, air in-leakage and pre-boiler/boiler corro- sion problems. The corrosion products generated in an unprotected pre-boiler system, typically oxides of iron and copper, are transported into the boiler dur- ing subsequent start-ups. It is not unusual to see suspended corrosion product levels in the part per million (ppm) range at the economizer inlet, for some time after a start-up, when the desired steady Figure 9 Cycle diagram with feedwater recycle line. state level is in the part per billion (ppb) range. In general, blowdown is relatively inefficient in remov- ing these corrosion products. They then deposit and accumulate on inside diameter heat transfer sur- faces. This increased inventory of corrosion pro- ducts, known to be the concentrating mechanism responsible for the under deposit corrosion BTF, makes the cycling unit more susceptible to water- side initiated corrosion BTF. This situation may mandate a more frequent chemical cleaning cycle to control deposit accumulation. During the transients of a start-up or a load change, the control of dissolved oxygen, pH, and other parameters are difficult, due to lengthy sam- ple lines, sensor lag, and treatment chemical injec- tion system control lag. One of the most difficult parameters to control during cycling operation is the boiler water phosphate concentration for drum type boilers on congruent phosphate chemistry con- trol. The detectable concentration of phosphate in the bulk boiler water will decrease with increasing load and pressure, and then return to the original level on subsequent load or pressure reductions. This phenomena, known as phosphate hideout, can make it difficult to maintain close control of the pH, and phosphate concentration. Both the magnitude of the phosphate hideout and the variation in load required to cause the phenomenon can vary consid- erably from one boiler to another."'!) In the past, phosphate hideout has been treated more as a nuisance than a real threat. Recently, a 100 x > 80 ra S 60 = 2 40 2-1/4.% Cr- 1% Mo a Ferritic Steel & g i 18 Cr-8 Ni Stainless Steel 0 538 593 649 704 760 816 871 (1000) (1100) (1200) (1300) (1400) (1500) (1600) Metal Temperature, C(F) Figure 10 Effect of temperature on coal ash corrosion rate. high pressure drum type utility boiler experienced severe internal furnace corrosion. Deposit analysis of the affected tubing detected sodium iron phos- phate, indicating that the sodium phosphate from the boiler water had participated in the corrosion mechanism. This was a relatively unique case, but for a unit in cycling service, phosphate chemical injection practice must be moderated to account for the chemical hideout behavior of the individual unit for its load cycling pattern.) Water chemistry monitoring and control difficul- ties that occur during transients also contribute to an increased susceptibility of the cycling unit to corrosion related BTF that are connected with deposit buildups. This can be caustic gouging, hydro- gen damage, and short or long term overheat ruptures. Recognizing the increased potential for corrosion BTF for a unit in cycling service, it is often justified to include or retrofit a pre-boiler recycle line and clean-up system (filter or condensate polisher) to pre-condition the feedwater purity prior to unit start-up (Figure 9). Also, if auxiliary steam capacity is available, the water could also be deaerated and used to mitigate the temperature shock associated with thermal shock damage to downstream compo- nents on rapid start up. Fireside Initiated BTF BTF mechanisms attributable to the outside or fire- side of tubes are a function of the local environment (gas composition and velocity, deposit composition and properties, and metal surface temperature), the static or dynamic stress state that exists in the tube, and the individual materials’ ability to withstand these. Proper original design, or corrective action of an existing problem, requires an appreciation of what these are and what part they play. The EPRI sponsored BTF manual has expertly summarized the failure types and mechanisms, and complimen- tary programs have left only waterwall circumfer- ential cracking as unresolved regarding cause and corrective action. As is true with the water and steam side, typical boiler design practice does not assume a fireside corrosion allowance. Rather, the water and steam circuits are located and sized, and materials selected, to result in conditions that will avoid appreciable wall loss. For anticipated severe envi- ronments in high temperature regions, the designer’s maximum flexibility is normally in materials selection. Other design approaches, such as burying highest temperature superheater tube legs in the center of pendant banks, or otherwise shielding them, are often employed. In water cir- cuits, options rely on such aspects as fuel and com- bustion conditions, as well as materials solutions which, based on Code restrictions, must of necessity allow for a pressure core of ferritic material. Mate- rial solutions to fireside problems in these boiler areas require consideration of bimetallic combina- tions or coatings. Coal and Oil Ash Corrosion These terms have been used to describe the external wastage mechanism causing BTF in steam- carrying-tubes excluding erosion. The phenomena of coal and oil ash corrosion are well studied. Coal ash corrosion results from the deposits of molten complex alkali iron sulphates. The severity of the attack varies with the corrosive- ness of the coal and with metal temperature. Corro- sion increases sharply between 593C (1100F) and 704C (1300F), and decreases at higher temperatures where the corrosive elements are not stable (Figure 10). The attack assumes a classic form of enhanced attack at the 10 and 2 clock positions, with 12 o’clock coinciding with the tube center toward the direction of oncoming flue gas (Figure 11).!°) Oil ash corrosion is manifest as a more uniform attack on the hot side of the tube, and is caused by a reac- tion of the metal and metal oxides with molten vanadates. Babcock & Wilcox has carried out a number of long-term corrosion research programs in boilers that operate at nominal steam temperatures up to 566C (1050F), burning particularly corrosive coal Gas Flow > ~<—- Gas Flow Left Side of Tube Right Side of Tube Typical Coal Ash Corroded Tube ~<«— Direction of Gas Flow Figure 11 Transverse sections of corroded tube. with high sulfur and alkali content. Test loops of bimetallic tube combinations of Type 310 stainless steel over Esshette 1250, and Inconel 671 (nomi- nally 50% Cr-50% Ni) over Incoloy 800H have been exposed under various conditions for up to seven years. This work has shown 310 stainless steel to suffer appreciable coal ash attack of up to 7 mils/ year at surface metal temperatures up to 607C (1125F). Under the same conditions, Inconel 671 clad tube samples exhibited only about 3 mils/year loss to corrosion. In a still higher temperature region of one boiler where direct furnace fireball radiation drives the surface metal temperatures to 704C (1300F), the Inconel 671 cladding was lost at a rate of 6-7 mils/year. Inconel 671 clad has been espoused for some years as the best available protection for coal ash corrosion of advanced plant designs. This work suggests that at the highest steam temperature under consideration in advanced plants, corrosion rates may be of such a magnitude as to force whole- sale tube surface replacement on fairly frequent intervals. (This presumes bimetal tube clad layer thicknesses on the order of 50-70 mils to avoid over thickening tube walls). The alloy is the best choice 1000 Relationship Between Na & V Content in Fuel Oil to Hot Ash Corrosion Rates Fe & Ni Based Alloys 50 Cr/50 Ni Alloy Corrosion Rate (MPY) 0 1 10 100 1000 Sodium & Vanadium Fuel Oil (PPM) Figure 12 Relationship between Na + V content in fuel oil and hot ash corrosion rates (14, 15). for nominal steam conditions up to 593C (1100F) and perhaps up to 621C (1150F), provided careful design practice is employed.4!5) Additional work is warranted in developing an improved corrosion barrier if steam temperatures are to achieve 649C (1200F) and long-term tube sur- face life is required. Alternately, only low sulfur coals, or coal blending should be considerated for those plants. One of the few new alloys to become available to the designer and to be applied in boilers is the mod- ified 9 Cr-1Mo (SA-213 T91). B&W has included this alloy in test loops in a boiler suffering chronic coal ash corrosion. As expected, it has been found that the alloy stands up better than the standard lower chromium ferritic grades, but is inferior to aus- tenitic stainless steels. It still has a niche in the designer’s arsenal of alloys, particularly if there is a desire to eliminate dissimilar metal welds, and fire- side corrosion conditions are not too severe. Fireside corrosion is aggravated by increasing metal temperature. It is important to recognize the part that tube inside surface conditions play with regards to metal temperatures. Both in water and steam service, ID deposits or oxides (respectively) can and will form. Both serve to insulate the metal surface and force average and peak surface metal temperatures upward. In the case of steam carrying tubes, the effect oxide buildup has on remaining creep life has gained wide appreciation, lately. In like manner, the increased metal temperatures exaggerate fireside corrosion damage. Likewise in water wall tubing, deposits can significantly 5 T ee »— Lab Tests * Service Components ‘| . = | NS E€ | £ 3h g aol, © | ao e con S ! S 2} I* 5 ° Oo e | e e 1 /,°? / e aya ° —_— e “TOU tse 0 o—syi |. § po 650 700 750 800 850 900 950 1000 1050 (1202) (1292) (1382) (1472) (1562) (1652)(1742)(1832) (1932) Operating Temperature - C(F) Figure 13 Summary of corrosion data for 50Cr/50Ni alloys in oil/ash environments (16). increase tube wall temperatures. Any furnace wall fireside corrosion that is obviously associated with high heat flux areas (e.g. at or above burner regions) can be partially alleviated if water side deposits are regularly removed and kept at minimal thicknessess. Our experience suggests that this may also play a significant role in the circumferential cracking mechanism being investigated under EPRI project 1890-8. Oil ash corrosion is dependent on the fuel’s sul- phur, vanadium, and sodium content and is temperature dependent. Corrosion rates increase with temperature above the liquation point of the vanadates. The utility industry has relied on aus- tenitic stainless steels, and in worst cases, the 50Cr- 50Ni alloy used in coal burning plants (Figure 12). However, even these alloys can suffer catastrophic wastage rates (Figure 13)."1® Fuel oil additives have been developed and used, but an EPRI sponsored study concluded that they have a rapidly diminished effect on oil ash corro- sion rates as component temperatures move above 632C (1170F)."7: 18) In addition to BTF due to oil ash corrosion, fail- ure of tube bank support systems must also be con- sidered. Their operating temperature may be signif- icantly above the heat transfer surfaces’ temperature. Oxidizing or Reducing Conditions Oxidizing flue gases are not corrosive to boiler materials at temperatures above the dew point. This Chromizing - A chromium diffusion process Alonizing - An aluminum diffusion process Table II Compositions of Materials Used In Low-NO, Corrosion Tests Material Cc Mn Ss P Al Si Cr Ni Mo Cu Co Fe Other Carbon Steel 0.27 0.93 0.058 0.048 0.10 bal (SA-210) Max Max Max Max Min Carbon Steel 0.50 0.75 0.050 0.040 bal (AISI 1045) Max Max Croloy 2-1/4 0.15 045 0.030 0.030 0.50 23 1.0 bal Max Max Max Max AIS! 304 0.08 2.00 0.030 0.045 1.00 19 95 bal Max Max Max Max Max AIS! 304L 0016 169 0.014 0.020 0.75 18.08 880 0.25 0.09 0.20 bal 0.072N AISI 309 0.054 162 0.008 0016 0.54 2210 1420 0.36 0.20 bal AIS! 310 0.033 105 0.008 0.029 0.50 2583 1920 0.24 bal Incoloy 800 0.03 0.84 0.010 0.32 1909 32.96 0.55 45.58 0.38Ti Incoloy 671 0.05 0.07 0.002 0.21 46.02 5292 0.32 Fe-Cr-Al Alloy 001 48 0.33 157 bal Y<0.01 Plasma 60 275 20 bal Coating A Plasma 70 90 bal 55 49 Coating B Plasma 19.0 13.0 35 bal Coating C Bal = Balance is even true when burning aggressive coals, includ- ing those containing appreciable quantities of chlor- ine (e.g. over 0.2% in the coal). Wastage of furnace tubing occurs when local reducing conditions occur; for example when fuel/air imbalances exist, result- ing in local substoichrometric combustion. Alter- nately, locally reducing conditions can occur against tube surfaces under deposits, particularly if they are fuel-rich. Normally, reducing conditions are limited to the lower portions of the furnace. Local substoichiome- try can be caused by burner misalignment with flame impingement on walls, and burner air regis- ter misadjustment, which will be worse with open versus compartmented windboxes. The same corro- sion concerns also must be extended to future Table II! Compositions of Simulated Low-NO,x Combustion Gases (Volume %) Ho9S Hp = O2,—s CO.— CO9_—— HSN Gas 1 05 05 50 16.0 80 700 Gas 2 20 05 5.0 160 100 665 Gas 3 05 1.0 215 80 690 staged combustion systems designed to meet low NOx guarantees. In order to address these challenges, B&W inves- tigated a series of materials and coating alterna- tives for resisting reducing environment corrosion (Tables II & III).!°) The results have been reported and offer a ranking of materials with regard to resistance to corrosion (Table IV). This work also identified that alternating oxidizing and reducing conditions result in more serious corrosion, in some instances, than continuous reducing conditions. Thermal cycling also aggravates corrosion of fer- ritic materials under continuous reducing condi- tions, apparently due to the resultant cracking and shedding of the scales that offer some protection under these conditions (Table V). Austenitic stain- less steels did not experience similar acceleration of attack under thermal cycling. This is apparently due to their bulk chemistry, and the plasticity of their thinner scales. A number of coatings, both the diffusion type and plasma arc applied, were also tested. Improved performance was noted in several of these protective coatings provided they were suf- ficiently free of cracks and holidays, and remained *Results from 1000 hours tests **Results from 3000 hours tests HI-15/HI-35 - Combination chromium and aluminum diffusion processes Table IV Corrosion Rates of Various Materials in Reducing Gases Containing 0.5% and 2% H2S at 260°, 371°, and 482°C HoS Corrosion Rate (mpy) Concentration 260°C 371°C 482°C Material Gas Volume, % (500°F) (700°F) (900°F) Conditions Carbon Steel SA-178 1 0.5 13** 4.0**, 8.5* - |sothermal Carbon Steel SA-210 2 2.0 o - 41.0*** Isothermal Croloy 2-1/4 1 0.5 0.9*,0.7** 4.4**,7/0* Isothermal Croloy 2-1/4 2 2.0 oo a 52.0*** Isothermal 304 Stainless Steel 1 05 0.3* 3.60* |sothermal 304 Stainless Steel 2 2.0 o 2 17.0*, 8.2*** Isothermal 304 L Stainless Steel 1 0.5 0.02** 0.17** o Isothermal 304 L Stainless Steel 2 2.0 2 12.0*** Isothermal 309 Stainless Steel 1 05 0.02** 0.03** a Isothermal 309 L Stainless Steel 2 2.0 a a 1.682% Isothermal 310 Stainless Steel 1 05 0.05** 0.05** 1.6* Isothermal Incoloy 800 1 05 0.2** 0.6** o Isothermal Incoloy 800 2 2.0 - a 13:0*** Isothermal Incoloy 671 1 0.5 0.01** 0.03** o- Isothermal Incoloy 671 2 2.0 -- 0.18*** Isothermal Carbon Steel SA-178 1 05 18.0* 80.0* Thermal Cycling 304 Stainless Steel 1 0.5 0.26* 7.6* Thermal Cycling 310 Stainless Steel 1 0.5 a 1.1* Thermal Cycling 16Cr-5AI-Fe Bal 1 0.5 0.3* Thermal Cycling Chromized Carbon Steel SA-210 2 2.0 0.25*** Isothermal Chromized Carbon Steel AIS! 1045 2 2.0 0.32*** Isothermal Chromized Croloy 2-1/4 2 2.0 0.28*** Isothermal Alonized Carbon Steel 2 2.0 - 2 0.02*** lsothermal Alonized Carbon Steel 1 05 0.02** 0.12** - Isothermal Aluminum-Dip Coated Carbon Steel 2 2.0 -- a 0.24*** Isothermal Aluminum-Dip Coated Carbon Steel 1 0.5 0.07** 0.03** = Isothermal HI-15 Coated Croloy 2-1/4 2 2.0 a a 0:03*** Isothermal HI-15 Coated Croloy 2-1/4 1 0.5 0.01** 0.01** 0.1* Isothermal HI-35 Coated Croloy 2-1/4 2 2.0 a o 0:7*** Isothermal HI-35 Coated Croloy 2-1/4 1 0.5 0.01** 0.03** Isothermal HI-15 Coated Croloy 2-1/4 1 0.5 o o 1.9* Thermal Cycling *** Results from 4000 hours tests 10 intact on metal surfaces. These tests, as well as actual field experience, indicate that local loss of coatings can act to aggravate corrosion or at least possibly mask corrosion effects that would nor- mally be more visually apparent spread over a larger surface area of tubing. Protective Barriers B& has been involved in an ongoing investigation of the products offered by the metals coating indus- try. Previous mixed or poor experience in the boiler industry has created a skepticism concerning coat- ing adherence and performance. However, there have been some impressive strides made over the last few years in coating technology. Expertise has been gained by the producers and users of these products, proper chemistries of coatings have been formulated for a number of applications, and most importantly, the criticality of coating application quality control has been appreciated and signifi- cantly improved. The skepticism of the past has been replaced with promise, combined with a mea- sure of caution. Only the most reputable of vendors of these services should be consulted, with proof of successful field experience being a strong factor. Plasma sprayed coatings continue to be the theo- retical application method of choice, but are the most sensitive to application operating variables (e.g. stand-off distance) and are thus difficult to apply uniformly in practice in boilers. This method is best applied to straight tube lengths before boiler fabrication, under controlled, automated conditions. Flame spraying with wire or powder feed is inher- ently less sensitive to spraying parameters, though it results in bonds of lower strength than the plasma method. The composition of coatings effective against var- ious fireside environments has been fairly well established. Success has been achieved with iron- based compositions with chromium and aluminum (Fe-Cr-Al), and those equivalent to ferritic and aust- enitic stainless grades (eg. 430, 304, 316), and high Cr-Ni combinations. Yttrium additions have proven useful in enhancing coating oxidation stability. In theory, the same alloying element logic applied to base metals can be applied to coatings when aim- ing for certain corrosion resistance characteristics. Diffusion-type coatings using aluminum or chro- mium or both have also been laboratory and field tested as fireside corrosion protection. These have proven to be sufficiently chemically inert to many of the environments of concern, but may suffer Table V Corrosion rates under reducing conditions, oxidizing conditions, and alternate oxidizing conditions, and alternate oxidizing and reducing conditions at 371°C and 482°C (mpy) and Reducing (c) *Note: The data is based on 1000 hour tests. (a) The reducing gas is Gas 1. (b) The oxidizing gas is Gas 3. (c) Gas 1 and Gas 3 alternated every 30 minutes in the test. Isothermal Thermal Cycling Material Temperature Gas No Deposit With Deposit No Deposit With Deposit Type 304 482°C (900°F) — Reducing (a) 3.6 43 7.6 77 482°C Oxidizing (b) - - 11 44 Type 310 482°C Reducing (a) 1.6 2.6 Ui 2.7 482°C Oxidizing (b) - - 04 13 HI-15 Coating 482°C Reducing (a) 0.1 0.1 19 1.2 on Croloy 2-1/4 482°C Oxidizing (b) = 14 1.2 Carbon Steel 371°C (700°F) Reducing (a) 8.5 6.3 18.2 11.9 371°C Oxidizing (b) - - 3.4 8.1 Alternate Oxidizing 32.5 25.8 S75C and Reducing (c) a - 32.5 25.8 Type 304 371°C Reducing (a) 0.3 0.24 0.26 0.23 Unsensitized 371°C Oxidizing (b) - - 0.05 0.06 371°C Alternate Oxidizing 0.09 0.08 and Reducing (c) Type 304 371°C Reducing (a) 0.19 0.25 0.20 0.21 Partially 371°C Oxidizing (b) - - 0.04 0.09 Sensitized Alternate Oxidizing 0.07 0.06 and Reducing (c) Type 304 371°C Reducing (a) 0.19 0.17 0.17 0.18 Full Sensitized 371°C Oxidizing (b) = - 0.02 0.07 Alternate Oxidizing 0.07 0.06 11 cracks and holidays that can lead to severe local- ized attack. Cracking of this sort can occur in the fabrication and handling of welded water wall pan- els and tube sections. These flaws have long been acknowledged in steam side diffusion-applied appli- cations (e.g. chromizing), but have proven accepta- ble in those cases due to the relatively benign nature of corrosion in pure steam. In aggressive fireside applications, such flaws may lead to serious metal penetration or undermining of the coating layer. Some Root Cause Investigation Guidelines (1) Failures that occur through an obvious loss of wall from the fireside may not necessarily be the result of abnormal fireside atmosphere con- ditions. Tube metal temperatures can be ele- vated by internal heat transfer surface condi- tions, deposition, or impaired circulation. (2) Stress aggravates and assists in BTF when act- ing with either water/steam side or fireside cor- rosive environments. Failures of this sort may often be characterized by multiple crack-like penetrations whose orientation can be very suggestive of fundamental cause. For example, circumferentially oriented cracks suggest stress contributions from thermal cycling or perhaps mechanical bending. Axially oriented cracks can be due to overpressure, or stress due to res- traint across tubes (e.g. panel walls). Defects adjacent to structural attachments or geometric discontinuities will often exhibit a complicated damage pattern, in response to the local stress fields. (3) Weld regions often act as sites for BTF because a geometric and metallurgical discontinuity exists at the weld edge, superimposed upon a heat-affected zone that possesses not only a hardened band, but also a slightly softened region. Since these welds often act as structural attachments, through which stresses are transmitted to tube surfaces, it is no wonder that weld regions are a common failure site. (4) Boiler modifications intended to improve stress level or weld location should not be attempted without soliciting input from the OEM to assure that boiler fit and function is not compromised. (5) Material upgrades can often improve BTF prob- lems, but these can be at the expense of thermal performance of the tube. If large scale material changes are contemplated, consider the effect 12 on the boiler functional performance. Even the change from a cold-finished to a hot finished tube can be a potential problem since ID surface roughness (and thus friction factor) can appre- ciably affect pressure drop through the boiler circuit. (6) Often, more than a single mechanism acts to cause a BTF. Commonly, environment and stress (cycling and static) act to cause failures combining creep and fatigue, corrosion assisted, usually with a stress dominated final phase. Careful examination of the physical evidence, combined with review of recent and long term operating data, location of failure, and BTF his- tory will be required to unravel the often com- plicated scenario involved to arrive at the pre- dominant cause of failure. References: 1. A. F. Armor, “Boiler Tube Failures: The Number One Availability Problem for Utilitites,” Failures and Inspections of Fossil- Fired Boiler Tubes: 1983 Conference and Workshop, EPRI CS-3272, Palo Alto, California: Electric Power Research Institute (EPRJ), December 1983, P. 14. 2. North American Electric Reliability Council, Generating Availability Data System: Component Cause Code Report, Princeton, New Jersey, North American Electric Reliability Council, 1980. 3. Angela V. Manolescu and P. Mayer, “Structure and Composition of Protective Magnetite on Boiler Tubes,” paper no. 174, presented at Corrosion/80 (National Association of Corrosion Engineers), Chicago, Illinois, 3-7 March 1980. 4. H. A. Grabowski and H. A. Klein, “Corrosion and Hydrogen Damage in High Pressure Boilers,” 2nd Annual Educational Forum on Corrosion, National Association of Corrosion Engineers, September 1964. 5. Latest Developments in Natural Circulation Boiler Design, Technical Paper BR-1085, M. Wiener, Babcock & Wilcox , Barberton, Ohio. 6. Once-Through Boiler Development, Technical Paper BR-1168, K. H. Haller, Babcock & Wilcox, Barberton, Ohio. 10. 11. 12. 13. . Current Waterside Corrosion Concerns in Fossil Utility Steam Generators, Technical Paper RDTPA 82-58, F. J. Pocock & A. Banweg, Babcock & Wilcox , Barberton, Ohio. . Manual on Chemical Cleaning of Fossil Fueled Steam Generation Equipment, EPRI Report CS- 289, Jan. 1984. . Consensus of Current Practices for Lay Up of Industrial and Utility Boilers, ASME Publication. A. Banweg, “Water Technology Concerns for Boilers in Cycling Service.” In Proceedings of the American Power Conference, Vol. 45, Chi- cago: Illinois Institute of Technology, April 1983. H. A. Klein, “Use of Coordinated Phosphate Treatment to Prevent Caustic Corrosion in High Pressure Boilers Combustion,” October 1962. Boiler Water Phosphate Chemistry, Plant Ser- vice Bulletin PSB-25 Babcock & Wilcox, Barber- ton, Ohio, Technical Paper by J. Stodola of Ontario Hydro, “Review of Boiler Water Alka- linity Control.” High Temperature Gas Side Corrosion in Coal Fired Boilers, Technical Paper BR-1157, 14. 15. 16. 17. 18. 13 A. J. Blazewicz, M. Gold, Babcock & Wilcox, Barberton, Ohio. J.J. Demo, “Hot Ash Corrosion of High- Temperature Equipment,” Corrosion/78, Paper 98, National Association of Corrosion Engineers, Houston, TX, 1978 J.J. Demo, “Hot Ash Corrosion of High- Temperature Equipment,” Materials Performance, March, 1980, pp. 9-15 G. L. Swales, and D. M. Ward, “Strengthened 50% Chromium, 50% Nickel Alloy (IN657) Refinery Heater Tube Supports to Combat Fuel Ash Corrosion - A Review of Service Case Histories, “Corrosion/79, Paper 126. National Association of Corrosion Engineers, Houston, TX, 1979. D. W. Locklin, H. H. Krause, D. Anson, and W. Reid, Electric Utility Use of Fireside Additives, Electric Power Research Institute, January, 1980. EPRI Report Number CS-1318. Hydrogen Sulfide Corrosion in Low NOx Combustion Systems, Technical Paper, RDTPA84-12, S.F. Chou, P. L. Daniels, A. J. Blazewicz, R. F. Dudek, Babcock & Wilcox, Barberton, Ohio. Technical Paper conversion of Forest Waste into Energy - The West Enfield Project R. F. Johns R. E. Washcer Babcock & Wilcox Barberton, OH 44203 Presented to ASME Winter Meeting Boston, MA December 13-18, 1987 Babcock & Wilcox BR-1329A a McDermott company Conversion of Forest Waste into Energy - The West Enfield Project R. F. Johns R. E. Washcer Babcock & Wilcox Barberton, OH 44203 Presented to ASME Winter Meeting Boston, MA December 13-18, 1987 INTRODUCTION Babcock-Ultrapower West Enfield is a wood burning power plant which generates 25 megawatts of electricity. It is located near West Enfield, Maine. The plant uses a Babcock & Wilcox circulating fluidized bed boiler to burn wood waste from nearby forest operations and from mills in the area. A sister unit, Babcock Ultrapower Jonesboro, is an identical plant located about 90 miles away and started up on almost the same schedule. By creating a market for in-forest waste wood products, the plant provides an economic incentive for improved forest management such as forest pruning operations. Babcock & Wilcox West Enfield Power, Inc. is a partner in the joint venture that owns the plant. B&W supplied and erected the boiler island and is a part of the company which operates the unit. The boilers were hydrostatically tested in June, 1986 and initial operation began in December, 1986 - two years after award. These two boilers have now been in operation for approximately twelve months. This paper will describe the design, manufacturing, and construction of these boilers. Additionally, the performance and operation of the two units during the first year of commercial operation will be reviewed. WOOD FUEL The plant burns about 260,000 green tons of fuel per year. This fuel is supplied primarily from wood waste material taken from forests within 30 to 50 miles of the plant. Mill residues also are a source of fuel to the plant. Waste wood chipped in the forest and the mill residue are transported to the plant in trucks. At the plant, the trucks are unloaded in a truck dumper. The wood is then carried by conveyor to a disk screen which passes two-inch minus wood. Oversize fuel is reduced in size by a hog and is stored together with screened fuel in a fuel yard. It remains stored until needed for burning in the boiler. A front-end loader transports the wood from. the storage piles and heaps it over a reclaimer. Two traveling screws under the reclaim pile load the wood on a conveyor which carries it to a metering bin. The four feed screws to the boiler are supplied by the metering bin. CFB DESIGN PARAMETERS The boilers were designed to achieve the following specified performance: Maximum Continuous Steam Flow...... cc ccneee 218,640 lb/hr Main Steam Pressure........1250 psig Main Steam Temperature.....955 F Feedwater Temperature..... .269 F Air Heater Outlet Gas Temperature............ ow 2I9-F So, Emissions............-.0.15 1b/MKB No, Emissions..........- -..-0.15 1b/MKB CO’ Emissions...... eeeeeees-0.15 1b/MKB Hydrocarbon Emissions......0.15 1b/MKB Particulate Emissions......0.5 grains/ DSCF At 3% oO, entering precipitator In addition, the normal, commercial boiler guarantees, such as boiler efficiency, power consumption, draft loss, air resistance, and steam and water pressure drop were required. These performance guarantees were based on firing wood waste with the following analysis (as fired, % by weight): ASN. ccccccccccceece eeeceeee 3.46 DESCRIPTION OF UNIT The CFB boiler (Figure 1) consists of a water cooled, membraned furnace which is 11 £t-10 in. deep by 17 ft-10 in. wide by 66 ft-O in. high from the furnace floor to the roof, a hot particle separator and solids recycle system, and a convection pass enclosed by membraned, water cooled tubes containing a two-stage superheater and an economizer. Primary air enters the furnace via a bubble cap air distributor located in the floor. Along with primary air, fuel and recycled bed material are introduced into the lower or primary section of the furnace. The expected primary air is about 50% of the total air required for combustion. The balance of combustion air is supplied through secondary air nozzles located about five feet above the floor and tertiary air nozzles located about half way up the furnace shaft. The turbulent mixing action of the bed material in the primary zone provides both air and fuel distribution. The combustion process continues and is completed in the secondary zone of the furnace. Entrained bed material (sand) and flue gas enter the U-beam hot particle separators located at the furnace exit. The U-beams separate the solids for recycle back to the furnace and allow the cleaned flue gas to flow into the convection pass. A horizontal, drainable, two-stage superheater arranged for counterflow is followed by a horizontal, bare-tube economizer. At the bottom of the convection pass, the gas flows through a multiclone dust collector before entering into a two-pass tubular air heater where it is cooled to 275° F. The multiclone dust collectors remove small quantities of sand and fly ash which were not collected by the U-beams. The primary purpose of the multiclone system is to collect and recirculate bed material back to the furnace in order to reduce the quantity of sand make-up. Incoming air from the forced draft fan is preheated by the tubular airheater before flowing to the primary and secondary air systems. Following the air heater, passes through a B&W Rothemuhle electrostatic precipitator (ESP) before discharging to the atmosphere via a 135 ft high stack. The ESP has’ three electrical fields and is designed to attain specified performance with only two of the three fields in operation. the flue gas J-BEAM HOT PARTICLE SEPARATOR The boilers at West Enfield and Jonesboro utilize the patented U-beam hot particle separator to disengage the solids from the flue gas at the furnace outlet (Figure 2). The U-beam separator is a labyrinth-type mechanical separator formed from a staggered array of high strength stainless steel channels. These channels collect the bed material along with the fly ash present in the flue gas stream. The cleaned flue gas continues to the convection pass and the solids are discharged into a storage hopper located directly beneath the U-beams cavity. The U-beams are formed from 1/4 in. thick stainless steel plate and are 6-1/4 in. wide by 6-3/4 in. deep. The clear spacing between adjacent U-beams is 5-3/4 in. The West Enfield and Jonesboro boilers have eleven rows (deep) of U-beams. In addition to separating solids from the flue gas stream for recirculation, the U-beams impart a turbulent, mixing action on the flue gas. This mixing action coupled with a one second gas residence time in the subsequent cavity is designed to complete the conversion of residual CO to co... PARTICLE STORAGE HOPPER Located directly below the U-beams is the hot particle storage hopper (Figure 3). This hopper is formed with membraned tube construction and is cooled with saturated steam from the drum. By functioning as an accumulator, the storage hopper decouples the collection process from the recycle flow process. The particle storage hopper tubes are covered with a layer of refractory material. The refractory is attached with anchors and was installed by a "gunning" process. The function of this refractory coating is to prevent hopper tube erosion. L-VALVE The L-Valve is a solids flow control device that has been utilized for many years in the process industry. B&W applied this knowledge to the CFB product in order to have a precise, non=mechanical method of controlling the solids recirculation rate (Figure 4). The boilers at West Enfield and Jonesboro include four L-Valves for each unit. These L-Valves discharge hot bed material at a controlled rate into the primary zone of the furnace. The L-Valves are located opposite the fuel feed points. The L-Valve device actually starts at the bottom of the particle storage hopper. Hot solids flow at low velocities through the stand pipe. In addition to conveying the solids to the primary zone of the furnace, the vertical section of the L-Valve provides a pressure seal between the primary furnace (operating at approximately 30 in. 4H,0) and the material storage hopper toperating at approximately - 2 in. H,0). The lower section of the L-Valve is where the flow control occurs. By injecting a small quantity of compressed air (10-20 SCFM @ 50 psig) above the 90° bend in the L-Valve, the solids in the stand pipe and L-Valve horizontal leg are aerated. By increasing or decreasing the aeration air flow, the resistance in the L-Valve is increased or decreased and, consequently, the solids flow is increased or decreased. Solids flow control as a tool to control furnace density and, therefore, heat transfer is made possible because the particle storage hopper/L-Valve system is designed to decouple solids recycle from boiler load and air splits, thereby providing independent control of furnace density. The construction of the stand pipe/L-Valve component consists of a high temperature stainless steel liner, surrounded by several inches of insulation and sheathed in a carbon steel outer casing. The West Enfield and Jonesboro L-Valves are 12 in. ID, and are equipped with a high temperature expansion joint at an _ intermediate elevation to account for differential expansion. MANUFACTURING The manufacturing process followed a standard B&W project (Figures 5, 6, 7). Pressure parts, including the steam drum, furnace walls with headers, U-beam and convection pass_ enclosures, and the convective superheater were fabricated in the B&W manufacturing facilities in Paris, Texas and West Point, Mississippi. Boiler accessories, such as fans, air heaters, steel, and feeders were purchased from qualified suppliers using B&W specifications. The components of the boiler that were new to the manufacturing process included the U-beams, the steam-cooled material storage hopper, and the L-Valves. Although new, the relative simplicity of the design resulted in no unusual manufacturing problems. CONSTRUCTION Construction of the West Enfield unit began in October 1985. At that time, work commenced to pour the foundation and erect buildings. Site clearing and rough grading had been done a year earlier in anticipation of starting construction after the spring thaw in early 1985. However, financing problems arose because of the impact of the Seabrook nuclear project on Bangor Hydro's’ financial health. This delay in financing delayed the start of construction by about six months. Even with this late start it was possible to get the turbine island and the administrative spaces under roof to permit work inside throughout the winter. However, it was not possible to get the boiler island under roof. Instead the boiler support steel was erected during the winter. This led to low productivity and extra cost in the boiler construction. But, by working through the winter and by getting materials delivered to the site at that time, it was possible to work during the spring thaw season which would otherwise have been lost. As a result, the boiler was hydrotested early in June of 1986. This effort by the B&W manufacturing facilities and B&W Construction Co. permitted the project to maintain the original schedule for start-up in December of 1986 (Figures 8, 9, 10). PERFORMANCE Table 1 shows a comparison of guaranteed performance vs data gathered during testing. OPERATION Several operational issues have been identified during the initial twelve months of operation which have resulted in modifications to either operating procedures or equipment. FUEL FEED SYSTEM The wood feed system on these units consists of a vibrating screen which discharges into a live bottom metering bin. Fuel from the metering bin passes through rotary seal feeder and into variable speed injection screws. TABLE 1 Guarantee Test Results NO, 145 ppm 105-120 ppm co’ 145 ppm 30-80 ppm Particulate 0.5 gr/DSCF 0.5 gr/DSCF Efficiency 76.7% 76.7% Early in the initial operation, fuel feed from the metering bin was frequently interrupted as bridging occurred in the bin. An equipment design evaluation was conducted, and B&W decided to replace the original variable diameter, fixed pitch live bottom screw flights with constant diameter variable pitch flights. The modified design has resulted in excellent fuel feed system availability as witnessed by the lack of forced outages due to the feed system since’ the modification. The rotary seal feeders have experienced blockage and binding occasionally. The problems, however, have been caused by improper fuel sizing. When fuel size is within specification, the rotary seal feeders have functioned well. L-VALVE Performance of the L-Valve has been close to predicted. During normal operation, each L-Valve requires approximately 10-15 SCFM of compressed air at 40-50 psig. During the first few months of operation, the L-Valves developed mechanical problems due to thermal stress. These problems were traced to the initial method of recycling relatively "cold" material from the multiclone dust collector. When this cold sand (approximately 500 F) came in contract with the stainless steel inner liner of the L-Valves (operating at about 1600°F) the resulting thermal shock produced cracking in the liner welds, this situation has been overcome by the modified multiclone recycle system which now has- this material routed directly to the furnace. Since this process revision, the L-Valves have operated without incident. REFRACTORY Some initial refractory problems developed in the L-Valve piping. These were attributed to improper curing during installation. In addition, the refractory in the lower or primary zone of the furnace has shown susceptibility to erosion. This is the area of highest turbulence - particularly where the secondary air is introduced. Proper installation techniques along with smooth and gradual transitions from refractory to bare tubes have greatly reduced refractory erosion. In fluidized bed boilers, including CFB's, this is an area that must be periodically monitored and maintained. SOLIDS INVENTORY CONTROL During preliminary operation of full load, it became apparent that the quantity of make-up bed material (sand) was greater than expected. By monitoring operations, it was concluded that most of this difficulty was the result of sand becoming re-entrained into theconvention pass from the particle storage hopper. To remedy this situation, two design changes were made. First, the free flow area of the U-beam openings into the particle storage hopper was reduced thereby reducing re-entrainment potential. Second, bed material collected in the secondary mechanical collector was rerouted to be recycled directly into the furnace. The original design recycled this material into the particle storage hopper where the finer size particles were re-entrained into the convection pass. This resulted in overloading of the multiclone recirculation equipment. The results of these two modifications made a dramatic improvement on furnace/ combustion performance. PRIMARY/SECONDARY AIR FLOW As dramatic as the above’ mentioned modifications were to furnace control, additional operating experience suggested that combustion control and fuel flexibility could be further enhanced by providing wider ranges of primary and secondary air flow splits. We also found that the tertiary air ports located in the upper furnace had little effect on performance. It was decided to relocate the tertiary air ports closer to the secondary air ports. SAND AGGLOMERATION The wood ash in the West Enfield unit has a fairly high level of potassium. As the percentage of potassium in the bed increases, it tends to react with sand bed material in zones with local temperatures above 1600°F to form agglomerates. These agglomerates can be easily shattered, however, they also can block the flow path of L-Valves. A great deal of testing has’ been conducted to better understand the mechanics of agglomeration formation. Two positive operating actions were developed from the testing. First, limestone is fed into the sand bed. Second, bed drain samples are tested for potassium concentration and when a threshold level approaches, some of the material is blown down and replaced with fresh make-up. The period of operation has been extended to an acceptable level by use of limestone and sand blowdown. SUMMARY The first year of commercial operation at West Enfield and Jonesboro has yielded a significant amount of design and operating feed back. Process variables, such as, bed material sizing and air flow splits and mechanical design enhancements, such as, refractory application L-Valve construction, and fuel feed systems have benefitted from our experience at these sites. Additionally, operating data has verified furnace design standards, heat transfer coefficients, emissions levels, and the ability to handle considerable variations in fuel, moisture, and size. REFERENCES 1. Whitney, S.A., Johnson, H.L. and Hemmingsson, L. "Wood Fired Fast Fluidized Bed Boiler". Proceedings of Energy Technology Conference and Exposition, Washington, D.C., March 25-27, 1985. 2. Knowlton, T.M. and Hirsan, I. "Solids Flow Control Using a Nonmechanical L-Valve", Proceedings of the Ninth Synthetic Pipeline Gas Symposium, Chicago, Illinois, October 31 - November 2, 1977. 3. Johns, R. F., and "Design Wascher, R. W. and Construction of a Wood- Fired Circulating Fluidized Bed Boiler", Proceedings of The Ninth International Conference on Fluidized Bed Combustion, Boston, MA., May 3-7, 1987 Figure 1 Side view of the West Enfield and Jonesboro CFB boilers. Babcock- Ultra Power U-Beam Configuration I 18-0 1 ro ——_—__ - —_+ JUUUUUUUUUUUUUUUY KUUUUUUUUUUUUUUUY UUUUUUUUUUUUUUUY WuuuuuuuuuuU uuu lbuuuUuUUUUUUUUUUUY UUYUUUNUUUUYUUUUNUUUUY 14.0 UUUUUUUUUUUUUUUY| UUUUUUUUUNUUUUUUD, JUUUUUUUUUUUUUUUUE: jp tte tet tt 1} jb type t tat tio 4 | 4, —_ — —. —. 6.17/32 fouou uuu UO UU |= peu ouuououuouuil 3 Detail U-Beam Figure 2 Plan view of the U-beam particle separators installed at West Enfield and Jonesboro. Figure 3 Installation of the material storage hopper and U-beam enclosure. 160 120 ao So a Oo Mass Flow Rate 1000 Ibs/hr 8 12 16 20 24 L Valve Aeration Flow SCFM Figure 4 L-valve characrteristic curve showing solids flow vs aeration air flow. Figure 5 Panel fabrication of the U-beam enclosure and material storage hopper in B&W West Point, Mississippi manufac- turing facility. Figure 6 Shop assembly of the boiler floor, water-cooled windbox and lower rear wall. The bubble caps were shop installed along with the dense stud pattern for refractory application. Figure 8 Construction progress at West Enfield. Figure 7 Shop assembly of the water-cooled convection pass. Figure 9 Construction at West Enfield showing boiler, precipitator, and stack arrangement. The U-beams are shown in the lower-right foreground. Figure 10 Aerial photo of the completed West Enfield job site. Technical Paper BR-1330 The DOE Sponsored LIMB Project Extension and Coolside Demonstration A. S. Yagiela Babcock & Wilcox Barberton, OH T. B. Hurst Babcock & Wilcox Barberton, OH R. M. Statnick Consolidation Coal Company Library, PA Presented to Energy Technology Conference & Exposition Washington, D.C. February 18, 1988 Babcock & Wilcox a McDermott company The DOE Sponsored LIMB Project Extension and Coolside Demonstration A.S. Yagiela Babcock & Wilcox Barberton, OH T. B. Hurst Babcock & Wilcox Barberton, OH R. M. Statnick Consolidation Coal Company Library PA Presented to Energy Technology Conference & Exposition Washington, D.C. February 18, 1988 Abstract The possibility of changes in the Federal air pollution regulations could require electric power plants “grand-fathered” under the New Source Performance Standards (NSPS) to address improved emission control strategies. For these facilities, maximum flexibility in developing compliance strategies is important to maintain cost within a tolerable limit. A low capital cost, easily retrofitted pollution abatement technology, with a moderate emission reduction capability, could provide some flexibility. The Department of Energy has selected Limestone Injection Multistage Burner (LIMB) Technology as a demonstration project under the Clean Coal Technology Program. The Department of Energy (DOE) LIMB Demonstration Extension is an extension of the ongoing Environmental Protection Agency (EPA) LIMB Demonstration at the Ohio Edison Edgewater station in Lorain, Ohio. This paper describes the DOE LIMB Demonstration Extension Project, the addition of flue gas humidification for improved electrostatic precipitator particulate control and the Coolside technology. The connection with the EPA project and development work performed therein which provides the basis of the DOE humidifier detailed design is presented. Also, engineering design and construction of the humidifier at the Edgewater Station of Ohio Edison are discussed. Introduction The U.S. Department of Energy, as a contributor to the “LIMB Demonstration Project Extension” is engaged in a demonstration program to develop improved control technologies for emissions from the combustion of fossil fuels. This DOE project is an extension of the ongoing EPA LIMB Project, together with a demonstration of the Coolside process. The Coolside process employs injection of sorbent for S02 control into the flue gas down- stream of the air heater prior to humidification. The purpose of the project is to extend the LIMB technology data base by investigating process performance using a number of sorbents and coals. It also expands the DOE’s list of Clean Coal Technologies by demonstrating Coolside. The LIMB extension and Coolside project is funded jointly by industry and government. The major participants in this effort are the U.S. DOE, the State of Ohio, Babcock & Wilcox (B&W), Ohio Edison, and the Consolidation Coal Company. Presently available commercial options to reduce sulfur oxides emissions include switching to a lower sulfur coal, utilizing a method of coal cleaning to reduce sulfur content, and flue gas desulfurization (FGD) processes. A low capital cost alternative could be useful, even at moderate S02 removal levels (e.g. 50 percent). This is espe- cially true for plants with a few years of remain- ing life. Such a technology will increase flexibility in control strategies. The LIMB and Coolside pro- cesses represent lower cost, easily retrofitted technologies for moderate control of SOx and, with the use of low-NOx burners, simultaneous control of NOx from coal-fired electric power plants. The LIMB process injects a dry sorbent into the boiler for direct capture of S02 from com- bustion gases. LIMB also utilizes low-NOx burners in which controlled combustion reduces NOx generation. The Coolside process consists of injecting dry sorbent into the flue gas down- stream of the boiler airheater followed by flue gas humidification using water sprays. As such, the Coolside process is essentially boiler independent. Project description Background Since the DOE LIMB Demonstration Extension is a continuation of the EPA Demonstration, a description of the ongoing EPA LIMB Demon- stration and Humidification Modification is of value. EPA LIMB demonstration The EPA LIMB project is designed to demon- strate the long-term operation of the LIMB pro- cess in a full size utility boiler. The project will provide economic information as a basis for commercialization of the technology. Ultimately, the goal is to demonstrate that LIMB is a low cost retrofit technology which could then be applied to a significant portion of the existing coal-fired boilers. The EPA demonstration is a four-year project which includes design, installation, and operation of a LIMB system at the 105-MWe Unit 4 boiler at Ohio Edison’s Edgewater Station in Lorain, Ohio. This effort is a scale up of LIMB technology to a representative full-scale, wall-fired utility boiler. The demonstration goals are to: 1. Reduce SOx by at least 50 percent and to reduce NOx to a maximum of 0.5 lbs/106 Btu at a fraction of the capital cost of add-on flue gas desulfurization systems. 2. Prove that boiler operability, reliability, and steam production can be maintained during LIMB operation. 3. Solve any technical difficulties attributable to LIMB operation in a cost effective manner. These may include slagging and fouling problems, changes in ash disposal require- ments and increased particulate loading in the flue gas. Briefly, the LIMB Process equipment consists of a sorbent storage and handling system and a feed system. The sorbent storage and handling system is responsible for supplying the sorbent for the process. The sorbent is delivered by bulk trans- port truck in a dry state suitable for pneumatic conveying from a main storage silo to a feed silo. The sorbent feed system provides a controlled feed rate of sorbent to the injection elevations in the upper furnace of the boiler. This system can be utilized at any two of three injection elevations, at one time. Material is conveyed pneumatically from the feed silo to distributor bottles from which the solids and air mixture enters the injection lines. At the boiler, a series of injection nozzles is provided to give the desired penetration and dis- persion of sorbent into the boiler. A booster air fan provides the energy necessary to develop the desired penetration and dispersion. Figure 1 is the LIMB system process flow sheet. The LIMB Process also includes retrofit of low- NOx pulverized coal burners to meet the NOx emissions reduction goal. The Dual Register and XCL low-NOx burners developed by Babcock & Wilcox are generally compatible with utility boil- ers. For LIMB, the XCL low-NOx burners were selected for installation at Edgewater. An Ohio bituminous coal having a nominal sulfur content of three percent or greater will be used to demonstrate LIMB. Thus far, EPA LIMB construction activities as well as start-up activities are complete . Sorbent injection began in July of 1987 with preliminary process optimiza- tion testing in August and September. Testing was discontinued in late September due to an extended Ohio Edison boiler outage to perform scheduled major boiler and turbine maintenance. The outage activities were completed in January 1988. Humidification As a modification to the EPA Base LIMB work, the applicability of flue gas humidification as a supplement to the LIMB process for the purpose of improving particulate collection and S02 cap- ture is being evaluated. The reason for this inves- tigation is that the LIMB process causes some loss of performance in the electrostatic precipita- tor (ESP). This performance loss results from three factors: 1. The dust loading to the ESP more than doubles. Injector System Distributor Baghouse Truck Delivery Storage \/ 1 Fuel Air Compressor Transport Blower Combustion Air Air Heater Superheater Waste Handling and and Disposal Economizer Figure 1 LIMB process flow sheet. 2. The particle size distribution of the injected sorbent is finer than normal flyash and there- fore more difficult to capture. 3. The sorbent calcium content increases the resistivity of the ash. Humidification should provide a low-cost option that would restore ESP performance on LIMB retrofitted boilers. In addition, humidification of flue gas has been shown to increase S02 capture by improving post-furnace sorbent particle reac- tivity. Although the mechanism by which this occurs is not completely understood, experience shows that S02 absorption efficiency increases as the final flue gas temperature approaches the adiabatic saturation temperature. The incentives to humidify a LIMB flue gas are: 1. Particulate emissions from power plants with small ESP’s designed for lower resistivity fly ash from high-sulfur coal can possibly be kept in compliance with flue gas humidification without the high cost of additional particulate collection area. 2. Additional S02 removal can be achieved by taking advantage of the unused sorbent leaving the boiler. However, the level of humidification required to improve S02 capture efficiency and to restore ESP performance may present operational problems. Concern about scaling, plugging and associated operating problems as the gas temperature approaches adiabatic saturation influence the humidifier design. The EPA Humidification program provides a humidification system design aimed at minimiz- ing operational problems and maximizing S02 capture and particulate removal enhancement. The effect of humidification will be determined during the LIMB Demonstration at Ohio Edison Edgewater Unit 4, during the remaining months of EPA testing. The potential for scaling and plugging of the humidifier at the Edgewater plant led to the decision to conduct the humidification demon- stration in a bypass flue. With the humidifier installed in a bypass flue, any initial problems with it will not interrupt boiler operation. Condi- tions at closer approaches to saturation can be tried without fear of shutting down the boiler should a wall deposition problem develop. Figure 2 is a simplified process flow diagram for humidification. DOE LIMB demonstration project extension overview The purpose of the DOE LIMB Demonstration Project Extension is to extend the data base on LIMB technology and to expand DOE’s list of Clean Coal Technologies by demonstrating the Coolside process. The main objectives of this pro- ject are: Flue Gas From Boiler Humidification Bypass Shield Air He Atomization Air Existing [F-> Gas Flow | LJ Flue Gas To Existing Stack Existing Water Solids To Existing Waste Disposal System Figure 2 Humidification process flow diagram. 1. To demonstrate the general applicability of LIMB technology by testing three coals and four sorbents (total of 12 coal/sorbent combi- nations) at the Ohio Edison Edgewater plant. 2. To demonstrate that Coolside is a viable FGD process which injects sorbent into the flue gas after the air heater rather than into the furnace. The Coolside S02 control process involves injec- tion of hydrated lime followed by flue gas hu- midification with water sprays. A water soluble additive such as sodium hydroxide can be injected with the humidification water to enhance the S02 removal performance. The process is installed downstream of the air preheater and upstream of the particulate control equipment. A unique feature of the Coolside process is the sorbent par- ticle/water droplet interaction for enhanced S02 capture by the sorbent. Removal of S02 occurs in the humidifier and in the particulate control equipment. Sorbent utilization and S02 capture increases as the humidifier exit gas temperature approaches the wet bulb temperature. The tar- geted Coolside operating temperature is a 20-30°F approach to adiabatic saturation. Figure 3 is a simplified process flow diagram for the Coolside process. To achieve the project objectives, Babcock & Wilcox is performing a three-phase project con- sistent with the DOE Cooperative Agreement: Phase I: Phase IT: Design and Permitting Construction and Start-Up (“Shakedown”) Flue Gas From Boiler Sorbent Injection Humidification Bypass Shield Air Atomization Air Water Solids To Promoter Blower Existing Waste Disposal System Recycle WO Transport Blower Figure 3 Coolside sorbent injection with recycle. Phase II-A: Site Preparation and Long-Lead Time Item Procurement Phase II-B: Coolside/LIMB Construction, Start-up and Shakedown Operation, Data Collection, Reporting and Disposition Phase III: Phase I began on May 11, 1987 and includes detailed design of the humidification system bypass for Coolside operation in addition to selec- tion of the various coals and sorbents for the LIMB extension testing. Permitting and licensing requirements for both Coolside and LIMB are also addressed in Phase I which is scheduled to con- tinue through November of 1988. Phase II-A - Site Preparation and Long-Lead Time Item Procure- ment began at the same time as Phase I. These activities consist of removal of a retired precipita- tor from the roof of the boilerhouse to make room for the humidifier and procurement of items hav- ing long delivery times. The construction activities should be complete in May 1988. At that time, humidification testing, as part of the EPA base LIMB demonstration, is scheduled to commence. Coolside testing under the DOE LIMB Extension should begin by Febru- ary, 1989 and the DOE LIMB Extension testing will begin once Coolside is complete. The DOE Extension testing is scheduled to be complete in August, 1990. Figure 4 is an overall schedule for the project. Technology development The humidification process is the controlled addition of water to the flue gas to change gas characteristics so that emissions control technol- ogy can be more effective. For LIMB technology, the addition of water to the flue gas is expected to increase the sulfur dioxide (S02) removal by pro- viding a more conducive environment for unreacted lime particles to capture S02. Also, precipitator performance is expected to improve with humidification since the resistivity of the ash will decrease with increased relative humidity of the flue gas, resulting in improved particulate collection. Moreover, a reduced gas volume due to the lower temperature will result in a lower veloc- ity which will allow more time for particle collection. The Coolside Process is designed to remove S02 at an entering flue gas temperature of about 300°F. At this temperature, the reaction of the injected lime with S02 would not occur to the extent required without the presence of liquid water. This fact makes humidification an integral part of the Coolside technology. As with LIMB technology, humidification is expected to provide the same precipitator advantages for Coolside. Goals of the humidification process The goals of the humidification process are to: 1. Improve S02 capture efficiency by approxi- mately 10 to 15 percentage points over the Base LIMB objective of 50 percent to a total of 60 --65 percent (same basis), and provide the necessary humidification for the Coolside pro- cess. (For Coolside, an overall S02 reduction of 70 to 80 percent is expected). 2. Improve electrostatic precipitator perfor- mance. Based on operating experience gained thus far during the base LIMB testing at Edgewater, there is a need to enhance precip- itator performance. The addition of hydrated lime to the boiler during LIMB operation increases the particulate loading of the flue gas and increases the resistivity of the ash particles to a point where a back corona effect develops. This problem reduced precipitator effectiveness and resulted in increased opacity. 3. Maintain a reliable operability level. Humidi- fying the flue gas by spraying water into a flue gas duct operating at 300°F is a complex process. The water must be atomized to very fine droplets to allow rapid evaporation. The nozzles must be oriented in such a way as to prevent direct impingement of water spray on Coolside Startup LIMB Starup Phase III Coolside Testing LIMB Testing Reporting Equipment Disposition 87 88 89 90 1] Ps YT SES ES Ae TT TW ee Phase | Coolside Design Rie LIMB Test Plan ee. Permitting & Licensing Bests | Phase II-A Long Lead Procurement 2a Site Preparation ba: Phase II-B Coolside Construction Bil Figure 4 LIMB demonstration project extension schedule. the walls. Wet spots on the duct walls will lead to an intolerably high build-up of ash and lime particles. Design issues The goals of the Humidification Process presented a number of design issues which needed to be resolved: 1. Cross-sectional area of the humidifier - The cross-sectional area of the humidifier deter- mines the gas velocity and hence the residence time of the water droplets in the humidifier. If the area is too small, droplets exit the unit before complete evaporation occurs. These droplets may impact a wall at the first elbow downstream, causing wall wetting and a sub- sequent ash build-up problem. An oversized cross-section, which would maximize resi- dence time, reduces gas velocities to the extent that the flue gas lacks enough momentum to deflect the spray jet away from the walls. The expanding jet will impact and wet the walls early in the humidifier, leading to a build-up problem. The optimum humidifier cross- section was required information. 2. Atomizer arrangement - Additional informa- tion needed for humidifier design consisted of the type and number of atomizers, the opti- mum distance between atomizers and the min- imum distance from the array’s outermost atomizers to the wall. 3. Inlet Flow Straighteners - Good distribution of the flue gas at the inlet of the humidifier is an absolute necessity to eliminate localized areas of high water concentration and resulting wet areas in the humidifier. The exact placement of flow straightening devices was necessary design information. 4. System Pressure Drop - At Edgewater, the humidifier and the bypass duct in which it will be installed, impact the gas side system pressure drop and the induced draft fan capac- ity. Assurance that fan capacity, a site spe- cific factor, would remain adequate was necessary. Approach The approach taken to develop the required design information was to evaluate the perfor- mance of a number of commercially available atomizers, choose the superior performer(s) and test it(them) in an array in a 6 feet square duct. Concurrently, a mathematical model was deve- loped to improve understanding of the humidifi- cation process and a 1/12 scale cold flow model was constructed and operated to investigate gas flow patterns and gas pressure drop. The work to be described subsequently was per- formed under the Humidification Modification to the EPA LIMB Demonstration contract. It is summarized here due to its importance in the design of the bypass humidification chamber for use in the EPA Humidification testing and the DOE LIMB Demonstration Project Extension and Coolside testing. Atomization evaluation The first phase of the LIMB Humidification work was the testing of commercially available and B&W-designed atomizers. This work was per- formed in the atomization test facility at Be W’s Alliance Research Center. Laser diagnostics were employed to measure droplet size distribution in the sprays of each candidate atomizer. This tech- nique utilizes the Malvern 2600 Particle Sizer which interprets the diffraction pattern created by the spray droplets in a beam of monochromatic light to generate a droplet size distribution. The mean droplet diameter used for atomizer comparisons was the Sauter mean diameter (a volume-surface mean) due to its significance for evaporation. This parameter was calculated from the droplet size distribution data obtained with the laser diagnostics equipment. The air/water ratio was used because it indicated the amount of air required to perform the desired atomization. This air requirement impacts process economics through the capital and operating costs of the compressor. Therefore, the air/water ratio must be confined within realistic limits. Another useful measure of spray quality is the number of large droplets produced by the atom- izer. The percentage above 112.8 microns was chosen because it was a convenient and represen- tative size for the atomizers tested. Also, it pro- vided significantly different values across the range of air/water ratios. This information was used for comparison at various air/water ratios. The data follow the same trends as the Sauter mean diameter, i.e., fewer large droplets at high air/water ratio. The B&W MK-12 atomizer was chosen for use at the Edgewater plant. Performance as reflected in Sauter mean diameter as a function of air/water ratio is shown in Figure 5 for the MK-12 anda number of other evaluated atomizers. The MK-12 is rated at a nominal flow rate of 0.8 GPM. 90 \3 NY — Babcock & Wilcox MK-12 @ 80 SA 5 \ XN ——— Commercial Atomizers § ~“ (A through E) S 70 = = ~N = 604 NN ~Y 8 \ NN ~ @ D a = — 504 \ “XN ~ — = N ~N CL — a son ~ i < 404 SSN ~ 3 SLT NE = 30 ST 2 204 =—— a n 10 0 T t r T 1 r 9 0.2 04 0.6 Air/Water Ratio (Ib/Ib) Figure 5 Atomizer performance results for five commercial atomizers and the B&W MK-12. System design validation The System Design Validation Test was con- ceived to develop an arrangement of full-scale atomizers to provide the required humidification without wetting the walls of the duct. This per- formance should be achieved under varying inlet temperatures and flue gas flow rates with nearly complete evaporation at the outlet. The test duct fabricated to conduct the work was 6 feet square and 80 feet long. Hot air from natural gas com- bustion simulated the flue gas at temperatures ranging from 250°F to 320°F. The test facility was specifically designed to accommodate measurements made with laser diagnostic systems. Nine sets of opposing win- dows were installed on the sides of the duct. A greater concentration of windows was located near the atomizer plane to allow jet deceleration to be studied. Lasers were used to measure veloc- ity profiles, droplet size distribution, and the extent of evaporation. A total of eighty-six thermocouples was in- stalled to monitor atomizer performance. More than half were located on the interior surface of the duct (skin thermocouples) to provide a measure of wall wetting. The remaining thermo- couples measured the temperature distribution and wet and dry bulb temperatures at the inlet and outlet of the duct. The skin thermocouple measurements directly correlated with the per- formance of the atomizers. Skin temperatures which fell below the bulk gas temperature indi- cated an unacceptable degree of wall wetting at the point where the depression occurred. Thermo- couples at the exit of the duct provided an indica- tion of completeness of evaporation. Significant depression of these temperatures below the bulk gas temperature indicated impingement of un- evaporated droplets reaching the exit of the duct. Figures 6 and 7 show schematic and isometric views of the facility, respectively. Several important parameters were studied dur- ing validation testing to provide humidifier design information. These included flue gas veloc- ity, inlet gas temperature, atomizer-to-atomizer spacing, atomizer-to-wall clearance, water spray rate, and approach to saturation temperature. Overall, validation test results indicate that the 14 feet 7 inches square duct size is preferable to either a 12 feet or 16 feet size. This conclusion is based on laser evaporation measurements at the duct exit, temperature profiles of the skin thermo- couples and visual observations of duct operation. Evaporation measurements indicate over 99.7 percent of the humidification spray water was evaporated with the best atomizer arrangement. In addition, temperature profiles show no signifi- cant wall wetting at a 20°F approach to satura- tion with the 0.8 GPM I-Jet. Based on the testing results, the humidifier cross-section was set at 14 feet 7 inches square, the B&W 0.8 GPM MK-12 atomizer was chosen, a minimum atomizer-to-atomizer spacing of 12 inches, and a minimum atomizer-to-wall spacing of 24 inches were determined. The array will con- sists of 11 rows with 10 atomizers in each row, for a total of 110 units. Cold flow model A flow model was used to determine location, dimensions and types of flow correction devices necessary to provide uniform gas flow, particu- larly at the inlet to the humidifier. This three-dimensional, 1/12 scale model was constructed of plexiglass and plywood as shown in Figure 8. The model begins at the intersection of the bypass duct with the existing flue work and ends when it rejoins the existing flue. Hoppers were modeled with the humidification chamber since they are already in existence under the retired precipitator which will be removed. These hoppers will initially be covered to demonstrate an in-duct process. However, the hoppers will serve as a readily available alternative to remove ash build up should it become a problem. WET/DRY SAMPLE WET/DRY BULB PORTS | PRESSURE PROBE ow FLOW \ tc Grin 4 TAP. wine / STRAIGHTENING SKIN TC TRANSITION AMBIENT BURNER SECTION 500 GAL. TANK BULBS ATOMIZING AIR ><} COMPRESSOR NATURAL GAS ORIFICE PLENUM FAN #1 FAN #2 Figure 6 Facility schematic. Exit To Atmosphere | Duct KO Observation Windows Air Supply and ; Fan No. 2 Laser Diagnostics Air Supply Fan No. 1 Figure 7 Isometric view of the 6 x 6 duct test facility. The results of the model study indicate that: 1. The bypass flue design is adequate to provide a uniform velocity distribution entering the humidification chamber. 2. Pressure drop through the bypass system should not cause flue gas system capacity problems. 3. If the existing retired precipitator hoppers are necessary, hopper baffles and extension plates are needed to reduce recirculation in the hoppers. The exact placement of these devices has been determined in the model study. Mathematical model Numerical analysis of the humidification duct proposed for Ohio Edison’s Edgewater Plant was conducted with B&W’s three-dimensional dry scrubber model, DRYMO. The goals of the model- ing effort were to: 1. Benchmark the DRYMO model against the data obtained from the validation test facility. 2. Evaluate the effect of duct cross-sectional area on formation of recirculation zones down- Existing Airheater stream of the atomizers by comparing pre- dicted performance for a 12 feet square duct with that of a 16 feet square duct. 3. Determine if flow recirculation occurs down- stream of the atomizers in the humidification duct recommended for Edgewater. The DRYMO model was successfully bench- marked against the velocity and temperature data obtained from B&W’s System Design Validation Test. The DRYMO flow predictions agreed well with test data. Evaluation of the 12 feet square and 16 feet square duct designs indicated that flow recircula- tion is not significant as a source of wall wetting. If wall wetting were to occur, the DRYMO model predicts that it would be more the result of droplet dispersion rather than direct impingement down- stream of a recirculation zone. Turbulent disper- sion was found to be less significant in the 12 feet square duct with higher gas velocities. However, the droplet evaporation expected in the 16 feet square duct with the lower gas velocities would be Atomizer Array Location Figure 8 Flow model schematic diagram. more complete. As a result, the 14 feet 7 inches square duct appeared to be the best compromise based on these results and those of the System Design Validation Test. Although the DRYMO predictions for the 14 feet 7 inches square duct show a recirculation zone in the corner between the roof and side wall, the impingement velocity at the reattachment point is negligible and gas temperature is rela- tively high. As with the other duct size predic- tions, the potential for wall wetting and deposi- tion at this location is believed to be insignificant. The potential for wall wetting and deposition will be greatest on the upstream side of hopper baffles, if hoppers are ultimately determined to be neces- sary for humidifier operation. Engineering design and construction General arrangement Humidification chamber bypass system The humidification chamber is a 14 feet 7 inches square duct located in a bypass system on the boiler house roof between the air heater outlet and the precipitator inlet. Although the anticipated industrial application of humidification is an “jn-duct” process, the humidification chamber itself was located in a bypass for the purposes of the project. This was done to dismiss operating concerns about possible boiler down time caused by potential humidifier start-up problems. Figure 9 is a plan view of the bypass system and humidification chamber. Turning vanes have been installed in the flue bends to minimize pres- sure drop and to distribute gas flow. The humidi- fier itself is located over four existing hoppers which were originally part of Ohio Edison’s retired precipitator. Baffles will be installed in the hoppers to minimize the effects of gas recircula- tion if the hoppers are used. The hoppers will initially be covered to simulate an “in-duct” process. Guillotine and louvered dampers will be used to extract the gas flow from Ohio Edison’s existing duct into the humidification chamber bypass duct. Initially, 100% of the flue gas will be bypassed into the humidifier. However, if ash handling problems are experienced in the ESP due to high moisture content, part of the hot flue gas can be diverted around the humidifier. This should improve ash handling by increasing the flue gas temperature entering the ESP. The guillotine dampers also provide a tight shut-off to divert flue gas away from the humidi- Stack New Bypass Flue With Humidification Chamber Existing Air Heater Outlet Flue From Outlet Flue To Precipitator Figure 9 Plan view of bypass system fication chamber to allow personnel to safely enter the area while the boiler is on line. Atomizing Water System Strained Lake Erie water will be used as the water supply for the humidification process. The water from the existing service water pumps is delivered to a spray water storage tank. An upgrade of Ohio Edison’s existing service water strainers used to filter the Lake Erie water, and the addition of a backflush system are required to meet the addi- tional water capacity necessary for humidifi- cation. The water for humidification is filtered through three new strainers before reaching the atomizer nozzles. The spray nozzles and the piping and fittings after the final strainer are stainless steel. Provisions have been made for the addition of NaOH to the water storage as part of the Coolside process. Compressed Air System An Ingersoll Rand compressor rated at approxi- mately 4600 ICFM provides the compressed air requirements for the humidifier. After passing through the three-stage compressor, the com- pressed air is piped to the air receiver tank. From there, the compressed air is filtered and trans- ported to the atomizer where it provides the energy necessary to atomize the humidification water. As with the water, the compressed air is filtered and stainless steel pipe is used after the final filter to prevent scale from entering the atomizer and having a detrimental effect on atomization. Lance Assembly The humidification atomizer array consists of a 10 x 11 array of atomizers supported inside the humidification chamber. A series of five carefully spaced atomizers is supported by each of twenty- two lances. Eleven lances enter the chamber from each side, meeting at the center, establishing eleven rows of ten atomizers. Lances can be indi- vidually isolated and removed from the humidi- fier for repair without interrupting humidifier operation. A shield air system designed into each lance provides each nozzle with an envelope of clean air to prevent solids buildup on the outside of the nozzle. Shield air The shield air is drawn from the forced draft fan room which typically runs between 100° and 115°F. This temperature assures that the shield air is warm enough to prevent water condensation as it enters the lance. The air is drawn into each lance by natural draft since the humidifier oper- ates at negative pressure. Reheater system A reheater has been installed at the precipitator outlet to reheat humidified gas prior to entering the stack. It will be used, as necessary, to increase the flue gas temperature to maintain plume buoyancy during operation. The reheater system is similar to several other successful reheat sys- tems designed by B&W. The boiler steam drum supplies steam to the reheater. Steam from the drum is regulated to a pressure of 350 psig to pro- tect the reheater from over pressure conditions. Coolside sorbent feed system Piping for Coolside sorbent delivery and distribu- tion will be installed to deliver sorbent to a point in the duct upstream of the humidifier. It is antic- ipated that a significant portion of the existing LIMB sorbent feed system will be used for Cool- 11 side. The equipment includes the LIMB sorbent storage silo and the pneumatic conveying system that brings sorbent to the feed silo. The present feed silo and the pneumatic conveying systems up to the present distributor bottles will also be util- ized for the Coolside demonstration. Coolside modifications include the installation of new piping from the end of the present pneu- matic conveying system to a point upstream of the humidifier. This piping will include another distributor bottle to assure good sorbent distribu- tion in the duct. Coolside Ash Recycle System A part of Coolside equipment requirements is the ash recycle system. This system will recycle a por- tion of the ash collected in the ESP back to the inlet of the bypass duct. It will consist of an ash recycle storage bin, a pneumatic conveying sys- tem, delivery piping and possibly another distrib- utor bottle. The details of this system are being developed’at the present time. Summary/Conclusions The DOE LIMB Demonstration Project Extension is an extension of the EPA LIMB Project, together with a demonstration of the Coolside process. The LIMB and Coolside processes represent lower cost, easily retrofitted technologies for moderate control of SOx emissions and, with the use of low NOx burners, simultaneous control of NOx emis- sions from coal-fired electric power plants. As part of the EPA LIMB Project, the applicability of flue gas humidification as a supplement to the LIMB process to improve particulate collection and SO, capture is being evaluated. A humidifier was designed, after extensive development work to resolve design issues, for the Edgewater station of Ohio Edison. The humidifier will be used in the EPA humidification test work at Edgewater as well as the DOE Coolside process demonstration, of which humidification is an integral part. Engineering and design of the humidifier, to be installed in a bypass, has been completed. Con- struction is under way and should be complete in May 1988. Humidifier operation is scheduled to begin in June 1988 in conjunction with EPA LIMB testing which will be complete in January 1989. Operation of the Coolside process, as part of the DOE Project, will begin in February 1989. The DOE LIMB Extension testing will begin once Coolside testing is complete. Although the research described in this article has been funded wholly or in part by the U.S. Environmental Protection Agency through contract 68-02-4000, it has not been subjected to the Agency’s required peer and policy review and therefore does not necessarily reflect the views of the Agency and no official endorsement should be inferred. Technical Paper upgrading a 1956 Vintage Recovery Steam Generator — Il M. W. Fridley Project Manager Mid-America Packaging Pine Bluff, AR 71611 J. A. Barsin Manager, Industrial Projects Power Generation Group Babcock & Wilcox Barberton, OH 44203 Presented to Canadian Pulp & Paper Association 74th Annual Meeting January 25-29, 1988 Palais Des Congres De Montreal Montreal, Quebec, Canada BR.1331 Babcock & Wilcox a McDermott company Upgrading a 1956 Vintage Recovery Steam Generator — II M. W. Fridley, Project Manager Mid-America Packaging Pine Bluff, AR 71611 J. A. Barsin, Manager, Industrial Projects Power Generation Group Babcock & Wilcox Barberton, OH 44203 Presented to Canadian Pulp & Paper Association 74th Annual Meeting January 25-29, 1988 _ Palais Des Congres De Montreal Montreal, Quebec, Canada Abstract BR-1331 Mid-America Packaging wished to increase steam production on their Recovery Steam Generator by 92% over the original design nameplate capacity. This upgrade was accomplished in two install- ments, one completed in 1986 producing 160% of nameplate steam and a second completed in November 1987 designed to produce 192% of nameplate steaming capacity. The original modifications (I) and the latest modifications (II) to the steam generator are given. Also before and after perfor- mance comparisons and operating observations are provided. Introduction Mid-America Packaging Company’s management decided to increase the total production of their existing Pine Bluff facility. A major requirement to attain increased production was increased steam generation and the associated liquor pro- cessing capabilities. The steam generator, conse- quently, was upgraded in October 1986 and, for one year, demonstrated an increase in steam pro- duction of 60% over nameplate, an increase in liquor processing capability of 20%, an increase in operational on-line time of 200% between dry cleanings, and a 99% reduction in total reduced sulfur (TRS) emissions. Another upgrade (II) exe- cuted in November 1987 had as its objectivé an additional increase in steam production to 91% over the original nameplate, with an associated 81% increase in liquor processing capability over nameplate. Objectives The Pine Bluff Mill management planned to increase production capacity by speeding up the paper machine. The existing pulp capacity of 250 t/d (276 bone dry tons/day) should be increased to 345 t/d (380 bone dry tons/day) after the upgrade. To support that production increase, management reviewed the pulp mill equipment and specifically the 1956 Kraft Chemical Recov- ery Steam Generator. The options reviewed included replacement of this 30-year-old generator with a new chemical recovery steam generator, or upgrading/life extending the existing nameplate rating of 226 t/d (250-ton) unit. An increase in steam production from the nameplate rating of 28.4 t steam/hour (62,800 lbs steam/hour) at 465 mPa (675 psig) design (operating at 400 psig) and (750°F) to a new capacity of 45.4 t/hour (100,000 lbs steam/hour at 414 mPa (600 psig), and 399°C (750°F) was identified as one objective. The nameplate solids processing capability of 226 t (490,000 lbs) of dry black liquor solids per day would be increased to 364 t (803,200 Ibs) - an increase of 64% over nameplate or 20% over pre- viously attained levels at the new steaming level of 45.4 t/hr (100,000 Ibs/hour). (Referred to as Modification I). The second or downstream objective was set at attaining 54.4 t/hr (120,000 lbs steam/hour) at 41.3 mPa (600 psig), and 399°C (750°F) at an equivalent increase in solids capability per day. (Referred to as Modification II). The third objective was to reduce emissions from this unit, presently equipped with a cascade- type direct contact evaporator. Baseline TRS emissions at the boiler outlet were in excess of 700 ppm, which were to be reduced by 90% to aid in reducing the TRS levels from the cascade evaporator. The fourth objective was to increase the contin- uous run time from an operational limit of 14 weeks, after a clean start, to 24 weeks. The objectives could all be met with a new, modern Kraft Recovery Steam Generator designed for the processing level specified. How- ever, the possibility of upgrading the existing recovery steam generator remained uncertain as did the cost-effectiveness of such a project. Design Approach In August of 1986, Mid-America, its consultant, T. King; and Babcock & Wilcox, together identi- fied a design approach for a retrofit to the exist- ing tangentially fired combustion system. This plan had a high probability of achieving the four objectives desired. The cost-effectiveness of upgrade versus replacement was evaluated by Mid-America, resulting in the decision to upgrade as the preferred way to go (Figure 1). The traditional, empirical standards of capacity limitations were far exceeded. For example, the cubic feet of furnace volume per ton of liquor pro- cessed would be 85 m?/t (30 ft?/ton) at the new rating, far exceeding conservative new boiler lim- its of 184 m3/t (65 ft?/ton). The pounds of solids per square foot of plan area at 41 Kg/M? (1020 lb solids/ft2) would also far exceed the 23 Kg/M2 (555 Ib solids/ft?) index utilized for new units. The upgrade approach would have to concen- trate on the combustion system and use the most recent experience gained, both by modeling tan- gential versus B&W firing systems and actual air flow upgrades tried on other operational B&W steam generators. These evaluations, and the actual performance data obtained from Pine Bluff, indicated that traditional indices were con- servative. The 1956 combustion system could be upgraded at a relatively minor cost, and the upgrade would concentrate upon improving air flow distribution and increasing the heat level of the bed. Reference No. 1 covers the background of the present B&W design and combustion philosophy and compares it to the as-found design to provide a basis for the modifications performed. Actual Modifications Combustion System - The lower furnace, from ter- tiary ports to the bottom, was equathermed (made gastight) by the addition of metal bars (TIG) welded to the existing tubes. Also, studs were added to reduce heat transfer. These studs, CO> 2500 Secondary Air Black Liquor Spray Nozzles ALE Figure 1 Mid-America Packaging, Inc., Original Design 9.65mm (3/8-inch) on a 10.16 x 10.16 cm (4x4-inch) matrix, were field applied to existing tubes up to the new gun location. From that elevation to the top of the tertiary air ports, 3/8 studs on a 2x3x2 pattern were utilized. In addition, pin studs were installed on the flat studs as well as on the pres- sure tubing to insure a frozen smelt layer over the total lower furnace. Air Distribution - A three-level air system was installed, which included new interlaced secon- dary ports with variable dampers and interlaced tertiary air ports with variable dampers, offset 90° from the new secondary ports. The primary port area was reduced by 30%, and the existing primary air supply duct was split in two to provide a duct for the new secondary air system. The auxiliary burners were isolated from the primary air system to increase primary air flow control and direction. The primary port size has been adjusted, as necessary, to optimize oper- ation by sliding in closure plates during the shake-down period (Figure 2). A 10cm (4-inch) H20 AP was found to be the minimum primary air/nozzle drop for acceptable operation. Five new secondary ports were created, located three feet above the modified primary. Ports are kept simple by creating nozzles from the spaces between tubes. Each port consists of two spaces fitted with a guillotine damper designed to permit variable port operation. The variable port guillo- Figure 2 Primary Air System Upgraded tine permits the creation of pressure drop at the nozzle, i.e., influencing velocities where it could influence penetration. Three secondary ports are located on the rear wall and two are located on the front wall. The ports are interlaced; i.e., per- mitting each to develop a 14 degree expansion profile in still air without, in theory, infringing on the adjacent or opposed projected flow path. Each feed duct to the secondaries is fitted with an air monitor to meter flow. The tertiary air system nozzles are located as low as possible but, due to building constraints, the ports are 1.37m (4-1/2 feet) above the new liquor gun elevation. Five ports in total are inter- laced, with three on one side wall and two on the other, as described for the secondary. Air moni- tors are utilized to meter total tertiary system flow. Each nozzle consists of one tube space 20.3 cm (8 inches) high, and each tertiary port is fitted with a variable pie-shaped guillotine damper at the nozzle. The tertiary feed required a new duct with expansion joints, control damper, and positioner. Existing fans were not modified and were util- ized “as is” for the new air system. Liquor Feed - The fixed nozzle was removed and a new B&W oscillating liquor gun and drive were installed physically 1.06m (3-1/2 feet) lower in the furnace. The oscillating gun now permits the ratio between wall dehydration/direct bed spray to be Table | Performance Following Upgrade | (SI) (USA) Predicted Test Predicted Test Liquor Solids 364 322 t/day 803,200 710,000 Ibs Steam Flow 45.4 t/day 100,000 92,000 Ibs Blowdown 8 6 % 8 6 % Excess Air In Furnace 15 15 % 15 15 % Total Heat Input 63.3 556 W 216.0 189.8 MM Flue Gas Leaving Economizer 92 78.6 t/hr 203,000 173,400 Ibs Steam Leaving Superheater 2.75 2.70 mPa 400 392 psi S.H. Steam Temperature 355 377 °c 671 710 oF Flue Gas Leaving Boiler 415 407 °c 719 766 °F Flue Gas Leaving Economizer 312 344 °C 594 652 oF Flue Gas Leaving Evaporator 154 149 °C 310 301 oF Water to Economizer 101 98 °c 215 208 °F Water to Boiler 154 143 °c 307 290 °F Air to Steam Coil 26.6 26.6 80 80 °F Air to Combustion 101 110 °c 215 230 °F Liquor to Evaporator 96 204 Liquor to Furnace 127 127 °c 260 260 oF % Solids to Furnace 58.5 64 % 58.5 64 Heat Content 14,003 13,770 J/g 6020 5920 Btu Smelt % Reduction 98.3 98.3 NOx, ppm 70 70 C02, % 116 11.6 CO, ppm 440 440 Thermal Efficiency % To Steam 54.8 59.8 54.8 59.8 Black Liquor Spray Droplets Stellite Material Oval Splash Plate Figure 3 Liquor Oscillator varied for this low solids 59.5% - 65% as-fired liquor. The final splash plate nozzle utilized was a No. 28 with a 49 degree splash plate angle (Figure 8). Performance Prior to Upgrade - Summarizing the as-found performance: TRS emissions had been running at 770 ppm and 14 weeks was the longest operational on-line period attained from a clean start, with a dry clean required every eight weeks after that to maintain output. Smelt reduction was found to be 91%; CO levels were found to be above 4000 ppm. Performance Following the First Upgrade - Perfor- mance following the upgrade modifications is presented in Table I and the predicted perfor- mance is tabulated as well. The on-line opera- tional period has been increased dramatically (twelve months at this writing) and the TRS boiler emissions have been reduced by over 99% to less than 7 ppm. The flow rate in gallons per min- ute to the oscillator has been reduced from 75 gpm to 65 gpm for the same pulp production rate. The Btu value of the liquor has increased by decreas- ing the “dead load” in the cycle by approximately 5%. That improvement in reducing dead load is, in turn, attributed to an increase in reduction (approximately 5%) and a decrease in particulate carryover. The reduced carryover (approximately 20% less now at higher steam output), in turn reduces the need for additional water per pound of solids to dissolve those recycled solids. In addi- tion, the decrease in carryover reduced the need for superheater sootblowing and has significantly increased the continuous operational period between forced waterwashes and dry cleans. The operational benefits are felt throughout the sys- tem and are cumulative. Summary - Modification | The relatively minor modifications made to the Pine Bluff combustion system, both air and fuel subsystems, have had a major impact upon steam generation, emissions, carryover, as-fired liquor Btu value, and continuous operation. All objec- tives have been achieved and all effects to date have been positive. Actual Modifications II Combustion System - The upper furnace from the tertiary air ports to the bottom of the mud drum elevation has been equathermed by the addition of metal bars (TIG) welded to the existing tube. Existing flat studs were dye checked for cracking and repaired as required. Air Distribution - The three-level air system installed during Modification I was modified to provide additional port area for the greater quan- tities of air required. Existing Secondary Air Port Modified Secondary Air Port 2 Tube Openings 4 Tube Openings Furnace Furnace = re 0 OO= a O00 O= Windbox Windbox feed Plan View Plan View —+| + 2.1/8" —+| |+-2-1/8" Naa a = | | Tube Tube Openings _| ++ Openings 15.5/8" 15.5/8" a }_}J \ +H i = J™eJ KI Front View Front View Figure 4 New vs. Old Secondary Air Ports Il vs. | Secondary Air Ports New 2-1/8" x8” Slots Existing 2-1/8” x 8” Slots — —— _ —~ Tr | : | p-— WA “SS | 4 Front Furnace = Wall Existing Ports (1 Slot) 5-1/8" Existing |+ 2-1/8" x8" HFA / Slots rH ry (Dampered) Typical Modified Ports 5-1/8" 2-1/8"—| |= New 2-1/8" x 8" Slot (Undampered) Figure 5 New vs. Old Tertiary Air Ports The primary air system was uprated by adding primary air flow monitors. No change was made to the number of primary air ports. A new steam coil air heater was installed to provide 204.4°C (400°F) air to both the primary and secondary systems. The five existing interlaced secondary air ports were doubled in size by using four tube spaces fit- ted with guillotine dampers to permit variable port operation (Figure 4). Using the spaces between furnace tubes permitted port creation with no pressure part modification required. The existing forced draft fan was reutilized and is operating in its test block. The five existing interlaced tertiary air ports were increased in size by doubling the area of the three ports closest to the centerline of the unit. The new ports created from existing tube spa- ces, as in Modification I, are not fitted with vari- able dampers, but the original five are, which will provide adequate velocity control. A new tertiary air fan providing cold tertiary air has been added and the existing flow monitor has been relocated to the new air duct (Figure 5). The existing induced draft fan was reutilized and was “tipped” to increase static pressure head. Liquor Feed - Provisions had been made to permit the addition of a second B&W oscillating liquor gun but operation to date has proven it unnecessary. Superheater - Due to the increased steam flow, the superheater pressure drop increased. Operating pressure was reduced to 570 psig to stay within drum pressure operational limits. Existing safety valves had sufficient relieving capacity and did not require valve or trim change-outs. The higher thermal loads, gas side temperatures, and flue gas velocities associated with the higher inputs required for the 91% steam flow increase made the existing superheater material and thickness mar- ginal for long-term reliability. Selected super- heater elements exhibiting metal loss were replaced during the November 1987 Upgrade II outage. Performance Following Second Upgrade II - At this writing, the unit has been operating for only two weeks at a steam flow of 49.8 t/h (110,000 pounds per hour). Preliminary performance is tabulated and compared to predicted as shown in Table II. Tertiary Top of Equatherm Air Ports o)<+—New Tertiary Fan Spray Stud Pattern Oscillator 2x3x2 Secondar: i Air Ports. 5 Stud Pattern _ 4x4 Primary J Air Ports —<—£ Figure 6 Mid-America Packaging, Inc., Modified Design Summary The unit continues to demonstrate the improve- ments realized from Modification I but at higher thermal outputs. The attainment of the secondary objective, with steam flow up to 54.4 t/d (120,000 lbs/hr), (92% over design nameplate) will be met. Figure 6 depicts the major changes made during both modifications. Additional operational history will be available for the presentation in January, 1988. References 1. Barsin, J.A., and Fridley, M.W.; “Upgrading the Combustion System of a 1956 Vintage Recovery Steam Generator’, presented at the 1987 TAPPI Engineering Conference, New Orleans, LA (BR-1317). Predicted Performance Following Upgrade I! (SI) (USA) Predicted Predicted Liquor Solids 415 t/day 915,000 Ibs/day Steam Flow 54.2 t/day 120,000 Ibs/hour Blowdown 6 % 6 % Excess Air In Furnace 19 19 Total Heat Input Flue Gas Leaving Economizer 110 t/hr 243,400 Ibs/hour Steam Leaving Superheater 3.93 mPa 570 psig S.H. Steam Temperature 382 °c 719 oF Flue Gas Temperature Entering the S.H. 1166 °c 2130 TE Flue Gas Leaving Boiler Flue Gas Leaving Economizer 499 °c 929 TE Flue Gas Leaving Evaporator 162 °C 325 ne, Water to Economizer 98 °c 208 rr Water to Boiler 151 °c 305 °F Air to Steam Coil 26.6 °c 80 °F Air to Combustion 204 °c 400 oF Liquor to Evaporator 96.1 °c 204 °F Liquor to Furnace 126 °C 260 oF % Solids to Furnace 64 64 Heat Content 14,584 J/g 6,270 Btu/Ib Smelt % Reduction 98 98 Thermal Efficiency % To Steam 55.2 55.2 Technical Paper BR-1332A Initial operating results of coal-fired steam generators converted to 100% refuse-derived fuel J. A. Barsin Manager, Industrial Projects Babcock & Wilcox Barberton, OH P. K. Graika Senior Production Engineer Northern States Power Company Minneapolis, MN J. A. Gonyeau Red Wing Plant Superintendent Northern States Power Company Red Wing, MN T. M. Bloomer Project Engineer CRS Sirrine, Inc. (North Carolina Division) Research Triangle Park, NC Presented to American Power Conference Chicago, IL April 18-20, 1988 Babcock & Wilcox a McDermott company INITIAL OPERATING RESULTS OF COAL-FIRED STEAM GENERATORS CONVERTED TO 100% REFUSE-DERIVED FUEL J. A. BARSIN Manager, Industrial Projects Babcock & Wilcox Barberton, Ohio P. K. GRAIKA Senior Production Engineer Northern States Power Company Minneapolis, Minnesota Jd. A. GONYEAU Red Wing Plant Superintendent Northern States Power Company Red Wing, Minnesota T. M. BLOOMER Project Engineer CRS Sirrine, Inc. (North Carolina Division) Research Triangle Park, North Carolina ABSTRACT The conversion of Northern States Power Company's (NSP) Red Wing and Wilmarth steam generators to fire refuse-derived fuel (RDF) is a unique project -- a retrofit repowering from coal to RDF is a first-of-a-kind. The use of the existing plant with the necessary modifications to the boilers has allowed NSP to effectively incinerate the fuel as required by Washington and Ramsey Counties. This paper covers the six-month start-up of Red Wing No. 1, commencing in May 1987, and the operating results since the plant went commercial in July 1987. BACKGROUND Two coal-fired power plants in Red Wing and Mankato, Minnesota, have been converted to burn 100% RDF. Four boilers, two each at Red Wing and Wilmarth steam plants, were selected for the conversion by the utility, Northern States Power Company. NSP's evaluation of future power generation needs indicated an opportunity to supplement conventional fuels, extend the life of older plants, and alleviate a waste disposal problem. The utility had capacity within its system (relative to more modern NSP plants) that was used for peaking service only. The two plants were 35 years old and had two steam generators each, rated at 125,000 lb/hr while firing coal on a stoker. They operated at pressures and temperatures similar to those of new units that were being sold to burn waste fuels. Burn trials conducted at both plants demonstrated the ability to burn RDF. Solid waste disposal is a_ particular problem in the Minnesota counties of Ramsey and Washington, which includes the city of St. Paul. A state mandate issued in 1980 declared that all waste must be delivered to a resource recovery facility by 1990. Existing landfills were full or nearly full. Carving out new ones is costly because of environmental restrictions and controls. Also, hauling distances increase as landfills are being located well away from metropolitan areas. It was clear to NSP that an energy and material recovery program involving municipal wastes was needed. Thus, they began to consider proposals for processing municipal solid wastes (MSW). THE RDF SOLUTION In mass burning, refuse in its as- received, unprepared state undergoes selective removal of bulky items such as refrigerators, mattresses, etc., with the remainder being fed into a furnace, where, finally, other non-combustibles are discharged for disposal. The RDF method further refines the as-received refuse by shredding, magnetically separating the iron, and by employing multiple stages of screening and air classification; after which it is reclaimed to yield recyclable products and a fuel. The fuel now sized can be burned in a conventional stoker-fired boiler dedicated to RDF combustion. NSP contracted with the two counties to provide solid waste processing service for 20 years. During the planning, design, and construction of the processing facility, the 35-year-old boiler plants were redesigned and modified. The utility also contracted with three boiler vendors to perform engineering studies covering the conversions from coal to RDF. Ultimately, Babcock & Wilcox was selected to redesign the plants and provide the material to permit conversion of these four boilers from coal firing to RDF. The objective of the repowering was to burn 15 tons per hour of RDF for each boiler. This paper discusses the plant design decisions, specifics of the steam generator design modifications, construction and operating problems/solutions, and presents the expected versus initial performance attained. FUEL PROCESSING SYSTEMS RDF FACILITY The Ramsey/Washington County Resource Recovery Facility, which is owned and operated by Northern States Power is an RDF processing plant located in Newport, Minnesota. Newport is approximately five miles southeast of the St. Paul metropolitan area. This location was selected primarily because it jis centrally situated to the MSW supply and convenient to both the Red Wing and Wilmarth steam plants. The Resource Recovery Facility is a "state-of-the-art" plant wherein 1,000 tons per day of MSW are received and processed into approximately 700 tons of fuel which is shipped to one or both of the power plants. MSW is received at the processing plant in the raw form from both the counties and private residents. The MSW is dumped from the county trucks onto a tipping floor, from where it is fed by front-end loaders onto a series of infeed apron conveyors. Grapple cranes are used to remove large nonprocessible, noncombustible materials (i.e., refrigerators, waterheaters and other large ferrous materials). The apron conveyors feed the MSW into a flail mill, which shreds the material and discharges it to a belt feed conveyor where a magnetic separator removes nearly all of the ferrous material. Then the feed conveyor moves the material to a series of disc screens, a secondary shredder, and air classifiers where most of the remaining noncombustibles are removed and the material is properly sized. The nonferrous, noncombustible materials (also called heavies) are conveyed from the air classifiers to a Residue Loadout Trailer for disposal at an off-site landfill. The ferrous materials are conveyed to a load-out trailer for off-site recycling. The remaining product is RDF, which is conveyed on a belt conveyor to the RDF loading area. The RDF is transported to the Red Wing and Wilmarth Steam Plants by a fleet of 20-ton, enclosed, RDF, transfer trailers. RDF TRANSPORT AND DELIVERY The RDF transfer trailers are specially designed for loading, transport, short-term storage and unloading of RDF. Each trailer can hold approximately 75 cubic yards of RDF. The trailers are loaded from the back. A ram-type compactor pushes the fuel into the trailers, compacting the fuel enough to ensure that the truck is completely filled to capacity, yet is not packed so tightly as to hinder unloading or handling at the plant sites. The fronts of the trailers are equipped with deflectors that prevent over compacting of the trailers when loading, and the trailers are equipped with hydraulic rams. A total of 50 RDF and ash disposal trucks arrive daily at the Red Wing Station. The RDF is delivered to a building which contains a live-bottom receiving pit. This receiving building is large enough to accommodate up to 20 RDF trailers at the same time. Unloading from individual trailers is automatic and is monitored, as necessary, by surveillance cameras or by receiving area operators. The receiving area presently accommodates 10 Compacted RDF eff reer 0] Apron Conveyor |] ROF Transter Trailers M FY - jagnet — Walking Floor Swing Door With Switch Transfer Conveyor Arch Breaker Roll 21/4Ton Capacity Screw Feeders, Variable Rate Furnace. Unit 2 Fig. 1 RDF receiving and conveying diagram. trucks. Filled trailers are stored at the Newport Facility with Red Wing deliveries established to meet actual firing rates. FUEL FEED SYSTEM Upon arrival at the facility the RDF trailers are unloaded, according to fuel demand, directly into the receiving pit. The bottom of the steel-lined receiving pit is a walking floor similar to those in the RDF transport trailers. These walking floors are operated when necessary to supply fuel onto the truck-unloading apron conveyor. A_ scalping roll over the apron conveyor effectively breaks apart the clumps of RDF into their original size distribution. The apron conveyor discharges onto the RDF transfer conveyor, which is located in the former coal gallery at the Red Wing Plant. The existing coal transfer conveyor was removed and replaced by a new, 54-inch belt conveyor. The transfer conveyor is approximately 324 feet (horizontal length) long and conveys the RDF from the receiving area into the boiler room, in the area of what was once the existing coal bunkers. The coal bunkers were removed to make room for the new fuel feed bins and distribution conveyor. The head end of the transfer conveyor is now equipped with a self- cleaning magnet. The prime separation is done at the RDF processing facility. The purpose of the magnet is to remove as much residual ferrous material as possible. Ferrous materials and aluminum can plug or restrict fuel feed equipment and/or plug undergrate air ports if they melt due to the intense heat on the grate. The transfer conveyor discharges onto the distribution flight conveyor (see Figure 1 for a schematic diagram of the fuel feed system) which is common to both units. This conveyor Superheater Hoppers Unit 2 Precipitator dtt1t Unit 1 Precipitator To Preciptator C11 Boiler No. 2 Air Heater Hoppers From Unit Inlet Dust 7 1 Collector joller Fl lyash ducaati ry Mixing Tank Sittin Water lone Conditioner joppers = Bottom Bottom Ash Ash t t t Ash Truck oO oou Bottom Ash Removal Conveyor Boiler Blowdown Tank From City Water Sump. Quench Water Quench Tank pia umps Fig. 2 Ash receiving and conveying system. feeds the fuel as necessary to the fuel feed bins. Slide gates in the bottom of the conveyor control the feed of fuel to each fuel metering bin. Fuel is fed to each boiler by two (per unit) fuel metering bins. Each bin has a capacity of 2 1/2 tons and each has a live bottom consisting of six variable speed screw feeders. The rate of fuel feed into the boilers is controlled by the speed of these screw feeders, which, in turn, is controlled by the new combustion control system. The metering bin screw feeders discharge into a pant-leg type section of chutework, where each leg discharges into an air swept spout-type fuel feeder. The four air swept spouts (per boiler) distribute the fuel in the furnace and are located on the stoker front, just above the grates. Each feed chute is equipped with a balance draft damper to provide a boiler seal and prevent back drafts and fires in the fuel feed system in the event of swings in the furnace pressure. ASH HANDLING SYSTEM Bottom ash, fly ash, and boiler ash are conveyed by a series of mechanical flight conveyors (Figure 2). The bottom ash conveying system consists of a three-foot submerged drag chain conveyor. The flooded upper trough quenches the ash as it falls off the grate and also provides an effective boiler seal. The ash is pulled from the submerged trough by the flights and dewatered on the 5C foot inclined (40° from horizontal) section of conveyor. The conveyor operates at 8 feet per minute to minimize wear and maximize dewatering of the ash by increasing the time the ash is on the dewatering section. Dewatering of the ash to approximately 25% moisture content is important 9 to minimize the potential for freezing in the trucks. All conveyor wear parts are of abrasion-resistant material (400 Brinnel). Boiler ash, including air heater, economizer, superheater hopper and siftings ash, is carried to the bottom ash conveyor by enclosed dry mechanical conveyors. STEAM GENERATOR DESIGN MODIFICATIONS Although mass burning of municipal solid waste has a long European operating history, RDF does not share a similar European heritage. RDF is largely a USA-developed technology driven by the goals of providing more efficient, more load responsive and more economical boiler facilities, coupled with a capability to recycle recovered materials (2). RDF has been burned in various ways since the late 1960s, both alone and in combination with other fuels. First generation plants adapted processing equipment from other industries and many lessons were learned. The learning process, typical of all new technology, has been continued and state-of- the-art RDF waste-to-energy facilities are being designed and built. The Red Wing and Wilmarth repowering projects have benefitted from the earlier experiences, which provided the design approach used for the repowering. Most combustion problems (slagging, fouling, erosion/corrosion) have been remedied by improvements in process facilities and increased conservatism in the steam generator design. The project is unique in that it is the first to utilize existing steam generators. BOILER MODIFICATIONS The existing boilers are balanced draft with variable speed ID fans. New electrostatic precipitators were installed in the early 1980s. The existing furnaces are of tube and tile construction. Neither plant was equipped with steam temperature control. The existing Wilmarth boilers were equipped with bare tube economizers. The existing Red Wing boilers (designed and supplied by Foster Wheeler) had extended surface economizers. None of the boilers were equipped with air heaters. Figure 3 shows the pre-conversion boiler outline that is typical of both plant sites. Meeting the objective of burning 15 tons per hour set the required furnace size, and operating experience dictated the addition of selective corrosion protection. The existing furnace had to be enlarged to meet the desired objective throughput, while at the same time insuring adequate tube wall life. Experience indicated that a furnace exit gas temperature (FEGT) of 1600°F is a practical maximum, given the economic trade-offs of alloy’ versus conventional gas-side tube materials for this corrosive environment. The relatively high levels of chlorides in the flue gas streams preclude the use of traditional stainless materials as the optimal tubing material Economizer Coal Bunker From Precipitator Induced-Draft Fan Boiler To Precipitator “—— Dust Collector Forced Draft Fan Fig. 3 Preconversion boiler outline. Economizer From ~ Precipitator Induced-Draft Fan Boiler From To ROF-Feed Precipitator — System Perera Giprel Air Heater —J Dust Collector Furnace Extension Forced Draft Fan Fig. 4 Post conversion boiler outline. choice. Ash materials in the RDF demonstrate slagging and fouling characteristics, not unlike coal, with the added challenge of removal, while preventing the exposure of new tube metal to corrosion. The cyclic removal of protective oxide coatings has not retarded corrosion. B&W's experience indicates that, at present, the best FEGT to balance the superheater corrosion- slagging-fouling/burnout concern is attained by limiting the furnace gas outlet temperature to 1600°F at maximum continuous rating (MCR) on RDF. It is desirable to reduce release rates for RDF below these rates for coal because RDF is a more difficult fuel to burn and generates corrosive gases. Longer residence times will aid combustion and reduce FEGT and associated furnace slagging. Additional furnace height Plan View Refractory O”W”O 3” Tubes on 6-1/2" Centers OO New Membrane 3” Tubes on 4” Centers Side Section Transition Header New Membrane Fig.5 Transition new furnace to old. could be gained by extending the furnace downward into the basement, removing the traditional stoker coal-bottom ash hopper and replacing it with a low-head submerged chain conveyor. Our evaluation indicated that the 15 tons per hour could be combusted and the FEGT of 1600°F not exceeded (Figure 4). OTHER FURNACE MODIFICATIONS The 14-ft furnace extension was fabricated from membrane panels and overlaid with 50 mills of high nickel alloy in selective areas to inhibit corrosion. The connection to the existing tube and tile "upper" furnace was accomplished, using a full periphery ring header, located outside of the heated area. The header permitted the former mismatch of tube spacings (old 6-1/2 inch to new 3 inch) to be matched. The water supply to the furnace extension is from the four new lower four-wall headers (Figure 5). The extension is supported by spring hangers from existing support steel. Natural gas firing capability was maintained to provide heat for start-up and for load carrying as a standby in the event RDF was not available. There were no plans to co-fire these fuels. Existing Loose Tubes The boiler outlet screen was redesigned to increase the side spacing (now 18-inch total), and thereby improve cleanability and increase the shielding of the superheaters from direct radiant heat transfer. The increased shielding was attained by rerouting two rows of generating tubes, turning them into new screen tubes. COMBUSTION SYSTEM The air, firing, and fuel feed systems all had to be modified to process the new fuel. The RDF design heating value of 5750 Btu/1b would vary due to the changes in moisture and constituents as a function of the time of the year. STOKER The furnace enlargement forced the lowering of the former coal traveling grate stokers by 14 feet. That relocation permitted opening the ash discharge height to 18 inches, dictated by B&W experience, to insure free ash discharge. That increase in turn required a new flat arch (to provide radiation shielding), new stoker front extension, and a new upper front support for the fuel feeders. Existing grade bars were retained and, as they failed, were to be replaced with ductile iron. Detroit Stoker furnished new rear tuyeres, removal of the rear coking section, new lower front seal, and new grate wear strips for the new fuel. FUEL FEED The distribution of the RDF on the grate is critical for proper combustion efficiency and to minimize the potential for lower furnace corrosion. The fuel feed was integrated with the air system to insure complete mixing, combustion, and to minimize corrosion due to reducing conditions. A sophisticated overfire air system (OFA) was installed to insure that the best possible mixing would be available so that all the advantages of the increased fuel preparation costs could be realized. The OFA system is intended to aid in keeping the RDF particle in suspension-enhancing dehydration, and by creating turbulence which increases its residence time. Four Detroit Stoker air swept spouts per unit were installed across the front face. Spouts measuring 18-in. x 30-in. had been used previously, and are the size applied to this project. The air swept spouts must distribute the fuel with some of the fines going into suspension, all the heavies on the grate, and the intermediates mixed. Grate speed has been increased from eight feet per minute to a 20-foot per hour maximum rate. Feed to the spouts is by two live bottom bins per boiler. AIR SYSTEM The existing unit had no air heaters and because RDF combustion requires hot air, North American Technologies plate-type air heaters were added to provide 355°F combustion air for both the undergrate and overfire air systems. The undergrate system is sized for 50% of the total air required. The OFA system is sized for 50% of the total air required. The objective of the combustion system, air plus fuel feed, is to maximize suspension burning of the fines while maintaining the combustion on the bed. The highly turbulent OFA system has _ been successfully demonstrated on RDF projects. The three rows of OFA nozzles plus the use of the air in the feed spouts has evolved over the years as an excellent mixing system. The OFA system required a fan to provide high static pressure to obtain the design nozzle velocities to assure penetration. The existing forced draft fan(s) had to be replaced with a higher static-pressure (now 5-inch H,0) fan to develop additional static to overcotfie pressure drops associated with the new air heater and the thicker ash bed on the grate. A new OFA fan using hot FD fan discharge air was supplied to provide the required overfire air. SUPERHEATERS The existing superheaters have been replaced with a new counterflow single-pass design, with increased clear side spacing. The coolest steam is in the hottest gas path and will act to lower tube metal temperatures. The original design had 850°F final steam temperatures; the redesign provides 700°F final steam temperature, which reduces metal temperatures, reducing both the fouling and corrosion potential. Spray attemperation is used to provide a final steam temperature control range, and alloy steel is utilized where tube metal calculations dictate. GENERATING BANK Internal tile baffles originally installed to promote cross-flow generating-bank heat transfer were removed. The generating bank is now a Single-pass, once-through, design that reduces erosion and plugging potentials. The resulting reduction in heat transfer has been made up by adding three rows of generating bank tubes which utilize the drum holes created by the relocated screen. The additional generating surface decreases gas temperatures entering the economizer. The existing economizers - one, an extended surface (Wilmarth); the other, bare tube with high gas velocities (Red Wing) - were replaced with bare tube types spaced to limit gas-side velocities at MCR to 30 feet per second to insure long life by minimizing erosion. Air heaters were described in the section on air system and are of the plate-type design. Regenerative air heaters had been used with RDF previously but were located in a clean gas stream after the electrostatic precipitators. Plugging of the hot, inlet side, due to oversize particulate carryover, was of high enough potential to eliminate a regenerative application. Tube and shell design would have been acceptable, but the plate type offered simplicity and equivalent performance. Gas temperature has been reduced to 450°F, which matches the existing electrostatic precipitator operating gas temperature. Gas-side erosion is of concern, and erosion shields are being used on the leading edge of each air heater plate to minimize this possibility. Existing variable speed ID fans and electrostatic precipitators have been reutilized. CONSTRUCTION Demolition of the boiler's bottom section and removal of asbestos were the first steps in construction. The stoker support steel was to be reused and had been match-marked for ease in reassembly. The match-marking, however, was not entirely accurate and time was lost during the reassembly. The sequence for erection had Unit 2 reconstructed first which forced all new pieces to be routed past Unit 1. Access was limited and space extremely constrained when components arrived. Certain interferences also became apparent; for example, cable trays, drain lines, blowdown piping, natural gas line, sootblowers for adjacent units, and both fuel and ash handling equipment all required interface connection, which demanded redesign time and additional costs. One specific example is the apron conveyor portion of the scalping conveyor, which is a device that takes RDF off a walking floor and transports it to the transfer conveyor. It could not be field assembled without torch-cutting the pieces apart, partially assembling the sections and re-welding the equipment. Though the procedure of assembling the apron conveyor appeared to be a four-day job, unsuccessful attempts consumed four man-weeks. START-UP Construction problems delayed start-up by approximately three months. Boiler hydrostatic testing was complicated by the combination of old, existing tubing and the new tubing. The boiler was brought to hydrostatic test pressure (1.5 x operating pressure). On boiler No. 2, only a few handholes and small valves were found to be leaking. However, on boiler No. 1, during a pre-hydro test, an old tube in the generating bank failed. Cause was attributed to localized tube erosion due to sootblowers Chemical cleaning and steam blows’ were successfully completed in one week for each unit. All existing plant systems were checked out prior to first fire of RDF. The turbine lube oil was found to be excessively dirty at both plants. At Red Wing the cause was attributed to the location of the reserve oil tank beneath the operating floor which allowed floor sweepings to enter the oil] tank. A filtration system was installed and the oi] cleaned to within guidelines. At Wilmarth, filtration failed to clean the oil. Further investigation revealed that the old oil piping was scaled internally, and had to be totally replaced. THE OVERFIRE AIR SYSTEM The overfire air system, as designed, supplied both the windswept spouts and the overfire nozzles. The windswept spouts required separate flow control to permit optimum distribution, and so the necessary dampers and damper drive will be retrofitted to permit this to occur. The screw feeders are encountering a larger RDF particle size than that for which they were designed. Initial RDF supply, running at 10-inch minus, was oversized as the feeders were designed for a 6-inch maximum. At higher loads, the screw would stop feeding and "plugging" occurred. RDF sizing is being adjusted at the preparation plant, and this problem is expected to be resolved when the feed size meets specification. OPERATION The operation of the RDF handling system with the walking floor and metering bins was a new experience for NSP. Consequently, checking out this system prior to the first RDF fire was prudent. A trailer of RDF was unloaded onto the walking floor, conveyed to the scalping conveyor, onto the transfer conveyor and into the plant. The fuel was fed into the metering bins and diverted from the pant legs (before the air swept spouts) onto a temporary conveyor which went outside to an open top trailer. In spite of some internal pessimism, this jury-rigged check-out system worked and permitted system checks to proceed without the need of actually firing the RDF. A new era for an old plant (Red Wing) was begun with the first fire of RDF in May 1987. Operations and start-up engineers were stationed at critical points along the RDF feed route. All systems were operated in a manual mode initially, which necessitated many extra people on shift. One operator directed truck traffic, trained truck drivers to hook up hydraulic lines and monitored the walking floor and scalping conveyor. The operator assigned to monitor the stationary magnet at the head of the transfer conveyor quickly discovered a flaw in the design. The RDF specification stated that the RDF would include no more than one percent metal. One percent of 700 tons per day is a large quantity. The RDF system had to be shut down every 15 minutes to manually clean the magnet. A self-cleaning magnet has since been installed and Red Wing recycled ferrous material now fills a 20 yard cubic dumpster in 1-1/2 days. Operators were also assigned to the distribution conveyor at the top of the metering bins. It was soon discovered that the bin level indicators were not reading accurately. Two causes were identified; (1) the distribution of RDF in the metering bin, and (2) the location of the electronics package in a high ambient temperature zone. The 90% angle of repose of the RDF was so steep that the initial placement of the bin level indicators could not read the actual bin level. The indicators and electronics have been moved, and an additional high level trip was added. A diverter plate was added to the bins to help direct the RDF flow to the opposite side of the bin. The feed to the bins was automated to keep the bins at 90% full. Another area discovered to require frequent monitoring by operators was the ash bed on the grates. Maintaining a bed of 6-8 inches insulates the grates from the high temperatures. Plugging of the holes in the grates has been noticed, and after three months of operation, 15% of the holes were found to be plugged. At low loads 3 to 4-1/2 MWS, natural gas is utilized to aid in keeping the furnace hot. Operation at the lower loads has indicated that the OFA and undergrate air must be adjusted to avoid chilling the bed. We are continuing to discover new ways to optimize boiler operation. Initially, 13 people/shift were required, at this writing 5 people/shift are sufficient with an intent to further improve efficiencies. This section has provided a candid review of the initial problems encountered and the solutions applied. The important fact is that RDF is being successfully converted to energy at design rates. EXPECTED PERFORMANCE VS ACTUAL PERFORMANCE Table 1 compares the performance of the original coal design with that of the predicted RDF design and with the actual preliminary RDF performance. SUMMARY The repowering of the Red Wing and Wilmarth power plants to fire refuse-derived fuel demonstrates how NSP, working closely with county agencies and selected vendors, was able to alleviate an increasing problem of solid waste disposal. The conversion project is on time, within budget, and presently has all four steam generators on Jline’ firing refuse. The extensive modifications to the steam generators will permit NSP to use the RDF it processes from the Ramsey and Washington Counties' trash Table 1 Predicted Actual Coal RDF RDF Design Design Perf. Performance Superheater Steam Outlet Temp. 825°F 700°F 720°F Superheater Outlet Steam Flow 125,000 Ib/hr 105,900 Ib/hr 109,000 Ib/hr Superheater Outlet Steam Press. 625 psig 625 psig 604 psig Economizer Inlet Feedwater Temp. 330°F 330°F 340°F Economizer Outlet Temp. 359°F 378°F Furnace Exit Gas Temp. 1820°F 1567°F 1480°F Boiler Outlet Gas Temp. 729°F 660°F Economizer Exit Gas Temp. 417°F 648°F 590°F Air Heater Exit Gas Temp. Not Applicable 450°F 420°F Undergrate Air Temp. 353°F 315°F Boiler Efficiency 80.15% 69% 70% Effective Furnace Volume 8,250 cu.ft. 12,086 cu. ft. Furnace Liberation Rate 23,600 Btu/hr/ft? 13,652 Btu/hr/ft? Grate Heat Release 638,954 Btu/hr/ft? 539,216 Btu/hr/ft2 Excess Air 30% 50% 65% Economizer Gas Flow 229,000 Ib/hr 234,400 Ib/hr Combustion Air Flow 188,000 Ib/hr 187,600 Ib/hr2 MW 9.9 Overfire Air Static 23” H20 Undergrate Delta P 1.3" H20 Data is from the Red Wing Unit 2. Wilmarth data is comparable. Overfire air/undergrate ratios of 51/49 typical to generate power for years to come. The steam generator modifications have resulted’ in applying conservatively designed upgrades to insure the high reliability necessary for NSP to meet the contracted commitments of trash acceptance. REFERENCES 1. Skinner, M.F., and Bloomer, T.M., "Red Wing and Wilmarth Steam Plants RDF Conversion." Presented to Joint Power Generation Conference, Portland, Or., 1986. 2. Kreidler, L. and Gibbs, D., "From Hamilton to Palm Beach: The Evolution of Dedicated RDF Plants." Presented to Processed Fuels & Materials Recovery from Municipal Solid Waste Symposium; sponsored by Resource Recovery Report and Gershman, Brickner & Bratton, Inc., Washington D.C., 1986. Available from B&W as BR-1294. | Technical Paper BR-1333 Waste wood combustion in circulating fluidized bed boilers F. Belin D. E. James D. J. Walker R. J. Warrick Design Engineering - AFBB Development Babcock & Wilcox Barberton, Ohio U.S.A. Presented to Second International Conference on Circulating Fluidized Beds March 14-18, 1988 Compiegne, France Babcock & Wilcox a McDermott company Waste wood combustion in circulating fluidized bed boilers F. Belin D. E. James D. J. Walker R. J. Warrick Design Engineering - AFBB Development Babcock & Wilcox Company Barberton, Ohio, U.S.A. Presented to PGTP-87-19 Second International Conference on Circulating Fluidized Beds March 14-18, 1988 Compiegne, France Abstract Babcock & Wilcox circulating fluidized bed (CFB) boilers with steam capacity 27.6 kg/s/219,000 lb/hr and 20.7 kg/s/164,000 lb/hr utilize waste-wood fuel to produce 86.2 bar/1250 psig, 513°C/955°F steam for electric power generation. The boiler design and fuel/bed material data are provided. Problems encountered during the initial boiler operation and the boiler start-up regime are described. The influence of fuel properties (moisture content, size consist and reactivity) on the combustion process is analyzed. Test data provided on the furnace temperature and gas concentration patterns as a function of fuel properties, combustion air distribution and solids circulation. The vertical fuel heat release profile is determined from experimental data using a zonal furnace performance model. Carbon conversion and CO/NOx emissions data are presented. Major factors involved in design of waste wood- fired CFB furnaces are discussed. Introduction The Babcock & Wilcox Company (U.S.A.) has designed and built three circulating fluidized bed boilers fired with waste wood. The circulating bed technology utilized is licensed by B&W from Studsvik Energiteknik AB of Sweden. Two of these boilers are owned by Babcock-Ultrapower and supply steam for 25 MWe net-output electrical generating stations located at West Enfield and Jonesboro, Maine (U.S.A.). A description of this project is provided elsewhere “). A third boiler, of slightly different design, supplies steam for the 16 MWe Feather River power plant located near Marysville, California (U.S.A.) and owned by Energy Factors. The B&W/Studsvik circulating fluidized bed boiler was selected to fire waste wood at these sites due to its ability to burn a wide range of fuels while meeting stringent air quality require- ments. A description of the boilers, associated systems, the fuels fired, important parameters involved in combustion of waste wood in a circulating fluidized bed and emissions levels attained are discussed herein. Boiler Description For each plant, the boiler scope consists of the CFB steam generator, combustion air and flue gas systems, including forced draft and induced draft fans, fuel feed equipment and solids han- dling equipment. The boilers are natural circulation single drum units. Performance of the boilers as designed and as determined from tests is shown in Table I. Sectional side views of the boilers and auxiliary equipment are shown in Figure Nos. 1 and 2. The following description of the Babcock-Ultrapower units applies to all three units, except as noted. Table | Babcock & Wilcox CFB Boiler Performance Data Babcock-Ultrapower Energy Factors Unit Design Test Design Test Electric Load (Gross) MW 27.5 28.3 19.5 19.6 Max Steam Flow (MCR) kg/s 27.6 26.4 20.7 21.5 1000Ib/hr 218.6 209.0 164.0 170.8 Steam Pressure bar 86.2 85.9 87.5 87.2 psig 1250 1245 1270 1265 Steam Temperature °C 513 511 513 509 °F 955 951 955 949 Feedwater Temperature °c 147 151 186 196 °F 296 303 367 385 Gas/Air Temperatures Furnace Exit Gas °c 857 873 849 823 °F 1575 1603 1560 1514 Flue Gas Leaving °c 135 128 150 152 Air Heater °F 275 263 302 305 Air Leaving Air °C 209 203 191 189 Heater °F 408 398 375 372 Thermal Efficiency (HHV Basis) % 78.8 79.81 81.3 81.28 Fuel Moisture % 40.0 38.0 30.0 46.4 Unburned Carbon Loss % 1.2 01 1.2 0.09 Excess Air % 16 24 21 19 Primary/Overfire Air Split % 50/50 50/50 60/40 25/75 Emissions at MCR Limits: Limits: NOx Ib/10° Btu 0.158 .155 175 0.110 co Ib/10® Btu 0.158 .025 .218 0.100 Furnace The furnace, which is rectangular in cross-section, has gas tight membraned enclosure walls. Additional heat transfer surface is provided by three membraned division walls which extend partially across the furnace from the rear wall. A thin layer of high conductivity refractory covers the walls in the lower portion of the furnace to protect the metal surface from erosion and corrosion. Fuel is introduced to the furnace at four locations in the lower front wall of the furnace (primary zone). Recirculated bed material and make-up solids also enter the furnace in the primary zone. Both primary and overfire air, supplied by a single forced draft fan, are heated in the tubular air heater. From the air heater the primary air is ducted to the primary air windbox which is divided into four compartments to allow biasing of primary air across the width of the furnace. Primary combustion air is introduced to the furnace through a bubble cap air distributor in the floor of the furnace. To heat the fluidized bed to the temperature required to ignite the main fuel, the primary air is heated by propane-fired in- duct burners located in the separate ducts to each windbox compartment. Overfire air is introduced through multiple nozzles in the front and rear walls of the furnace at two elevations. The distribution of the overfire air is controllable between the front and rear walls and each elevation. Primary Particle Collection and Recirculation System Flue gases and particulate matter leaving the furnace pass through a high temperature primary Flue Gas To ESP, ID Fan & Stack Figure 1 Babcock-Ultrapower CFB Boiler 1-Tertiary Air, 2-Secondary Air (original), 3-Secondary Air (added), 4-Refractory (original), 5-Refractory (added), 6- Metering Screws, 7-Duct Burners, 8-Particle Storage Hopper, 9- Rotary Valve, 10 U-Beams, 11-L-Valves & Standpipes, 12-Fuel Metering Bin, 13-Fuel Injector Screws, 14-Bed Drains, 15- Primary Air Duct, 16-Air Heater, 17-Secondary Superheater, 18-Main Steam Outlet, 19-Primary Superheater Multiclone, 20-Dust Collector, 21-Division Wall, 22-F.D. Fan, 23-Economizer. 21 10) © 7 1 Flue Gas To ESP, ID Fan & Stack os ® @ Figure 2 Energy Factors CFB Boiler 1 through 17 per Figure 1, 18-Primary Superheater, 19-Economizer, 20-Multiclone Dust Collector, 21-Wing Walls, 22-F.D. Fan, 23-Overbed Burners. collector where nearly all of the particulates are separated from the flue gas. A unique feature of B&W CFB boilers is a primary collector made up of a labyrinth type array of beams having a shape similar to the letter “U”. The beams utilized in the B&W CFB boilers are made from a readily available stainless steel alloy. The collector is designed to remove solids from dense gas-solids suspensions. The solids impact on the beams and fall by gravity into a particle storage hopper located directly beneath the array of beams. From the particle storage hopper the material is recirculated to the furnace at a controlled rate through four non-mechanical L-valves. Each L-valve system consists of three components: a standpipe, the L-valve proper and a horizontal leg. The L-valve system is capable of controlling the flow rate of material recirculated to the furnace by means of air injected into the solids in the L-valve through an aeration port which is located near the bottom of the standpipe. The flow rate of solids is proportional to the amount of air injected. With this system, a large flow rate of solids can be maintained with a very small volume of aeration air. The ability to control the rate at which the bed material is recirculated to the furnace is a distinctive feature of the BeW/Studsvik CFB technology. It allows the furnace inventory to be controlled independently from the boiler load and air split by redistributing the total inventory between the furnace and the particle storage hopper. This feature provides flexibility in adjusting combustion and heat transfer condi- tions to accommodate variability of fuel quality, bed material size and boiler load. Boiler Convection Pass and Back-End Equipment The flue gases leaving the primary solids collector flow through a refractory covered cavity provided for residual CO burnout and then into the boiler convection pass which includes the horizontal tube superheater and economizer banks. The gas velocity in these tube banks is designed specifi- cally to minimize tube erosion. The flue gases then enter a multiclone dust collector for addi- tional solids separation and continue on to the tubular air heater, electrostatic precipitator, the induced draft fan and the stack. Fuel Feed System The fuel feed system is designed to accommodate the diverse fuels specified for each plant site. The wood fuel, after passing through a screen, is delivered to a live bottom fuel metering bin installed at the boiler front. The fuel is fed to the boiler by four parallel feed trains, consisting of a triple screw metering feeder, a rotary seal valve and an injection screw feeder. The rotary seal valve is installed in the feed chute to seal against the positive furnace pressure which may reach 10 KPa (40 in. H20) at the feed point location. Solids Handling System The solids handling system includes particle separators, solids recycle system, bed drains and transport systems. The orginally installed system with sand as a make-up material is shown in Figure 3 which is mostly self-explanatory. Dense phase transport systems are used to move recircu- lating (except furnace recirculation) and make-up solids, while drag chain conveyors move solids to ash disposal. The bed drains (one per primary air compartment) discharge to a water-cooled screw. After cooling, the solids are screened to remove over-sized bed material and then transported to the receiving bin. Differences, Between The Babcock-Ultrapower and the Energy Factors Boilers The boiler at the Feather River plant differs from the units in Maine in the following aspects: @ ® Flue Gas e Number of fuel feed trains is three. e Number of L-valve systems is three. e Two natural gas-fired burners are located in one sidewall of the furnace to assist the duct burners raise bed temperature during start-up and to stabilize combustion during fuel upsets. e Wing walls are used in the furnace rather than full height division walls. e The bare tube economizer is located downstream of the multiclone dust collector. e The solids discharged from the multiclone hoppers are cooled by a water-cooled screw to a temperature suitable for handling. Fuel and Bed Material The diversity of fuel characteristics present challenges in storage, handling and combustion to achieve high boiler reliability while meeting strict performance and emissions requirements. Design of the boiler required consideration of the fuel type, sizing, moisture and ash constituents. The range of fuel types and characteristics is shown in Table II. Flue Ga: aoe to Stack to ESP. Ash Disposal Figure 3 Solids Recirculation/Make Up System of Babcock-Ultrapower CFB Boilers 1-Receiving Bin, 2-Holdup Bin, 3-Furnace, 4-U-Beam Separator, 5-L-Valve - Primary Recycle, 6-Water-Cooled Screw, 7-Screen, 8-Particle Storage Hopper, 9-Dense Phase Transport System, 10-Multiclone Dust Collector, 11-Rotary Seal Valve, 12-Electrostatic Precipitator (ESP). Fuel moisture, % by wt. Table Il Fuel Characteristics Babcock- Energy Ultrapower Factors Fuel Type Waste Wood Waste wood, natural gas Form Whole trees, hogged or Dry mill waste & chipped, and sawdust, Saw mill sawdust residue, . Agricultural waste, Urban waste Species West Enfield Jonesboro Cedar Larch Douglas fir Pines Poplar Oaks Spruce Beech Alder Hemlock Soft Maple Incense cedar Hardwoods White Birch Pines Design range 40 to 50 30 to 40 Operating range 35 to 55 20 to 40 Ultimate analysis, % by wt. Ash 3.17 0.21 Ss 0.02 0.01 H 3.37 3.76 Cc 28.78 35.17 H20 45.00 30.00 N 0.10 0.06 0 19.56 30.79 HHV, kJ/kg 11,910 (5120) 14,003 (6020) (Btu/Ib) Redwood (sawdust) Orchard prunings Due to the low sulfur content of the specified fuels, a sulfur capturing sorbent is not required to meet SOx emission limits, so the bed material chosen for these projects was sand. Diversity of sand properties and the relationship between these properties and combustion system perfor- mance necessitated several changes of the make up sand at each site. Start-up and Initial Operation During the initial start-up each of the CFB boilers suffered from problems normally encountered in the start-up of a new plant. The fuel feed system at Feather River required modifications to the fuel metering bin to prevent very fine fuel available at this site from packing against the discharge end of the metering bin which caused eratic fuel flow. The modifications, combined with better bin level control, resulted in improved fuel flow. The Maine units operating with coarser fuel did not experi- ence this problem, however, extremely wet (ice and snow) and fine material did tend to plug the feed chutes beneath the rotary seals. Operation also becomes more difficult if a significant portion of the fuel is larger than two inches. Tests conducted in the laboratory indicated that bed temperatures in the range of 593°C/1100°F to 649°C/1200°F were required to ignite wet wood chips. For this reason the duct burners were designed to provide 760°C/1400°F air to the boiler windbox for start-up. The difference between the duct burner temperature and the bed temperature results from heat losses from air to the water- cooled windbox walls and from the bed to the furnace. If the bed temperature is lower than about 593°C/1100°F and wood with about 40% moisture is used, initial fuel ignition is slow, and care is required to prevent over feeding to avoid excessive CO generation and bed quenching. After ignition the bed temperature increases rapidly. With bed temperatures over 649°C /1200°F, initial light-off and bed temperature control are very smooth. A considerable amount of carbon monoxide is generated during the light-off process and until the bed temperature exceeds about 677°C/1250°F. Above this temperature carbon monoxide produc- tion decreases dramatically. The units have operated at low loads, with high moisture chips and with bed temperatures in the low 700°C/1300°F range with carbon monoxide concentrations less than 100 ppm. In the EFFR unit with above-bed burners, the temperature produced by the duct burners is limited to about 593°C/1100°F. The duct burners expedite the initial bed warming up to around 427°C/800°F. The above-bed burners are used to reach the final temperature required for smooth and safe fuel ignition. These burners also help to sustain stable combustion and prevent high CO formation during the start-up operation. They are also being used for boiler operation in a stand-by mode, when problems are experienced with other plant equipment. The cold start-up regime for a CFB boiler is shown in Figure 4. An important part of the cold boiler start-up is warming up the solids inventory. Before start-up, bed material is loaded into the boiler furnace and particle storage hopper. During the bed warming and low load conditions the bed behaves as a bubbling bed with high solids elutriation. The bed inventory is replenished by intermittent operation of the L-valves. As load is increased to about 20% MCR the solids circulation becomes continuous. The whole solids inventory gradually warms up and approaches the furnace exit temperature. The duration of the cold start-up is determined by the allowed rate of the drum pressure increase and the time required for warming up the particle storage hopper inventory. The load increase, after bed inventory warming up, may be done at a rate of about 3% per minute. After the initial start-up problems were solved and periods of sustained high loads were achieved, operating difficulties with solids circulation were encountered at Babcock- 100 Temperature. °F Drum Press. : Normal Time ——> © Average Windbox Tem, © Average Bed Temperat © Drum Pressure © Average Standpipe Temperature 1 Megawatts in Primary Zone Figure 4 Cold Start-up Regime - West Enfield CFB Boiler Ultrapower units. Considerable quantities of sintered sand were discovered in the particle storage hopper. Large pieces of the sintered material made solids flow through L-valve unsteady and caused problems with solids inventory and temperature control. The bed material in the furnace did not agglomerate. A program was initiated to determine the cause and solution of the problem. Tests were carried out in B&W’s Alliance Research Center, and a comprehensive sampling and analysis program was established at the site. As a result of this effort, the sintering problems were traced to the potassium content of the fuel which normally analysed at between 10% and 16% of the fuel ash. The potassium from the ash combined with the sand bed material to form a potassium-calcium silicate which “glued” the sand particles together in stagnant zones. Several measures were taken to control the problem including improvement of the level control in the particle storage hopper, changes in solids recirculation from the multiclone, addition of limestone to the bed material, and tuning of combustion and solids circulation through the furnace. While more work is required to complete- ly understand the sintering problem, boiler operation has improved to the point where sintering is manageable and does not result in forced outages. During the initial stages of operation it became obvious that it was necessary to provide addition- al flexibility in furnace operation for the combus- tion of fuel with variable properties within the wide load range (4:1 turndown). To achieve this flexibility, an additional level of overfire air nozzles was installed in the lower furnace above the existing nozzles and the refractory extended up to the new nozzles. The bed material size was adjusted to provide the circulation rate required. The solids handling system was changed to improve controllability of the solids inventory, solids size and size distribution in the unit and include means for limestone addition. During the 30-day acceptance tests conducted at the Babcock-Ultrapower units, the availability was 100%. Today these units are considered commercial and they are on the local utility system’s load dispatch. Problems are still experi- enced at the Feather River unit due to the extreme changes of fuel quality, but the boilers proved to be capable of burning a phenomenal variety of fuel with moisture content ranging from 10% to 55% and varying combination of sizes from fine sawdust to 2 inch chips. Though a CFB boiler is well-suited for firing fuel with variable properties it is not immune to abrupt changes and extreme deviations of the fuel quality (like wood with large fines content and 60% or 5% moisture). Configura- tion of the control system to accommodate such variations becomes very complex, and the control responses to some changes of operational conditions become inappropriate and difficult to predict. Proper wood yard management improves boiler operation in cases where fuel properties are widely variable by limiting the range and rate of the changes. Wood Combustion Development of the Babcock & Wilcox wood-fired CFB boilers was based on data from small-scale laboratory models and prototypes. The capacity of the commercial units is about 40 times greater than that of the Studsvik 2.5 MWt prototype. Due to the large scale-up factor and lack of long- term operational experience a considerable effort was needed to follow the start-up of the first commercial units, to find optimal combustion/ solids circulation conditions and to collect field data required to verify the functional design of commercial-size CFB boilers. An extensive test program was implemented at the West Enfield CFB boiler with supplemental data collected at other units. Interrelated processes of waste wood combustion, solids circulation and furnace heat transfer were studied together. The test data provided a basis for the boiler performance analy- sis and for investigation of the impact of fuel properties and operational upsets on the boiler performance. The following sections describe the scope of the tests, summarize process factors important for wood firing in a CFB boiler, and illustrate the response of the CFB combustion system to the process variables. Carbon burnout and gaseous emission data are also provided. Test Measurements The West Enfield boiler was equipped with test instrumentation needed for complete evaluation of boiler performance. This included multi-point gas sampling and thermocouple grids at the economizer and air heater outlets. Continuous Og, CO», CO, NOx gas analyzers were used to monitor the excess air and gaseous emissions. Additional test measurements included: 1. Traversing in the furnace and the U-beam region at five planes. Water-cooled probes were used to collect gas samples, measure gas temperature (bare and high velocity thermo- couples), and velocity. 2. Furnace vertical temperature profiles measured at 6 elevations by stationary thermocouples. 3. Furnace vertical pressure distribution measured at 15 elevations using DP transmitters. 4. Wall heat flux measurement at 66 points throughout the furnace by different means (Land fluxtubes, chordal and differential thermocouples). 5. Solids flow rates for the make-up and recirculating streams (16 load cells) 6. Solids sampling points at all streams for chemical analysis and size distribution. The data aquisition and on-line performance analysis system installed for the tests used two dedicated Hewlett-Packard computers and a computer interface unit which tied directly into the plant computer data loop. The data were collected from 328 local test instruments and 170 plant instruments at approximately 90 second intervals. Selected signals such as from the load cells and furnace pressure DP transmitters, were read at 1.5 second intervals. Process Considerations The major requirement to the waste wood combustion system performance is the ability to accommodate variability of fuel properties while providing: a) High combustion efficiency b) Maximum burnout of gaseous combustibles within the furnace c) CO/NOx emissions within the guaranteed limits d) Proper temperature distribution within the furnace While items a, b, and c are self-explanatory, the last item needs to be clarified. Three temperature values should be considered in discussion of the proper furnace temperature distribution: primary zone (TBI), furnace maxi- mum (TMAX), and furnace exit (FEGT). For wood-fired units, where sulfur capture is not an issue, TBI should not exceed the maximum value dictated by mechanical design of the elements penetrating the furnace and/or determined by solids agglomeration in the dense bed. TBI should not decrease below a certain value, dependent on fuel properties, to prevent an increase of CO concentration at the furnace exit. For the West Enfield unit, the limits for TBI values were found to be from 788°C/1450°F to 885°C/1625°F. The vertical temperature distribution in a CFB furnace with high solids circulation rate is uniform compared to conventional furnaces, but usually includes a region of elevated tempera- tures, located above the overfire air nozzles. In wood-fired units limiting of TMAX becomes important when properties of fuel ash and make- up material are conducive to sintering. In this case the temperature in any point of the bed should be kept below a limiting value above which the sintering is aggravated. The FEGT should not exceed the upper limit dictated by the allowed U-beam temperature and superheater heat absorption. Due to the large solids circulation at high loads, FEGT usually does not differ much from TBI. With load reduction FEGT tends to decrease. Reduction of solids circulation rate helps to limit FEGT decline with the load decrease and to extend the superheat control range. Understanding the effect of waste-wood properties on combustion conditions is the key to proper combustion system design and tuning. The most important properties are the moisture content, size distribution and reactivity. The first two are readily measurable and, therefore, can be managed. The reactivity depends on the type of wood and is difficult to manage, especially when the waste-wood is supplied from mulitple sources. The wood reactivity is determined by the kinetically controlled rate of wood pyrolysis which may change an order of magnitude depending on wood type °). Variations of fuel reactivity can explain the well known upsets of the waste-wood combustion conditions (bed temperature, CO burnout) occuring without significant changes in fuel moisture and size. Much work remains to be done to understand and characterize wood reactivity for the prediction of furnace performance. For the tests described herein these variations could not be controlled. Wood combustion consists of four consecutive and partially overlapping phases: dryout, pyrolysis, combustion of pyrolysis products, and combustion of char. For wood combustion in CFB, the first three phases are the most important. Due to comparatively low char output, its high reactivity and considerable solids recirculation, the unburned carbon loss in CFB wood-fired furnaces is low and is not a primary object of combustion optimization. In the discussion below completeness of combustion will refer to the combustion of gaseous pyrolysis products. The residence time of gaseous pyrolysis products in the furnace depends on location of the dryout and pyrolysis zones. The latter is determined by the fuel moisture content, reactivity, the residence time of fuel particles and bed temperature. The fuel particle residence time depends on the particle size, density, and shape, gas velocities and bulk density distribution in the furnace. Wood particles may be divided in two categories: coarse particles - the ones that cannot be entrained and carried out of the primary zone before their mass is reduced by dryout, pyrolysis and char combustion processes; fine particles - the ones immediately entrainable. For coarse wood particles, the extent of pyrolysis in the primary zone depends on the moisture content in fuel. Since the wood char residual is about 25% of the initial dry mass, most of the coarse particles leave the primary zone before their pyrolysis is completed. The pyrolysis continues throughout the furnace, but mostly occurs in its lower part due to the higher fuel particle refluxing (residence time) and higher pyrolysis rate in this region. The burnout of the gaseous combustibles is controlled by the gas-air mixing (for temperatures above 700°C/1300°F) and may be completed within the furnace if a proper mixing is provided. Coarse particles leaving the furnace are caught by the U-beam separator and are recirculated to the primary zone. Their pyrolysis, if not completed in the furnace, may continue in the particle storage hopper and standpipes where the residence time greatly exceeds the furnace residence time. The generated gaseous combustibles are mostly burned above the hopper level in the gas stream by-passing the U-beams. The residual CO burns out in the CO burnout zone. For fine wood particles, the residence time in the primary zone is determined by the gas velocity and the bed inventory in the zone. The finer and dryer the particle is, the greater will be the extent of drying and pyrolysis in the primary zone, and better are conditions for complete fuel combustion in the furnace. For coarser (but readily entrainable) particles, the residence time in the primary zone and the shaft is not substantially different from the very fine particles, but the time required for completion of dryout and pyrolysis is considerably higher. Due to the high surface/volume ratio and low density, the residence time of the fine particles in the shaft may approach the gas residence time. If these particles have high moisture content, their active pyrolysis may be shifted to the middle and upper furnace which would not leave enough time for the burnout of the pyrolysis products in the furnace. Their combustion will continue in the U- beam and CO-burnout cavities. Due to the predominantly once-through nature of wood fines combustion, the wood reactivity has a strong influence affecting the pyrolysis time and, therefore, the length of the combustion zone. Fine char particles passing through the U-beams are carried by flue gases to the multiclone. Most of this char is recirculated to the furnace; the finest particles which represent a very small fraction of the carbon input leave the boiler to be collected in the precipitator. Test Results The results of combustion tuning and testing at the Babcock & Wilcox waste wood-fired CFB boilers illustrate the effect of the controllable process parameters on combustion of waste-wood of vari- able properties. During the tests at the West Enfield boiler there were basically three typical variations of the fuel moisture and fines content: a) Freshly cut and chipped logs - coarse chips with 38-45% moisture content. b) Dried and chipped logs - coarse chips with the moisture content in the low 30% range. c) Reclaim wood - partially decayed fuel with greater fraction of fines and sometimes higher moisture content. The means to control the temperature distribution and fuel burnout in a CFB furnace are the primary/ overfire air split, solids circulation rate, overfire air distribution, and primary air distribution among the compartments. a) Furnace temperature and heat release profiles. When coarse chips with 32-33% moisture content were burned and a coarse sand was used as a bed material, problems were experienced with keeping TBI below the maximum allowable value. As a first attempt, it was resolved by reducing the primary air fraction to 0.25-0.30 and this way suppressing the fuel heat release in the primary zone. As it is shown in Figure 5 (curve 1), this resulted in a sharp © 25/25/50 Air Split; Coarse Sand 94 0 50/15/35 Air Split; Coarse Sand 50/15/35 Air Split; Fine Sand Height From Distributor Plate, Ft 1400 1500 1600 1700 1800 Temperature, °F Figure 5 Furnace Temperature Profile - Effect of Air Split and Solids Circulation 75% Load; 30% Excess Air temperature increase (TMAX = 999°C/1830°F vs. TBI = 843°C/1550°F) above the overfire air ports due to the intensive combustion of the gaseous combustibles generated in the primary zone. This peak temperature was considered unacceptable because of its influence on bed material sintering in the particle storage hopper. When the primary air fraction was increased to 0.5 (Figure 5, Curve 2), TBI increased to 888°C/1630°F while TMAX dropped to 924°C/1696°F. This happened because of the shift of the heat release to the primary zone from the zone above the overfire air ports. The solids circulation rate increased but was not adequate to keep TBI below the desirable upper limit. Further increase of the solids circulation rate was achieved by switching to a finer bed material. This resulted in a decrease of both TBI and TMAX (Figure 5, Curve 3). The temperature distribution data shown in Figure 6 were collected during combustion of coarse chips with about 35% moisture content (Curve 1) and coarse chips mixed with fines with 40% moisture content (Curve 2). © 40% Moisture; Coarse & Fine Chips O 35% Moisture; Coarse Chips 50 w des o oO th oO Height From Distributor Plate, Ft we oO 1400 1500 1600 1700 1800 Temperature, °F Figure 6 Furnace Temperature Profile - Effect of Fuel Quality 75% Load; 40% Excess Air The shift of fuel dryout and pyrolysis from the primary zone to other furnace zones for the fuel with the higher moisture and fines content resulted in a lower TBI value, but caused a greater differ- ence between TMAX and TBI, delayed combustion of pyrolysis products and produced an increase in CO concentration in flue gases. To predict and analyze the furnace performance, B&W has developed a zonal non-isothermal comput- er model for CFB furnaces. The vertical fuel heat release profile in a CFB furnace may be evaluated based on the test heat and material balance, temperature measurement and heat transfer data. The results of the heat release profile calculations for the two test conditions displayed on Figure 6 are shown in Figure 7. They indicate a substantial shift of the fuel burnout from the primary zone to upper furnace zones as described above. The heat release in the primary zone for the dryer, coarser wood was 52% with the zone stoichiometry 0.68, while for the moister and finer wood it was 43% for 0.67 zone © 40% Moisture Coarse & Fine Chips 9 35% Moisture Coarse Chips Furnace Height, Feet 10 20 30 40 50 60 70 80 90 100 Heat Released, % Of Total Figure 7 Predicted Heat Release Pattern - Effect of Fuel Quality 75% Load; 40% Excess Air stoichiometry. For the furnace region above 35 ft. from the distributor plate, the higher moisture fuel produced a noticeable heat release while for the dryer fuel the combustion was practically completed. A mixture of coarse and fine wood particles with high moisture content exemplifies a difficult case for control adjustment due to the conflicting require- ments for the primary/overfire air optimal distribu- tion. To improve fines retention in the primary zone, the primary air fraction could be reduced, which in turn would cause an increase of TMAX similar to conditions described above (Curve 1 on Figure 5). At the Feather River CFB boiler, where the waste wood is predominantly fines, the optimum primary/ overfire air split is different from the one determined for the West Enfield unit where mostly coarse wood chips are burned. Comparison of the typical com- bustion parameters in both furnaces after tuning is shown in Table III. One can see that at the Feather River boiler combustion is not completed within the furnace and continues in the U-beam region caus- ing a temperature increase at the U-beam outlet. Turbulent mixing and temperature level in the U-beam and CO burnout cavities provide for virtually complete combustion, and the final CO concentration in flue gases is low. b) Fuel/air mixing. A non-uniform fuel/air mixing in the furnace may be a source of incomplete combustion within the furnace even when coarse wood chips are burned. There are two fuel-related sources of non- uniformity of the fuel/air distribution. One is wood classification in the metering bin which may produce a difference in the fuel heat input per a feeder. Another source is the volatile matter release and fines entrainment from the fresh fuel feed in the region adjacent to the front wall. The dryer, more reactive and finer the fuel is, the more pronounced is this effect. To correct side-to-side non-uniformity, biasing of primary air distribution among the compartments is used. Front-to-rear non-uniformity may be corrected by biasing of the overfire air between the front and rear wall nozzles. Another source of non-uniform fuel/air mixing is inadequate penetration of the overfire air. The penetration is a strong function of the furnace bulk density. A considerable vertical density gradient in the lower furnace makes the air/gas mixing very sensitive to the nozzle elevation. The proper selec- tion of the overfire air distribution between the nozzle levels, which may be dependent on the boiler load and fuel quality, improves the uniformity of the fuel/air mixing and provides for a maximum fuel burnout within the furnace. During the tests of the West Enfield CFB boiler, furnace traversing was used to measure Og and CO concentration profiles and optimize the overfire air distribution. Table III Combustion Performance of BaW CFB Boilers at Full Load Parameter Unit West Enfield Feather River Wood Size - Predominantly — Predominantly Coarse Fine 100% less than 90% less than 51mm/2 inch 90% more than 13mm/0.5 inch 13mm/0.5 inch Primary Air Flow (Fraction of Total) - 0.45-0.55 0.20-0.35 Primary Zone °C 788-870 788-843 Temperature oF 1450-1600 1450-1550 Average U-Beam °c 816-870 816-843 Inlet Temperature oF 1500-1600 1500-1550 Average Temp. °C ( 8-14) 22-44 Increase (Drop) in oF (15-25) 40-80 U-Beams QO, Content in Flue Gases % 3.5-4.5 3.5-4.5 CO Content in Flue Gases ppm 25-40 20-100 10 A comparison of O2/CO front-to-rear profiles in the furnace plane 45 ft above the bubble caps measured during boiler operation at 100% MCR with two different overfire air splits is shown in Figure 8. With the shift of secondary air from the lower to upper elevation (Figure 8, Curve 2), the CO concentration was reduced considerably in the central furnace region due to the better jet penetra- tion at the upper elevation. An increase of the CO concentration near the front and rear walls was observed after this change. To correct this problem, lower secondary air flow was increased at the expense of primary air. This partially improved the air/gas mixing (Figure 9, Curve 2 vs. Curve 1), but high CO concentrations were still observed near the front wall. Further improvements in combustion uniformity were achieved by biasing overfire air towards the front wall at both nozzle elevations (Figure 9, Curves 3). As a result of these adjust- ments the CO content at the furnace exit was minimized. c) Carbon burnout and gaseous emissions. Overall combustion performance of the Babcock & Wilcox waste wood-fired boiler exceeded the design level. Due to the high solids circulation rate through the furnace, recirculation of the multiclone catch, and the bed drain, carbon in wood char was burned out almost completely. The typical distribution of unburned carbon among different solids streams of the West Enfield boiler was: Bed Drain 1.10% L-Valve 0.06% Multiclone Catch 0.18% Precipitator Catch 0.21% The unburned carbon loss did not exceed 0.1% and was practically all from the precipitator ash. Gaseous emissions were below the design guarantees limits. Concentrations of CO and NOx (corrected for 3% Og) as a function of Og (dry) concentration measured after the economizer at the West Enfield boiler at full load are shown in Figure 10. During operation within the normal operating excess air range (Og = 3.5 to 4.5%) and with the average nitrogen content in the fuel 0.1%, CO and NOx concentrations were 25-40 ppm and 120-135 ppm respectively. For the Feather River boiler during operation at full load with O2 = 3.5 to 4.5% in stack gases, the emission data typically are CO = 20-100 ppm and NOx = 90 ppm. For a comparison, typical CO and NOx 11 Carbon Monoxide, ppm _’ Neo © 50/15/35 Air Split; CO = 54 ppm (Stack) 50/5/45 Air Split; CO = 77 ppm (Stack) Oxygen, % 0 1 ST SI AS Gi 7 8 9 10 11 12 Furnace Depth, Feet Figure 8 Furnace oxygen/Carbon Monoxide Distribution Effect of Air Split 100% Load; 30% Excess Air; Fuel: 38% Moisture, Coarse Chips 5000: 40004 Eg 4 = 30004 \ g 3 20004 1000 o4 mr © 50/5/45 Air Split; CO = 77 ppm (Stack) 4 © 45/10/45 Air Split; CO = 104 ppm (Stack) ‘© 45/10/45 Air Split, Front Bias; CO = 51 ppm (Stack) 7.04 - 6.04 x ¢ $ = 5.0: 6 404 3.04 aL eerie eae 0 1 2 3 4 5 ||| F 8 9) || (90 || 52) || da Furnace Depth, Feet Figure 9 Furnace oxygen/Carbon Monoxide Distribution Effect of Air Split 100% Load; 30% Excess Air; Fuel: 38% Moisture, Coarse Chips 180 * NOx (Corrected For 3% O2) 160 ee * es we FF ee ai Pera e sd ** Sj * “* * * a * x t, eee Pe e120] ae ot g ° 100 8 < 5 3 & 80 z z © 60 8 i ° 40 ° ° isl aI 0° Po 8880008g55°8°0,8° oo 20 ° 3 4 5 Oxygen Concentration, % (Dry Basis) Figure 10 Carbon Monoxide/Nitrogen Oxides Emission - West Enfield CFB Boiler 100% Load; Air Split 45/10/45; Fuel: 37% Moisture, Coarse Chips concentrations in flue gas of conventional stoker furnaces firing a similar quality wood are 200-400 ppm and 200-300 ppm respectively. d) Fuel feed and L-valve flow upsets. An important issue for the CFB boiler operational flexibility and scaling-up is the influence of fuel feed and L-valve flow upsets on CFB furnace performance. Primary zone bed temperature data were collected during furnace operation with one fuel feeder and two L-valves out of service. The bed temperatures during these periods were compared with the corresponding data during “normal” operation. The results are shown in Figures 11 and 12. One can see that the bed temperatures did not change significantly during either the fuel feed or L-valve upsets. This indicates a good mixing of solids in the lower bed despite presence of three division walls at the West Enfield unit. When a fuel feeder is out of service the primary air distribution between the compartments may be adjusted if a local increase in the U-beam temperature is observed. The effect of the L-valve upsets on temperature and gas concentration distribution at the furnace exit is not discernable. Design Considerations The experience of operation and testing of the B&W 12 Compartment B Feed Screw Off Compartments. Bed Temperature, °F x 1073 m A B Cc D 4 8 12 Time, Hours 16 20 Figure 11 Primary Zone Bed Temperatures During a Fuel Feed Upset Weed Temperature BXL-Vaive Temperature Temperature, °F x 10 Temperature, °F x 10 LLL CLL LLL A N N N N N N N N NI VLLLLLLLLLLLLLLLLILA A, VLLLLLLLLLLLLILLLLLLLLLLLL \) N N N N N N N N N N N A 8 c D Compartments All L-Valves On > 8 c Compartments B and C L-Valves Off ° Figure 12 Primary Zone Bed Temperatures During a L-Valve Flow Upset waste wood-fired CFB boilers allows one to define the major factors which should be taken into consideration during design of this boiler type. 1. Fuel Fuel specifications should be carefully studied with attention paid to the size distribution, moisture content, type of wood (reactivity), and ash content/composition. When wood wastes with different quality and type of wood are used, the expected mixture of fuels should be specified. Proper woodyard management is needed to avoid extremes in the fuel quality and abrupt changeovers between the extremes. 2. Fuel Feed The fuel feed system should be designed for uninterrupted fuel supply and uniform fuel distribution among the feeders considering variations in fuel properties. The number of the feed points is dependent on wood size, moisture and reactivity. Closer spacing is required for finer, dryer and more reactive fuel. . Bed Material The bed material should be selected depending on the fuel ash composition and expected temperatures throughout the furnace. The most important material properties include chemical composition, mineralogical make-up, hardness, stability to thermal and mechanical decrepitation/ attrition, particle size distribution, and shape factor. . Furnace Sizing The gas velocity in the furnace is selected within the 5-8 m/s (17-26 ft/s) range normally used in CFB’s. The required gas residence time is determined based on expected fuel properties. The coarser, dryer and more reactive the fuel is, the smaller is the residence time needed for completion of the fuel combustion. Gas velocity in the primary zone is determined by the desirable dense bed inventory with considera- tions given to the boiler turn-down capability. The furnace plan aspect ratio is selected taking into consideration fuel feed spacing, overfire air penetration and, in case of using U-beam particle separators, by the U-beam height and velocity requirements. . Bed Density Distribution Bed material size and solids circulation rate are selected based on desired furnace bulk density distribution and bed temperature requirements within the expected load range. Increase of the maximum solids circulation rate reduces the furnace surface required, increases fuel particle residence time in the furnace, improves the temperature uniformity and broadens the boiler turndown ratio. The negative effects are: an increase in the required FD fan pressure, larger solids handling system, greater solids loading in the convection pass and larger solids losses to the final dust collector. A controllable solids circulation rate allows to vary the furnace inventory by distributing the total solids inventory between the furnace and the particle storage hopper. This improves furnace adaptability to varying operation conditions. The dense bed inventory is selected to provide intensive heat and mass exchange in the primary zone, prevent coarse fuel from accumu- lating and causing bed temperature excursions, and to effectively remove the tramp material through the bed drains. When wood fines are 13 burned, an increase in the dense bed inventory is needed to increase the particle residence time in the primary zone. 6. Combustion Air System The primary/overfire air split is selected depend- ing on fuel properties to provide the desirable fuel combustion and temperature profiles in the furnace. A means should be provided for monitor- ing and controlling the primary air distribution among the furnace compartments (side-to-side). The overfire air should be admitted to the furnace through multiple nozzles installed at two or more elevations on opposite furnace walls. A means should be provided to control air distribu- tion among the different nozzle levels and among nozzles in the different walls. The overfire air split is a function of the fuel properties and bulk density distribution. The overfire air nozzle size and number are determined by the desirable gas/air mixing pattern and the air splits. The nozzle direction and jet penetration should be selected to avoid jet impingement upon the furnace walls. 7. Start-Up Burners A combination of in-duct and overbed start-up burners provides the best conditions for quick and safe CFB boiler start-up. The in-duct burners are used during the cold start up for initial bed warming up. The overbed burners are used for raising the bed temperature to the level required for safe and quick main fuel ignition. The burner capacities are selected based on the desirable start-up time and stand-by operation on auxiliary fuel in case of the loss of the main fuel. The list of design requirements above may look overwhelming, but, as B&W’s design and operating experience has shown, they are all essential for reaching stable and efficient combustion of waste- wood fuels with widely different and variable properties similar to those successfully burned in the B&W CFB boilers. References 1. R. F. Johns, R. E. Wascher, “Design and Construction of A Wood-Fired Circulating Fluidized Bed Boiler”, Proceedings of the Ninth International Conference on Fluidized Bed Combustion, Boston, MA, May 3-7, 1987. 2. Roberts, A. F., “A Review of Kinetics Data for the Pyrolysis of Wood and Related Substances”, Combustion and Flame, 14, 1970, pp. 261-272. Technical Paper BR-1337 Operation of 80 MW Bubbling Bed Heskett Station Montana- Dakota Utilities D. L. Kraft Project Engineer, Design Engineering Babcock & Wilcox Barberton, OH Presented to American Power Conference Chicago, IL April 18-20, 1988 Babcock & Wilcox a McDermott company OPERATION OF 80 MW BUBBLING BED HESKETT STATION MONTANA-DAKOTA UTILITIES DAVID L. KRAFT Project Engineer, Design Fngineering Babcock & Wilcox Rarberton, Ohio INTRODUCTION This paper describes the start-up activities and initial commercial operation of the Montana- Dakota Utilities (MDU) 8&0 MW Bubbling Red Retrofit. The primary objective of this paper is to give a summary of the successes, problems and design solutions that occurred during the initial commercial operation of this retrofit. General design considerations and detailed background of this conversion are not covered in this paper but can be found in previous papers (1-3). For a very brief overview, the following is a chronological listing of activities for the past twelve months: 1. January 87 to Final construction, 2. April 87 to July 87: and initial operation. April 87: Start-up checkouts 3. July 87 to Mid-October: Commercial operation at dispatcher's demand. 4. Mid-October 87 to Mid-January 88: Extended fall outage due to the FD fan failure. 5. Jenuary 88: Return to commercial operation and start of performance testing. motor BACKGROUND This two drum power boiler was originally designed to operate at 650,000 lb/hr (72 MW) firing Beulah Lignite on a travelling grate stoker. Since start-up, the boiler could not operate continuously at full load due to severe slag build-up on the furnace division and enclosure walls. To achieve continuous operation, the boiler was derated and cleaned every Spring and Fall outage. The slagging problem was caused by an excessively high heat release rate or, in other words, the furnace volume was too small and the three in-furnace division walls did not compensate for this smaller volume. The bubbling bed technology was selected as a solution to this problem because of the low 1500°F bed temperature, which is below the ash softening temperature. Therefore, the furnace slagging would be eliminated and the boiler could operate continuously at full load, thereby justifying a retrofit based on increased generating capacity. TABLE I FINAL DESIGN CONDITIONS WITH THE RETROFIT Maximum Steam Flow Installed Station Capacity 700,000 lb/hr. Approximately 80 MW Bed Velocity at Full Load 12 ft/sec. Bed Temperature 1500°F Bed Material Sand Coal Feed Overbed Overall Bed Dimensions 40 ft Wide x 26 ft Deep No. of Independent Beds 8 Compartments (Air Side) including one start-up compartment 2/3 Generating Surface 1/3 Secondary Superheater Surface 54 inches In-Bed Surface Split Bed Depth at Full Load START-UP The initial start-up activities were completed without many problems. During April and May, 1987, the boiler was chemically cleaned and the steam lines were blown. By mid-May the first attempt at coal firing confirmed one of the concerns of firing lignite in a sand bed. The bed drains stopped flowing within hours of initiating a coal fire. For fear of plugging the in-bed tube bank, the boiler was shut down to determine the extent of the plugging. Surprisingly, the bed was very clean except for agglomerate* piles on top of the plugged bed drains. It was encouraging to know agglomerates the size of softballs would migrate to the bed drains, however, too many of these agglomerates were reaching the 8 inch drains at the same time. In addition to revising the start-up sequence to minimize the agglomerate formation, some means of on-line rodding capability was required. Otherwise, if the drains could not be kept open, it was just a matter of time before the sodium content in the bed would build up to the point where severe agglomeration would plug the tube bundle. Several attempts of connecting nozzles at the top of the drain pipe and blowing the piles apart using plant air proved ineffective. Reluctantly, a mechanical rod port was attached to the top of *An agglomerate is a mass of fused sand. The binding agents are sodium based compounds that adhere to the surface of the sand particles. In the presence of high local temperatures and perhaps reducing zones, the coated sand particles bind together to form low density sandstone. Superheater Boiling Bank Al Start-up Compartment YL Lf... fit fo —- (10) Spreader Feeders a Compartment Division Plates Figure 1 Plan View of Bed. the drain pipe and extended down through the windbox to the outside casing. With safety as the primary concern, the rod port was equipped with a ball valve followed by a plant air connection, a packing gland, and finally a rod which could not pass through the packing gland. by the first of June, 1987, the bed drains in the center four compartments were retrofited with the mechanical rods. On June 5, the boiler was on-line with 1/3 of the bed compartments in-service. As in all of the other start-up attempts, the bed drains were plugging, but this time the mechanical rodding would re-establish flow, allowing the firing to continue. However, 24-hour-a-day rodding was not an acceptable operation. In the interest of proceeding with start-up activities, it was decided to continue operating in this mode until the June 17 plant outage or until the bed plugged up. Surprisingly, by the fifth day, the bed drain plugging ceased and the drains flowed freely for the next seven days. The conclusion at this time was that the agglomerates were caused by the start-up sequence and, after more stable conditions were established, the bed would eventually clean up. As previously indicated, the start-up sequence was modified to minimize agglomerate formation. The final sequence which eventually proved successful was basically: 1. Establish a 12 inch to 14 inch fluidized bed in the start-up compartment. 2. Heat this bed to 00°F with the duct burner. 3. Start coal tlow to the start-up compartment, limiting the throw (over bed feed) to the hot active bed as much as possible. 4. Once the coal ignites, bring in other surrounding beds and establish a deep bed as quickly as possible. Subsequent start-ups and prolonged operation demonstrated that this type of agglomeration was limited to start-up, after which the bed drains flowed freely. Also, the safety concerns of the mechanical rod ports were adequately addressed by the previously described design. COMMERCIAL OPERATION After the short June outage, commercial operation was established. Though the boiler operated at dispatcher's demand, the maximum load was limited to 65 MW (80 MW full load) due to mechanical problems. The non-metallic expansion-joint outer-layer splices failed, and the higher static pressure required at full load would cause the inner layers to rupture. Also, the seal between the bed enclosure walls and the existing furnace cracked open in one corner allowing bed material to blow out on the operating floor at high bed levels. Commercial considerations dictated continued operation at this reduced load conditien until the Fall outage when the repairs could be made. Even though the boiler was not operated at full load for several months, the combustion system was adjusted and the bed performance steady state LH Division Center Division RH Division Wall Wall Tubes to Wall Center Division Enclosure Wall Wall Horizontal Tubes > ~~ > ~—+ > 54” a + Tube Supports 6 . eae ™ ye Al Ag Bi Ci Di De 7 1 4 Thermocouples’ { Tubes to Bubble Cap Floor Windbox Outboard Division — Water cooled ~ Water Supply Compartments Walls Water Supply Typ. Figure 2 Section Front View In-Bed Boiling Surface. checked. Basically, this amounted to. air calibrations, adjusting fuel distribution and comparing expected versus actual performance. One of the biggest problems encountered during these checkouts was the difficulty of controlling uniform bed temperatures from front to rear. Note that the existing front-wall-mounted spreader feeders were retained for the retrofit coal feed system (Figure 1). Without a special effort to even the coal distribution from front to rear, the in-bed thermocouples near the front wall would read as much as 170°F higher than the rear wall in-bed thermocouples. By adjusting the feeder speed to maximum, maintaining constant load, and varying the coal size over several days, the bed temperatures evened out. However, after the load was allowed to vary, the bed temperatures were no longer uniform. The ability to throw coal 26 feet was stretching the capability of the feeders, and uniform bed temperatures were very sensitive to the coal size. What was gained by retaining the original feed system was offset by poorer operating flexibility and possibly lower efficiency as compared to an underbed feed system, During the brief operating period when the bed temperatures were uniform throughout the entire bed, the steam flow was approximately 30% less than what was predicted for this set of bed operating conditions. The bed conditions were as follows for that period, which will be discussed in a later section of this paper: 48 inch Bed Depth All Beds In-Service 1500°F Eed Temperatures 52 MW 460,000 lb/hr. Since it was decided to continue to operate under the dispatcher, there was not much that could be done until after the Fall outage when additional instrumentation could be installed and full bed depth could be achieved. Therefore, for most of August, September, and October the boiler was base loaded. To help the customer with higher daytime peaks, the bed temperature feeder trips were raised to 1625°F or approximately 65 MW. By late September, maximum obtainable load gradually dropped off to the point where, just prior to the October outage, the maximum load was 50 MW. Continued operation at the higher bed temperature was probably promoting agglomeration in the tube bundle which decreased the effective heating surface. In summary, the boiler operated from late June to Mid-October 1987 at 94.5% availability and logged over 3,000 hours of on-line operation. FALL OF 1987 OUTAGE In mid-October, the boiler was taken off-line for an acid cleaning, as well as to install additional instrumentation and to repair expansion joints, bed to furnace seals, etc. Unfortunately, the FD fan motor literally blew up during a restart after the acid cleaning, which extended the Fall outage through late January, 1988, The remainder of this section is a description of the major findings from the Fall outage boiler inspection. BOILING TUBE BUNDLE AGGLOMERATION Figure 2 is a sectional front view of the in-bed boiling surface. The water enters the two outboard floor headers and flows through every other floor tube to the two in-board floor headers. From there, the water enters the in-bed tube bundle and exits to the three in-furnace division walls. The design criterion for arranging the in-bed surface is equal heating surface per foot of plan area. Note that the tubes to the two outboard division walls loop back and forth in the middle of the tube bundle and the tubes to the center division wall by-pass the center loops. Therefore, the surface is equal on a plan area basis and the tubes are all the same length for circulation requirements. As expected, agglomerate formation had practically filled the cavities in the outboard compartments, which explains why the boiler capacity was dropping off during September and October. On the other hand, the center compartments were very clean, which led to the conclusion that the original surface arrangement should have been distributed equally from top to bottom. Another indication to support this observation occurred during the commercial run, Keeping bed temperatures constant, the center compartment coal flows were 40 to 50% higher than the outboard compartments at any given steady state condition. This suggested that the cavities should be filled with heating surface, the subject of which will be discussed in a later section of this paper. In general, the remainder of the boiler tube bundle was very clean (exceptions to this are noted in the next section). The bare tubes had no ash deposits, and there were no signs of agglomeration caused by poor fuel/air mixing. EROSION Erosion protection was a key design consideration for this bubbling bed operating at 12 feet/second velocity with a sand bed. Based on B&W's TVA 20 MW pilot experience, the expected areas of high erosion were tube bends and the bottom row of tubes facing the bubble caps. Some thought was given to protecting the top row of tubes facing the furnace, but the decision was made to simply monitor the erosion in this area. To protect the tube bends, pin studs and refractory were applied to the outside bend radius and to any inside bend radius that faced the bubble caps. These small patches of refractory did not fall off nor show any signs of spalling. However, a thick, hard ash deposit formed on all of the refractory surfaces. The bottom row of tubes were protected with 120° cast shields, which faced the bubble caps. Once again, a thick, hard ash layer coated all of these shields. The erosion protection was more than adequately addressed, but these deposits adversely affected heat transfer, which will be addressed in a later section of this paper. The rest of the tube bundle was visually inspected for locally high erosion areas and none were found. CIRCULATION Waterside circulation was another key design concern due to the horizontal circuitry of the floor, enclosure sidewalls and in-bed boiling tubes. The 3 inch 0.D. tubes on 8 inch centers floor construction showed no signs of warpage. The cooler underbed combustion air on _ the backside and the dead sand layer on the gas side adequately protected the tubes and wide membrane from overheating. The horizontal enclosure walls were refractory covered to lower the heat absorption and to provide erosion protection. Except for a few minor cracks, the refractory was intact. To ensure proper cooling of the in-bed boiling tubes, multilead ribbed tubes were used to prevent steam separation. There were no signs of warpage or overheating, indicating the total tube circumference was running at the _ same temperature. From these observations, it was concluded that the circulation pumps were sized properly for the heat absorption. TUBE SUPPORTS The B&W designed in-bed tube supports at the TVA 20 MW pilot plant were a constant maintenance item. The TVA design used welds to connect all of the scallop plates and collecting bars, which were a constant source of failure. There were several theories on the weld failure mechanism, but regardless of why they failed, the MDU design criteria was to use bolted connections. This resulted in a very massive arrangement using ? inch x 4 inch stainless steel vertical bars, bolted to collector bars top and bottom, and attached to the floor through 2-1/2 inch bolts in a pin and clevis arrangement. There were no bolt failures, and the pin and clevis allowed supports to move with the bundle without bending the tubes in the vertical bars. Also, there were no visible signs of tube thinning where the tubes pass through the support holes. This design met all of the requirements of a low maintenance support system, although it was expensive. SUPERHEATER The arrangement is basically the same as the boiling tube bundle (Figure 2) except the middle loops extend all the way to the sidewalls which eliminate the cavities. The tube bundle occupies the rear 1/3 of the bed (Figure 1) with tubes running from sidewall to sidewall, and the inlet and outlet headers are in the windbox below the floor. The full compliment of tubes maintains the equal heating surface per square foot of plan area. The major finding in the superheater was a very hard ash coating over the entire superheater. The deposit was uniform around the tube circumference and varied in thickness from 1 inch in the front middle of the bundle to 1/8 inch in the rear corners. The varying thickness was due to fuel distribution and higher fuel input to the center compartments. Inspection of the deposit showed layers of different colors through the cross-section indicating the mechanism for ash deposition was variable and not a steady state occurrence. Also, the 740°F to 955°F superheater tubes had a consistent ash deposit from inlet to outlet, as contrasted with the 5&5°F boiling tubes, which were very clean, indicating that the deposits start to form on surfaces between saturation and 740°F. The erosion shields and refractory coatings also supported this observation. The magnitude and the exact mechanics of the deposition are both unknown at this time. The bed temperature control point is set back to 1500°F, and presently the superheater is being monitored over time to determine the longer term effects. FURNACE SLAGGING As previously mentioned in the Background section, the justification for this retrofit was based on increased generating capacity. To obtain this additional capacity, severe furnace slagging had to be eliminated. Though there were no large slag formations on the furnace enclosure and the division walls, these tubes were covered with a thin white ash deposit. The results of operating at 1625°F bed temperature for several months in relation to this thin ash deposit are unknown at this time. However, by lowering the bed temperature control point, the furnace conditions can be monitored for improvements in the ash deposit. RETURN TO COMMERCIAL OPERATION On January 21, 1988, the boiler went back on-line, and by the middle of February the performance tests were complete. The following is a discussion of those test results. The MCR design conditions were established for the performance test as follows: All beds in service 54 inch bed depth 1500°F bed temperature Fstablish uniform bed temperature Under bed air set at 20% excess or less These conditions should have produced 700,000 lb/hr, but instead only produced 470,000 lb/hr. To explain the difference in the actual versus predicted performance, the following summarizes the testing results: 1) Thermocouples (TC) were installed et three elevations in Al and Bl compartments (Figure 2) to compare the temperature distribution from top to bottom. The bottom TC's both read 1500°F, but’ the middJe TC in Al read 125° to 150°F higher than the middle TC in Bl (Bl middle TC read 1500°F). The combination of the surface arrangement in Al and poorer mixing in that compartment proved inadequate to maintain a uniform temperature from top to bottom. Also, the agglomerate formation in the cavity was directly related to these elevated temperatures. 2) During the initial commercial run, the boiler was set up to operate at a uniform temperature throughout the entire bed (same condition as previously discussed in the Commercial Operation section). During those combustion checks, the bed depth was 48 inches. This test set-up was nearly identical except for a 54 inch bed depth, and the maximum steam flow output increased only 10,000 lb/hr. The bed expansion at 12 ft/sec. resulted in a lower than predicted bed density at a level of 54 inches above the bubble caps. This resulted in an actual lower heat transfer coefficient than the design value used for setting the surface, 3) As previously discussed, the erosion sheilds on the bottom row of in-bed tubes were coated with a thick, hard ash deposit. Due to this deposit, the heat transfer was significantly reduced. Since an effectiveness factor was not used during the original design, approximately 15% of the in-bed surface was under-absorbing. To summarize, the boiling bank is under-surfaced. Currently, B&W is designing a surface addition which fills in the cavities. By doing so, the bank should be slightly over-surfaced and the agglomeration in the outboard compartments should no longer be a problem. CONCLUSION Although the findings in this paper are based on a relatively brief operating period, no major problems have been experienced with accelerated local erosion, uncontrolled tube bank plugging and short-term overheat failures, which would have occurred during 3,000 hours of operation. The past year's operation presents no barriers to the technology and should be viewed as a success. There are long-term issues such as availability, maintenance frequency and type, and long-term erosion, but these will be left to future analysis. REFERENCES 1. Imsdahl, B., Gorrell, R.L., Johnson, H.L., "Mont ana-Dakota Utilities 8C-MW AFBC Retrofit", Presented at Jt. ASME/IEEE Power Generation Conference, Portland, OR., October 1986. 2. Gorrell, R.L., Strong, D.N., "Description of 80-MW Fluidized Bed Retrofit at Montana-Dakota Utilities Co.", Presented at FPRI Seminar on Atmospheric Fluidized Bed Technology’ for Utility Application, Palo Alto, Ca., April 1986. 3. Johnson, H., Imsdahl, R., Spada, R., "Montana-Dakota 80-MW Retrofit - An Integrated Approach to Design and Erection". Technical Paper BR-1338A Chemistry guidelines for cycling service of Babcock & Wilcox - fossil power plants A. Banweg Fossil Operations Division N. J. Mravich Research & Development Division F. J. Pocock Consultant Babcock & Wilcox Barberton, Ohio Presented to EPRI Conference on Cycle Chemistry in Fossil Plants Seattle, WA August 30-September 1, 1988 Babcock & Wilcox a McDermott company CHEMISTRY GUIDELINES FOR CYCLING SERVICE OF BAEBCOCK & WILCCX - FOSSIL POWER PLANTS A. Banweg - Fossil Operations Division, Babcock & Wilcox N. J. Mravich - Research & Development Division, Babcock & Wilcox F. J. Pocock - Consultant, Babcock & Wilcox INTRODUCTION With the increased base-load nuclear steam generating capacity and base-luading of only the newest, most efficient fossil generating equipment, many older fossil units are being placed into cycling service. These new operating conditions have required study and expansion of B&W water chemistry control guidelines. The study also produced recommendations for improvements in the reduction of corrosion product transport through downtime waterside surface storage protection and in start-up procedures. Control of corrosion in cycling boilers is @ problem that has plagued the utility industry for many years. The magnitude of the problem was not considered severe because formerly, most cycling units were the oldest ones in the system. They operated et relatively low pressure and temperature and were usually small. Now, however, many larger units have been placed in cyclic service. Cycling boiler cperating problems of these larger units, are aggravated by higher pressures, higher temperatures, complex tube circuitry, and extensive pre-boiler systens. Utilities suon found that boilers that operated essentially trouble-free as base loaded units were experiencing frequent problems and low availability. One major problem with these cycling units is tube failure from corrosion and deposition. BOILER WATER CREMISTRY CONTROL Maintenance of the protective oxide film of magnetite on the tube metal is critical to the protection of the steam generator tubes from waterside corrosion. The purpose of boiler feedwater conditioning and interna’ boiler weter treatment is to maintain an environment conducive to the formation and repair G? this protective oxide film (1). Babcock & Wilcox's recommendations for steady state full-load operation for once-through and high pressure recirculating utility steam generators are shown in Tables 1 and 2. For most boiler pressure part materials in contact with water, a minimum corrosion rate is exhibited when the pH of the water is between 9.0 and 11.0(2). In addition to total dissolved solids (TDS), which are typically dictated by steam purity considerations of the turbine, and pH control, the dissolved gas (oxygen and carbon dioxide) content of the feedwater must be controlled to very low levels. Dissolved oxygen can be very corrosive to boiler pressure parts and is usually specified to be jess than 7 parts per billion (ppb) at the economizer inlet. Actually, during steady state operation, this value is normally attained at the ceaerator outlet, and chemical oxygen scavenging reduces it to near zero at the economizer inlet. In North America deaerated all volatile treatment (AVT) chemistry using ammonia and hydrazine is used for once through type boilers exclusively. In contrast, a significant number of once-through boiler operators in Eurepe (West Germany) and the Soviet Union have adopted an oxidizing (oxygen addition) chemistry using either a neutral or alkaline pH (3). This chemistry has been reported to exhibit very low corrosion product transport characteristics, as confirmed by the reported extended interval between chemical cleanings. The success of this chemistry also depends on its compatibility with the materials of construction of the boiler/feedwater system which vary significantly from US design practice. For high pressure recirculating boilers in North America al] volatile treatment and various forms of pH-phosphate chemistry predominate. In Europe boiler water treatment with sodium hydroxide is very popular. The success of any water treatment chemistry requires compatibility with the materials of construction of the system, the consistency of its epplication, and operator understanding of allowable variations during transients. Provided that the basic boiler water parameters, alkaline pH, TDS, aria deaeration, are met, the waterside corrosion rate of pressure parts should be negligible. Therefore, no waterside corrosion allowance is either warranted or included in the design of fossil units. WATERSIDE DEPOSITS Operational waterside corrosion failures in utility boilers are not due to general or uniform-corrosion, but rather are cue to localized corrosion phenomena. These localized phenomena are a result of a concentrating mechanism that elevates the normal trace boiler water contaminants to levels that are corrosive to the boiler materials. The most common condition that causes waterside-initiated boiler tube failure is the accumulation of porous deposits on waterside heat transfer surfaces (Figure 1). Deposits can act as an insulating barrier to heat transfer, causing a tube metai temperature elevation, and ultimately an overheat failure (4). The deposit can also act as the concentrating mechanism that elevates the originally low-level boiler-water solids to corrosive levels at the deposit-tube interface (Figure 2). The nature of the boiler water dissolved solids dictate whether the under-deposit corrosion mechanism is of the strong-alkaii type, leading to caustic-gouging, or the hydrogen-damage type, associated with under-deposit acidic conditions (5). EPRI has published a report, "Manual for Investication and Correction cf Boiler Tube Failures (CS-3745)", which addresses the origin, character and corrective action of these two waterside tube failure mechanisms. The report also describes the 20 other classic failure modes experienced by fossil? utility boilers, and their associated root causes. Control of waterside tube failure must also include the control of deposit accumulation on heat transfer surfaces. This control is economically accomplished through an appropriate maintenance chemical cleaning schedule. Appropriate is the key word. Many guidelines for chemical cleaning exist, from the boiler manufacturer, consultants, architect-engineers, or EPRI (6). However, they are just that, guidelines. Many parameters specific to individual plants must be considered so that each plant must determine its own unique (appropriate) chemical cleaning frequency. Periodic tube sampling and waterside deposit analysis is a common method of determining this frequency. Further, an appropriate cleaning process is necessary and should be suited to both the compusition and extent of deposition, and to the materials of construction and configuraticn of the boiler component. CYCLING OPERATICN Start-Up The frequent start-ups associated with a cycling unit, whether joad cycling er on-off cycling, make it inherently more susceptible to out-of-service corrosion damage, air in leakage and pre-boiler and builer corrosion problems. The corrosion products generated in an unprotected pre-boiler system, typically oxides of iron and copper, are transported inte the boiler during each start-up. For some time after a start-up, the level of corrosion products, which normally exist as suspended solids, appear in the part-per-million (ppm) range at the economizer inlet, instead of at the desired steady-state level in the part-per-billion (ppb) range. During the transients of a start-up or a load change, the control of dissolved oxygen concentration, pH, and other parameters is difficult. This is due to lengthy sample lines, sensor lag, and treatment chemical injection system control lag. Oxygen or air in-leakage is greater for a cycling or tow load plant since condensers and deaerators are less efficient at low load, while the potential for air in-leakage at pump seals is greater. The current B&W recommendations for transient chemistry control for once-through and recirculating steam generators are shown in Tables 3 and 4, respectively. These, of course, will be adjusted to reflect the results of the EPRI Fossil Cycle Water Chemistry Monitoring Project (RP-2712-3), which is presently being conducted by Babcock & Wilcox and Sargent & Lundy. The guidelines for once through boilers were developed during the introduction of once-through boilers in the mid 1950's. The interim guidelines for recirculating boilers are the result of applying a sliding pressure chemistry philosophy to high pressure boilers operating at reduced pressure. Chemical injection parameters, such as the volatile pH additives and dissolved oxygen control chemicals, must be examined to preclude overfeed. These injection systems may have been designed for the constant injection rate associated with base-!oaded operation. Overfeed could lead to excessive concentrations which could be detrimental to copper-based alloys in the condensate-feedwater system. Every effort should be made to place on-line instrumentation in service as soon as possible during a start-up; specifically instrumentation to measure feedwater purity, such as cation conductivity, and boiler water pH, or specific conductivity, and feedwater dissolved oxygen concentration. Feedwater corrosion product concentration is a measurement of interest that is not readily accomplished with on-line instrumentation but can be performed on grab samples. One of the difficulties of quantifying corrosion product transport is obtaining a truly representative sample. Sampling nozzles, isokinetic sampling rates, and short sampling lines are required to obtain a representative sample. Membrane filter analyses, (Fig. 3) are easily accomplished during start-up, to monitor corrosion product transport. The color of the stain qualitatively reveals the oxidation state of the circulating fluid. A critical design review should be made of existing water sampling equipment to insure that all samples are as representative of cycle fluid as possible (7). This becomes more critical when less time is available to verify and cross check measurements. Sampling and analysis frequency must be adjusted to be appropriate for the load pattern cf the unit. One of the nost difficult parameters to control, during cycling operation, is the boiler water, pH-phosphate relationship fur drum-type boilers using congruent phosphate chemistry. This is a common boiler water treatment chemistry used by North American utilities. The detectable concentration of phosphate in the bulk boiler water may decrease with increasing load and pressure, and then return to the original level on subsequent load or pressure reductions. This phenomenon is known as phosphate hideout. Phosphate hideout can make it difficult to maintain close control of the desired pH and phosphate concentration. Both the magnitude of the phosphate hideout and the variation in load required to cause the phenomenon can vary considerably from one boiler to another (8). For a unit in cycling service, phosphate chemical injection practice must be moderated to account for the chemical hideout behavior of the individual unit for its load cycling pattern (9). Deposit Control Options The increased inventory of corrosion products, known to be the concentrating mechanism responsible for the under-deposit-corrosion tube failure, makes the cycling unit more susceptible to such waterside initiated failures. If uncontrolled, this situation may mandate a more frequent chemical cleaning cycle to control deposit accumulation. Recognition cf this increased potential for under-deposit corrosion failures often justifies retrofit of a pre-builer recycle line and clean-up system, such as a filter or condensate polisher, to pre-condition the feedwater prior to unit start-up (Figure 4). An evaluation of cycle design and corrosion product transport is needed in order to properly identify the optimum location of any proposed filter. Additionally, depending upon the composition of the metal oxides, either a media-type filter or an electromagnetic filter may be identified as appropriate. The known corrosiveness to pressure part materials of water containing an excessive concentraticn of dissolved oxygen makes it desirable to deaerate the water as soon as possible to minimize damage and subsequent corrosion product generation. If auxiliary steam capacity is available, the water could also be dearated in this recycle loop. However, insufficient steam capacity to the deaerator may actually steam-blanket the water and maintain its aerated condition, or worse, aspirate more air into the system. Proper design and cperation is critical to successful deaeration. Thermal shock and excessive rates of temperature change are believed to play a significant role in some of the cracking-type tube failures affecting cycling units. The action of thermally deaerating the water prior to start-up should also mitigate any thermal stress damage to downstream components on rapid start-ups. Hydrazine Substitutes Recently, a number of chemical-oxygen scavenger substitutes for hydrazine have become available. Some of these are reported to be more effective than hydrazine at lower temperatures. In theory then, they should be more effective in situations such as wet lay-up at ambient temperatures, and, operationally, in the low pressure section of feedwater train. To date, the results of increased use of hydrazine substitutes is inconclusive with respect to removing higher levels of oxygen in cycling boilers. The information available indicates that one or two substitutes could perform better than hydrazine, but more study is needed in this area. In general, all of these substitutes are more compiex organic molecules than hydrazine. Therefore, their decomposition products will, at a minimum, contribute carbon dioxide and, in many cases, multiple short-chain organic compounds, to the system. In a high-purity water system, these constituents will affect the pH and conductivity measurements taken in the system compared to the nornal ammonia hydrazine system. This effect must be recognized. This is also an area requiring further research. OUT-OF-SERVICE CORROSION The importance of controlling dissolved oxygen concentration during operation has been noted previously, but the need to control dissolved oxygen in the out-of- service boiler condition is also crucial and cannot be over emphasized. Out-of-service internal corrosion damage usually is caused by dissolved oxygen pitting, and is a very common problem in reheaters (Figure 5). When a boiler is taken out of service, and as it cools, condensate can form and accumulate in the low areas such as in the bottom bends of pendent tubing. Corrosion camage, due to dissolved oxygen attack, can occur on any wetted internal tube surfaces. This can jead not only to a tube failure, but also to the generation of corrosion product oxides. Though this problem has been most often associated with reheaters, ai] the ferritic pressure part materials of construction exposed tu oxygenated condensate are equally susceptible to this mode of corrosion attack. The reheater tubing generally has the thinnest wall and, therefore, is the first to be perforated by the resultant pitting attack. Since reheaters are seldom hydrostatically tested, leaks from out-of-service corrosion start small and are often not detected. However, they can cause many outages and severe availability losses because the small leaks can cause additional leaks in adjacent tubing. Dry storage under nitrogen, or wet storage with chemically treated demineralized water, are means to prevent this type of corrosion. The appropriate storage procedure recommendations must be made on an individual unit basis. These recommendations must consider: the anticipated duration of the storage period, the physical arrangements of the circuitry, the fuel fired, and potential temperature variations during the storage period. However, plant configurations and unit start-up schedule requirements often interfere with the implementation of these techniques on the reheaters in many plants {10). Even drairable reheater surfaces have been susceptible to dissolved oxygen pitting corrosion damage, since, with time, these horizontal surfaces may sag between supports, allowing condensate to collect. An improved drainable reheater has recently been designed and installed by B&W (Figure 6). Though this improved design is appropriate for horizontal configurations, a material solution for corrosion in pendant arrangements is stiil being pursued. CONCLUSION When a unit begins cycling service operation, a critical review of the increased potential for waterside corrosion problems must be made to evaluate the most cost effective approach to minimize preboiler corrosion, corrosion product transport, and tube failures, and optimize the unit's chemical cleaning frequency. In summary, the following actions are recommended: 0 Lay-up practices (10) should be instituted that are appropriate to the duration of the out-of-service period and the projected circumstances of the unit restart. Many details of the practices will be plant or component specific. They may include either wet or dry lay-up practices. c Feedwater deaeration should be instituted as soon as possible in the start-up. An appropriate steam flow capacity will be required. The process should be monitored to verify its effectiveness. A feedwater recycle loop may be required. G Sampling and monitoring equipment design must be examined, to insure that all samples are representative of the cycle fluid composition. 0 Corrosion product transport in the condensate/feedwater system should be quantified. System specifics will identify the potential locations and effectiveness of candidate locations, corrosion-product-transport controls, and possibly, of a feedwater recirculation line. 0 The frequency of boiler tube sampling should be increased to facilitate the early detection of an increase in heat-transfer- surface waterside-deposit accumulation compared to previous base- loaded operation practices. Additionally, changes in deposit composition and any incipient underdeposit corrosion can be monitored. Early detection of any required change in chemical cleaning frequency is preferred. ro REFERENCES Manolescu, A. V. and Mayer, P., "Structure and Composition of Protective Magnetite on Boiler Tubes, "Paper presented at Corrosion/80 (National Association of Corrosion Engineers), Chicago, Illinois, March 1980. Grabowski, H. A. and Klein, H. A., "Corrosion and Hydrogen Damage in High Pressure Boilers," 2nd Annual Educational Forum on Corrosion, National Association of Corrosion Engineers, September 1964. Bursik, A.: "Eight Years of Modified AVT with Elevated Oxygen Level for Once-Through Steam Generators. Proc. Int.Water Conf. 47, 226-230 (1986)". Haller, K. H., "Once-Through Boiler Development, "Technical Paper BR-1168, Babcock & Wilcox, Barberton, Ohio. Pocock, F. J. & Banweg, A., "Current Waterside Corrosion Concerns in Fossil Utility Steam Gerierators, "Technical Paper RDTPA 82-58, Babcock & Wilcox, Alliance, Ohio. Manual_on Chemical Cleaning of Fossil Fueled Steam Generation Equipment, EPRI Report CS-289, January 1984. Coulter E. E., "Sampling Steam and Water in Thermal Power Plants", Operating Symposium Thermal Utilities Boiler Reliability McMaster University, Hamilton, Ont., May 4-5, 1983, Klein, H, A., "Use of Coordinated Phosphate Treatment to Prevent Caustic Corrosion in High Pressure Boilers, " Combustion, October 1962, "Boiler Water Phosphate Chemistry, "Plant Service Bulletin PSB-25 Babcock & Wilcox, Barberton, Ohio, Consensus of Current Practices for Lay Up of Industrial and Utility Boilers, ASME, New York, NY. Table 1 Feedwater specification for universal pressure boilers (copper alloy) 9.2-9.5 (copper free) Carbon Steel Copper Alloy Heaters Heaters pH @ 25°C (77°F) 9.3-9.5 8.8- 9.2 Total solids, ppb (ug/L) 50 -50 Cation cond., nu»mho/cm(uS/cm)25°C <0.5* <0.5* Fe, ppb (ug/L) 10 10 Cu, ppb (ug/L) 2 2 SiO, ppb (ug/L) 20 <20 O., ppb (ug/L) 7 7 *Normal maximum Table 2 Water quality criteria - drum boilers (> 2000 psi) _ Boiler water Recommended No With feedwater condensate condensate Item values Item polishing polishing Oxygen, ppb 7 Max Total solids, PPM 15 Max 0.5 Iron, ppb 10 Max Na3PO4, PPM 3-10 - Copper, ppb 5 Max OH, PPM 1.0 Max - Hardness, ppb 0 pH @ 25°C(77°F) 9.0-10.0 9.0-9.5 CO2, ppb 0 Silica by drum by drum pressure pressure Organic, ppb 0 pH @ 25°C(77°F) 8.8-9.2 Table 3 Water quality limits for various operating conditions for once-through type boilers. WATER QUALITY LIMITS FOR —T BEFORE RAISING PRIM SH DIFFERENT OPERATING BEFORE FIRING OUTLET TEMP. ABOVE S50 F. NORMAL OPERATION IMITS SHOWN BELOW ELOW, SHOULD BE MET IF ANY OF THE LIMITS, BELOW. ARE | Sioupaewer 1. | Barone pm sh OUTLETS es dreamer toy fear eens | eee aaN ceMe OR TO THE CORRECTIVE ACTION. IF ANY. SHOULD. BEGIN ° 3 7 3 . zZ iz z |¢ w |e z |¢ w fo a Salis ee [So0lzZ Zz ia Sauls Z lo Zz le S8]8s] = |Z |8else 5.12 8gse Sa/e % z e2|32|_ |2el23|22| 5 [28] 28 |22|22|22| = |2e| 22152] fe | 3e| Se Ss]e8} ze ]2°1S3]os 21g | ssies S12 [PS |S 13 SAMPLE LOCATION 5 |é * 15 18 = 16 & 1S * Ia 4 <> HOTWELL PUMP DISCH {-|-] [- [2] sio]os | ge w [of s | <4) DEMIN. EFFLUENT o| : | - | <O1 | : | - Jao | | - [o | 2 | © ggpcvern ool Ele fel —Tetst fl beret f-[ | > DEAERATOR INLET | L | - [2 | - wfof2 | - <8> DEAERATOR DISCH . | : | : - | : | 5 | 5 | 10 | 2]- ECONOMIZER INLET sio] 10 [32 20 [510] -10 | 32 | - | 100 sio| os 82 8 | 10 | 2] - SUPERHEATER BY-PASS - | 10 | - | . | _ |. | & SUPERHEATER OUTLET - : | [os | 2 (1) pH for cycles having carbon stee/ feedwater heaters. Notes 1. In start-up situations the boiler should not be fired until the cation conductivity 1s below 1 Oy MHO. 2 While ining, the feedwater O> cation conductivity exceeds 2yMHO for five minutes or 5uMHO for two minutes the boiler fires should be tipped During start-up. if the conductivity exceeds 1 O~MHO (normally should be less than 0.5yMHO) investigation of the cause and correction action should be taken 3. pH and hydrazine concentrations are controlled by chemical feed. 4 Dictated by turbine vendor requirements, Table 4 Drum-Type Boilers Operating at 2000 psi or Higher Cycling Service Feedwater* 1500 and over Item 0 to 600 600 to 1000 1000 to 1500 Recommended Oxygen, ppb U 7 7 7 Max. lron, ppb 100 50 10 10 Max. Copper, ppb 50 30 5 5 Max CO2 Organics 0 0 0 0 Silica, ppb - 250 100 20 Max. pH @ 25°C (77°F) 8.8-9.2 8.8-9.2 8.8-9.2 8.8-9.2 (Copper Alloy) pH @ 25°C (77°F) 9.3-9.5 9.3-9.5 9.3-9.5 9.3-9.5 (Copper Free) *There are many differences between specific plants that impact corrosion product transport, so these interim limits should be used as target goals rather than specific requirements. Recycle clean-up capability is recommended for pre-boiler corrosion product accumulation removal along with steam or nitrogen blanketing of feedwater heater shell- sides during downtime. If in-cycle polishers are available, utilize at full flow during pressure transients. Direction of flow (0.1 inch) Transverse section through internal magnetite layer in a once-through boiler furnace tube. <— Nickel plating ag Porous overlay Protective oxide layer s— Base metal ~~»! _ Transverse section through internal 0.25 mm magnetite layer in a once-through (0.01 inch) boiler furnace tube. Figure 1 Transverse section through internal layer in a once-through boiler furnace tube. Steam escaping from mouth of steam chimney by successive formation Capillary channels drawing and release of steam liquid to the base of the steam bubbles chimney [ | sues Deposit Thickness Metal Surface Figure 2 Model of “wick boiling” in magnetite deposit. 25 100 250 500 1000 This chart is for suspended Black Iron Oxide (Fe304) only and is based on a passage of one liter of water containing the indicated concentration of (Fe30,) in terms of parts per billion . . . iron (Fe) through a 0.45,u pore size membrane filter. Figure 3 Membrane filter comparison chart. ppb (Fe;0,) The Babcock & Wilcox Company Power Generation Division Figure 5 Pendant return bend exhibiting out of service dissolved oxygen pitting. Normal Horizontal Reheater Improved Drainability Reheater Design RH Outlet Header RH Outlet Header RH Inlet Header RH Inlet Header Figure 6 Normal vs. improved drainability design for horizontal reheater surface. Technical Paper Combined cycle using PFBC F. L. Kinsinger Manager, PFBC Projects Babcock & Wilcox Barberton, OH D. K. McDonald Manager, PFBC Engineering Babcock & Wilcox Barberton, OH Presented to American Power Conference Chicago, IL April 18-20, 1988 Babcock & Wilcox BR-1339 a McDermott company COMBINED CYCLE USING PFBC F. L. KINSINGER Manager, PFBC Projects and D. K. MC DONALD Manager, PFBC Engineering Babcock & Wilcox Barberton, Ohio ABSTRACT American Electric Power (AEP) and ABB Carbon (ABBC), formerly ASEA PFBC, of Sweden began studies in 1976 which confirmed the advantages of Pressurized Fluidized Bed Combustion (PFBC) and led to the construction of a 15 MWt Component Test Facility (CTF) in Sweden which has provided over 5,000 hours of operating experience since 1982. In early 1985, Babcock & Wilcox (B&W), also involved in PFBC development since 1978, joined the project and formed a partnership, ASEA Babcock PFBC (ABP), with ABBC to serve the United States and Canadian markets. In 1986, ABP proposed a 70 MWe combined cycle PFBC demon- stration plant to AEP and a contract was signed for a unit scheduled for start-up in mid-1990. In 1987, the U.S. Department of Energy (DOE) and the Ohio Coal Development Office (OCDO) selected the Tidd project for funding under their Clean Coal Technology programs. Since that time, the detailed design work has been nearly completed and component fabrication is in progress. This paper describes the basic configuration and advantages of PFBC as applied to the Tidd project. INTRODUCTION A number of technologies are available today or are currently being developed that offer the opportunity to burn coal cleanly and efficiently. Among them are both bubbling and circulating atmospheric fluidized bed boilers (AFB), coal gasification combined cycle, and pressurized fluidized bed combustion. The comparative economics of these various technologies, though not discussed in this paper, continue to be extensively studied and debated. However, a number of attributes of the PFBC will be dis- cussed which support the conclusion that the combined cycle PFBC offers an environmentally acceptable and economically attractive source of electricity. It is generally accepted that combined cycle applications yield higher efficiencies than con- ventional steam cycles. This is certainly true with the combined cycle PFBC which employs both a gas turbine and a steam turbine for generating electricity. PFBC employs a fluidized bed boiler which can be either a drum or a once-through design operating in an environment of 12 to 16 atmospheres. Two different PFBC cycles have traditionally been considered, turbocharged and combined. In the turbocharged cycle, heat is extracted from the gas prior to entering the gas turbine, using steam cooled convective surfaces, leaving sufficient energy to drive a compressor to pressurize the system but not enough to generate electricity. The combined cycle maintains the high temperature of the combustion gases between the fluidized bed and the gas turbine allowing the gas turbine to have suf- ficient energy to drive both the compressor and an electric generator. The major technical uncertainties associated with PFBC are in-bed tube erosion and the hot gas clean-up system before the gas turbine. Both turbocharged and combined cycle bubbling bed boilers are subject to in-bed tube erosion though the lower fluidizing velocity of PFBC has greatly reduced the risks relative to bubbling AFB. The risk associated with the hot gas clean-up (HGCU) system has been debated. The turbocharged system with lower gas temperatures has been viewed as having lower risk relative to HGCU development. However, ASEA Babcock's two-stage high efficiency cyclone system is existing technology. Based on the testing completed to date, it offers an effective means of protecting the gas turbine. Another consideration is the higher proportion of SO, in the gas produced by the higher partial pressures relative to a conventional PC unit. If moisture is present and the gas temperature falls below the dew point (250°F to 300°F), acid can form. This can occur in the gas turbine for turbocharged systems and in the back rows of the turbine exhaust gas heat exchanger for the combined cycle. Since methods exist to control and/or protect the economizer, the combined cycle is more attractive than either operating the turbo expander in an acidic mist environment or suffering the power loss associated with a higher turbine exit gas temperature. Thus, when weighed against the relative risks between the two PFBC cycles, the higher cycle efficiencies of combined cycle PFBC make it more attractive. potentially WHY A FLUIDIZED BED IN A PRESSURIZED ENVIRONMENT? In PFBC, as in any bubbling bed Fluidized Bed Boiler, crushed coal and limestone or dolomite are fed into a boiler in which air, entering from the bottom, maintains the essentially inert bed material in a highly turbulent suspended state, called fluidization. This turbulence creates good contact between the air and particles and between the particles themselves, allowing for good combustion efficiency and excellent absorp- tion of SO, by the sorbent to form calcium sulfate. The solid material removed is dry and essentially benign and is much more easily managed than the wet sludge by-product from a wet FGD. By controlling the fuel feed rate, the temperature of the bed can be maintained at the relatively low temperature of 1580°F. The cooler combustion zone temperatures are below the ash fusion temperature of virtually all known coals. Therefore, slag is not formed and the process is relatively insensitive to the type of coal burned. Also, the low combustion temperature minimizes the formation of thermal NOx to less than 50% that of a conventional boiler. Another advantage of a bubbling fluid bed is that the boiler tubes submerged in the bed have overall heat-transfer rates of 4 to 5 times higher than those in a convective environment resulting in more economical and compact steam generating surface. By pressurizing the process, the boiler exit gases contain enough energy to drive a gas turbine-generator. The gas turbine, in concert with the steam turbine, results in a very efficient combined cycle configuration. Since the combustion process takes place in a pressurized environment, a deeper bed may be utilized without unduly penalizing the overall system pressure drop. In fact, the deeper bed of a PFBC uses a proportionately smaller amount of the total system pressure drop and allows 50% of the total residence time to be in the bed where it is more effective compared with 10% for the shallower beds in bubbling AFB. Pressurizing also permits the use of much lower fluidizing velocities which lead to lower erosion rates. The resulting high in-bed residence times produce combustion efficiencies in excess of 99% and good sorbent utilization. Notes: 1. Furnace shaft size is bed plan area and height 2. Based on’ AFBB | CFBB| PFBB {eS Pressure Latm | 1atm | 12 atm Velocity 10 tps | 20tps| 3 tps es | Total | | Residence 6sec | 45sec] 6sec ‘Atmospheric ‘Atmospheric Pressurized Bubbling Circulating Bubbling Bed Fluid Bed Fluid Bed Figure 1 Comparison of furnace shaft sizes for equal heat input. Pressure also results in a _ considerably smaller bed plan area (A). For the same air mass flow (M), a bubbling PFB at 12 atmospheres with 3 fps fluidizing velocity (V) will require 28% of the bed plan area of an atmospheric bubbling fluid bed and about 56% of that required for an atmospheric circulating fluid bed due to the much higher air density (P) where M = PVA. The lower velocity also significantly reduces the total height required for the bed and freeboard since the height is the product of the velocity and the total residence time. The result of these factors on the PFBC boiler is illustrated in Figure 1. This compact design can result in a physically smaller power plant leading to a lower capital cost and the ability to modularize major components. Maximum use of off-site fabrication and assembly can also further shorten field construction time. TESTING To prove the process and components being utilized for Tidd with a larger commercial plant in mind, AEP and ABBC have cooperated in the development of PFBC technology and its application to combined cycle power generation since 1976. The early feasibility studies, hot and cold model tests, and involvement in a DOE sponsored combustion test program at the CURL pilot plant in Leatherhead, England led to the 1982 construction of ABBC's 15 MWt Component Test Facility (CTF) formerly in Malmo but now located in Finspong. Those early tests, and the en- couraging results of over 5000 hours of testing at the CTF, have culminated in the Tidd Demonstration Project. The CTF is designed to permit testing at pressures up to 17 bara and bed temperatures up to 1650°F (900°C) at an air flow of about 56,000 lb/hr (7 kg/s). Coal and sorbent can be fed either pneumatically or in paste form at up to 7,900 lb/hr (1 kg/s) each. Use of this large scale PFBC test facility has fulfilled the need to develop and test systems and learn how they can best be integrated into the process. Two sets of tests essentially identical in operating conditions and using the same coal and sorbent specified for Tidd have since been completed and a third is in progress. Considerable testing using a variety of European coals and sorbents in support of ABBC's Vartan and Escatron PFBC projects has also been completed. One of the most’ significant areas of investigation at the CTF has been understanding how to operate and control the process. In preparation for the 200 MWt P200 plant being supplied for Tidd, virtually all of the systems have been included. The CTF boiler is about 3 ft. 3 in. deep by 6 ft. 6 in. wide at the top of the bed, 1/15th of Tidd's bed plan area, and nearly full scale in height. It, too, is of membraned waterwall construction with all of the heat transfer surface located in the bed. It operated at the same process pressure with the same 10 ft.-6 in. full load bed height as Tidd. The steam cycle is once-through with 1523 psia (105 bara) 995°F (535°C) outlet conditions compared to 1300 psia 925°F for Tidd. Like Tidd, the boiler bottom is of waterwall construction and contains the sparge air distribution system. Though one coal water paste (CWP) feed point is used, the same coal prepara- tion requirements and the piston type injection pumps being supplied at Tidd have been proven at the CTF. The CTF also utilizes a pneumatic sorbent feed system. Bed ash and cyclone ash removal are also accomplished with the same component types Tidd will use. A two-stage string of cyclones has been tested with the same entry velocities and component design and at nearly the same size relative to Tidd. The CTF has also proven all of the primary control principles being employed in the Tidd design. Load is changed by changing bed level. The method of measuring bed level and temperature used is the same as that planned for Tidd. The bed reinjection system components and function are also identical. Both the CIF and Tidd use firing rate to control bed temperature and the feedwater flow is varied to protect the in-bed surfaces and remove an appropriate amount of heat from the bed at a given load. Thus load changing as well as start-up and shutdown at the CTF are accomplished in the same way it will be at Tidd. Since wastage of the in-bed tubing is of concern, extensive testing of tube materials, sizes, spacings, arrangement patterns and erosion inhibiting methods has been accomplished. In fact, much of the CTF tube bundle has _ been present during all the 5,000 hours of operation, although modifications have been made to _ the arrangement and other types of tubing materials and erosion resisting methods have been incorporated. On several occasions, tube platens, consisting of evaporator and superheater tubes have been lifted into the freeboard for FLY ASH,* CYCLONE ASH AND GAS TURBINE, AND COAL & DOLOMITE BED ASH SILOS DOLOMITE — STORAGE BED ASH EXISTING, STORAGE NOT UTILIZED FOR PFBC COAL STORAGE TEST COAL PILES PUMP STRUCTURE WASTE WATER SETTLING POND Figure 2 Tidd PFBC site layout. inspection. The results indicate that the much lower fluidizing velocities employed in PFBC relative to AFB greatly reduce the rate of erosion but the cool evaporator tubing remains sensitive to wear while the higher temperature surfaces are protected by the formation of an oxide layer resulting in virtually no significant wastage. Development of a combination of judiciously selected tube materials and employment protective methods has resulted in immeasurably low wear rates for evaporator tubing in the CIF leading to tube life predictions in excess of 40,000 hour. Testing of materials at the higher metal temperatures has also led to a better understanding of the relationship between the oxidation rate of various alloys and their sensitivity to erosion and corrosion when exposed to PFBC conditions. The results of these findings have been incorporated into the design of the tube bundle for Tidd. Since hot gas filters capable of protecting the gas turbine and meeting New Source Performance Standards (NSPS) simultaneously are not yet commercially available, blade erosion has also been extensively studied at the CTF and the results factored into the new GT-35P design. The first gas turbine rotor at the CTF was installed downstream of the cyclones after about 600 hours of operation and was subsequently removed after about 1000 hours of exposure. The unclad blade showed slight localized corrosion, but the clad blades showed none. A second rotor was installed which has been removed and examined after nearly 3000 hours of additional operation. Rotor blades made of Udimet 500 have been tested unclad and with a variety of cladding. None, including the unclad blade, has shown any sign of erosion. Particulate deposition on the gas turbine test components has also been low, being typically less than 10 mils. CARDINAL UNIT1 STACK [~ COMBUSTOR, * PRECIPITATOR* PREPARATION BUILDING ECONOMIZER*, FLUE GAS DUCT* TEST DOLOMITE PILES * NEW EQUIPMENT TO BE INSTALLED AT EXISTING SITE INTERCOOLER CIRCULATION PUMP =} BED BED ASH REMOWAL! PREHEATER ECONOMIZER BOWLER FEED T-___.-. To PRECIPITATOR AND STACK PUMP i SE / wet A yh 6 DAMPER, Y REJECT CHUTE | CYCLONE ‘ASH COOLER BELT WEIGHER AND CESS Srosrine weren TO CYCLONE ASH SYSTEM MIXER AND ‘QUALITY ANALYSIS chute WATER Figure 3 PFBC system schematic for TIDD. DESCRIPTION OF TIDD Preliminary design work for Tidd was completed in 1986. Currently, work is proceeding with the detailed engineering and design in parallel with material procurement and the permit and licensing application process. Construction at the site is scheduled to begin in the Spring of 1988, with first fire in the Summer of 1990. At this time, planned funding includes a three-year operation and testing period extending through 1993. The Tidd Project is a repowering of Tidd Unit 1 near Brilliant, Ohio. This mid-1940 vintage 110-MW plant, employing two PC fired boilers supplying a single non-reheat steam turbine, was de-commissioned in 1976. The new PFBC island, including fuel and sorbent preparation and feed systems, a gas turbine, ash removal systems, a combustor containing the boiler, bed reinjection and gas cleaning systems, will be installed adjacent to the existing building (Figure 2). The existing steam turbine will be operated at about 50% capacity. Other balance-of-plant equipment such as the deaerator, condenser, feedwater heaters and pumps and coal handling and storage equipment will also be refurbished and used. These factors will result in a projected total net electrical output of 70.5 MWe, a net efficiency of 34.5% and net heat rate of 9890 Btu/kWh for the Tidd P200 compared to a total net electrical output of 83 MWe, a net efficiency of 39.8% and net heat rate of 8573 Btu/kWh for a new P200 installation utilizing a 2400 psi steam cycle with reheat. Figure 3 illustrates the basic components and the cycle for Tidd. In the air-gas cycle, ambient air enters the gas turbine and passes through the low pressure compressor and is cooled via an intercooler. It is then further pressurized to 12 bara and heated to about 572°F in the high pressure compressor. The hot compressed air is directed through the outer annulus of a coaxial air/gas crossover pipe into the pressure vessel. The air from the top of the pressure vessel proceeds downward in a duct which directs the air through ash coolers prior to entering the bottom of the bubbling fluidized bed boiler through a _ sparge duct distribution system. Coal and sorbent are injected into the bottom of the fluidized bed containing submerged boiling and superheating surface, and combusted at a controlled tempera- ture of 1580°F. The gases from the bed pass through the boiler freeboard and into seven parallel strings of two stages of gas cleaning cyclones. After the cyclones, the gas is col- lected in a common manifold and conveyed via the inner pipe of the air/gas coaxial pipe to the gas turbine. The pressurized and heated gas is expanded through the high pressure and low pressure turbines providing not only the energy to drive the compressors, but also a generator which produces about 20 to 25% of the plant electrical output. After the gas flows through an economizer, where it is cooled by feedwater to approximately 350°F (177°C), and a precipitator, it is finally discharged to the atmosphere from the existing nearby Cardinal plant stack. In the water-steam cycle, water from the condensate polishing system is warmed in the feedwater heaters and forced by the boiler feed pumps into the economizer at about 1900 psia at full load. From the economizer the slightly subcooled water enters the boiler bottom hopper enclosures. After passing through the boiler bottom, bed and freeboard enclosures, it enters the in-bed evaporator surface where boiling occurs. Fluid from the evaporator outlet, which is two phase up to about 40% load and slightly superheated at full load, is conveyed to the vertical separator. At lower loads, the drain from the separator is recirculated through the boiling surfaces with pump assistance. The steam from the separator then enters the in-bed primary superheater. From the primary superheater outlet, spray attemperation is used to control final steam temperature leaving the in-bed secondary superheater to the turbine. Pressure is controlled by the steam turbine throttle valves. During start-up and in the event of a steam turbine trip, a turbine bypass system to the condenser and a pressure control valve to the atmosphere serve to dispose of the excess steam while controlling the boiler pressure/temperature decay and conserving as much of the boilerwater as practical. In the event of a loss of plant power or the boiler feed pumps during operation, a backup feedwater system is also provided to maintain water flow to the boiler circuits exposed to the heat contained in the slumped bed. COMPONENT DESCRIPTION The three primary components in a PFBC plant are the gas turbine, the steam turbine and the boiler. Though the other systems support one of the primary components, they must be carefully designed since their energy losses can significantly impact plant efficiency and are often major maintenance concerns. This section describes the components being supplied for these components at Tidd with insight into the considerations and philosophy incorporated into their design. GAS TURBINE The gas turbine compressor acts like the fans in a conventional plant to supply the combustion air and fluidize the bed. However, unlike the fans which are independent of the combustion process, the gas turbine is powered by the exhaust gases from the boiler. The energy remaining after the air is compressed is used to generate electricity. Since the gas turbine is driven by the hot pressurized gases from the boiler, and simultaneously supplies the combus- tion air to the boiler and generates electricity, certain characteristics are desirable to ef- ficiently accommodate the subsequent un- conventional conditions it must meet. A gas turbine ideally suited for PFBC throughout the load range would: 1) accept the relatively low inlet gas temperatures associated with the fluid bed process (assuming no topping cycle); 2) be minimally affected by changes in ambient air conditions; 3) provide a pressure- to-air mass flow characteristic which would permit near constant fluidizing velocity, excess air ratio and velocity into the HGCU (especially cyclones); 4) optimally compromise the opposing desires at low load for a low air flow to the boiler and a high air flow to the gas turbine; 5) and withstand some dust loading in the gases without significant damage. To achieve these characteristics, ABB STAL has modified their conventional GT-35 gas turbine and renamed it GT-35P. The conventional three shaft design was changed to two shafts to provide operating flexibility throughout the load range, and additional stages were added and turning angles reduced to minimize erosion potential. The constant speed high pressure (HP) shaft mechanically couples the four-stage HP turbine to the eleven-stage HP compressor and is connected via an epicyclic gear to a motor/generator which is synchronized with the grid. The free running low pressure (LP) shaft mechanically couples the single stage LP turbine to the twelve stage LP compressor. The power split between the HP and LP turbines is controlled to compensate for changes in ambient conditions and the LP compressor conditions can be adjusted to extend the flow capacity. STEAM TURBINE In order to minimize the project cost, the existing steam turbine-generator will be utilized. The shaft water seals on this 1946 vintage 1800 RPM condensing General Electric machine will be replaced with new steam seals and will be thoroughly inspected and refurbished as necessary. This turbine-generator was designed to produce an electrical output of 110 MWe for a non-reheat 1300 psia, 925°F steam cycle. Though the steam conditions will be the same for this project, it will operate at reduced load, producing only 55.9 MWe of gross electrical output. The use of this oversized relatively low pressure steam turbine- generator and other original balance-of-plant equipment accounts for the lower expected cycle efficiency and electrical output of Tidd relative to a new PFBC installation of the same 200 MWt size. BOILER A PFBC boiler must produce steam at the desired conditions, provide sufficient controllability to accommodate the steam turbine and the grid, provide the exhaust gas conditions corresponding to the power required by the gas turbine for a given air flow, and contain and regulate the fluid bed combustion process throughout the load range. It must also be designed to withstand the abnormal conditions resulting from equipment trips and plant power loss. The boiler for Tidd is a once-through design. It is enclosure which contains the fluidized bed, evaporator, primary and secondary superheater surfaces. A water attemperator, start-up, steam turbine bypass and backup feedwater systems are also provided. The enclosure is top supported from rods and a conventional buckstay system supports the walls against the relatively high pressure difference between the air and gas sides. It consists of three distinct sections: the boiler bottom, the bed enclosure and the freeboard enclosure, each having their own unique functional and design considerations. The boiler subcritical comprised of an bottom supports the air distribution system and collects the bed ash, which is cooled by combustion air and conveyed out of the pressure vessel. It is constructed of membraned waterwalls arranged into two inverted pyramidal hoppers which fill with bed ash during operation. An air inlet duct which contains the air preheating system for start-up and conveys the air to the sparge distributors during normal operation is located between the hoppers. The bed enclosure, where the combustion process and the majority of the heat transfer to the steam cycle take place, contains the fluidized bed and the in-bed heat transfer surfaces. Its membraned waterwalls are tapered to enhance the process and are spiral-wound to provide the high mass flows per tube required to withstand the high heat transfer rates in the fluid bed without requiring internal refractory or insulation. Though no internal insulation is used in the zone containing the tube bundle, a short section below it is internally insulated to minimize start-up bed-preheating requirements. The freeboard enclosure is also of membraned waterwall construction and is internally insulated and lined to minimize heat absorption above the bed and maintains the gas temperature as close to the bed temperature as practical at full load. The in-bed tube bundle is supported within the bed enclosure. It consists of evaporator, primary and secondary superheater surfaces. Between the primary and secondary superheaters, a water attemperator is provided. The tubing and its supports are arranged to achieve the spacing and density desired to promote good mixing in the bed and minimize erosion potential. It is designed to resist the dynamic forces of the bed material and absorb the correct amount of heat from the bed at a given bed level while providing a temperature leaving the boiler which is compatible with the gas turbine at that load. A moisture separator and circulating pump have also been incorporated. At low loads, the moisture separator, like the systems B&W provided with other subcritical UP once-through boilers, acts similar to a steam drum to separate the steam-water mixture leaving the evaporator. The water separated is combined with the feedwater flow, which matches the steam flow (minus at- temperation), and is recirculated at a rate to achieve the minimum flow requirement to protect the water circuits submerged in the bed. Thus as load increases and steam quality leaving the evaporator is sufficient, the vertical separator and circulating pump are no longer needed and the boiler operates once through. During these low load conditions and in the event of a steam turbine trip, a turbine bypass is also necessary. AUXILIARY SYSTEMS PFBC requires several auxiliary systems, the most significant of which are: the pressure vessel, fuel preparation and feed, sorbent feed, ash removal, bed material reinjection for load control, hot gas cleaning, economizer and balance- of-plant equipment. Figure 4 Pressure vessel fabrication. For the process to take place under pressure, the fluidized bed must be contained in a pressure boundary. In addition, it is expedient to maintain several of the other systems at the high pressure. By locating equipment inside, it can be designed for the much lower air-to-gas side pressure difference, and any heat losses are recovered to the combustion air. For Tidd, the boiler, reinjection system, cyclones and cyclone ash coolers are contained within a 44-ft. diameter, 68-ft.-high cylindrical pressure vessel. The vessel is 2-7/8 in. thick with a shipping weight of 1340 tons, which includes most of the contents. It is being constructed at B&W's Mt. Vernon, Indiana plant, and once completed, stress relieved, hydrostatically tested and contents assembled within, will be shipped up the Ohio River to the Tidd plant site. The fuel preparation and feed system must convey coal from the coal yard into the boiler in a safe manner while overcoming the high process pressure. The most common methods developed to accomplish this have been pneumatic and coal water mixtures (CWM). While the pneumatic methods require inerting, a source of boosted air for conveying, and can be erosive, the CWM are inherently inert and ‘'slippery'. Assuming the mixture can be pumped and the moisture content is not too great, (less than about 30% by weight), the moisture will have little impact on efficiency since about 1/3 of the energy used for vaporization can be recovered in the gas turbine. CWM have also been shown to burn more evenly producing more uniform bed temperatures and requiring fewer feed points than pneumatic systems. However, the optimum system/design de- pends on the ash and sulfur content of the coal. For Tidd, a coal water paste (CWP) system using 25% water by weight is being provided. A variable speed roller crusher is utilized to reduce the raw 3/4 inch top size coal to the proper combination of large (1/8 in. top size) and fine material to produce a pumpable paste. From the crusher, coal and water are brought together in a controlled manner into a mixer. When sufficiently mixed, the CWP will be stored in a continuously agitated surge hopper prior to being fed into the boiler by six parallel hydraulic-dual-piston fuel injection pumps (similar to those used in the concrete industry). Each pump feeds its own nozzle and compressed air is introduced near its tip to properly break up the paste stream as it enters the bed. Since sorbents are not combustible, the greatest challenge in system design is overcoming the process pressure. Generally, pneumatic feed systems have been used. However, in some cases where the coal ash and sulfur content is low and relatively low sorbent feed rates are required, it has been combined with the coal in a CWP to simplify the feed systems. A pneumatic feed system has been selected for Tidd. Sorbent crushed to 1/8 inch top size and dried is fed into a lockhopper system. The lockhoppers are sequenced so that as one fills, the other is pressurized and is feeding material into a rotary feeder which meters it into the boiler via a light phase transport system. Air for the system is taken from the combustor vessel and boosted by a compressor. Two ash removal systems will be installed at Tidd; one for the ash from the cyclones and one for bed ash. The cyclone ash is continuously removed and conveyed pneumatically using a small amount of high pressure gas from the cyclones to an atmospheric silo. To overcome the difficulty of handling the 1580°F ash and the need to reduce it to atmospheric pressure, a device has been developed which cools and depressurizes the ash while still inside the combustor vessel without using lockhoppers or valves’ for _ pressure reduction. The bed ash, collected and cooled in the boiler bottom hoppers and removed at a rate coordinated with the sorbent and fuel feed rates to maintain bed level, is metered from the boiler bottom into a conventional lockhopper system where it is reduced to atmospheric pressure. It is then transported by conveyors to a _ storage silo. Load is controlled by changing bed level via a bed material reinjection system. This system pneumatically conveys material from the boiler during load reductions, stores it in a pair of vessels located inside the pressure vessel which are designed to maintain the material temperature, and reinjects it back into the boiler via an "L" valve when an increase in load is desired. In PFBC, the dust laden hot gases leaving the boiler must be cleaned to an acceptable level before entering the gas turbine. This task must be accomplished at the high gas temperature associated with the combined cycle. Though several methods of hot gas cleaning are in the experimental stage, high efficiency cyclones have been successfully tested. Thus, seven parallel two-stage strings of cyclones made of micro-alloy stainless steel rather than an internal refractory lining will be used at Tidd. The inlet velocities will be about 100 ft/sec for the first stage and 140 ft/sec for the second stage to remove all particles larger than 10 microns. Though the cyclones will clean the gas suf- ficiently for the gas turbine, further cleaning is necessary to meet NSPS. After leaving the gas turbine, it passes through a once-through economizer where heat is recovered to the feedwater prior to entering the boiler. The economizer is of modular construction utilizing vertical in-line tubes spirally finned to maximize heat transfer while remaining compact. The spiral fins are permitted by the low dust loading in the gas relative to conventional PC conditions. From the economizer, the gas passes through a conventional electro- static precipitator before flowing to the stack. Most of the remaining equipment such as with the feedwater system, including heaters and pumps, plant air and service water, switchgear, condenser and deaerator from the existing plant, will be refurbished as necessary and reused. A condensate polishing system and a_ source of auxiliary steam have been added. OPERATING PRINCIPLES Due to the unique synergistic relationship between the boiler and the gas turbine, the steam cycle will normally be operated in a turbine- following mode. Load changing of up to about 2% per minute (4% per minute can be provided) is accomplished by changing bed level. This will expose more or less of the in-bed surface to the lower convective heat transfer rates above the bed simultaneously reducing heat input to the steam cycle and reducing the gas temperature to the gas turbine. The reduction in power to the gas turbine results in a corresponding reduction in air flow to the combustor. Firing rate is adjusted to maintain bed temperature at a constant 1580°F (860°C) at a specified excess air ratio. Feedwater flow is regulated to protect the heat transfer surface, absorb the proper amount of heat from the bed and, along with attemperation, control steam temperature. Steam pressure is maintained by modulating the steam turbine throttle valve. To start the process, a _ shallow bed is fluidized and heated by the oil fired preheater. During this stage, a portion of the air flow is temporarily diverted through the preheater where fuel oil is burned at high excess air ratios. The hot gases are directed into the start-up bed through the air distribution system. When the bed material has reached the required ignition temperature, CWP is injected and the preheater is ramped off. During start-up, the gas turbine is motored to produce the required air flow. As load increases and power to the turbine increases, a decreasing amount of power is used from the grid until the turbine becomes self sustaining. Above that point, the gas turbine generates electrical power. The moisture separator and circulating pump are used up to about 40% load above which the boiler becomes once-through and the pump is shut down. Feedwater flow is established at 10% of full load flow and the pump recirculates water through the boiler enclosure and evaporator to maintain minimum cooling flow. Until the steaming rate reaches 10%, the separator drains will remove excess water to the condenser. While the boiler circulating pump is operating, feedwater flow is regulated to maintain separator level. Once it is shut down, feedwater flow is regulated to control steam temperature and a minimum water level is maintained to permit a restart if necessary. Until sufficient steam flow, temperature and pressure are available to roll the steam turbine, the pressure ramp is controlled by the bypass system. Once rolled, the turbine throttle valve regulates pressure. Shutdown is accomplished by reducing the bed level to the minimum level, stopping the fuel and allowing the gas turbine to coast down thus cooling the remaining bed material to prevent combustion. At the appropriate time, the intercept/bypass valve is closed and the gas turbine is tripped. In the event of a gas turbine trip, the combustor will be immediately isolated from the gas turbine by an intercept/bypass valve at the turbine inlet. The loss of air flow will cause the hot bed to slump. Sufficient feedwater flow will be maintained to protect the in-bed surface. In addition, special systems have been designed to depressurize the combustor vessel, prevent generation of combustible gases, cool the bed material and remove it from the bed in preparation for a restart. A steam turbine trip will result in bypassing up to 50% of the steam flow to the condenser while venting the remainder to the atmosphere using a pressure control valve to regulate steam pressure. Heat input is rapidly reduced and the unit is shut down in the normal manner. A station blackout is accommodated similar to a gas turbine trip with all essential service valves and equipment on standby power. Since the feedwater pumps are not on standby power, a special backup feedwater system is provided which will supply the necessary water flow to protect the heat transfer surface at compatible conditions. This system would also be engaged in the event of a loss of the main feedwater pumps. MAINTENANCE CONSIDERATIONS With major systems contained within a large pressure vessel, maintenance oriented design philosophies and conventional designs must be applied whenever practical. For example, whenever possible, the temperature and pressure boundaries have been separated. The crossover pipe carries the cooler air in the outer annulus, which must resist the full process pressure, while the inner pipe is internally insulated and lined to withstand the high gas temperatures but only resists the much lower air-to-gas_ side differential pressure. Likewise, the boiler containing the combustion process is surrounded by the cooler air in the pressure vessel and must only resist the lower air-to-gas side differential pressure. Since the gas_ side operates at a lower pressure than the air side, any leakage would result in the cool air entering the gas path rather than the gases affecting equipment not designed for the high temperatures. Access to equipment inside the pressure vessel is impossible during operation, thus moving parts are minimized and all control valves are located outside where they are accessible from platforms. For the purpose of servicing equipment inside the combustor vessel, platforms and lifting lugs have been provided, and removal methods have been considered for each of the major pieces of equipment. Within the boiler, the tube bundle has been specifically designed with maintenance in mind. The platens, containing either evaporator or secondary superheater surface, can be removed by cutting only six tubes and either lifted into the freeboard for servicing or removed from the boiler and/or combustor vessel through an access door in the boiler roof and the upper head of the combustor vessel. For inspection and minor maintenance, access doors in the boiler sidewalls below the tube bundle allow inspection and servicing of the bottom rows of the bundle and servicing of the air distribution system. Access doors above the tube bundle permit entry into the freeboard space, and cable doors are provided to allow lifting and handling of the platens. To facilitate construction and maintenance, even the GT35P gas turbine is of modular construction whose main components are of horizontally unsplit barrel design. The major sections of the turbine can be removed and replaced in a matter of hours. THE NEAR FUTURE ABP, its partners and AEP continue to look to the future. In addition to the many potential retrofits and new site applications for the 200 MWt size PFRC, work on the next step, an 800 MWt plant, is underway. As part of the Tidd project, studies and preliminary designs were developed for all of the major components’ of the larger plant. A supercritical steam cycle with reheat employing two combustors supplying a single steam turbine (680 MWe) has been examined with encouraging results. In addition to the environmental advantages, net efficiencies in excess of 402% have been predicted for this PFBC compared to 36.6% for a supercritica] PCF unit with a FGD system. Even considering contingencies of 25% for the boiler related equipment, 10% for the gas turbine, 5% for balance-of-plant items and 6% for the process itself, advantages of 11% for capital cost, 8% for production costs and 10% for bus bar power cost, were indicated. Table | Comparison of Commercial Plant Options Item Option 1 Option 2 Combustor/Gas Turbine Units 2 1 Net Electrical Output (MWe) 680 330 Steam Cycle Pressure (psig) 3700 2400 Main Steam Temperature (F) 1000 1000 Reheat Steam Temperature (F) 1000 1000 Net Efficiency (%) 41.2 39.7 Net Heat Rate (Btu/kWh) 8360 8600 Plans are now being considered to develop a subcritical 340 MWe reheat steam cycle to a high degree of detail with the intent of building such a plant sometime in the mid-to late 1990's. Table I shows a comparison of the two P800 options studied to date. Should PFBC continue to prove itself attractive, designs for additional sizes (dependent on gas turbine frame sizes available), utilizing both sub- and supercritical once-through, drum type, reheat and non-reheat steam cycles, could be made available. Tidd is a major step in the commercialization of this new and promising technology for efficient and environmentally attractive combustion of coal for electrical power generation. CONCLUSION Pressurized fluidized bed combustion is advancing coal-fired power generat on technology by providing high efficiency, lower emissions and lower capital cost than competing technologies. The continued corroboration of process and component testing at the CTF and the innovative yet simple solutions to the unique technical problems presented during the design of Tidd are increasing confidence that combined cycle PFBC will be successfully demonstrated. The equipment design for Tidd has been essentially completed, and the manufacture of hardware is underway. With first fire only two years ahead and con- siderable work on the next size behind us, ASEA Babcock is now convinced that combined cycle PFBC is an attractive option in the power generation market for both new and retrofit installations. BIBLIOGRAPHY 1. Bauer, D. A., Stogran, H. K., Hemenway, L. F., "Update on the Tidd PFBC Demonstration Project". Fourth Annual Pittsburgh Coal Conference, October 1987. 2. Huryn, J. B., Wickstrom, B., Rederstorff, B., "Design Concept for a PFBC Commercial Plant". Proceedings of the Ninth International Conference on FBC, 1987. 3. Jansson, S. A., “Update on Test Experience From the ASEA PFBC Component Test Facility". Proceedings of the Ninth International Conference on FBC, 1987. 4. Jensen, A. D., "Progress Report on the Tidd PFBC Repowering Project and Comparative Economics of the P-800 System". Comparative Economics of Clean Coal Technologies Conference, November 1987. 5. Mc Donald, D. K., Weitzel, P. S., Thornblad, P., "PFBC Design Integration - ASEA Babcock PFBC's Approach". Proceedings of the Ninth International Conference on FBC, 1987. 6. Mudd, M. J., Guha, M. K., "AEP's Pressurized Fluidized Bed Combustion Development Program". Joint Power Generation Conference, Milwaukee 1985. 7. Mudd, M. J., Bauer, D. A., "Tidd PFBC Demonstration Plant: Combined Cycle PFBC Clean Coal Technology". Proceedings of the Ninth International Conference on FBC, 1987. Technical Paper Initial test results of the limestone injection multistage burner (LIMB) demonstration project P. S. Nolan Babcock & Wilcox 20 S. Van Buren Barberton, OH R. V. Hendriks U.S. Environmental Protection Agency Air and Energy Engineering Research Laboratory Research Triangle Park, NC 27711 Presented at the 81st Annual Meeting of the Air Pollution Control Association Dallas, Texas June 20-24, 1988 Babcock & Wilcox BR-1343 a McDermott company Initial test results of the limestone injection multistage burner (LIMB) demonstration project P.S. Nolan Babcock & Wilcox 20S. Van Buren Barberton, OH R. V. Hendriks U.S. Environmental Protection Agency Air and Energy Engineering Research Laboratory Research Triangle Park, NC 27711 Presented at the 81st Annual Meeting of the Air Pollution Control Association Dallas, Texas June 20-24, 1988 Introduction Control of emissions of sulfur and nitrogen oxides, S02 and NOx ,from existing coal-fired util- ity boilers is a significant concern of the U.S. Environmental Protection Agency (EPA), the utility industry, and other private and public organizations. These two pollutants are generally believed to be major precursors of acid rain. Coal- fired utility boilers account for about 65 percent of the S02 and 29 percent of the NOx emissions in the United States.! The strategy for reducing acid rain has yet to be adopted by Congress; the debate centers on reduc- tion of current S02 emissions by 5 to 12 million tons* per year and on reduction of NOx as an S02 offset or by a fixed amount nationally. The choice of control techniques for limiting S02 and NOx emissions from utility boilers will most likely include a mix of technologies to achieve the desired reduction at minimum cost. Of the exist- ing commercial control options, flue gas desulfur- ization (FGD) is costly, and switching to a lower sulfur content fuel may have severe economic and logistic impacts. Only 10 percent of the boilers east of the Mississippi River, where the impact of acid rain is felt most, are subject to New Source Performance Standards.” Most of these are older units having a useful life of 10 to 30 years. An effective control technology for these boilers would produce a significant reduction of national S02 and NOx emissions. There is a need, there- fore, for lower cost alternatives for control which can be retrofit to this existing boiler population. One such technology is LIMB (Limestone Injec- tion Multistage Burner), which is based on the injection of a dry sorbent into the boiler for direct capture of S02 in the flue gas, combined with the use of low-NOx burners for reduction of NOx emissions. Building on attempts made a decade earlier, EPA initiated a research and development pro- gram in 1981 to improve the LIMB process and bring the technology to a commercial scale. Efforts are concentrated in four major areas: 1. Generic research to provide an understanding of the controlling factors in the technology. 2. Prototype testing to confirm research data and develop design criteria. 3. Full-scale demonstration to show cost effec- tiveness under actual retrofit conditions. 4. Application studies to develop criteria for LIMB application to a wide range of boilers. The current focal point of the LIMB program is the demonstration phase, with much of the more recent research activity oriented toward its support. LIMB demonstration project The EPA LIMB demonstration is a 4-year project that includes design and installation of a LIMB system at the 105 MW, Unit 4 boiler at Ohio Edison’s Edgewater Station in Lorain, OH. The primary fuel being used is an Ohio bituminous coal having a nominal sulfur content of 3 percent. The project plan calls for long-term operation and testing to demonstrate its capabilities. EPA is sharing the cost of the project with the major par- ticipants: the State of Ohio, Ohio Edison, Bab- cock & Wilcox (B&W), and Consolidation Coal. The basic goal of the LIMB demonstration is to extend technology development to a full-scale application on a representative wall-fired utility boiler. The successful retrofit of LIMB to an exist- ing boiler is expected to demonstrate that: 1. S02 reduction of 50 percent or more and NOx emissions of less than 0.5 lb/10® Btu can be achieved at a fraction of the cost of add-on FGD systems. 2. Boiler reliability, operability, and steam pro- duction can be maintained at levels existing prior to LIMB retrofit. 3. Technical difficulties attributable to LIMB operation, such as additional slagging and fouling, changes in ash disposal require- ments, and an increased particulate load, can be resolved in a cost-effective manner. The demonstration project consists of several dis- tinct phases: a preliminary phase to develop the LIMB process design specifically for the host boiler, a construction and start-up phase, and an operating and evaluation phase. The process design was completed and the LIMB system installed at Edgewater in 1987. The system was started up in July 1987 and was operated inter- mittently until late September 1987, when Unit 4 was shut down for a scheduled maintenance outage. Edgewater LIMB design In the LIMB process a calcium-based sorbent is injected into the boiler where it calcines to active calcium oxide, and then reacts with sulfur dioxide and oxygen in the flue gas to produce calcium sul- fate. The product is a solid which is removed with ash in the particulate collection device. Research has shown that maximum sorbent reactivity and sulfation is obtained in the temperature range of roughly 1600 to 2300°F, so the injectors are located where the temperature in the boiler is at Sorbent Injection System Distributor Baghouses Booster Air Feed Humidification System Sf Shield Air \—/ 7 — Atomization Air Silo Truck Delivery of Sorbent Low NOx Burner Storage Silo 1 Fuel Air Compressor Transport Blower Combustion Air Air —4 Heater ~~T\__ Water Stack P+ Dampers Steam Coil Flue Gas Reheater ( Precipitator ) Waste Handling and Disposal Figure 1 L/MB process flow diagram. the upper end of this range. The NOx control component of LIMB is implemented by the use of low-NOx burners which keep flame temperatures relatively low to reduce NOx formation. A gener- alized process diagram is shown in Figure 1. A brief description of the Edgewater equipment is provided here; a more detailed review of LIMB design issues and the basis for the Edgewater design was presented in an earlier paper.® At Edgewater, commercial, calcitic hydrated lime is delivered to the plant by truck and kept in an outside storage silo before it is pneumatically transported to a day silo within the boilerhouse. A controlled rate of sorbent is then metered into conveying lines through a differential weight loss feeder which provides an accurate feed rate. Dried air conveys the sorbent to a distribution bottle which disperses the mixture into the designated number of injection lines connected to the boiler. A flexible injection scheme has been installed consisting of three rows of injectors on the front wall at and just above the nose of the boiler. Two sets of feeders, conveying lines, and distribution bottles are installed to enable sorbent feed through either one or two rows of injectors at one time. The 12 existing B&W circular burners at Edge- water were replaced with B&W XCL burners because of their ability to operate with the desired NOx reduction while fitting within the space con- straints of the Edgewater boiler. Particulate control was a concern at Edgewater because LIMB increases particulate load by a fac- tor of two to three, and changes particulate prop- erties such that the ash is more difficult to collect. The Edgewater electrostatic precipitator (ESP) was retrofit to the unit in 1982 and is substan- tially oversized for its present duty (the specific collection area is approximately 700 ft?/1,000 acfm under normal high load operating condi- tions), sono ESP modifications were made. LIMB ash collected in the ESP will contain approxi- mately 30 percent free lime (CaO) and is highly alkaline. The existing dry ash handling and stor- age system is being used, but a controlled amount of water is added to the ash as it is unloaded from the silo in order to hydrate the free lime for safe handling and disposal in an existing ash landfill. Recent research by EPA and others has led to development of a downstream flue gas humidifi- cation step which has the effect of increasing S02 removal through the utilization of unreacted free lime in the LIMB ash. Humidification therefore appears to offer a significant economic benefit to the LIMB process through incremental sorbent utilization. Because of this technological de- velopment, the demonstration contract was expanded to include the addition of a humidifier at Edgewater to supplement the sorbent injection system. The humidifier consists of a horizontal chamber installed in a bypass duct between the air heater and ESP, together with supporting ductwork, flow devices, water nozzles, and auxil- iary equipment. With the bypass system the flue gas can be diverted back to the main gas duct in case of humidifier malfunction or outage. The humidifier is now being installed and is expected to be operable by June 1988. As will be discussed in a later section, the humidifier will also lower the flue gas temperature and flyash resistivity and thus restore lost efficiency in the down- stream ESP. LIMB demonstration start-up goals During the initial operating period, a number of activities were planned to bring the LIMB unit on line and up to an operating level where continu- ous, full-time operation at a high level of depen- dability was possible. The primary start-up goals were to: 1. Verify the mechanical operation of all process equipment. Particular emphasis was placed on the reliable, steady, and accurate delivery of sorbent to the boiler and on the accuracy of boiler and emission monitoring equipment. 2. Observe and understand the effects of the new technology on full-scale boiler equipment which, to date, had only been observed at smaller scale and predicted for larger scale. Questions of concern included the effects of extremely fine sorbent particles on material handling equipment; the effects of ash modi- fications on boiler tube cleanliness and soot- blowing requirements; the effects of increased amounts of finer ash with higher electrical resistivity on particulate removal equipment; and the effects of altered ash chemical com- position on removal and disposal equipment. 3. Optimize the sorbent injection parameters so that maximum S02 removal at high load could be accomplished. Prior EPA develop- ment work‘ had shown the importance of good sorbent distribution in an optimum temperature region to maximize removal effi- ciency. The Edgewater injection system with three rows of injectors and a range of injec- tion velocity and tilt capabilities was designed to allow fine-tuning of parameters to achieve optimum performance. 4. Determine if a modified sorbent could improve S02 removal performance over that seen with the commerical hydrated lime [Ca(OH)2] chosen for use in the demonstra- tion program. Recent studies had shown that hydrated lime, modified by the addition of a small amount of calcium lignosulfonate dur- ing hydration, improved LIMB S02 removal performance by 8 to 15 percent during injec- tion into a pilot combustor.® Start-up experience and results The demonstration unit started up, as scheduled, in July and operated through September 25, 1987. The early operating period was characterized by intermittent operation due to problems encoun- tered with the sorbent feed system and the neces- sity of incorporating design changes to attain full flow capability. Once this had been achieved, a particulate removal problem was identified and related to the high electrical resistivity of the ash produced with sorbent injection. This problem limited LIMB operation to 3 to 8 hours per injec- tion period and made long-term, continuous operation impossible due to the need to maintain particulate emissions within the opacity limita- tions mandated by the Ohio EPA. Despite these time-consuming setbacks, a series of test runs were conducted over a 2-week period, and valu- able operating information and experience was obtained. The emissions data gathered during this period show that the S02 removal capability of the LIMB technology exceeds the 50 percent removal goal originally set for the demonstration. The start-up experience is described in more detail in the following sections. Operating conditions The coal fired during the 2-week test period was a high sulfur (2.5 to 3.0 percent) bituminous coal mined in Ohio. The sorbent was a calcitic hydrated lime obtained from a commercial vendor; the modified sorbent was prepared by the same vendor by adding 1.5 percent by weight of calcium lignosulfonate to the lime during hydra- tion. Twelve tests were conducted under constant operating conditions at full boiler load. Due to lack of time to optimize the various injection parameters, one injection condition -- use of only the lowest level of injectors at a set booster air flow rate and no injector tilt -- was selected for most of the tests and held constant over the 2-week period. The only exceptions to this were one test equally using the two lower levels and one test with the injection ports tilted down 15° from horizontal. The overall performance of the total system -- boiler, sorbent injection, and emissions analyzers -- is monitored by a computerized data acquisition system (B&W’s Boiler Performance Diagnostics - System 140™). In addition to its regular function of tracking normal boiler parameters, it is also programmed to perform a variety of calculations specific to the technology, such as Ca/S stoichi- ometry and momentum flux ratios, as well as to reduce the data according to specified procedures. The continuous emission monitoring system analyzes several gases in the duct following the ESP. An extractive system is used to sample flue gas, which is then dried and split among the var- ious analyzers which include SO2, NOx, Oz, COz2, CO, and total hydrocarbons. The filters in the sampling system are maintained at high (approx- imately 300°F) temperature to avoid capture of S02 in the sampling system. Since the tests conducted were primarily directed at the S02 removal efficiency of the LIMB system as a function of the Ca/S stoichiometry, the calculations of particular importance are those associated with “inlet S02,” sorbent injec- tion rate, and SOz outlet emissions. During con- tinuous LIMB operation, inlet sulfur to the boiler is calculated by the data acquisition system from the input coal analysis and a series of heat and material balances derived from boiler measure- ments. This inlet sulfur calculation is the basis for the sulfur term in the Ca/S stoichiometry ratio and is compared with the outlet S02, as measured by the continuous emission monitoring system, to calculate removal efficiency after appropriate conversions to a lb/10® Btu basis. This removal efficiency is calculated as a 10-minute average by the data acquisition system. (Stoichiometry and numerous other parameters are likewise calcu- lated as 10-minute averages.) This was considered the only practical method of determining S02 removal efficiency during the demonstration, since continuous measurement of SO2 concentra- tion in the lower furnace is impossible. However, during the 2-week test period described in this paper (when intermittent, short-duration tests were performed), the removal efficiencies reported o S TTTTI 1 tt | ! > T Current Density, uA/ft2 oO TT } J i 4 ° Without Lime - With Lime 0 PA ee 20 24 28 32 36 40 44 Secondary Voltage, kV nD I T | Figure 2 ESP current/voltage relationship with and without lime injection. are further compared to measured S02 emissions before and after sorbent injection. This calcula- tion method was considered to provide an even more accurate assessment of removal efficiency for these tests. Process equipment Start-up was hampered by problems associated with the material handling and feeding equip- ment. The initial maximum feed rate was limited to approximately 4,000 lb/hr of hydrated lime, far below the design feed rate of 18,000 lb/hr. This problem was traced to the inability of a rotary valve to seal against the high back pressure of the transport air into which the lime feeds, causing air to leak back through the feed system and dis- rupt flow. The problem was solved by installing a vent system on the rotary valve to prevent the air leakage and by reducing the back pressure by switching to larger diameter lines. At first the feed rate was found to be somewhat difficult to control. This was primarily due to the flow properties exhibited by the finely sized lime (mass mean diameter of 3 to 5 um) and pressure changes when the silo filled. Consistent feed con- trol for the short-term tests was obtained by utilizing a single feed silo full of sorbent for each test. Rotary valves have since been installed above each feeder as a long-term solution to the problem. Electrostatic precipitator operation Once sorbent feed rates corresponding to a calci- um/sulfur ratio of 2.0 (10,000 to 13,000 lb/hr) were achievable, operation was characterized by steady degradation of the ESP as indicated by an increase in stack opacity. Current/voltage rela- tionships such as that shown in Figure 2 indicated that the high electrical resistivity of the ash was producing back corona in the ESP. The condition was so severe that all attempts to find an effective operating voltage were unsuc- cessful. Measurements of ash resistivity using an in situ point-plane probe showed that sorbent injection resulted in an increase from approxi- mately 3 x 10!° ohm cm to 3 x 10!2 ohm cm. As expected, mass train data showed that the partic- ulate loading almost doubled, while particle size measurements indicated the decrease in size dur- ing sorbent injection. As a result of the decreased ESP performance, continuous sorbent injection was limited to 3 to 8 hours, depending on the feed rate. Typically, injection would be terminated when stack opacity would begin to rise rapidly into the 10 to 15 per- cent range in order to avoid exceeding the opacity standard of 20 percent. Upon termination of sor- bent injection, stack opacity would gradually drift downward, and ESP electrical conditions would return to normal. Approximately 4 to 6 hours were usually required for return to the original condi- tion where another sorbent injection test could be conducted. One method of reducing LIMB ash resistivity and improving ESP performance during lime injection is to condition the flue gas. Past work has shown that both SOs conditioning and humidification are effective means of lowering flyash resistivity. In the case of a LIMB ash, however, S03 reacts with the lime-rich ash, requir- ing a very large excess of S03. A more practical method is water addition to the flue gas to increase humidity and reduce gas temperature, which in turn reduces resistivity to a level (about 1 x 10!! ohm cm) where normal ESP operation is possible. Figure 3 shows how the resistivity of LIMB ash varies with flue gas humidity and temperature.® Low-NOx operation A short series of tests were conducted with the low-NOx XCL burners prior to lime injection in order to allow comparison with data obtained on the circular burners they had replaced in 1986. As can be seen in Figure 4, NOx emissions with the XCL burners achieved the project goal of 0.5 lb/10® Btu over the full operating range of the boiler (for the purposes of the project the maxi- 1016 © 5.0 Volume % H20 15 9.8 Volume % HO 10 © 15.0 Volume % H30 1014 ° o ran co) Resistivity, ohm cm 6 Nn - oO ° oO © we ° 0 3.0 28 26 24 60 84 112 144 182 227 283 352 °C 140 183 233 291 359 441 541 666 °F Temperature 2.2 20 18 1.61000/K Figure 3 LIMB ash resistivity as a function of temperature and humidity. mum continuous rating of the boiler’s main steam flow is 700,000 lb/hr with a peak load of 770,000 lb/hr). This performance represents approxi- mately 40-50 percent reduction when compared to data on the circular burners. Further burner tests are planned after optimization of sorbent injec- tion. The data will be examined for interactions between S02 removal and NOx reduction, though none are expected at this time. Sulfur dioxide removal Because of the impending boiler outage, the test period for determining S02 removal performance was limited to the 2 weeks prior to the outage. The first test week was devoted to operation with a commercial, hydrated calcitic lime sorbent and the second to operation with the lime modified by the addition of calcium lignosulfonate. During this period, one or two tests of 1 to 5 hours dura- tion (depending on stoichiometry) could be con- ducted during a day. Typically, the first few tests of each week were characterized by inconsistent sorbent feed and operational upsets which had an impact on performance, while the tests during the latter part of each week were characterized by FAN en TET }— + Circular Burner +— 2° XCL Burner S © o a NOx Emission, Ib/10® Btu | 400 500 600 700 800 Main Steam Flow, lb/hr x 10°3 Figure 4 NO, emission as a function of boiler load. steady-state operation. Once handling problems were overcome, it appeared that a steady-state condition was reached very quickly after sorbent injection was begun and that this condition was representative of long-term operation. The results reported in this paper focus on the data consid- ered to be most reliable during the 2-week period and do not include transitional periods or times when operational upsets were known to occur. It should be understood that the 2-week test period does not provide a data base developed under rigorous statistical design, nor does it provide a definitive comparison between the normal com- mercial lime and the modified lime. However, the data is considered to be a reliable indicator of the type of results expected when more thorough test- ing resumes following the boiler outage. The overall relationship between S02 removal efficiency and Ca/S molar stoichiometry, as determined by the tests conducted, is shown in Figure 5. In this graph each point represents the average value of all of the 10-minute averages taken during a test with a well-controlled feed rate. Each point represents data taken over a 1 to 5 hour period. The error limits shown for both S02 removal and stoichiometry reflect plus or minus twice the standard deviation calculated from the individual 10-minute averages making up each point. A removal efficiency in the 55 to 60 percent range was obtained at a Ca/S stoichiometry of 2; the project goal of 50 percent removal appears obtainable at a stoichiometry of approximately 1.6. Future optimization of injection parameters is expected to improve the performance. Preliminary statistical analysis suggests that the modified 100 80 port 15° below horizontal 70 60 50 SO, Removal, % 40 30 a ay 20 10 0 0.4 0.8 12 90 OQ Commercial lime; injection at lowest elevation % Commercial lime; injection at two lower elevations © Lignosulfonated lime; injection at lowest elevation 4 Lignosulfonated lime; injection at lowest elevation with ge 16 2.0 24 28 Ca/S Stoichiometry Figure 5 Sulfur dioxide removal efficiency as a function of calcium/sulfur stoichiometry. hydrated lime exhibits slightly better perfor- mance. However, the nature of the tests conducted thus far does not support this as a strong conclu- sion concerning the relative merits of the two. Conclusions Despite its being based only on short-term test data, LIMB technology appears capable of exceeding the SO2 control goal of 50 percent re- moval predicted by earlier research and develop- ment work at smaller scale. The high degree of emission control performance gives very strong impetus to continuing the demonstration program at Edgewater despite the operational problems that were encountered. The 2-month start-up and test period described in this paper provided suffi- cient time to identify problems and determine how they can be overcome through design and opera- tional changes. The most significant operational problem is associated with downstream particulate control. The high resistivity ash produced during sorbent injection significantly reduces Edgewater ESP performance to the point that injection must be terminated after several hours to avoid violation of opacity limitations. Since the Edgewater ESP is unusually large for boilers of this vintage, it is probable that most LIMB retrofits would face the same or worse problems. In light of this difficulty, a flue gas humidification system has been designed and engineered, and is presently being added to the LIMB demonstration unit at Edge- water to serve the dual function of improving both S02 removal and ESP operation. The humidifier is designed to operate at a temperature 20 to 30°F above the adiabatic saturation temperature where S02 removal and good operability are anticipated. Preliminary results from very recent studies sug- gest that humidification to reduce ash resistivity requires only a 100°F approach temperature (i. e., much less water than needed for S02 removal) and, thus, can be accomplised easily in the Edge- water humidifier. With the addition of the humidi- fier, LIMB can be operated in a long-term, con- tinuous mode with modest water conditioning to enable ESP operation, or with additional humidi- fication to enhance S02 capture. * Readers more familiar with the metric sys- tem may use the following conversion factors: From To Multiply By acfm m3/min 0.02832 ft? m2 0.09290 lb/hr kg/h 0.45359 Ib/10® Btu. ng/J 429.94 ton t 0.90718 pA/T2 nA/cm?2 1.07639 oF °C °C=(5/9) (°F - 32) Acknowledgments The authors gratefully acknowledge the support of Ohio Edison, Radian Corporation, and South- ern Research Institute personnel in conducting the tests, operation and maintenance of the con- tinuous emission monitoring system, and particu- late sampling and resistivity measurements, respectively. References 1. J. K. Wagner, et al., “Development of the 1980 NAPAP Emissions Inventory,” EPA-600/7- 86-057a (NTIS PB 88-132121), 1986. 2. G. B. Martin and J. H. Abbott, “EPA’s LIMB R&D Program - Evolution, Status, and Plans,” In Proceedings: First Joint Sympo- sium on Dry SO2 and Simultaneous SO2/NOx Control Technologies, Vol. 1, EPA-600/9-85- 020a (NTIS PB 85-232353), 1985. 3. R. V. Hendriks and P. S. Nolan, “EPA’s LIMB Development and Demonstration Pro- gram,” JAPCA, 36: 432 (1986). 4. B. M. Cetegen, et al., “Effective Mixing Pro- cesses for SOx, Sorbent, and Coal Combus- tion Products,” EPA-600/7-87-013 (NTIS PB 87-188 892/AS), 1987. 5. D. A. Kirchgessner, and J. M. Lorrain, “Lignosulfonate-Modified Calcium Hydrox- ide for Sulfur Dioxide Control,” I&EC Research, 26: 2397 (1987). 6. J. P. Gooch, J. L. DuBard, and R. Beittel, “The Influence of Furnace Sorbent Injection on Precipitator Performance, and Methods of Improving Performance,” presented at the Third International Conference on Electro- static Precipitation, Aborno, Italy, October 1987. Technical Paper BR-1307-A Recovery boiler design considerations J.A. Barsin Manager, Industrial Projects Member ASME, Member TAPPI The Babcock & Wilcox Company 20 S. Van Buren Avenue Barberton, Ohio 44203 Presented to Kraft Operations Seminar Orlando, Florida 1989 Babcock & Wilcox a McDermott company Recovery boiler design considerations J.A. Barsin Manager, Industrial Projects Member ASME, Member TAPPI The Babcock & Wilcox Company 20 S. Van Buren Avenue Barberton, Ohio 44203 Presented to Kraft Operations Seminar Orlando, Florida 1989 Abstract BR-1307-A The fundamental design principles from the first successful Kraft Recovery Furnace developed by Babcock & Wilcox and G.H. Tomlinson, which were commercially applied in 1929, are reviewed with specific emphasis placed upon present day applications and B&W’s development plans for the future. Introduction The first section of this paper gives an overview of the external effects that have influenced the major changes in design to the KRAFT Process Recovery steam generator. The second section dis- cusses the design principles applied to the major components such as, fuel and combustion sys- tems, water/steam circulation system, gas side furnace, superheater and convection pass design. The final section is devoted to future develop- ments presently underway that will be commer- cialized and applied within the next 24 months. Background The Kraft Recovery process evolved initially in Danzig, Germany some 25 years after the soda process was developed in Great Britain in 1853. In 1907 the Kraft process was tried in North America and from its inception, a variety of furnace types, including rotary furnaces and stationary furnaces, all competed for a successful commercial design. During the late 1920’s and early 1930's, significant developments in furnace design were achieved by G.H. Tomlinson, work- ing in conjunction with B&W engineers. The Tomlinson design evolved with a technique of spraying black liquor onto the walls of the furnace. The liquor is dehydrated both in flight and as it builds up on the wall, at which time pyrolysis begins with release of volatile combus- tibles and organically bound sodium and sulfur. The weight of the resulting mass causes it to break off and fall to the hearth where pyrolysis is completed and the char is burned, providing the heat and carbon required in the reduction reac- tion. A smelt consisting primarily of Na,S and Na,COs; is produced and continuously drained off to a dissolving tank. The world’s first Tomlinson Recovery system was installed in 1929 at the Canada Paper Com- pany, Windsor Mills, Quebec plant. This first Black Liquor Recovery Boiler, supplied by B&W Canada, had a water-cooled roof with refractory furnace walls. The early refractory furnaces proved costly to maintain and the amount of steam generated was much less than the amount theoretically possible. Tomlinson decided that the Black Liquor Recovery Furnace should be com- pletely water-cooled with tube sections forming an integral part of the furnace. This new concept boiler was designed in cooperation with The Babcock & Wilcox Company and was installed at Windsor Mills in 1934. This water-cooled type was a complete success and, in fact, this first unit con- tinues in operation today (Figure 1). The first U.S. 28'-8" Figure 1 Windsor Mills Unit, 1929. sale occurred in 1935 to the Southern Kraft Com- pany for its Panama City, Florida, mill, and the first sale external to North America occurred in 1936 to Mo och Domsijo AB Mill in Husums, Sweden. The Kraft Recovery Boiler has a dual mission: 1. Recovery of the sodium and sulfur from the spent pulping liquor in forms suitable for regeneration of the cooking liquor and, 2. Efficient heat recovery from the burning of the liquor to generate steam for process use. Over the years, the B&W designers have been challenged to increase the chemical recovery; decrease the cycle dead loads by increasing the reduction efficiencies, obtaining more chemical in the useful form of Na,S; and to generate more steam at higher pressures and temperatures to permit increased cogeneration of electrical power from black liquor fuels. Design changes Furnace Size The dominant influence on the physical size of the furnace is the black liquor solids processing capability specified by our clients. The unit for Canada Paper Company at Windsor Mills in 1929 was capable of processing 60 air dried tons/day (Figure 1) as compared in scale to our present design, which is capable of processing 1500 B&W Btu tons/day (Figure 2). The need for furnace size increases to process greater quantities has been adequately demonstrated and is accepted by an industry that continues to obtain a return from applying economics of scale principles (1). The larger furnaces (1500 B&W Btu tons/day), re- Figure 2 2000 ton kraft recovery boiler. quired to achieve the dual missions, challenge the combustion system designer to obtain air/fuel mixing in a furnace now 38 feet wide by 37-1/2 feet deep (11.55m x 11.4m) and 162 feet (49.25m) high as compared to a 20 ton/day unit, 5-1/2 feet (1.67m) wide by 6-3/4 feet (2.05m) deep and 29 feet (8.82m) high. The oxygen level must be controlled to provide an overall hearth zone stoichiometry well below that required for complete combustion and to ensure that the reduction of Na,SO, can be maximized. Furnace Shape The question of having two distinct combustion zones, isolated one from the other, such as the NSP (2) is not necessary to achieve predicted per- cent reduction, percent chemical recovery or carbon utilization performance. The present rec- tangular shape with separate combustion zones created by segmented air zones is cost effective and does provide the desired performance. How- ever, it is recognized that as the unit tons pro- ==] iy HALA Figure 3 CCZ process recovery. cessed per day increase, the difficulty of main- taining good air/combustible mixing at an acceptable power penalty, will increase. A concept that is currently undergoing flow model testing would utilize lower furnace arches to aid in (1) containing the reducing zone, (2) improving the secondary and tertiary air penetration of the flame core, (3) reducing the possibility of black liquor droplet carryover (entrainment), (4) increas- ing the lower furnace temperatures, and (5) main- taining the consistently high reduction efficien- cies traditionally obtained. This concept has been successfully applied to our high moisture bark- fired power boilers and offers an alternate cost- effective way for larger units to achieve the goals without utilizing a totally separate furnace (Figure 3). Corrosion Protection The application of pin studs in the recovery boiler reducing zone does protect that zone from corro- sion as long as the studs are maintained. An application history is available (3). As operating pressures were increased to levels above 1,000 psig (6.89 Mpa), corresponding to a furnace tube outside metal temperature of 600°F (316°C)), fur- nace tubing corrosion became a problem. Conse- quently, pin stud size, application densities, and application zones are now all designed to minim- ize corrosion. Composite tubing, also referred to as bimetallic or duplex, provides an alternative, with a eight-year history in the U.S.A. That expe- rience indicates the bi-metallic approach is an alternative to the pin studs and does provide cor- rosion protection and reduced maintenance (no pin studs). The furnace construction has evolved over the years from refractory only, initially, to tube and tile construction using 3-1/4 inch (8.26cm) tubes on 6-inch (15.24cm) centers with flat studs, to today’s 2-31/32 inch (7.62cm) tubes on 4-inch (10.16cm) centers with the membrane furnace. Air infiltration is greatly decreased, permitting air to be placed where it is required, and total air levels - reduced to more closely parallel fuel requirements. Field evaluations are now complete and confirm design predictions, and insure that this construc- tion does provide adequate cooling margins at 1500 psig (10.34Mpa) and above operating pres- sures, ensuring that membrane fin temperatures are below those associated with high corrosion potentials. Environmental Effect A second major impact upon the size and design of a Recovery Boiler installation is the effect of environmental and economical forces applied dur- ing the past 15 years. The direct contact evapora- tor was applied to evaporate water from the black liquor by utilizing the latent heat in the flue gases. The stripping of Total Reduced Sulfur (TRS) from the liquor in a direct contact evapora- tor is now an environmental problem due to the odor released into the atmosphere. Additionally, it was desirable to improve the thermal efficiency of the boiler by generating more steam and utilizing exhaust (waste) steam to concentrate the black liquor. These desires led to the development of the “Low Odor Recovery Boiler.” The “Low Odor Rec- overy Boiler” was first applied commercially in 1969 at American Can’s Halsey, Oregon, Plant, now owned by Pope & Talbot at the James River mill, which utilizes a cross flow economizer as the last heat trap to maximize steam generation. The economizer is a large structure and has, by its addition, radically changed the side sectional ele- vation picture of a Recovery Boiler. Competitive Pressures Finally, competitive pressures play a major role in the shaping of a Recovery Boiler. Active invest- ment in development by several suppliers and many users ensures that the technology will pro- gress; and that is healthy. In the electric power industry during the 1960’s, power plants were pur- chased strictly on first-cost bases and designers were forced to remove conservatism to stay com- petitive. Specifications were met but margins were removed if client acknowledgement and evaluation of the “conservatism” could not be obtained. A similar phenomenon has been occur- ring in the pulp and paper industry since the late 1970’s. Competitive pressures are forcing optim- ized designs with diminished conservatism since the conservation cannot obtain a dollar evalua- tion. Unit solids processing capability at 150 per- cent of the nameplate Maximum Continuous Rat- ing were not uncommon in the past, in fact overloads have been the expected norm for the majority of our units placed in service since 1931. Today’s units require improvements in operation to minimize fume carryover, and improvements in combustion, and in hearth temperatures, in order to obtain capacity additions over nameplate, at least until circulation limitations are reached. Specific design principles Fuel The prime factor which affects all recovery perfor- mance parameters is the specific black liquor being fired and the large deviations which may occur with this fuel. Black liquor is a unique fuel in that no other fuel can match its ash content of approximately 25-35 as-fired percent and its mois- ture, of 30 to 40 percent. The fuel varies with the type of wood being pulped, whether carried by salt water or fresh, if oxidized or not and, as a result, demonstrates variable solids, viscosity, pH, organics, and various chemicals relating to the tightness of the cycle. Both in the wood and the chemical make-up, all of the following can ride along and wind up in the black liquor: potas- sium, aluminum, iron, silicon, manganese, mag- nesium and phosphorous (Figure 4). Our practice in designing a specific unit is to calculate a Kraft Recovery unit’s air and gas weight on the basis of an elemental analysis and gross heating value furnished by the client. When no analysis is provided by the client, we utilize a liquor analysis which corresponds to the mill’s geographical location and the wood species being pulped at the mill in question. This analysis is reviewed with the client and agreement obtained to use the analysis as the basis for the design. This analysis is “salt cake free” with the Na,SO, makeup and precipitator hopper recycle excluded. The design analysis heating value (Bomb Calorimeter) is then corrected as required for: effect of oxidation and heat of re-action to arrive Liquor Components Hard Soft Carbon 39 426 Cc Hydrogen 34 3.6 Hp Oxygen 31.6 317 0, Sulfur 43 3.6 Ss Sodium 215 18.3 Na Total 99.8 99.2 Heating Value (Btu/Ib)} 6200 6600 Solids 60.7 63.0 H,0 39.3 37 Btu/Ib 3850 4160 Figure 4 Analysis of black liquor - hardwood/softwood. at the heat of combustion, resulting in the burn- ing of black liquor in a furnace. This is referred to as the heat available (HA). The heat available is then corrected for losses in the process, such as: e The latent heat required to evaporate water formed from combining H, in solids. e The latent heat required to evaporate water added to the black liquor. e One-half the calculated boiler radiation losses. e Heat required to reduce the salt cake feed to Na,S. e Heat lost in smelt and the heat of fusing smelt. e Heat of reduction of the sulfur in the liquor. e Heat absorbed in the furnace. The application of these corrections generates the heat available for superheater absorption. Furnace absorptions are affected by gas tempera- tures, soot formation, percent solids fired, wall cleanliness, droplet carryover, excess air and sensible heats. The black liquor viscosity is an important vari- able as it affects droplet size and, therefore, rate of dehydration. Viscosity is affected by the solids concentration, temperature, pH and the organic- to-inorganic ratio (4). Solids concentration and temperature are the variables that we fix when designing the liquor feed system - these are the dominant variables, but pH has been identified as an important factor in some circumstances. Extensive experience obtained on the many Kraft liquors utilized have provided the design flexibility which permits fairly wide variations in the fuel properties to be processed, while still meeting the specified performance. The capacity of a recovery unit should be based upon its ability to burn completely, in 24 hours, the dry solids contained in the liquor recovered in the pulp produced in 24 hours. B&W measures the capacity of a recovery unit as we do with all boil- ers of our design, i.e. using heat input to the fur- nace rather than a solids processing capability. B&W has established as a unit of capacity a heat input of 19,800,000 Btu (20,890MJ) in 24 hours. This unit, known as the B&W Btu ton, corre- sponds to the heat input from 3,000 pounds (1,362Kg) of solids (approximately equivalent to one ton of pulp produced) having a heating value of 6,600 Btu/lb (15,337KJ/Kg) of solids. B&W’s Btu tons are constant and the tons per day of capacity specified by clients are converted by us to B&W Btu tons. This is achieved by correcting for solids recovery lb/ton of pulp (A), heating value of solids Btu/lb (B), and pulp output of mill, tons/24 hr (C). Combustion System Preparation The low Btu, high ash and high moisture Kraft black liquor fuel, plus the heat absorbing chemi- cal reaction involved, requires the best prepara- tion economically possible. Direct contact cascade evaporators (flue gas utilized to evaporate mois- ture) were capable of increasing solids to about 63 percent and direct contact steam heaters could heat liquor to about 230°F before solids dilation and steam hammer combined to force a practical limit. Presently, technology has advanced and permitted “as-fired” solids to gradually increase to a generally available 70 percent solids (with falling film concentrators) and to “as fired” temperatures of up to 290°F (indirect falling film heaters). These improvements in fuel preparation permit alternative firing techniques to be applied. Delivery System The fuel is sprayed on the walls by means of a splash plate nozzle and oscillator. Formerly, with the low solids, direct black liquor heaters and low firing temperature, it was absolutely critical to dehydrate the liquor in flight and on the walls, but not on the bed, to aid in maximizing the bed temperature. Not all the liquor droplets end their flight on the walls; some fall directly to the bed and some are entrained in the flue gases rising in the furnace. It is the designer’s intention that the dehydration and pyrolysis occur in the lower fur- nace under reducing conditions. The control of droplet size is not as critical using wall drying as it has proven to be with total bed-only dehydra- tion, i.e. “stationary firing.” If the black liquor NTI NN Spray Ahh \ H/,/; Pattern ui ! iy SS Top View Oval Splash Plate Figure 5 Typical black liquor spray nozzle. viscosity was controlled, the droplet size could be controlled and the number of droplets entrained could be reduced. Herngren, T., et al. (5) have presented a control system and an online mea- surement and control of droplet index (DIX) that has demonstrated performance improvements related to droplet size variation on a recovery unit in Monsteras, Sweden. Oscillator burners, located in the center of the front and rear walls (and on all four walls of larger units) emit from the spray nozzle, Figure 5, a flat sheet spray of coarse droplets. The nozzles are continuously both rotated and oscillated to cover a wide band on all walls above the hearth. The degree of movement is adjustable to cover a greater or smaller area of wall surface, as required to compensate for variation in the solids concentration of the atomized liquor and the consequent residence time that is needed for dehydration. The temperature, pressure, viscosity and solids content of atomized liquor are important to Kraft Recovery Furnace operation. It is desirable to create a large droplet of atomized liquor. This maximizes the amount of liquor sprayed on the wall to minimize entrainment of liquor particles and mechanical entrainment of sodium chemicals in the combustion gases passing to heat absorb- ing surfaces. A liquor temperature of approxi- mately 240°F (115°C) and pressure of about 25 psig (172Kpa), with solids as high as practical to 70 percent, would aid in maximizing lower fur- nace temperatures. The object continues to be the largest droplet, major spray pattern on wall for dehydration with some liquor at the high solids, and high temperature directly on the bed. Units, having a capacity from 500 to 1000 B&W tons, usually utilized two spray nozzles, one in each the front and rear wall; whereas, smaller capacity recovery units have a single oscillating burner in the center of the front wall distributing liquor on the side and rear walls. Units larger than 1000 tons have four oscillating burners, one in the center of each of the four walls. Consider- able variation in the quantity of liquor introduced through a nozzle location is accomplished by vary ing nozzle sizes. Currently, many of our larger units are successfully combining wall spraying with inflight drying and bed spraying (station- ary). Additional nozzles have been provided and a common arrangement on several 800 ton units is to have two nozzles on each wall providing par- tial wall spray, and partial bed spray. Evalua- tions of stationary firing indicate it to be a good tool if high solids, high temperature and, most critical, the proper spray angles are utilized. Wall Dehydration and Pyrolysis The liquor that does land on the walls (the major portion) is dehydrated, removing heat from the fireball. When dehydration is complete, the drop- let temperature increases and some volatile organics evolve. These become part of the gas stream and are ignited and burned. The particle volume increases during both the dehydration and pyrolysis phases (drastically). The tempera- ture further increases and pyrolysis reactions occur producing volatile combustibles from com- plex organic compounds. The pyrolysis phase is important because it is a major factor in the release of sulfur gases; generates volatile com- bustibles; and provides the fixed carbon and hy- drogen that is burned off. Hearth Eventually, the mass of the built-up char on the walls causes it to break off and fall to the furnace hearth in a highly reactive state, joining the drop- lets that had dropped directly upon the bed. As particle volume was increased during dehydration and pyrolysis, porosity increased due to the vola- tile evolution. This porosity greatly increases the total surface--internal and external--of the parti- cle, and thereby, increases reactivity. During the char-burning phase, some of the carbon is utilized in the reduction reaction and the excess gasified, forming volatile combustible compounds such as CO. The reduction of Na,SO, to Na2S occurs on the char bed and the atmosphere is controlled to maintain a reducing condition. High temperatures are important to the rate of char burning and it is desirable to have only fixed carbon and inorganics landing on the char bed. The fuel, carbon, is on and in the bed but in a highly porous state and air is discretely directed at it using relatively low head (4 inch H,0 100mm ). If the porosity of the bed could be further increased, more carbon and hydrogen could be burned in the bed. The oxygen concentrations in the gas boundary layer next to smelt in the char bed must be high enough to burn the carbon and low enough to provide an adequate atmosphere for reducing Na,SO, to Na,S. Experimental evi- dence indicates that additional primary air would be beneficial to increasing bed temperatures as long as the amount of oxygen supplied does not exceed the requirements for burning the amount of char landing on the bed (6). The references further indicate that if carbon is in the smelt, reduction will be maintained even in an oxidizing atmosphere such as that of the smelt spout. Suffi- cient heat is liberated in the hearth to support the reduction reaction and generate gases hot enough to aid in the dehydration phase. The floor of the hearth is sloped to provide posi- tive and continuous drainage of smelt from the char bed and minimize the inventory of smelt within the furnace. Molten smelt is drained from the furnace through smelt spout openings located in the lowest part of the floor to assure positive smelt removal. A removable water-cooled smelt spout assembly designed to match the “V” shaped wall opening, is externally mounted over the opening and inhibits air infiltration into the furnace. The tight-fitting smelt spout is designed to prevent smelt from infiltrating in the dead space between casing and tube wall, which should reduce maintenance in this normally high main- tenance area. Air system Each air flow stream: primary, secondary and tertiary, is metered individually and can be sepa- rately controlled to ensure that the furnace is optimized for the fuel actually being processed. Steam coils provide preheat to both the primary and secondary air streams. Popular cycle pres- sures and maintenance concerns, in the past, have kept steam coils at fairly low pressures which have limited resulting air temperatures to about 320°F (160°C). Air temperatures up to 400°F (204°C) have been utilized when poorer quality fuel is specified. Temperatures in the lower furnace could be enhanced by increasing combustion air temperatures which would aid dehydration and reduce bulk gas temperature losses caused when the combustion air is brought up to gas temperature in the furnace. Preheat of 450°F would be helpful and if cycle steam or water is available at that temperature, it should be applied. Water coil air preheaters utilizing recycled economizer feedwater have been benefi- cial in permitting higher air temperatures while reducing feed water temperatures to the drum - a positive benefit on some units where it is desir- able to increase the saturated water head. Distri- bution, penetration and mixing of combustion air design philosophies are under continuous review to insure they match the demands of greater solids throughput capability, improved fuel preparation and maximized reduction while min- imizing TRS emissions. Primary Air The reducing condition is maintained by introduc- ing up to 50 percent of the total air (including excess) required for complete combustion through the primary air ports. Primary air quantities can be varied from 30 percent to 50 percent of the total air to permit optimization for the actual fuel fired and the actual combination of delivery utilized-- total oscillatory; combination or total stationary-- recognizing that each system/fuel will require dif- ferent amounts of primary. The primary air ports’ size and location are critical to attaining the high reduction percentages we have successfully dem- onstrated. They are rectangular in shape, sized 2 inches wide by 12-3/8 inches high (5x8.6cm). They are located on all four walls surrounding the periphery of the char bed and are vertically located 0.9m above the furnace floor following the smelt drainage slope. The windbox supplying the primary air is sized to obtain excellent distribu- tion around the full furnace periphery and each set of primary air ports is equipped with a damper to permit group adjustment of air flow. The port is pointed downward and imparts a downward deflection to the air stream. The air interacts with the carbon at the outer periphery of the bed with a maximum penetration of approximately five cen- timeters. The depth of bed is variable, but the active smelting region is on the top. Furnace imaging systems indicate that the thickness of the hot part of the bed is less than 5cm which, in turn, indicates that the active layer may be less than 5cm thick (7). The primary air provides the oxygen for the carbon burnout, which provides the heat for the reduction reaction to proceed. The multitude of small ports producing discrete air streams is similar to our successful approach for burning heavy fuel oils. The fuel, in that case, is atomized into discrete streams of droplets to ena- ble the oxygen to surround the individual droplet and permit combustion. The primary ports are subjected to plugging and, once plugged with smelt, the ideal air distri- bution is lost. This results in loss of activity in that area and a cooling of the bed. In addition, the restriction on the air path changes distribu- tion all around the primary zone which gets worse as more ports plug. Port plugging is a fact of life and it, as well as the hand rodding required to clear it, is disruptive to the bed, steady state reduction and oxidation. Steady state conditions and bed stability have been greatly improved by the retrofit of automatic port rodders. They have now been retrofitted to six units in the U.S.A. with measurable benefits. We offer this improve- ment on all our new and upgraded recovery unit proposals because they do an excellent job. Secondary Air Located one meter above the highest primary air ports are the secondary air ports providing up to 50 percent of the total air required for combustion. They are located there to control the top of the char bed as the momentum is sufficient to provide turbulence across the top. These ports are evenly spaced around the periphery on all walls and pro- vide the air required to burn the volatiles to pro- vide the heat for aiding in the dehydration of the ee Port Opening Figure 6 Variable port damper. droplets. The secondary air ports are larger than the primary ports and are 2x14 inches (5x35.5cm) on the 500 ton and larger units, with sizing dependent upon obtaining the desired velocity. The static pressure available is double that util- ized for the primary system and each port is fitted with its own adjustable damper control (Figure 6). Adjustable dampers permit port velocities to be adjusted to match actual fuel/bed requirements. The full penetration of these secondary air jets into the bulk furnace gas is critical to completing burnout low in the furnace; to break up the com- bustion gas cone; to assure mixing and combus- tion of the air with the volatile gases rising from the char bed, and to utilize the furnace effectively. Tertiary Air Located above the black liquor spray nozzles are the tertiary air ports, which provide up to 30 per- cent of the total air required for combustion. They are located on only two opposing walls either front/rear or side/side. The sizing varies with sol- ids processing capability, the largest presently is 5x31 inches (12.7x79cm) at 1,500 tons, but the siz- ing is dependent upon obtaining the required exit velocity. Individual dampers at each port provide velocity control and the static pressure is more than double that available to the primary system. The momentum developed at this level is also crit- ical to complete combustion of the volatiles, to complete the break up of the flame cone, and insure oxygen availability throughout the upper furnace to eliminate TRS. Overload Experimental data taken from units operating at 115 percent MCR and above indicate that many original equipment, secondary and tertiary air systems, are not capable of being utilized fully due to lack of sufficient fan static pressures and flows. Probing of furnaces indicated that gas recirculation patterns were present; therefore, the full furnace area was not effectively utilized for gas flow. This resulted in increasing peak gas velocities, which in turn increased solid particle entrainment. It has been determined that the flue gas coming off these overloaded beds forms a column in the center of the furnace which inhibits tertiary air penetration. The latest air system review has been initiated and has taken the form of creating 1/8 scale isothermal three-dimensional models of actual units which have permitted the establishment of a verifiable baseline. A 1000-ton B&W-design unit, scheduled for a low odor conversion, provided the first uprate modeling opportunity. It was decided to build a 1/8 scale model of this unit and isothermally test it for various air port configurations, both tertiary and secondary. The starting point or baseline was the existing large (3x30-inch) tertiary air ports originally supplied on this 1968 design. The ter- tiary air system was common with the secondary system on this unit. Therefore, the tertiary system needs could be served only after the secondary was satisfied. Baseline data indicated that obtain- ing a positive static pressure reading in the ter- tiary windbox was difficult at the overload condi- tion. Load was limited by I.D. fan capacity to 123 percent of MCR. The isothermal model provided possibilities for visual evaluations using smoke patterns, styrofoam pellets, neutral buoyancy bubbles, as well as hot wire anemometer velocity traverses. Video taping provided smoke flow records while the velocity data generated quantifi- able readouts. Our objective was to find the best way to attain furnace gas column penetration, reduce furnace gas recirculation and maximize the positive use of the furnace area; i.e., reduce spot peak velocities. The existing configuration, resulting furnace flue gas pattern, and velocity data from two planes - one below the tertiary ports and one taken above the tertiary ports- are shown in Fig- ure 7. Many variations and iterations were run and the “as recommended” configuration, with resulting flue gas patterns in the furnace and velocity traverses, are shown in Figure 8. The “as recommended” configuration resulted in reducing gas peak velocities from +1200 fpm to less than 450 fpm - a 70 percent decrease in peak furnace flue gas velocities. Once the optimum configuration was deter- mined, the nozzle itself was modeled full scale in our Research and Development Center. Ten retro- fits in the past 24 months have confirmed the air flow model development indications. Three are briefly described: A 400-ton retrofit upgrade to 560 tons (140 percent of nameplate) staying with oscil- lating firing has demonstrated the additional pro- cessing of solids capability and an increased oper- ational period. The retrofit of a high static tertiary air system, separated from the secondary, permitted the additional processing capability. A second, of 900 tons, with oscillating firing has demonstrated less superheater deposition as manifested by decreasing the rate of superheater final temperature decay from 7°F/day pre- [Rear Wait Port Right Side View Front View Front Wall Port yy yy ru Rear ‘Traverse Plane 1 Primary Ports 7 ae t rt) ‘ Pee Ports va ts . yf) Secondary Zh scot su fleece EA eu Ports: if atl Ports ii Figure 8 1100 ton as will be. upgrade to less than 1°F/day post-upgrade. The third, a 250 ton upgrade converted from 2-level tangential air firing to 3-level wall air with oscil- lator, has demonstrated an operational period more than tripled at solids processing levels, and steam production higher than pre-upgrade with TRS reduced at the boiler outlet from 700 ppm to less than 5 ppm (8). Furnace construction Floor and wall construction of furnace: The lower half of the furnace, Figure 9, is operated as a chemical retort. Partial combustion of the char, in a reducing atmosphere at the surface of the po- rous bed in the furnace, releases carbon monoxide and elemental carbon which act as reducing agents to convert the sulfate in the smelt to sul- fide. The heat evolved melts the inorganic sodium compounds of the smelt. The molten smelt filters Figure 9 Floor and wall construction of furnace from steam. through the char bed to the sloped furnace floor, and drains to spouts, from which smelt flows into the dissolving tank. To withstand the erosive and penetrating char- acteristics of the smelt, furnace walls and floor are constructed to assure tightness. The surface of the floor and wall tubes must also be protected against the corrosive potential of the smelt in the reducing atmosphere of the hearth. Chilled smelt acts as a refractory, cooled and retained by pin studs on the water-cooled tubes, and prevents molten smelt from contacting the tube metal. Pin stud construction is used for the floor and walls in the retort zone up to the tertiary air ports. The length of the studs and thickness of refractory are limited to that which can be effectively cooled. A positive membrane barrier at the centerline of the tube prevents gas side constituents from penetrat- ing beyond that point. The pin studded surface, even when covered by char, continues to permit some heat transfer due to the conductivity of the pin studs. Furnace wall tubes above the pin-stud zone are bare, with membrane closing the space between tubes. The construction provides a gas-tight, fully water-cooled metallic surface, forming a barrier to furnace combustion products and air infiltration. The composite tube furnace bottom provides a good alternative for lower furnace corrosion pro- tection at higher operating pressures. The corro- sion rate on densely studded carbon steel tubing is acceptable at fireside skin temperatures up to 600°F. However, metal temperatures in excess of 600°F, created by internal operational tempera- tures related to the pressure and waterside depos- its, have demonstrated accelerated corrosion with specific liquors and internal deposition condi- tions, or if studs are not maintained properly. The composite tube comprised of a nominal .065 inch clad thickness of Type 304 or 304L stainless and an inner pressure bearing tube of SA-210 Al has been found to be an effective corrosion deterrent. The additional capital costs for composite tubing are in most cases offset by the decreased stud maintenance costs when operating at tube wall temperatures in excess of 650°F. B&W’s first composite furnace bottom was sold in 1979 and went into service in 1980. Presently, we have 11 in service throughout the United States and Canada with operating times ranging from 10 months to six years. Seven of the 11 units operate at 1000 psig (6890 kPa) or lower, and four operate above 1000 psig. During a routine outage inspection in early 1985, localized corrosion of the stainless cladding was observed on the composite tubing in the vicinity of the primary air ports. Subsequent inspections revealed similar corrosion in both secondary and tertiary air ports as well as the ports for burners, oscillators and observation doors. Depending on port construction, each of these areas is not necessarily entirely affected in every unit. No evidence of stainless layer corro- sion has been associated with smelt spout open- ings. The affected surfaces are generally asso- ciated with crevices created by stud plates and/or port castings and locations where gastight construction is not maintained at the furnace wall centerline. Membrane Figure 10 Secondary port design (old and new). 10 Eliminating crevices and non-gastight con- struction at the furnace wall centerline eliminates caustic corrosion of the stainless cladding. Inspec- tions of modified secondary air port openings that have eliminated port stud plates have shown no corrosion after 28 months in a 1500 psig unit (9) (10). The construction of the original and updated secondary air port design is illustrated in Figure 10. Flow modeling indicates an improvement in nozzle action, and, therefore, this design will be applied to all retrofits and is available for carbon steel upgrades to eliminate those maintenance costs associated with stud plate repair. Smelt spouts and smelt openings Smelt spouts and openings are subject to the most severe and aggressive duty of all the many boiler components: Direct contact with 1700°F corrosive smelt, furnace radiation and flame impingement created by oxygen infiltration, impact loading during rodding, and major thermal variations in small spans combine to stretch the design chal- lenge. B&W has established design goals of a) obtaining a 5-year life in openings, and b) obtain- ing a 2-year life in spouts for excessive duty units. It is realized that many recovery units already attain and exceed these goals, but if we design for the worst case, i.e., those now having a present life of less than 12 months because of overloads or super aggressive smelts, there will be an im- proved design and longer life for all users. Openings Instrumented carbon steel and composite open- ings were field tested and provided good actual temperature measurements for use in the analyti- cal finite element model constructed to better identify critical areas. Actual temperatures were found to be lower than expected, but significant differences were noted among various units, and attempts have been made to quantify the influen- ces upon spout and opening life. The major effect other than design and fabrication appears to be smelt loadings rather than chemistry. Field exper- iments currently underway indicate that the open- ings being evaluated will meet the design objec- tives for five years life in the most aggressive smelts. Spouts Instrumented spouts with upgraded cooling capa- bility are presently in service and providing base- line data for that finite modeling effort. As an alternative, improved ceramic spouts, initially to be water-cooled and ultimately to operate uncooled, were installed in April 1987 at two test sites (water cooled), and July 1987 at one site (uncooled). The ceramic spout program continues in operation at this writing. Water Circulation The natural circulation recovery boiler is de- signed with circulation adequate to at least 110 percent MCR when firing oil. The auxiliary fuels such as oil and gas have the highest spot heat absorptions and, if a circuit is limiting under that extreme condition, it would be modified. The unit is rechecked at MCR on black liquor to ensure that all circuits meet our criteria. Gas temperatures leaving the boiler on a new unit are limited by the need to maintain sufficient saturated water head to ensure circulating in the boiler bank, and provide water to the supply circuits. The large steam drum is fitted with suffi- cient steam cyclone separators, to ensure mini- mum liquid droplet entrainment in the steam to the superheater. Our experience permits cal- culating superheater pressure drop with good accuracy. In the retrofit case, where solids capacity is increased, all the water circuits, steam separating capacity and superheater pressure drop are poten- tial choke points and circulation must be rechecked. Heat input limits for black liquor have been empirically established over the years and for new units relate to: Plan Area Index: Hearth heat input - 900,000 Btu/ft2 (10.221 MJ/M2) of furnace floor plan. Average Tube Absorption: Average heat flux (studded tube zone) 40 to 60,000 Btu/ft? (454 to 681 MJ/M2) of wall flat projected surface. As heat inputs are increased, the bed tempera- ture increases, the heat flux will increase and the air flow splits will have to be adjusted to contain the fireball. As solids are increased above the typical 3500 Ibs dry solids/day ft? (17,100 Kg/M2), and exceed the plan area index, the potential for increased carryover does exist if the air system or firing system is not optimized. If not optimized, particulate carryover will probably increase and establish the next choke point which usually is high temperature surface deposition. 11 Convection pass and deposition The furnace volume per unit input is a measure of residence time and is calculated as the cubic feet of volume up to the level of the furnace arch. Another empirical index in common use is the volume index composed of furnace volume (ft?) per B&W Btu tons/day (TPD). Values are estab- lished and range from 65 ft?/TPD (9,940 M3/B&W J TPD) suitable for an 800 TPD unit to 80 ft?/TPD (2,388M3/B&W J TPD) suitable for a 1200 TPD unit. Exceeding these indexes will increase solids carryover, which will accelerate plugging of con- vection surfaces if all other conditions are not optimized (hot bed, air mixing, etc). Staying within the index ensures that sufficient residence time is available for combustion, and sufficient heat transfer surface is available to cool the gases to below 1700°F (927°C) entering the superheater. The furnace arch baffle serves two important functions. First, the nose baffle shields most of the superheater from the radiant heat of the active burning zone of the furnace; the high- temperature steam loops at the rear of the super- heater are completely protected. Secondly, pene- tration of the nose into the furnace distributes the gas to enter the furnace screen (if used) and super- heater at a uniform temperature and velocity. An eddy above and behind the tip of the nose causes the gas to recirculate in the superheater tube bank, preventing hot gas from bypassing super- heater surfaces. Convection Heating Surfaces In the recovery unit shown in Figure 2, the flue gases leaving the furnace are not cooled by a furnace slag screen. In our newest units, the re- duction in FEGT required to minimize deposition is obtained by increasing furnace volume. The drive to remove screen stems from the experiences of several users where screen tube failures did or could have resulted in a smelt water catastrophic reaction. A properly designed screen continues to be a good element of conservatism in aiding the control of superheater fouling. When a screen is provided or exists, the extent of screen surface is determined in part by the quantity of heat required to be absorbed in the superheater. Where high final steam temperatures are required, the screen must be reduced so as to provide suf- ficient heat. The superheater is arranged for parallel flow of gas and steam whenever permitted by the final temperatures specified. Saturated steam enters the front tubes of the superheater in contact with the hot gas and flows through successive loops, so that the final tube with the hottest steam is in contact with the cooler gas. There is a dual advantage with this arrangement: first, cooling of the gas is most rapid at the front of the super- heater, where the need for cooling of the ash is greatest; second, the parallel-flow arrangement results in a lower average tube metal temperature at both the high-steam temperature and high-gas- temperature end of the superheater. Careful coor- dination of alloys for superheater tubes permits final steam temperatures up to 950°F (510°C). Fireside deposits are a variable mixture of inor- ganic components. As a result, they melt over a broad temperature range (11). Eutectic points (first melting point of a mixture) of a solid mix- ture containing Na,SO,, NasCO; and NaCL decreased from 620°C to about 520°C as the per- centage of NaCL was increased. As the NaCL, potassium and sulfide contents in the deposit increase, the eutectic point decreases. The exam- ple is still above the Top temperature of a clean superheater tube and the mixture should remain a solid. However, if deposits are already on the tube, the depth might be great enough to have the deposit surface at or above 520°C, which would permit melting to occur, stickiness to be created, and additional deposit added, ad infinitum. The designers’ defense is to limit gas temperatures entering the superheater, reduce particulate carry- over, widely space the superheater, and apply sootblowers on conservative centers. In the past, superheaters were designed on 0.15m (6 inch) side spacing. At present, we are supplying new units on 0.3m (12 inch) side spac- ing. When potassium or chlorine are in the black liquor in a total greater than one percent, the drastic effect upon the eutectic points is recog- nized by limiting the gas temperature to approxi- mately 1150°F (621°C) entering the 5-inch (12cm) boiler bank side spaced tubes. The boiler bank spacing can be increased to 10 inch (25cm) side spacing to relieve that choke point if firing/air adjustments cannot control fume carryover. On new single drum units, the modular generating bank surface has been opened up to 8-inch (20 cm) side spacing. The improvements in controlling solid particulate carryover and its beneficial effect upon superheater fouling would indicate that additional conservatism, with the air flow upgrades, is now possible. As the desire for improved cogeneration heat rates gets stronger, the recovery unit will be 12 pushed towards higher pressure cycles and higher temperatures. This progression ended in the elec- tric utility area about 15 years ago with the 3500 psig (24.11Mpa), 1000°F (538°C), 1050°F (566°C), 1050°F cycle. It became necessary for designs to go to wider and wider convection side spacing to permit continuous operation on severe slagging coals. Pendant superheaters are presently on 1.5m side spacing and finishing superheaters are on 1.0 meter side spacing centers. This successful stra- tegy applied to utility units inhibited the deposits’ ability to bridge as they would fail in shear and return to the ash hopper once grown beyond their ability to support. We have proposed and sold superheaters on 18-inch (46cm) side spacing on retrofits and believe that it is an effective way to increase solids throughput capability by moving the deposition choke point out further along the increased solids processing capability load line. Our work in micronized coal has confirmed the importance of low gas velocities to minimize impact deposition. Gas velocities are designed to be low at design MCR; a limit of 25 fps (7.6m/s) is utilized in the convection pass’ tightest flow zone. Another requirement met to extend superheater life is to structurally tie the superheater sections together to maintain alignment while leaving enough flexibility to allow for differential expan- sion of the sections. This is accomplished with a unique arrangement of front-to-rear and side-to- side ties utilizing both D-link and TG-link ties (Figure 11). The additional benefit is reduced side to side swing which reduces the possibility of pre- mature failures due to fatigue near the roof pene- trations, but not total elimination as swing helps to shed slag. 12” 12" 12" 12" 4 L}\ | —— g 1 | or lt AO? Front View Side to Side Ties Side to Side Superheater Tie Tubes Bent Out of Line to Provide Rigid Tie and Maintain Wide Tube Spacing. Front to Rear Superheater Tie Plan super i Figure 11 Superheater structural ties. The boiler section is of single-pass design with- out gas baffling, thus providing maximum cleanability of boiler surfaces. The boiler bank is provided with two cavities through which sootblowing and maintenance access are accom- plished. An additional design feature is that the furnace rear wall tubes are carried into the steam drum instead of the lower drum, forming a screen section in front of the boiler bank. As units get older and generating tubes are replaced, it gets more difficult to attain a tightly rolled seal at the drum. A solution is seal welding, but to be effec- tive, all the tubes should be seal welded at the same time. An alternative design - a single steam drum with all welded supplies and risers has been applied successfully to several new process recov- ery steam generators and several hundred power steam generators. This construction eliminates all expanded (rolled) tubes and is available for new units. A boiler hopper is located below the lower drum providing positive removal of ash from all areas of the boiler bank. Two large downcomers supply water from the lower drum to the lower furnace circuits. The emergency shutdown rapid drain connections are located in these downcomers for fast, positive drainage to recommended water lev- els during emergency situations. Economizer The economizer used is either the vertical bare- tube type, generally baffled to establish crossflow of gases or the long flow with no internal baffles and finned tubes (Figure 12). King and Blackwell indicate that in boiler banks and economizers, the dominant mechanism for deposit formation is the diffusion-driven deposition of chemical fume par- ticles (major) and condensed vapors (minor). The deposits are enriched with potassium and chlo- rides. They expect the enrichment of NaCL in deposits to increase as flue gas temperatures decrease. Deposits in these regions of the boiler are usually soft and fluffy and easy to remove with sootblowers. A design velocity of 35 fps (10.64m/s)is our limit for this heat trap. The eco- nomizer gas outlet design temperature is subject to a practical limitation because of the sulfur con- tent in the products of combustion from Kraft black liquor. To prevent rapid corrosion and fail- ure as a result of condensation and formation of dilute sulfurous acid, the metal temperature of ordinary steel in contact with the gas must not be below the dew point of the gas. This means that the temperature of the feedwater to the econom- 13 Figure 12 Economizer - long flow. izer should be above the dew point temperature of the gas. To minimize external corrosion, expe- rience in recovery unit operation indicates that feed temperature should not be less than 275°F (135°C) and the associated gas temperature should be not less than 350°F (177°C). Sootblowing The buildup of chemical ash on recovery-unit heat-transfer surfaces is related to design and operation. Entrainment of ash in gases ascending the furnace is affected by gas velocity, air distri- bution, and liquor properties. Care is taken in the design of all recovery-unit heating surfaces to assure that such surfaces are arranged for soot- blower cleaning. Cavities are left at optimum locations in superheater, boiler, and economizer banks to provide for insertion of sootblowers. Gas temperatures are calculated to make certain that velocities and particular tube patterns are com- patible with good cleaning characteristics. Steam sootblowers are universally used in mod- ern recovery unit installations. Sootblowers are arranged for automatic sequential operation, con- trolled from the main recovery unit panel board. When a modern recovery unit is operated at or near its rated capacity, no hand lancing should be required to keep the gas passages open. As load is increased on a unit, however, mechanical entrain- ment of ash and volatilization of sodium com- pounds increase and invariably lead to cleaning problems. In addition to excessive quantities of ash in the flue gas, velocities and temperatures at all points in the unit are increased, and ash deposits become more difficult to remove. Future developments New products and redesigns are under develop- ment to improve operation, increase reliability, increase solids processing capability and extend operational periods. New products Pyrosonic 2000® is a non-intrusive gas tempera- ture measuring device that utilizes the change in the speed of sound as a function of temperature as its driving force and primary sensor. The Pyro- sonic 2000 has been field tested and verified on steam generators fired by coal, municipal solid waste, refuse-derived fuel, oil, gas and black liquor. It has proven to be an excellent real-time accurate average gas temperature indicator at any “upper” furnace location (Figure 13). Figure 13 Pyrosonic transmitter. Application to process recovery steam genera- tors is used as an aid in: 1) Setting combustion by measuring gas temperatures at the furnace exit and at the oscillator level, and (2) monitoring ash deposition. BLRBAC approved the supervised use of water-cooled High Velocity Thermocouple (HVT) probes in 1985; in 1986, we were field calibrating the Pyrosonic 2000 temperature measuring device with a watercooled HVT on a 900 ton process recovery steam generator. Our first commercial application, currently under shakedown, was to a 1,500 ton unit and used a total of 6 senders/receivers at each of two furnace elevations to provide a deconvoluted out- put. One output from that system is shown in Figure 14 where the furnace flue gas isotherms are measured in 14a; an operator adjustment is 14 Isothermal Contours 100 Deg. Intervals BaW Pyrosonics July 1, 1987 8:41:2 AM 34.9 27.9 Left Wall Length in Feet 168 252 335 Front Wall Length in Feet 00. «84 41.9 Figure 14a 1500 ton recovery - plane just above the tertiary windbox. Isothermal Contours 100 Deg. Intervals BaW Pyrosonics July 1, 1987 11:23:20 AM 34.9 27.9 21.0 14.0 7.0 Left Wall Length in Feet 0.0 16.8 25.2 33.5 Front Wall Length in Feet 419 Figure 14b 1500 ton recovery - plane just above the tertiary windbox. then made to the air flow splits at the tertiary; and the effect is demonstrated on the CRT shown in 14b. A 200°F drop in peak gas temperature has been obtained by a knowledgeable operator con- trolling the mixing. This first Pyrosonic recovery application will have 24 months on line to be com- pleted in June of 1989. Ports lower in the furnace have been plagued with plugging which affects the accuracy of the temperature readout. Manual rodding works but the frequency (once per hour) is not acceptable and automatic rodding is under development. Smelt Bed Imaging System The SBIS has been an excellent tool to aid us in determining bed formation and related cause/ effects from air flow and other combustion variables. Several commercial products are now available and all which we have tested have dem- onstrated excellent bed viewing potential when two cameras are utilized. Performance Diagnostic System It has long been our goal to apply a real time, online performance diagnostic analysis system to permit continuous actual operational performance to be compared to the design performance. The performance diagnostic System 140® performs heat balances around each heat trap starting at the economizer outlet and moving towards the furnace exit. Until the advent of Pyrosonic 2000, the system relied on thermocouples to measure the gas temperature entering the generating bank. Measuring the gas temperature here is required because the steam quality of the fluid in the generating bank is unknown, thus making the heat balance equation unsolvable. The System 140 monitors design type information such as gas weights, absorption, and surface cleanliness and presents it to the operator in a form that can be easily understood. A software package is created specially for each application and online analysis of heat absorption in each heat trap—-superheater, generating bank and economizer--is now avail- able and operational on a process recovery system. The commercial application of System 140 to a recovery unit will have 12 months on line completed in August of 1989. At that time, we anticipate releasing the System 140 for general recovery applications. The use of these three systems: SBIS, Pyrosonic 2000, and System 140 greatly expanded our ability to obtain real time performance measure- ments as operating variables are deliberately changed. The first field evaluation program offered a novel opportunity to apply two SBIS’s and the Pyrosonic and then evaluate real time effects upon bed temperature and FEGT as com- bustion variables were altered to create the two extremes - a “hot” bed and a “cold” bed. The two extremes were then compared to the baseline; i.e., normal operation. The present 1500-ton com- mercialization effort offers an additional challenge - pre-upgrade operational periods of 72 days maximum as the rule. The upgrade: air flow, firing system, Pyrosonic and System 140 are targeted on attaining a 360-day goal. Acoustic Leak Detector The Acoustic Leak Detector for improved process 15 recovery safety is presently being applied com- mercially. This tool uses the gas-borne noise generated by a steam or water leak to activate a highly sensitive pressure transducer pick-up and provide an early warning for a potentially critical exposure incident. We are working jointly with Westvaco applying the leak detector for the furnace, superheater, generating bank and econ- omizer applications. Over 200 channels of acous- tic leak detection have been placed in successful service on utility steam generators in the U.S.A. and the first 13 channels on a process recovery application are currently in place. Life Extension Spiraling capital costs for new equipment have made the economics associated with upgrades to extend life an increasingly attractive alternative. Smelt spouts and smelt spout openings are pro- ducts that operators have identified as having major impacts on availability. Our program to improve the life of these compo- nents has been actively underway for 2-1/2 years. We have created a finite element model of a spout opening; obtained chordal thermocouple data from instrumented openings to know actual spot heat fluxes; completed our review of alternative cooling and alternative materials; and created revised designs that are currently installed, undergoing field trials and evaluations. Successes thus far are very encouraging. Some examples are: Our new VEVR spout opening which has demonstrated an increase in life of six times greater than the VEV openings they replaced. Also, ceramic cooled, and ceramic uncooled spouts, are presently under commercial evalua- tion in the field at three sites and, at this writing, 1-1/2" Refractory Liner 1/4" Ceramic Fiber Insulation Steel Spout Figure 15 Ceramic smelt spout. the uncooled ceramic has 72 days of service. Composite openings utilizing materials other than 304 as the outer skin are demonstrating outstanding performance on a recovery boiler that has super aggressive smelt. The original inserted opening life was as short as three months; the redesign is now entering its second year. The VEVR spout has had internal baffling installed to improve cooling and these are under- going field evaluation (Figure 15). Pressure Part Design Zero leaks are the objective in any of our pres- sure part design efforts. T.P McGee (12) using BLRBAC data summarized the five most numer- ous Recovery Boiler pressure part failures during 1979 to 1987 as follows: Economizers 87 - incidents Waterwall Tubes 55 - incidents Generating Bank 35 - incidents Superheater Tube 43 - incidents Screen Tube 10 - incidents T.P. McGee then identified the number of inci- dents per manufacturer and the respective market share of the five boiler vendors currently serving the U.S. Recovery Market. Economizer B&W designs represent 54.8 percent of the operat- ing Recovery Boilers in domestic service. B&W did identify a manufacturing weld problem (orbital welder) that contributed to the major portion of B&W economizer tube-leak incidents and that defect has been corrected. Furthermore, we have initiated a major review of our long flow design and have redesigned it to further reduce stress levels and attain the zero leak objective. Waterwall Tube Leaks The second most common cause of an emergency shutdown procedure (ESP), waterwall tubes have been under review to reduce stresses in those assemblies. We have modified our designs of buck stay attachments, windbox attachments and air ports with the objective of reducing stresses. This alternative design approach has proved prudent, as during the past 12 months internal waterwall tube cracking at attachment welds has been uncovered on approximately 25% of the units inspected. The observations have been labeled “waterwall stress assisted corrosion” and are related to acid cleaning/feedwater chemistry 16 upsets. Field inspection, using x-ray techniques at sidewall-to-floor attachments, windbox attach- ments, buckstay and access door attachments on membrane, non-membrane, old and new (eight years service) units, indicates damage can and does occur in various designs originated by differ- ent vendors. Generating Bank Most leaks typically are a result of corrosion/ erosion. Thicker walls are recommended and improvements in combustion have reduced carry- over and the need for sootblowers at the MCR rat- ing. Nondestructive testing techniques have been invented by B&W, which permit wall thickness mapping from the inside of the tube, even when filled with water. Superheater Tubes Our design approach has been to design for coun- terflow to maintain minimum tube metals in the maximum gas temperature passage locations, reduce mechanical fatigue failure potential by applying improved side-to-side and front-to-rear ties, and permanent thermocouples to ensure total boil-out (clearance) prior to going on-line. Screen Tube Designs have been strengthened by using the nose as a support point, increasing the wall thick- ness of the horizontal runs, and passing directly through the roof at the exit. Strapholders (prone to burn up) have been eliminated. Our new boilers are designed with or without screens depending upon the final steam temperature required, but screens have been and can be designed to survive in the environment. Screens are most commonly destroyed when a low water incident continues uncorrected. If we permit operation with no water in the boiler, eventually, if there is no screen, a furnace water wall tube will overheat and fail in the short- term mode. We are constantly looking for ways to improve our pressure part design to attain zero leakage from a design and workmanship objective. BLRBAC, by providing hard industry data, has had a positive effect on those efforts. Automated Process Recovery Operational Control System (APROCS) Background The need to tie together all the various pieces of Process Recovery operational control is evident. The Pyrosonic 2000, the on-line diagnostic System 140 for performance analysis, plus the individual control loops which exist as separate islands of control are not yet fully integrated. APROCS ties together and integrates the fol- lowing islands of automation into a whole: e Real time on-line comparison of design performance to actual e Combustion control - air flow zone control e Solids e Liquor flow e Liquor spray control e Furnace Exit Gas Temperatures e Fouling e Sootblower utilization e Auxiliary burner/automatic burner control system e Leak detectors e Liquor droplet size control (new) Approach Integrating the various islands and applying the proven B&W process combustion control logic to the APROCS System will be field tested, evaluated and commercialized during 1989. Upgrades Older units designed with tube/tile non- membrane furnace enclosures offer an oppor- tunity for updated wall construction. An oft-repeated complaint is justifiably leveled against the flat stud and refractory closures commonly utilized in the older (pre- 1960’s) units for furnace closure. Water from waterwashing gets behind the studs, wets the refractory, accelerates corrosion of the casing, and permits air infiltration. One approach is economically attractive, and that is to field- install a membrane type closure, called “equa- therm”’. It has been successfully field applied in lower furnaces to augment the older pre- membrane flat-stud-design. The benefits of getting the combustion air where it is required for combustion, dehydration and pyrolysis results in proven capacity upgrades and extended operational periods. The upgrade is relatively inexpensive when a cost benefit analysis is performed. Major Capacity Upgrades Recently, it has become more cost-effective to upgrade present capacity and avoid the major 92" Le Figure 16 Comparison base to 170% MCR increase. 17 Old 319 tons 71-4" Update 638 tons 150% MCR Solids 185% MCR Heat aL Figure 17 Comparing PR-27 to new PR-27. capital outlay required for new green field equipment. One recent upgrade proposal indi- cated that 170 percent over MCR could be attained with an innovative approach. Expanding the front wall outward by eight feet and raising the roof by seven feet were the major physical changes, but the combustion system and fuel preparation system were also thoroughly updated at the same time. Figure 16 shows the major effect upon furnace cross section and upon the side-sectional elevation picture of this Process Recovery steam genera- tor, effective with the proposed upgrade. An alternative approach when sufficient hearth area is available for the upgraded con- dition is to just raise the roof. The attached graphics (Figure 17) indicate the effect of adding 32 feet of height to an existing furnace when performing a low odor conversion at the same time. Resulting capacity increase (solids processing base) is 181% of original MCR. Summary There is great opportunity to further improve KRAFT Process recovery steam generator opera- tion and maintenance. It is our objective to con- tinue to work closely with operators to uncover problems and initiate improvements to aid clients in increasing solids throughput capability, decreasing downtime, and increasing both relia- bility and availability. Improvements have been and will continue to be made as we work together to solve those problems. We hope that clients will continue to share these problems with us. Acknowledgments Work reported here is a summary of the contribu- tions of many talented people within B&W. I wish to acknowledge the efforts of some of the major contributors. John Kulig and John Monacelli of Design Engineering have directed the Air Flow Program; Joan Barna of Design for the Smelt Spout/Smelt Opening Programs; and Byers Rogan of Atlanta Service for the Composite Tube Program; Jerry Blue, Manager, Industrial Boiler Design; Rich Mattie, Sam Allison and Ed Hichel- berger of Field Service for the field data collection and analysis, comparing the pre-to post- upgraded performance. References 1. Blue, J.D., Dudek, R.F., and Suda, S., “Some Considerations in the Use of Kraft Recovery Boilers for High Temperature and Pressure Application,” TAPPI Engineering Conference, Atlanta, GA, September 1981. 2. Bjorklund, H., et al, “The NSP Cyclone Fur- nace for Black Liquor Reductive, Combustion Status Report,” International Chemical Conference, New Orleans, LA, 1985. . Dickinson, J., Murphy, J.A., and Wolfe, W.C., “Kraft Recovery Boiler Furnace Corrosion Protection,” TAPPI - Engineering Conference, Atlanta, GA, 1981. . McGillivray, S., Harris, L.E., and Blackwell, B.R., “Black Liquor Evaporation,” Project Memo V535017; “Dead Load Reduction in the Kraft Pulping Process,” Environment, Canada, Ottawa, 1984. . Herngren Torbjorn, et al, “Control of Black Liquor Supply and Furnace Conditions,” International Chemical Conference, New Orleans, LA, 1985. . Grace, J.M., Cameron, J.A., and Clay, D.T., “Role of the Sulphate/Sulphide Cycle in Char Burning - Experimental Results and Implica- tions,” International Chemical Recovery Conference, New Orleans, LA, 1985. . Blackwell, B.R., and King, “Chemical Reac- tions in Kraft Recovery Boilers,” Sandwell & Company Ltd., 1985. 19 10. 11. 12. Barsin, J.A., “Upgrading the Combustion System of a 1956 Vintage Recovery Steam Generator” TAPPI Engineering Conference, New Orleans, 1987. Barna, J.L., and Rogan, J.B., “Corrosion of Composite Port Opening Tube in Recovery Boilers: Appearance and Occurrence,” 1986 TAPPI Engineering Conference, Seattle, WA. Ingevald, S., and Bruno, F., “Forty Years Fight Against Corrosion in Recovery Boilers,’ International Chemical Conference, New Orleans, LA, 1985. Barynin, J.A., Dickinson, J.A., “Considera- tions for the Updating of Recovery Boilers,” International Chemical Conference, New Orleans, LA, 1985. McGee, T.P. “The Recovery Boiler Dilemma” 1987 TAPPI Engineering Conference, New Orleans, LA.