Loading...
The URL can be used to link to this page
Your browser does not support the video tag.
Home
My WebLink
About
Wales, Alaska High Penetration Wind-Diesel Hybrid Power System Theory of Operation 2002
National Renewable Energy Lab Wind-Diesel Hybrid System Publications Drouilhet, S.; Shirazi, M. (2002). Wales, Alaska High Penetration Wind-Diesel Hybrid Power System: Theory of Operation. 77 pp.; NREL Report No. TP-500-31755. Drouilhet, S. (2001). Preparing an Existing Diesel Power Plant for a Wind Hybrid Retrofit: Lessons Learned in the Wales, Alaska, Wind-Diesel Hybrid Power Project. 13 pp.; NREL Report No. CP-500-30586. Randall, G.; Vilhauer, R.; Thompson, C. (2001). Characterizing the Effects of High Wind Penetration on a Small Isolated Grid in Arctic Alaska. 11 pp.; NREL Report No. CP-500-30668. Muljadi, E.; McKenna, H. E. (2001). Power Quality Issues in a Hybrid Power System: Preprint. 12 pp.; NREL Report No. CP-500-30412. Bialasiewicz, J. T.; Muljadi, E.; Drouilhet, S.; Nix, G. (1998). Modular Simulation of a Hybrid Power System with Diesel and Wind Turbine Generation. 12 pp.; NREL Report No. CP-500-24681. Shirazi, M.; Drouilhet, S. (1997). Analysis of the Performance Benefits of Short-Term Energy Storage in Wind-Diesel Hybrid Power Systems. 13 pp.; NREL Report No. CP-440-22108. Bialasiewicz, J.T.; Muljadi, E.; Nix, G.; Drouilhet, S. (2001). RPM-SIM: A Comparison of Simulated Versus Recorded Data (Preprint). 14 pp.; NREL Report No. CP-500-29174. Gevorgian, V.; Touryan, K.; Bezrukikh, P.; Bezrukikh, P. Jr.; Karghiev, V. (1999). Wind-Diesel Hybrid Systems for Russia's Northern Territories. 12 pp.; NREL Report No. CP-500-27114. Barley, C. D.; Flowers, L. T.; Benavidez, P. J.; Abergas, R. L.; Barruela, R. B. (1999). Feasibility of Hybrid Retrofits to Off-Grid Diesel Power Plants in the Philippines. 12 pp.; NREL Report No. CP-500-26927. McKenna, E.; Olsen, T. (1999). Performance and Economics of a Wind-Diesel Hybrid Energy System: Naval Air Landing Field, San Clemente Island, California. 109 pp.; NREL Report No. SR-500-24663. May 2002. « NREL/TP-500-31755 Wales, Alaska High-Penetration Wind-Diesel Hybrid Power System Theory of Operation S. Drouilhet and M. Shirazi fe. GAs ” + National Renewable Energy Laboratory 1617 Cole Boulevard Golden, Colorado 80401-3393 NREL is a U.S. Department of Energy Laboratory Operated by Midwest Research Institute e Battelle e Bechtel Contract No. DE-AC36-99-GO10337 May 2002. * NREL/TP-500-31755 Wales, Alaska High-Penetration Wind-Diesel Hybrid Power System Theory of Operation S. Drouilhet and M. Shirazi Prepared under Task No. WER2.1750 a « »NREL - National Renewable Energy Laboratory 1617 Cole Boulevard Golden, Colorado 80401-3393 NREL is a U.S. Department of Energy Laboratory Operated by Midwest Research Institute e Battelle « Bechtel Contract No. DE-AC36-99-GO10337 NOTICE This report was prepared as an account of work sponsored by an agency of the United States government. Neither the United States government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States government or any agency thereof. Available electronically at http://www.osti.gov/bridge Available for a processing fee to U.S. Department of Energy and its contractors, in paper, from: U.S. Department of Energy Office of Scientific and Technical Information P.O. Box 62 Oak Ridge, TN 37831-0062 phone: 865.576.8401 fax: 865.576.5728 email: reports@adonis.osti.gov Available for sale to the public, in paper, from: U.S. Department of Commerce National Technical Information Service 5285 Port Royal Road Springfield, VA 22161 phone: 800.553.6847 fax: 703.605.6900 email: orders@ntis.fedworld.gov online ordering: http://www.ntis.gov/ordering.htm ae fa Printed on paper containing at least 50% wastepaper, including 20% postconsumer waste About this Document This document represents the first chapter in the Operations and Maintenance Manual for the Wales Wind-Diesel Hybrid Power System. The entire manual is organized into many chapters with multiple appendices, totaling hundreds of pages, most of which are quite specific to the Wales system and not of general interest. The entire manual will therefore be produced in only a limited number of copies for those individuals and organizations with a direct role in operating and maintaining the system. The first chapter of the manual, however, deals with the system theory of operation, which is of more general interest, and is being published as this National Renewable Energy Laboratory (NREL) Technical Report. This report will also serve as a record, in summary fashion, of the hardware and software hybrid system controls technology developed by NREL under the Wales project. The authors intend that it be used as a reference not only by those directly involved in the operation of the Wales system, but by anyone with an interest in high-penetration, wind-diesel hybrid technology. ii TABLE OF CONTENTS INTRODUCTION... ccccsccesessseeseesseeeeeaeseseseneseseatacaeasacaeseesaeacaesesesasaeaeseeeeatseaeacaeseseass 1 1 SYSTEM OVERVIEW isiiiisscsssscsssscossssoransacessscsoncsssassasnsascasresteessessoosersearesevovassniacesnore 2 2 SYSTEM OPERATING MODE. .............sccssssscssssssssssssscssssssessssesssenessiacssesesscorsesssesees 9 3 SYSTEM OPERATING STATES ...0.....cccsssssessseeeeeeseeeseseeeeseeeseessesetsesesesseeesaeseees 13 4 COMPONENT DISPATCH oiiccccesscccsscacsocsssetsesceessniensnesosersnseccnensesssnsossesocenssssesasesns 15 ik.» PRRRRC EAA scsi cicinscscevcccessesssecusacesecsesesecsescessssteesscscnssesetevessonsstecacsensioncasesnensadeseesenees 15 4.1.1 Determining Diesel Capacity Required we 16 4.1.2 Diesel Combination Selection (Diesel Run Select BD Sosa ctocsceciteicetssavdoctusttaiusssstontessise 27 AD WAR ETAT ING TUS T UCI sissci.niadacsesssncdnssotssnssssssessessssenosessiceserssscsssanscnssesssssestssssepesseiass 4.2.1 Determining Wind Turbine Capacity Allowed a 4.2.2 Wind Turbine Run Select .......cecscsssssssssesesesesesesesesesesesesesesssesessseseseseseseseseecesseeceseneseees BS AC ARCHIE EMSC Foo ssseccscconsnoscsesnsnssusessesesscousocesusstecscsssassosasisostasssetssastnasitacsvacdes 35 4.3.1 Statistical AC Machine Dispatch .....c.cccscccsssssssessesessseesesesesesesesesesesesueneseseseseseacseeees 36 4.3.2 Instantaneous AC Machine Dispatch ......cccccccccccccssessesseseseeseesessessesseaessesessesseseacsseaees 38 4.4 DC Machine Dispatch ...........ccccscsssecsssssssssssssssssssesssseesessesesessssesessesssesessssssssesssesesseeesase 38 4.5 Auxiliary Battery Charger Dispatch ...........csssssssssssssssscsesessssssesessssssesessssesessssesessesesesee 39 4.5.1 Statistical Boost Charge Dispatch Criterid ......c.cccesecssesesesesesesessseneecssesesesesescsseneneees 40 4.5.2 Instantaneous Boost Charge Dispatch Criterid ........cccccscsssessescseesesesesesessenensessesees 40 5 POWER FLOW MANAGEMENT ..........:cccccecssssssseesecsesseesessessesseesessnssessesseeaeeareeees 42 SF BrGdtenCy Regiibsrtbo no isi685.256s82.sassccosusvcsoscsusvosssvssatossoseccsscsssessttveadesetactestescoteusetstsdtocneses 42 5.2 Voltage Regulation 5.3 The Power Flow Management Algorithm.............scsccsscscssescscscesescscssecescssessssssesssseseses 43 5:3.1 Diesel. ON Side s.eiiccdssvessvsabeescssesceevats we 44 5.322 Diesel OFF State... 44 5:33 Other System State Variables ..........ccccccceeeeee: we 4S 5.3.4 Power Flow Management Algorithm FIOWCHALE ........cccsecssesesssesteesesesesesesescsseneeseaees 46 G6 DIESEL LOAD CONTROL... ceeceeeesseeseeeeesseeseeeaesseesaeeseeesaeseaeesasseesseeeesnees 48 in it =| > io S E io iz) 2, kt ey is eu in ico = mS iS e 5 Fag a « iii 6:2)2, Dieseli-oad Control M Oded sesccscssccsccsrcccs rss 49 Z NONDIESEL POWER COMPONENT CONTROL ...........ccccsssssesesesssessesesessestscecsees 50 eM DUMP COMM OME ON ceesececesseecescsscsserrseercsreesreen stot el Determining Dump Load Required: Dump Load Control Modes. es ae le) Local/Remote:Dump:load Apportionments teva nena ene eS Dump ioad Element SWLCHInG mrcrrtcrecetee erecta nner err ree eee a AC Machine Control po | AC Machine Startup Sequence... MDD, Qn-Line Control Modes of the AC Machine OD 2 A CiMachinel Shutdown! SCG Ucn Ce sane ce reais eerie anc een eee ere a5 73. D€ Machine Control................. 56 deh DC Machine Startup Sequence ..... el 132 On-Line Control Modes of the DC Machine... iT oe DG: Machine Shutdown Sequences ee 58 Si (BATTERYiMANAGEMEND tircssscceccecceccscssotceccccrsreresrensctercrenratcsteen cana eneceseeaceecs 59 S315 Battery: Characteristics cccccccccccsscscscsccsssessestesecessusecstescssesesvosetesstesseesetsoeeeset eens 59 8.2 State of Charge Tracking Sie Battery: Stats! Flags ciccccccscccsesessscscssascsscsscscesevccesecsvactssotsscseesecssisceieetsseviececse ate etestteeestets 62 3.422 NormaliGharging/and(Discharpingeercncncrcnsnce en 64 BD ie BOOS CHALDIN Deeccsceresececscscersescececseenesceesescrerecereesecsnsesrecencesteteescocsaceseeenceeettecn em nantasseeseaes 65 SF FAVUETDETECTIONIANDIHANDEING ter crrcrrsrerrrererrrerterierecrercser cet cercerenertcetenrirs 66 OE DATAEOGGINGIANDIREROR TING Setter cere ranean ee enrernrernars 68 10.1 Data Acquisition, Logging, and Transfer ...............cccccscccscccscscscsesesnsnsecescoesesecssecscessosees 68 BL OFZ Data PC COG Geeeessccesrescteneseessscssersrensseseecvommecencomeeoeserat cnc ea serene ncaa nant annette enna 69 OS Fe] DatarA nalysis AMG! REPOrtim G ceccreccccacscaccesscacscesscsceccececscssscceeneeseeotocesesusesacrerensescecesetatestes 70 Introduction Alaska has over 200 rural villages that have no link to the power grids serving the main urban areas. The majority of these villages are served by diesel-driven generators. Because of the extreme remoteness of most of these communities, and the lack of roads, the delivered cost of diesel fuel is high, ranging from $1.00 to $3.00 per gallon. The high operating and maintenance costs of diesel generating stations result in electric generation costs that average nearly $0.40/kWh and can be as high as $1.00/kWh. There are also significant environmental hazards associated with diesel power generation, including fuel spills during transport, leaky bulk fuel tanks in the villages, and carbon dioxide (CO2) and other emissions. To reduce the cost of rural power generation and the environmental impact of diesel fuel usage, the Alaska Energy Authority (AEA), Kotzebue Electric Association (KEA, a rural Alaskan utility), and the National Renewable Energy Laboratory (NREL), began a collaboration in late 1995 to implement a high-penetration wind-diesel hybrid power system in a village in northwest Alaska. The project was intended to be both a technology demonstration and a pilot for commercial replication of the system in other Alaskan villages. Significant financial contributors to the project were the U.S. Environmental Protection Agency (EPA), the U.S. Department of Energy (DOE), and the Alaska Science and Technology Foundation. Kotzebue Electric Association and the Alaska Village Electric Cooperative also contributed some of their own resources to bring the project to fruition. During the first several years of the project, NREL focused on the design and development of the electronic controls, the system control software, and the ancillary components (power converters, energy storage, electric dump loads, communications links, etc.) that would be required to integrate new wind turbines with the existing diesels in a reliable highly automated system. Meanwhile, AEA and KEA focused on project development activities, including wind resource assessment, site selection and permitting, community relationship building, and logistical planning. Ultimately, the village of Wales, Alaska, was chosen as the project site. Wales is a native Inupiat village of approximately 160 inhabitants, with an average electric load of about 75 kw. A variety of obstacles, both technical and project development related, were encountered in the years 1996-1999, delaying the installation and commissioning of the system several years beyond the date originally projected. Testing and demonstration of proper operation of the wind- diesel control system was completed at NREL’s National Wind Technology Center (NWTC) in spring of 2000. The wind turbines, control panels, and ancillary components were installed in Wales in summer of 2000. The hybrid power system began operation in fall of 2000, with successful demonstration of all operating modes not occurring until fall of 2001. 1 System Overview The Wales Wind-Diesel Hybrid Power System combines diesel generator sets, wind turbines, energy storage, power converters, and various control components into a single highly integrated system. The primary purpose of the system is to meet the village electric demand with high- quality power, while minimizing diesel fuel consumption and diesel engine run time. The system also directs excess wind power to several thermal loads in the village, thereby saving heating fuel as well as diesel fuel for electric generation. Figure | is a one-line diagram of the system showing the principal power components, which are itemized in Table 1. Although there are currently only two wind turbines installed, the figure shows three because a third turbine one is planned. Figure 2 shows the components and interconnections of the controls that govern the operation of the entire wind-diesel system. Table 1. Roster of Power Components in the Wales Wind-Diesel System Qty Component Rating Make & Model 2 _| Wind Turbine 65 kW Atlantic Orient Corp. 15/50 1_| Diesel Generator 168 kW Cummins LTA10 1 Diesel Generator 75 kW Allis-Chalmers 3500 1__| Diesel Generator 168 kW Cummins LTA10 1 | Local Dump 89 kW NREL Design Load Controller 1 Remote Dump 144 kW NREL Design Load Controller 1 Rotary Converter 156 kVA NREL Design (AC Machine: Kato Engineering) (DC Machine: Reliance Electric) 200 | Battery Cell 1.2 VDC SAFT SPH130, Nickel Cadmium 130 Ah 1 | Auxiliary Battery 300 VDC NREL Design Charger 30A The Atlantic Orient (AOC) wind turbines use 480 volt 3-phase induction generators and connect directly to the medium voltage village distribution system through step-up transformers. The diesel generators are of the synchronous type, also 480 volt 3-phase. The dump load controllers each consist of a computer-controlled bank of solid-state relays, each of which controls power flow to a 480 volt 3-phase heating element in an electric boiler. The relays may be switched on and off rapidly, thereby providing precise real time control over electric power to the dump loads. The rotary converter is an electromechanical bidirectional AC/DC power converter. It consists of an AC synchronous generator shaft coupled to a DC motor. When in use, the AC machine is connected to the 480 VAC bus of the power plant. When the DC machine is in use, it is connected to the battery bank. As will be explained later in detail, by controlling the field current in the AC and DC machines, one can control the flow of both real (kW) and reactive (KVAR) power between the AC bus and the rotary converter. Being shaft-coupled, the DC machine always spins at the same speed as the AC machine. Electrically, however, the AC machine can operate independently from the DC machine. The AC machine can on-line (connected to the AC bus) without the DC machine being on-line (connected to the battery bank). In this state, the rotary converter is operating simply as a synchronous condenser. There is no equivalent operating state involving only the DC machine, which cannot be on-line unless the AC machine is also on-line. Though not shown in Figure 1, there is also a small 10 HP pony motor used to spin the rotary converter up from rest to synchronous speed so that the AC machine may be connected to the AC bus. The pony motor is connected to the AC machine by a large timing belt. How these components interact to provide continuous high-quality power is explained in the various chapters of this report. If the reader is unfamiliar with concepts of real and reactive power and the basic methods of frequency and voltage control, it may be helpful to first read Chapter 5 on power flow management before Chapter 4 on component dispatch. WALES POWERHOUSE 480V, 3PH = STATION Q — ‘STATION SERVICE METER cee a2 #1: 168 kW Diesel —1 (Cummins LTA 10) ENERGY STORAGE SUBSYSTEM Ea — = i let eau Ee #2: 75 KW Diesel Rotary (Allis-Chalmers 3500) Pi deel = TOTALIZING HO $6 — : ees — +l] #3: 168 kW Diesel CONVERTER end (Cummins LTA 10) Battery Bank METERS + 3 a t+ ; PoweRHoUse DUMPLOAD POWERHOUSE (3) AOC 15/50 Wind Turbines ee eaoe! CONTROLLER Oe LOE 3 x 65 = 195 kW METER (90 kW) (480V, 3-PH) VILLAGE ELECTRIC LOADS A (120V, 1-PH) SCHOOL HYDRONIC LOOP DUMP LOAD DUMP LOAD + os (6X24=144kW) —- CONTROLLER ar ar } oe ! 4 4 VILLAGE LOADS WNT OTALIZING Hoa INDIVIDUALLY } HYDRONe toes OF] METERS METERED TOTALIZING 9 METER 8 7.2 kV, 3-PH DISTRIBUTION SYSTEM Figure 1. Wales wind-diesel hybrid power system AOC 15/50 TURBINE CONTROLLER} (PLC MODEL 250) AOC 15/50 TURBINE CONTROLLER (PLC MODEL 250) Figure 2. Wind-Diesel system control layout WALES POWERHOUSE DIESEL PLANT DIESEL GENERATOR CONTROL PANEL HYDRONIC LOOP DUMP LOAD CONTROLLER ANALOG SIGNALS DISCRETE CONTROL SIGNALS SHIELDED TWISTED PAIRS (24VDC, 120VAC) ‘SLICE VO NETWORK (4-20 mA) TWISTED PAIR SLICE PROTOCOL 614 KBAUD eh eee WIND-DIESEL ETHERNET HYBRID POWER SYSTEM PLC-PLC NETWORK HYBRID POWER SYSTEM |_____FIBEROPTICLINK | eer VIA RADIO MODEM MODBUS PROTOCOL ROTARY CONVERTER S00 BAUD MAIN CONTROL EEL je CONTROL CABINET (PLC MODEL 450) | (PLC MODEL 450) | REMOTE VO NETWORK VIA RADIO MODEM PHONE CONNECTION PHONE CONNECTION acne FOR DIRECTSOFT LINK FOR DATA REPORTING SCHOOL HYDRONIC LOOP DUMP LOAD CONTROLLER (REMOTE I/O SLAVE) 2 System Operating Modes There are five possible system operating modes: Manual, Mode 0, Mode 1, Mode 2, and Mode 3. Each of these modes implies the availability of a particular set of power system components. The current operating mode says nothing about the actual operating state of the systems (i.e., which components are actually on-line), only which components could be on-line. The various operating modes and their characteristics are summarized in Table 2. Once the system operator places the system in a particular mode, it will remain in that mode until the operator requests a different operating mode, or until a required component becomes unavailable. In that case, the control system will automatically drop down to the next lowest operating mode that does not require that component. Structuring the control system in terms of distinct operating modes based on component availability makes the controller more robust and fault tolerant. Mode 3 is the normal and intended operating mode for the system. This mode requires the availability of all system components. Suppose there were a fault in the battery bank or DC machine, making the energy storage function unavailable. In that case, if there were a single rigid system control algorithm that assumed the availability of energy storage, then the entire automatic control system could not function, and the power system would be reduced to manual diesel operation until the fault was repaired. With the distinct operating modes, however, the system would simply drop into Mode 2 and continue to function, now as a no-storage wind-diesel system. Thus, the failure of a particular component does not necessarily render the system inoperative. Note that the system is intended to be operated from the Wind-Diesel Control Panel, which is the plant supervisory controller. Even in Manual Mode, the diesel start/stop commands are issued from the Wind-Diesel Control Panel, and the diesel automatic synchronization and load-sharing remains fully functional. Table 2. Wales Wind-Diesel System Operating Modes MODE DESCRIPTION MANUAL Diesel genset mode. Diesels dispatched manually from operator interface. System defaults to this mode when bus de-energized. Diesel genset mode. At least one diesel runs continuously. No wind turbines available. Diesel gensets dispatched automatically to meet load. AUTO DISPATCH At least one diesel always running. Wind turbines run whenever wind is available. Diesel genset controls system frequency and voltage. Dump load ensures minimum load on diesel. Diesel runs only if wind turbines cannot meet the load with adequate margin. When diesel is ON, diesel genset controls system frequency and voltage, and dump load ensures minimum average load on diesel. If not needed for reactive power, AC machine turned off to eliminate parasitic loss. When diesel is OFF, AC machine controls system voltage, dump load controls system frequency. Same as Mode 2, except: When diesel is ON, excess diesel power is used to charge battery. If load briefly exceeds wind power plus on-line diesel capacity, power is drawn from battery as needed to prevent another diesel from coming on. The diesel genset controls system voltage and frequency. When diesel is OFF, excess wind power is first used to charge the battery. Additional excess power is sent to the dump load. If wind power is briefly insufficient to meet the load, power is drawn from the battery as necessary to keep a diesel from being turned on. The AC machine controls system voltage. The DC machine controls system frequency, unless dump load is required, in which case dump load controls frequency. DIESEL GENSET 11 COMPONENT STATUS BATTERY SIAIE Ac | ANDDC| FREQUENCY DIESEL WTG MACHINE |MACHINE CONTROL 0 (system de-energized) 1A X diesel iC A 1B Xx x diesel . ~\ [2A X X diesel ol jg Y 2B Xx x Xx diesel Cepuene 3 X X dump load 7! 4 x X X diesel . 5 Xx X X X diesel Mt 6 X X X battery behing 7 x x battery 13 3 System Operating States The system operating state refers to the set of power sources that are actually on-line at any given time. 10 different operating states are possible, as shown in Table 3. The integer state designations refer to the various possible combinations of real power sources. Note that the AC machine by itself (without the DC machine) is not considered a real power source. Thus, both States 1A and 1B refer to the situation where diesel is the only generating source on-line. In State 1B, the AC machine is acting only as a synchronous condenser. State 2, in which both the diesel(s) and the wind turbine(s) are on-line, is similarly divided into States 2A and 2B, ‘ depending on whether or not the AC machine is on-line. Certain combinations of components are not possible and therefore do not have state designations. For example, the DC machine cannot be on-line unless the AC machine is on-line, so there are not defined states where that is the case. The system state usually is not something that the operator need be concerned with. System state information is used internally by the system controller to determine which set of component dispatch algorithms to use at any given time. Table 3. Definition of System Operating States Only certain states are possible in each of the system operating modes, based on component availability. Table 4 indicates the possible operating states in each of the various system operating modes. Table 4. Possible Mode/State Combinations 14 STATE 1A | 1B | 2A | 2B 3 7 MANUAL 1 MODE x MODE 0 x MODE 1 x x MODE 2 x x x x MODE 3 x x x xX Why Mods O — |<< Z 15 4 Component Dispatch Note to the reader: In this chapter, capital letters are used to refer to Algorithms or Procedures in the control program. Italics are used to refer to Parameters or Quantities used by the algorithms, and underlining is used for emphasis or when introducing new terms. Component Dispatch refers to the process of determining when to turn on/off individual diesel generators, individual wind turbines, the AC machine, the DC machine, and the auxiliary battery charger. This is distinct from Component Control, which refers to the process of actually starting/stopping each component and bringing it on-line/taking it off-line. The dispatch process depends on whether there is only one unit in a particular component group or multiple units. For single-unit component groups, such as the AC machine, DC machine, and battery charger, dispatch is only a matter of determining when to turn the component on and when to turn it off. For multiple-unit groups, such as the diesels and wind turbines, dispatch consists of two phases: first, determining the quantity, in terms of kW capacity, of that component required or allowed to be on-line at any given moment, and second, determining which of the individual units to turn on/off to meet that quantity. Two different types of criteria are used to make the determinations shown above for single-unit component dispatch and the first phase of multiple-unit component dispatch: statisticaland _ instantaneous. Statistical dispatch criteria serve as the foundation of component dispatch. The objective of statistical dispatch criteria is to predict wind power generation and electric load in the near future, based on the recent past, and to turn components on/off in response to any predicted imbalances. Statistical dispatch criteria are evaluated once every minute or in response to special events that indicate an imminent loss of a system component. Instantaneous dispatch criteria serve as a backup to statistical dispatch criteria. The objective of instantaneous dispatch criteria is to monitor instantaneous power values and to immediately turn components on/off to prevent sudden and unexpected system power imbalances. Instantaneous dispatch criteria are evaluated every scan of the programmable logic controller (PLC) program, or approximately 20 times per second. The following sections describe both the instantaneous and statistical dispatch criteria for each of the components in further detail. In addition, for diesels and wind turbines, these sections also describe the methods used to determine which individual generators/wind turbines to turn on/off to meet capacity requirements (the second phase of multiple-unit component dispatch). 4.1 Diesel Dispatch Because the diesel generators are a multiple-unit component group, Diesel Dispatch consists of two phases: 1. Determining diesel generating capacity required to be on-line at any given moment 16 2. Determining the optimal combination of diesel generators to supply the required capacity (this process is called Diesel Run Select) Diesel Dispatch is thus distinct from Diesel Control, which is the process of turning on/off actual control signals to start/stop, synchronize, and load/unload a generator and to close/open the generator breaker. Diesel Dispatch is enabled if the system is operating in Auto Mode 0, 1, 2, or 3. It is disabled if the system is operating in Manual Mode. The following sections describe the two phases of Diesel Dispatch. 4.1.1 Determining Diesel Capacity Required Diesel Capacity Required is defined as the minimum amount of diesel capacity that must be on- line to ensure that the primary load is always met. This amount depends on the primary load on the bus (the load that must be met) and the power available from all other sources. Diesel Capacity Required may be significantly greater than the instantaneous load on the diesels at any given instant. Load and wind power fluctuations, combined with the fact that diesel capacity cannot be added instantaneously, require that a certain reserve capacity be maintained. At any instant, the wind power could drop, increasing the load that must be met by the diesels. Without any reserve capacity, this would result in a power outage because another diesel could not be started and brought on-line instantaneously. Diesel Dispatch is based on both statistical and instantaneous dispatch criteria. Statistical dispatch criteria can increase or decrease Diesel Capacity Required in order to start or stop a diesel. Instantaneous dispatch criteria can only increase Diesel Capacity Required to immediately start a diesel if there is insufficient power available on the bus to supply the load. 4.1.1.1 Statistical Diesel Dispatch The objective of Statistical Diesel Dispatch is to predict that portion of the primary load that must be met by the diesels in the near future. Statistical diesel dispatch criteria are evaluated every minute or in response to special events that indicate an imminent loss of a system component(s), e.g., the occurrence of system warnings, component warnings, component disable requests, and/or mode change requests. Statistical Diesel Dispatch consists of the following steps: 1. Determining the appropriate statistical diesel dispatch mode 2. Evaluating the corresponding criteria to determine Diesel Capacity Required 3. Determining whether or not all diesels may be shut off 4. Evaluating any additional special dispatch considerations and determining whether or not to execute Diesel Run Select. 17 The following sections describe these steps. 4.1.1.1.1 Statistical Diesel Dispatch Modes The first step in Statistical Diesel Dispatch is to determine the appropriate dispatch mode based on the system state as defined in Section 3. The system state is the combination of power sources on-line at any given moment. There are four different sources of real and/or reactive power: Diesels Wind turbines AC machine Battery bank/DC machine. eee In many cases, the system controller will have advanced notice that a generating component will soon go off-line. For example, a component warning, a component disable request, or mode change request all may indicate the imminent (but not immediate) loss of a component, possibly changing the system operating state. Diesel Capacity Required will increase when a power source component goes off-line. To prevent possible loss of load when a component goes off- line, Statistical Diesel Dispatch must determine Diesel Capacity Required based on the projected system state rather than the current system state. The projected system state is the combination of power sources expected to be on-line in the near future. There are statistical diesel dispatch modes corresponding to each projected system state, with a separate set of dispatch criteria for each mode, to handle the various possible projected states. Statistical Diesel Dispatch will determine the appropriate statistical diesel dispatch mode and evaluate the corresponding criteria once every minute, or whenever conditions indicate a system component(s) is about to be taken off-line, e.g., the occurrence of system warnings, component warnings, component disable requests, and/or mode change requests. There are six statistical diesel dispatch modes: Statistical Diesel Dispatch projected state 1A Mode Statistical Diesel Dispatch projected state 1B Mode Statistical Diesel Dispatch projected state 2A Mode Statistical Diesel Dispatch projected state 2B or State 3 Mode Statistical Diesel Dispatch projected state 4 or State 7 Mode Statistical Diesel Dispatch projected state 5 or State 6 Mode. Note that several statistical dispatch modes apply to more than one projected state. That is because the criteria used to evaluate required diesel capacity are driven by the non-diesel generating components on-line. For example, the same dispatch mode is used for projected states 4 and 7, which both have wind turbine(s) and the battery on-line. 18 4.1.1.1.2 Statistical Diesel Dispatch Criteria Once Statistical Diesel Dispatch has determined the appropriate dispatch mode, the next step is to evaluate the corresponding dispatch criteria to determine Diesel Capacity Required. There are four criteria corresponding to each statistical diesel dispatch mode. Each criterion quantifies a different dispatch criterion for Diesel Capacity Required: average kW, peak kW, average kilovolt ampere reactive (kVAR), and peak instantaneous kW. The actual Diesel Capacity Required value used to determine which diesels to turn on/off will be the largest of the Diesel Capacity Required values predicted by each of the four criteria. The Diesel Capacity Required given by some or all of the criteria may be negative, which simply means that according to those criteria, no diesel capacity is required. If Diesel Capacity Required is less than zero according to all criteria, and certain other conditions are met, the system may switch to (or remain in) diesel- off operation. Note: In the following sections we use the term primary kW load to distinguish this load from the secondary kW load. The primary kW load consists of those kW loads that must always be met, whereas the secondary kW load consists of those loads that can be turned on/off at any time and therefore need only be met when there is excess wind power. The primary kW load always includes the village kW load. It also includes the kW required to spin the rotary converter if the AC machine is on-line and kW required to boost charge if the system is boost charging. The secondary kW load consists of local and remote dump load and rotary converter DC kW. We do not make the same distinction for the total KVAR load. This is because the entire kVAR load must always be met — there are no kVAR loads that can be turned on/off at any time. The kVAR load consists of the village kVAR load plus the kVAR consumed by the wind turbines’ induction generators. Statistical Diesel Dispatch Average kW Criterion The objective of the statistical diesel dispatch average kW criterion is to dispatch sufficient diesel capacity to ensure that the average near-future primary kW load can be met without exceeding a user-specified percentage of the diesel capacity on-line, called the Maximum Allowed Continuous % Diesel Loading. Predicting the average kW load that must be supplied by the diesels requires predicting the average future village kW load and average future wind power. Although not perfect, the recent past is the best available predictor of the future. The average kW criterion uses the most recent 20-minute average values of village kW load and wind power to determine the expected average primary kW load and expected average wind power. The following is an example of a statistical diesel dispatch average kW criterion. It is the criterion that applies to projected states 5 (Diesel + Wind + Battery) and 6 (Wind + Battery). 19 Diesel Capacity Required = (Village kW 20m Average + kW to Spin Rotary Converter) - Wind kW 20m Average Maximum Allowed Continuous % Diesel Loading Note that diesel capacity is dispatched to meet the entire difference between expected average load and wind power, regardless of whether the projected state includes the battery bank or not. This is because the batteries are not intended to supply the average load. As described in the following section, they are only intended to meet short-term power requirements caused by fluctuations in wind power. Statistical Diesel Dispatch Peak kW Criterion The objective of the statistical diesel dispatch peak kW criterion is to dispatch sufficient diesel capacity to ensure that the peak near-future primary kW load can be met without exceeding the diesel capacity on-line. This requires predicting the maximum kW load that must be supplied by the diesels in the near future. This amount depends on whether or not the projected state includes the wind turbines and the battery bank. Wind power can be used to supply the peak primary kW load. In addition, if the batteries are on-line, then the batteries can also be discharged to meet the peak primary kW load. By supplying power during short-term peak loading conditions caused by transient dips in wind power, the batteries can prevent the necessity of starting another diesel. The maximum amount of AC kW power that can be delivered from the batteries through the rotary converter is called the Rotary Converter kW Limit and is based on either the maximum discharge rate of the battery bank or the kW capacity of the rotary converter, whichever is smaller. Predicting the maximum kW load that must be supplied by the diesels requires predicting the maximum future village kW load and minimum future wind power. The peak kW criterion applies predictive safety factors to the most recent 20-minute maximum village kW load and minimum wind power values to determine the expected maximum primary kW load and expected minimum wind power. The predictive safety factors are included to accommodate the fact that the wind power may be dropping and the load may be rising relative to the recent past. The following is an example of a statistical diesel dispatch peak kW criterion, the one used for projected state 5 (Diesel + Wind + Battery) and 6 (Wind + Battery): Diesel Capacity Required = Village kW 20m Peak X Village kW Safety Factor - (Wind kW 20m Minimum X Wind kW Safety Factor + Rotary Converter kW Limit) Note that the rotary converter power losses (the power required to energize and spin the rotary converter) are not included in this criterion, because they are embedded in the Rotary Converter kW Limit. When the rotary converter is supplying power to the bus, the batteries supply the rotary converter losses. 20 Statistical Diesel Dispatch kVAR Criterion The kVAR capacity of a generator is limited by saturation and/or heating effects in the windings and therefore can be exceeded on a transient basis without harm. Thus, it is only necessary to ~ dispatch diesel capacity to meet the average kVAR load, not the peak kVAR load. The objective of the statistical diesel dispatch kVAR criterion is to dispatch sufficient diesel capacity to ensure that the average near-future kVAR load can be met without exceeding the kVAR capacity on- line. This requires predicting the average kVAR load that must be supplied by the diesels in the near future. This amount depends on whether or not the projected state includes the AC machine. If the AC machine is on-line, it will share VARs proportionally with the diesel generators. Predicting the average kVAR load that must be supplied by the diesels requires predicting the average future total KVAR load. The average kVAR criterion uses the most recent 1-minute average kVAR load to determine the expected average total kVAR load. The criterion uses 1- minute averages instead of 20-minute averages in order to respond more quickly to an overload situation (because there is no peak kVAR criterion). Once the criterion has determined the kVAR capacity required, it must then convert this into an equivalent kW capacity required value. This is because the diesel combination table used to determine which diesels to turn on/off to meet the Diesel Capacity Required rates each combination in terms of kW capacity only. The criterion uses two different diesel capacity ratio factors to accomplish this conversion, the Diesel kVAR:kVA Capacity Ratio and the Diesel kW:kVA Capacity Ratio. The Diesel KVAR:kVA Capacity Ratio expresses the kVAR to kVA capacity ratio of the diesel generators. It is used, along with the Diesel kW:kVA Capacity Ratio, to convert kVAR capacity required to a kW equivalent value and to determine kVAR capacity ready from kW capacity ready. The Diesel kVAR:kVA Capacity Ratio should be approximately the same for all generators. Most generators are rated at 0.8 power factor. If such a generator were supplying rated kW (80% of kVA rating), the most kVAR it could supply would be 60% of kVA rating in order to remain within its kVA rating. By setting the Diesel kVAR:kVA Capacity Ratio to 0.6, we ensure that we will never exceed the kVA capacity of the generator as long as we do not exceed the kW capacity and calculated kVAR capacity. Therefore, there is no need for diesel dispatch to consider kVA criteria. The Diesel kW:kVA Capacity Ratio expresses the kW to kVA capacity ratio of the diesel generators. It is used to convert kVA capacity required to a kW equivalent value and to determine diesel kVA capacity ready from kW capacity ready. Depending on the size of the engine relative to the size of the generator, each diesel on-site may have a different kW:kVA capacity ratio. By using the largest of these ratios for all gensets on-site, we can ensure that any given diesel combination that can provide the equivalent kW capacity required can also provide the kVA capacity required. 21 The following is an example of a statistical diesel dispatch kVAR criterion, the one used for projected state 5 (Diesel + Wind + Battery) or 6 (Wind + Battery): Diesel Capacity Required = Total kVAR Load \m A - Rotary C i ota oa lm verage - Rotary ONCE kVAR Capacity X Diesel kW:KVA Capacity Ratio Diesel kVAR:kVA Capacity Ratio Statistical Diesel Dispatch Peak Instantaneous Diesel Capacity Required Criterion Instantaneous Diesel Dispatch will start a diesel immediately if the instantaneous power quantities used by its criteria exceed user-specified limits (see Section 4.1.1.2). To ensure that Statistical Diesel Dispatch does not turn off a diesel after Instantaneous Diesel Dispatch has just started one, the statistical diesel dispatch peak instantaneous diesel capacity required criterion prevents this from happening by ensuring that Diesel Capacity Required is never less than the most recent 20-minute peak instantaneous diesel capacity required. The criterion is the same regardless of the projected state. Statistical Diesel Dispatch Peak Instantaneous Diesel Capacity Required Criterion: Diesel Capacity Required = Instantaneous Diesel Capacity Required 20m Peak 4.1.1.1.3 Statistical Diesel Dispatch Diesel-Off Criteria Once Statistical Diesel Dispatch has determined the appropriate dispatch mode and evaluated the corresponding dispatch criteria to determine Diesel Capacity Required, the next step is to determine whether or not all diesels may be shut off. There are two conditions that must be met before diesel-off operation is allowed: 1) the projected state must include wind turbines and the AC machine, and 2) there must be sufficient wind power to shut all the diesels off. The first condition is straightforward. The second condition involves a number of different considerations. The most obvious requirement to ensure sufficient wind power to shut all the diesels off is that Diesel Capacity Required be less than zero. If the projected state does not include the battery bank, this implies that (1) the expected average wind power is sufficient to cover the expected average primary kW load, (2) the expected minimum wind power is sufficient to cover the expected maximum primary kW load, and (3) the rotary converter can supply the expected average kVAR load. If the projected state includes the battery bank, this implies that (1) the expected average wind power is sufficient to cover the expected average primary kW load, (2) the expected minimum wind power plus the Rotary Converter kW Limit is sufficient to cover the expected maximum primary kW load, and (3) the rotary converter can supply the expected average kVAR load. 22 Diesel Capacity Required being less than zero is not by itself sufficient to shut off all the diesels. There are additional requirements that must be met, one pertaining to the minimum instantaneous excess wind power, the other to the average excess wind power. Statistical Diesel Dispatch Diesel-Off Minimum Excess Power Requirement The minimum excess power requirement only applies if the projected diesel-off state does not include the battery bank. In this state, frequency control is accomplished by modulating dump load power to maintain an instantaneous real power balance. This frequency control mode requires that there is some amount of dump load on-line at all times. If, in the course of trying to regulate frequency, all dump load is removed, then controllability is lost. Any additional load on the system, or drop in wind power, will create an imbalance that cannot be corrected by further decreasing the amount of dump load. Therefore, a safety margin of excess wind power is necessary to prevent outages caused by larger-than-anticipated load fluctuations. Thus, wind- only operation (no diesel or battery) will only be allowed if the expected wind power is sufficient to cover the maximum primary load and maintain a user-specified amount of excess power, which is called the Wind-Only Dump Load Margin. This is in contrast to the statistical diesel dispatch rules for states in which either a diesel or the battery bank is on-line. Even when these components reach their full power rating, some additional power is always available on a transient basis to respond to wind and load fluctuations. However, with dump load frequency control, once all dump load has been removed, no additional power is available, even on a short-term basis. It is this increased vulnerability to power fluctuations in the wind-only state that requires the excess wind power margin. Statistical Diesel Dispatch Diesel-Off Average Excess Power Requirement The statistical diesel dispatch diesel-off average kW requirement applies regardless of the projected state. The Alaska Village Electric Cooperative (AVEC) power plant in Wales relies on waste heat from the engines to keep the off-line diesel engines warm and to heat the power plant. Once the diesels are shut off, this heat must be supplied by the local dump load. Therefore, diesel-off operation will only be allowed if the expected average wind power is sufficient to ‘cover the expected average primary kW load and maintain enough excess power to keep the _ diesel engines warm and heat the power plant. This amount of excess power is called Plant Heat Required... An empirically determined linear expression, in terms of a Plant Heat Constant and a_ Plant Heat Loss Factor, is used to estimate Plant Heat Required. Note that if the linear expression yields a value less than the user-specified Minimum Plant Heat Required, the latter is used instead. Plant Heat Required = MAXIMUM{ Minimum Plant Heat Required, Plant Heat Constant + [Plant Heat Loss Factor X (70 - Outside Temperature) ]} 23 4.1.1.1.4 Statistical Diesel Dispatch Additional Considerations Once Statistical Diesel Dispatch has determined the appropriate dispatch mode, evaluated the corresponding dispatch criteria to determine Diesel Capacity Required, and determined whether or not it is all right to shut all the diesels off, the last step is to evaluate any additional dispatch considerations and determine whether or not to execute Diesel Run Select. Specific system conditions existing at the time Statistical Diesel Dispatch executes may require modifications to the diesel dispatch process. It may be necessary, for example, to adjust the Diesel Capacity Required value determined by the main statistical diesel dispatch criteria, or to modify, delay, or prevent the execution of Diesel Run Select (defined in Section 4.1). These additional considerations are described below. Imminent Loss of Non-Diesel Generation Component Because the statistical diesel dispatch criteria are based on projected state, the Diesel Capacity Required value determined by these criteria will have already taken into consideration imminent loss of a system component due to the occurrence of system warnings or operator requests to disable a component. (The only exception is diesel warnings and operator requests to disable a diesel generator. These do not cause a new state to be projected.) Because power transients may occur when a component goes off-line; however, it would be risky to allow a diesel to go off-line at the same time, even if the Diesel Capacity Required value as determined by the statistical diesel dispatch criteria would otherwise allow that. Therefore, Statistical Diesel Dispatch will not allow a diesel to be taken off-line if the system is facing imminent loss of a non-diesel generation component because of a warning or operator request to disable. This is accomplished ‘by temporarily making Diesel Capacity Required equal to the larger of the Diesel Capacity Required (as determined by the statistical diesel dispatch criteria) or the diesel capacity currently on-line before executing Diesel Run Select. Diesel Disable Requests If the operator has requested to disable a currently running diesel, Statistical Diesel Dispatch will instruct Diesel Run Select to select a new diesel combination that can supply the Diesel Capacity Required but does not include the specified diesel. Execution of Diesel Run Select will be suppressed; however, if the system is facing imminent loss of any other generation component because of the risk involved with taking a diesel off-line at the same time another component may be going off-line due to a warning or disable request. Operator Mode Change Requests Switching from one system operating mode to a lower operating mode may involve taking a component off-line. Therefore, operator requests to change to a lower operating mode will project a new state. Because the statistical diesel dispatch criteria are based on projected state, the Diesel Capacity Required value determined by these criteria takes into consideration operator requests to change to a lower operating mode. However, if the operator-requested mode change would require additional diesel capacity, the operator may wish to cancel the mode change request. Therefore, if the operator has requested to change to a lower operating mode, Statistical 24 Diesel Dispatch will determine whether there is sufficient diesel capacity already on-line, sufficient diesel capacity available but not on-line, or insufficient diesel capacity available to enter the new mode. This information will be provided to the operator via the touchscreen along with prompts to either continue with or cancel the mode change request. Statistical Diesel Dispatch will allow Diesel Run Select to execute only when the operator requests to continue with the mode change. Diesel kVA Support for a Wind Turbine Start The Atlantic Orient Company (AOC) wind turbine has a large inrush current w when its contactor closes. To prevent this inrush from causing an unacceptable voltage flicker, it is necessary to “ensure that there is sufficient kVA capacity on-line to provide this inrush current before allowing a wind turbine to start. The amount of kVA capacity required is defined to be the most recent 1- minute average kVA load plus a user-specified amount of excess kVA capacity required for a wind turbine start. If the amount of kVA capacity required is less than the total kVA capacity on-line, then additional kVA capacity must be provided, either by the! AC machine or another diesel. If the system is in Mode 2 or 3, the AC machine is not already on-line, and there is sufficient diesel capacity to run the jpony motor (at least 10 kW excess diesel capacity on-line to provide the power to spin the rotary converter up to speed), then the AC machine will be dispatched to provide the additional kVA capacity. However, if any of these requirements are not met, then a diesel(s) will be dispatched to provide the additional kVA capacity. This is accomplished by making Diesel Capacity Required equal to the larger of the Diesel Capacity Required as determined by the main statistical diesel dispatch criteria or the total kVA capacity requirement (including the excess required by the wind turbine start), minus the rotary converter kVA capacity (if the AC machine is already on-line), the result converted to a kW equivalent. This revised Diesel Capacity Required is determined before executing Diesel Run Select. 4.1.1.2 Instantaneous Diesel Dispatch Statistical diesel dispatch peak kW criteria uses recent statistical values of wind power and load to predict the maximum kW load that must be supplied by the diesels in the near future and dispatches diesel capacity accordingly. The recent past is a good predictor of the near future; however, it is not perfect. There may be cases where an unanticipated change in wind power or village load causes the load that must be supplied by the diesels at any given instant to exceed the diesel capacity on-line. Therefore, Instantaneous Diesel Dispatch is provided to serve as a backup to Statistical Diesel Dispatch. The objective of Instantaneous Diesel Dispatch is to immediately dispatch additional diesel capacity if the kW load that must be met by the diesels at any given instant approaches or exceeds the amount of diesel capacity on-line at that instant. This is accomplished by monitoring instantaneous power quantities and immediately dispatching additional diesel capacity if they exceed specified limits. Instantaneous diesel dispatch criteria are evaluated on every PLC scan. Instantaneous Diesel Dispatch consists of the following steps: 1. Determining the appropriate instantaneous diesel dispatch mode 25 2. Evaluating the corresponding criteria to determine Instantaneous Diesel Capacity Required and determining whether or not the resulting value warrants executing Diesel Run Select. 4.1.1.2.1 Instantaneous Diesel Dispatch Modes The first step in Instantaneous Diesel Dispatch is to determine the appropriate dispatch mode. The current System State dictates which instantaneous power quantities should be monitored to determine whether or not additional diesel capacity is required. Therefore, the instantaneous diesel dispatch modes are based on the current System State. There are four instantaneous diesel dispatch modes: e Instantaneous Diesel Dispatch State 1A and State 1B e Instantaneous Diesel Dispatch State 2A and State 2B e Instantaneous Diesel Dispatch State 3 e Instantaneous Diesel Dispatch State 4, State 5, State 6, and State 7 4.1.1.2.2 Instantaneous Diesel Dispatch Criteria Unlike statistical diesel dispatch, there is only one criterion associated with each instantaneous diesel dispatch mode. The criterion is based on instantaneous kW quantities. There is no need to monitor instantaneous kVAR or kVA quantities because the kVAR and kVA capacities of a generator are limited by heating effects in the windings and can therefore be exceeded (within saturation limits) on an instantaneous basis without harm. Once Instantaneous Diesel Dispatch has determined the appropriate dispatch mode, the next step is to evaluate the corresponding dispatch criterion to determine Instantaneous Diesel Capacity Required. Statistical diesel dispatch peak kW criteria and instantaneous diesel dispatch criteria are similar, but there is an important difference. The statistical diesel dispatch peak kW criteria determine Diesel Capacity Required based on the predicted the maximum kW load that must be met by the diesels in the near future, while the instantaneous diesel dispatch criteria determine it based on actual instantaneous power measurements. The statistical criteria use certain power quantities, e.g., village kW and total wind kW, which are calculated and averaged over a several-second interval and cannot be measured instantaneously.’ However, the instantaneous diesel dispatch criteria must be based exclusively on quantities that can be measured instantaneously, such as diesel kW and dump load kW. * Each wind turbine controller is polled sequentially via radio link to determine its power output. The individual turbine power levels are summed to determine total wind power. Because of communication delays and nonconcurrent sampling, this total must be low pass filtered to eliminate sampling errors. 26 As previously stated, the objective of Instantaneous Diesel Dispatch is to immediately dispatch additional diesel capacity if the generation capacity on-line is insufficient to meet the instantaneous primary load with adequate safety margin. The specific instantaneous power quantities used depend on whether or not the current state includes the diesel generators and the battery bank. Instantaneous Diesel Dispatch: State 1A, 1B, 2A, and 2B In states that do not include the battery bank but do include a diesel generator, then the kW load that must be met by the diesels at any given instant is that portion of the primary kW load that is greater than the wind power at any given instant. This can be calculated, using instantaneous quantities, as the total diesel kW minus the total dump load kW. (Recall that diesel capacity need only be dispatched to ensure the primary kW load is met. Dump load is subtracted because it is a secondary load and need not be met.) The instantaneous Diesel Capacity Required is then obtained by dividing this primary diesel kW load value by a user-specified Maximum Allowed % Diesel Loading. If the resulting Diesel Capacity Required is greater than the diesel capacity on- line, Diesel Run Select is executed. The Maximum Allowed Instantaneous % Diesel Loading parameter provides a safety margin to allow for the fact that a diesel cannot be started instantaneously. There are actually two Maximum Allowed Instantaneous % Diesel Loading values, one which applies if there are wind turbines on-line and one which applies if there are no wind turbines on-line. The former is more conservative to provide a larger safety margin to accommodate fluctuations in wind power. Instantaneous Diesel Dispatch Criterion for Operation in State 1A (Diesel Only/AC Machine Off), State 1B (Diesel Only/AC Machine On), State 2A (Diesel + Wind/AC Machine Off), or State 2B (Diesel + Wind/AC Machine On): Instantaneous Diesel Capacity Required = Diesel kW - Dump Load kW Maximum Allowed Instantaneous % Diesel Loading Instantaneous Diesel Dispatch: State 3 During operation in a state that does not include the battery bank or a diesel generator (wind-only operation), the system is most vulnerable to power outages. Therefore, wind-only operation will only be allowed if the wind power is sufficient to cover the primary kW load and maintain a user-specified amount of excess power, called the Wind-Only Dump Load Margin. The Wind- Only Dump Load Margin acts as a safety margin should the primary kW load exceed the wind power. A diesel must be started if the excess wind power drops below this safety margin. In wind-only operation, the excess wind power is equal to the total dump load kW. If this value is less than the Wind-Only Dump Load Margin, Instantaneous Diesel Dispatch will cause a diesel to start by setting Diesel Capacity Required equal to the Instantaneous Diesel Capacity Required and executing Diesel Run Select. 27 Instantaneous Diesel Dispatch Criterion for Operation in State 3 (Wind Only): Instantaneous Diesel Capacity Required = Wind Only DL Margin - Dump Load kW Instantaneous Diesel Dispatch: States 4, 5, 6, and7 If the current state includes the battery bank, then the batteries can also be discharged to meet the primary kW load. By supplying power during short-term peak loading conditions caused by transient dips in wind power, the batteries can eliminate the need to start another diesel. The maximum amount of AC kW power that can be delivered from the batteries through the rotary converter is called the Rotary Converter kW Limit. Therefore, during operation in any state that includes in the battery bank, the instantaneous Diesel Capacity Required is that portion of the primary kW load that is greater than the sum of the wind power and the Rotary Converter kW Limit. The expression for instantaneous Diesel Capacity Required is given below. In this case, it is not necessary to provide a safety margin in the form of Maximum Allowed % Diesel Loading because the Rotary Converter kW Limit provides that safety margin by being less than the actual transient capacity of the battery bank/rotary converter. If the instantaneous Diesel Capacity Required exceeds the diesel capacity on-line, Diesel Run Select is executed. Instantaneous Diesel Dispatch Criterion for Operation State 4 (Diesel + Battery), State 5 (Diesel + Wind + Battery), State 6 (Wind + Battery), or State 7 (Battery Only): Instantaneous Diesel Capacity Required = Diesel kW + Rotary Converter AC kW - Dump Load kW - Rotary Converter kW Limit 4.1.2 Diesel Combination Selection (Diesel Run Select) Once the amount of diesel generating capacity required is known, the second phase of Diesel Dispatch is to determine the optimal combination of diesel generators to provide the required capacity. This process is called Diesel Run Select. This process executes differently, depending on whether or not Statistical Diesel Dispatch has determined that it is OK to shut off all diesels. 4.1.2.1 Diesel-Off Operation Not Allowed The primary objective in selecting a diesel combination to meet the required capacity is maximum efficiency, therefore the ideal approach would be to use the fuel curve of each diesel generator to find the combination of diesels with the lowest total fuel consumption rate at that load. However, this method has several drawbacks. It is computation intensive, and more 28 importantly, it does not allow the system operator to run certain generators above others for reasons other than efficiency. A plant operator may prefer to run a certain generator over another to concentrate run-time on the first generator. This is often done with equivalent or nearly equivalent generators so that they do not require overhaul at the same time. The “ideal” approach described above would not accommodate this plant operating strategy, since it dispatches diesels strictly according to maximum fuel efficiency. The Wales wind-diesel control system uses a practical approach to diesel combination selection. This method is based on an operator-prioritized table of diesel combinations and therefore allows for operator preferences. The operator can setup and change the diesel combination table at any time using the Wind-Diesel Control Panel (WDCP) touchscreen. Diesel Run Select will search through the operator-prioritized table of diesel combinations for the first diesel combination that meets the following requirements: Combination capacity > Diesel Capacity Required All diesels in the combination are available None of the diesels in the combination have a warning’ The combination would not shut off any diesels that are currently on-line but have not met their minimum run-time. ee A fifth and final requirement applies only if the system is not facing imminent loss of any other component: 5. None of the on-line diesels in the combination have a pending operator request to be disabled by the operator (but are still enabled because they are still on-line)."* If any combination fails to meet all these requirements, Diesel Run Select will skip that combination and resume the search. If none of the diesel combinations meet all these requirements, then there is insufficient diesel capacity available to meet the Diesel Capacity Required. In this case, Diesel Run Select will require only that all the diesels in the combination be available and will select the combination that has the largest capacity of all the combinations that meet this requirement. " Ifa diesel generator has a warning condition, a controlled shutdown of that diesel will be initiated. However, while it is still on-line, it is considered available, even though it has a warning. It will not become unavailable until it goes off-line. * When the operator requests to disable a diesel generator that is on-line, the control system will first ensure that there is sufficient generating capacity on-line without that diesel. Only then will it be taken off-line and disabled. 29 4.1.2.2 Diesel-Off Operation Allowed If Statistical Diesel Dispatch has determined that it is acceptable to shut all the diesels off, Diesel Run Select will allow all diesels to be shut off except for those that have not met their minimum run-time. 4.2 Wind Turbine Dispatch In a low-penetration wind-diesel system any and all available wind turbines may be operated whenever there is sufficient wind. In a high-penetration system, however, it is sometimes necessary to limit wind turbine capacity on-line to avoid producing more power than can be controlled by the system. This is the function of Wind Turbine Dispatch. Because the wind turbines are a multiple-unit component group, Wind Turbine Dispatch consists of two phases: 1. Determining wind turbine generating capacity allowed to be on-line at any given moment 2. Determining which wind turbines to allow to start/take off-line in order not to exceed the allowed capacity (this process is called Wind Turbine Run Select). Wind Turbine Dispatch is thus distinct from Wind Turbine Control, which is the process of turning on/off actual control signals to release/engage the turbine brakes and close/open the turbine contactor. Wind Turbine Control is performed by each individual wind turbine controller. Wind Turbine Dispatch is enabled if the system is operating in Auto Mode 1, 2, or 3. It is disabled if the system is operating in Manual Mode or Auto Mode 0. The following sections describe the two phases of Wind Turbine Dispatch. 4.2.1. Determining Wind Turbine Capacity Allowed Wind Turbine Capacity Allowed is defined as the maximum amount of wind turbine capacity that can be on-line without risk of overpowering the bus. This amount depends in part on the amount of wind power the system can absorb, which, in turn, depends on the primary load, the amount of secondary load available, and the required minimum power contribution from the diesels.” However, Wind Turbine Capacity Allowed is not simply equal to the amount of wind power the system can absorb, or wind turbine power output allowed. At lower wind speeds, Wind Turbine Capacity Allowed will be greater than the actual wind turbine power output allowed because each turbine is producing less than rated power. Similarly, at higher wind speeds, Wind Turbine Capacity Allowed may be less than the actual wind turbine power output allowed because each turbine may be producing more than rated power. * If any diesels are on-line, they must be operated at least at the user-specified Minimum Diesel % Loading. 30 To convert wind turbine power output allowed to Wind Turbine Capacity Allowed, the wind turbine power output allowed is divided by the recent average wind turbine capacity factor, which is the ratio of the actual power output of the wind turbines to the rated capacity of the wind turbines. In this way, the calculation of Wind Turbine Capacity Allowed takes into account prevailing wind conditions. Wind Turbine Dispatch is based on both statistical and instantaneous dispatch criteria. Statistical dispatch criteria can increase or decrease Wind Turbine Capacity Allowed to allow a wind turbine(s) to come on-line or take a wind turbine(s) off-line. Instantaneous dispatch criteria can only decrease Wind Turbine Capacity Allowed in order to immediately take a wind turbine(s) off-line if there is insufficient load available on the bus to absorb the wind power. 4.2.1.1 Statistical Wind Turbine Dispatch Statistical wind turbine dispatch criteria serve as the foundation of wind turbine dispatch. Their objective is to predict the amount of wind power the system can absorb as well as the actual wind turbine power output in the near future, and to dispatch wind turbine capacity accordingly. Statistical wind turbine dispatch criteria are evaluated every minute, or in response to special events that indicate an imminent loss of remote dump load, or if a diesel has just been brought on-line or taken off-line. (The last condition is necessary because the required power contribution from the diesel generators changes when the amount of diesel capacity on-line changes.). The process of evaluating statistical wind turbine dispatch criteria is called Statistical Wind Turbine Dispatch. Statistical Wind Turbine Dispatch consists of the following steps: 1. Determining the appropriate statistical wind turbine dispatch mode . Evaluating the corresponding criteria to determine Wind Turbine Capacity Allowed 3. Determining whether or not it is acceptable to allow a (or another) turbine to start and executing Wind Turbine Run Select. The following sections describe these steps. 4.2.1.1.1 Statistical Wind Turbine Dispatch Modes The first step in Statistical Wind Turbine Dispatch is to determine the appropriate dispatch mode. The amount of wind power the system can absorb depends on total available load on the bus, which consists of the village load and the total available dump load.” The total available dump load consists of local dump load and remote dump load if it is available. “In reality, both the kW required to spin the rotary converter (if the AC machine is on-line) and the kW required to boost charge the battery (if the system is boost charging) also contribute to the total load on the bus. However, these loads are negligible and have been excluded from 31 Local dump load must be available for the system to be in any of the operating modes in which wind turbine operation is allowed.” Sometimes the system has advance warning that one of the dump loads will soon become unavailable, as when the operating temperature of its corresponding hydronic loop is reached. If there is an imminent loss of local dump load, then the system will drop into Mode 0, and all operating wind turbines will be shut down. This is not considered part of wind turbine dispatch. The standard mode of Statistical Wind Turbine Dispatch assumes that local dump load is available and that if remote dump load is available, it will remain available. However, in the case of imminent loss of remote dump load, there is a second dispatch mode, which adjusts the Wind Turbine Capacity Allowed to reflect the soon-to- be-reduced dump load capacity available. Thus, there are two statistical wind turbine dispatch modes: e Statistical Wind Turbine Dispatch Standard Mode e Statistical Wind Turbine Dispatch Lose Remote Dump Load Mode. 4.2.1.1.2 Statistical Wind Turbine Dispatch Criteria Once Statistical Wind Turbine Dispatch has determined the appropriate dispatch mode, the next step is to evaluate the corresponding dispatch criteria to determine Wind Turbine Capacity Allowed. There are three criteria in each statistical wind turbine dispatch mode: Average kw, Peak kW,/and, Minimum Instantaneous. The final Wind Turbine Capacity Allowed value will be the smallest of the values given by each of the three criteria. Statistical Wind Turbine Dispatch Average kW Criterion The statistical wind turbine dispatch average kW criterion determines the maximum amount of wind turbine capacity that ensures that the average near-future total available load is sufficient to absorb the average near-future wind power plus the required contribution from the diesel generators. Predicting the average future total available kW load requires predicting the average future village kW load. Although not perfect, the recent past is the best available predictor of the future. Therefore, the average kW criterion uses the most recent 20-minute average value of village kW load to determine the expected average village kW load. The expected average wind power output is predicted using the most recent 20-minute average wind turbine capacity factor. The following is an example of a statistical wind turbine dispatch average kW criterion, the one that applies if the system is not about to lose remote dump load: Statistical Wind Turbine Dispatch. The only effect this simplification has is to make Statistical Wind Turbine Dispatch slightly more conservative. * However, sudden loss of remote dump load could cause there to be insufficient load available on the bus to absorb all the wind power and could result in an overfrequency shutdown. 32 Wind Turbine Capacity Allowed = Village kW 10m Average + Total Dump Load kW Available - Minimum Diesel kW Load Wind Turbine Capacity Factor 10m Average Statistical Wind Turbine Dispatch Peak kW Criterion The objective of the statistical wind turbine dispatch peak kW criterion is to determine the maximum wind turbine capacity that ensures that the minimum near-future total available load is sufficient to absorb the maximum near-future wind power while maintaining a user-specified amount of unused dump load. The statistical wind turbine dispatch peak kW criterion does not subtract the minimum diesel load from the total available kW load to determine the amount of wind power the system can absorb. This is because the minimum diesel load is considered available to absorb wind power (by unloading the diesel) on a transient basis. However, this leaves the system without a load buffer to absorb unanticipated wind power spikes. This buffer can be provided in the form of unused, or excess, dump load. Because the magnitude of the possible wind power spikes depends on the amount of wind turbine capacity on-line, the required excess dump load is calculated as a user-specified percentage of the wind turbine capacity on-line. This percentage is called the Dump Load Headroom %. The following is an example of a statistical wind turbine dispatch peak kW criterion, the one that applies if the system is not about to lose remote dump load: Wind Turbine Gapacity Alowed = (Village kW 20m Mininum + Total Dunp Load kWAvailable-Wind Turline Capacity OnLineX Dump Loal kW Headoom %) Wind Turbine Capaciy Factor 2m Maximum Statistical Wind Turbine Dispatch Minimum Instantaneous Wind Turbine Capacity Allowed Instantaneous Wind Turbine Dispatch will take a wind turbine off-line immediately if the instantaneous power quantities on which the instantaneous wind turbine dispatch criteria are based exceed user-specified limits (see Section 4.2.1.2). However, the Wind Turbine Capacity Allowed determined by the main statistical wind turbine dispatch criteria may not have changed by the time those criteria execute next. This could result in Statistical Wind Turbine Dispatch adding a wind turbine after Instantaneous Wind Turbine Dispatch has just removed one. The Minimum Instantaneous Wind Turbine Capacity Allowed Criterion is to prevent this from happening by ensuring that Wind Turbine Capacity Allowed is never greater than the most recent 20-minute minimum Instantaneous Wind Turbine Capacity Allowed. The criterion is the same regardless of whether the system is about to lose remote dump load or not. 33 Statistical Wind Turbine Dispatch Minimum Instantaneous Wind Turbine Capacity Allowed Criterion: Wind Turbine Capacity Allowed = Instantaneous Wind Turbine Capacity Allowed 20m Minimum 4.2.1.1.3 Statistical Wind Turbine Dispatch Wind Turbine Start Criteria Once Statistical Wind Turbine Dispatch has determined the appropriate dispatch mode and evaluated the corresponding dispatch criteria to determine Wind Turbine Capacity Allowed, the next step is to determine whether or not it is alright to allow a turbine to start. Wind Turbine Capacity Allowed greater than wind turbine capacity currently on-line does not necessarily mean that a wind turbine can be started, even if the difference is greater than the capacity of a single turbine. There are two conditions that must be met before a turbine will be allowed to start: (1) the system must not be facing imminent loss of remote dump load, and (2) there must be sufficient available load to start a turbine. The first condition is straightforward. The second condition involves several additional considerations. First, the required excess dump load included in the statistical wind turbine dispatch peak kW criterion, and therefore used to help determine Wind Turbine Capacity Allowed, was based on the amount of wind turbine capacity currently on-line. There must be an_additional turbine’s worth of required excess dump load to allow a turbine to start. Second, when a wind turbine comes on-line, it initially generates a surge of extra power before dropping to a lower power level consistent with the wind speeds. There must be sufficient excess dump load to absorb this power surge to allow a wind turbine to start. This requires quantifying the largest expected power surge, called the Wind Turbine Start Overshoot. This value is equal to the maximum wind power the particular wind turbine is expected to produce multiplied by a user-specified power surge factor, called the Wind Turbine Start Power Surge Factor. The maximum wind power the particular wind turbine is expected to produce is based on its capacity and the recent 20-minute maximum capacity factor of other turbines on-line, if any. The Wind Turbine Start Overshoot is given by the following expression: Wind Turbine Start Overshoot = Wind Turbine Capacity X Wind Turbine Capacity Factor 20m Maximum X Wind Turbine Power Surge Factor 4.2.1.2 Instantaneous Wind Turbine Dispatch There is only one instantaneous wind turbine dispatch mode and one instantaneous wind turbine dispatch criterion, therefore Instantaneous Wind Turbine Dispatch consists of evaluating the criterion to determine Instantaneous Wind Turbine Capacity Allowed and determining whether or not the resulting value warrants executing Wind Turbine Run Select. 34 Statistical wind turbine dispatch peak kW criteria attempt to ensure that the minimum near-future total available kW load is sufficient to absorb the maximum near-future wind power. These criteria rely on predictions of expected wind power and load. Instantaneous Wind Turbine Dispatch is provided to serve as a backup to Statistical Wind Turbine Dispatch. Instantaneous Wind Turbine Dispatch immediately takes a wind turbine off-line if the wind power exceeds the total available kW load minus the user-specified amount of required excess dump load. Instantaneous Wind Turbine Dispatch criteria are evaluated every PLC scan. As with diesel dispatch, the statistical wind turbine dispatch criteria use certain power quantities that are calculated and averaged over a several-second interval and cannot be measured instantaneously. However, the instantaneous wind turbine dispatch criteria must be based exclusively on quantities that can be measured instantaneously, such as diesel kW and dump load kw. Excess wind power can be expressed in terms of instantaneous power quantities by the Secondary Load Request (as determined by the diesel load control algorithm in the system controller). The objective of instantaneous wind turbine dispatch criteria expressed in terms of instantaneous quantities is then to immediately take a wind turbine off-line if the Secondary Load Request plus the user-specified amount of required excess dump load exceeds the total dump load available plus the total diesel load. Diesel load is added because it is considered available to absorb wind power (by unloading the diesel) on a transient basis. Instantaneous excess wind power is converted to an equivalent wind turbine capacity using the Instantaneous Wind Turbine Capacity Factor. Instantaneous Wind Turbine Dispatch Criterion: Instantaneous Wind Turbine Capacity Allowed = Wind Turbine Capacity On Line - (Secondary Load Request + WT Capacity OnLine X Dump Load kW Headroom% - Total Dumpload kW Available -Diesel kW) Instantaneous Wind Turbine Capacity Factor If the Wind Turbine Capacity Allowed is less than the current wind turbine capacity on-line, Instantaneous Wind Turbine Dispatch will cause a wind turbine to be removed by setting Wind Turbine Capacity Allowed equal to the Instantaneous Wind Turbine Capacity Allowed value just determined and executing Wind Turbine Run Select. 35 4.2.2 Wind Turbine Run Select Once the amount of Wind Turbine Capacity Allowed to be on-line is known, the second phase of Wind Turbine Dispatch is to determine which wind turbines to allow to start or take off-line in order not to exceed the allowed capacity. This process is called Wind Turbine Run Select. Wind Turbine Run Select determines how many wind turbines are allowed to be on-line by dividing Wind Turbine Capacity Allowed by the capacity of a single wind turbine.” Wind Turbine Run Select then compares the number of wind turbines allowed to be on-line to the number of wind turbines actually on-line. To approximately equalize run time on the wind turbines, the wind turbines are started and stopped in a/first-on, first-off (FIF O) manner. If a turbine needs to be taken off-line, Wind Turbine Run Select will search the list of turbines for the turbine that was the first to come on- line and select that one to be taken off-line. If more than one turbine needs to be taken off-line, Wind Turbine Run Select will repeat the process, searching for the next turbine that is on-line. If a turbine can be allowed to start, as determined by Statistical Wind Turbine Dispatch, Wind Turbine Run Select will search the list of turbines for the turbine that has been off-line the longest and is ready to run (available and sufficient winds) and select that one to start. If more than one turbine can be allowed to start, Wind Turbine Run Select will repeat the process, searching for the next turbine that is ready but not already on-line.’ 4.3. AC Machine Dispatch Because there is only one AC machine, AC Machine Dispatch simply consists of deciding when to run the AC machine and when to shut it off. AC Machine Dispatch is distinct from AC Machine Run Control, which is the software module that controls the starting and stopping of the AC machine. AC Machine Run Control is performed by the Rotary Converter Control Cabinet (RCCC) PLC. AC Machine Dispatch is enabled if the system is operating in Auto Mode 2 or 3. It is disabled if the system is operating in Manual Mode, Auto Mode 0, or Auto Mode 1. AC Machine Dispatch is based on both statistical and instantaneous dispatch criteria. Statistical dispatch criteria determine both when to bring the AC machine on-line and when to take it off- line. Instantaneous dispatch criteria act only to prevent the AC machine from coming on-line when there is insufficient diesel capacity to run the pony motor. The following sections describe these two criteria. * This algorithm assumes that all the wind turbines have the same capacity. This method would need to be modified to accommodate different-sized turbines. * To avoid the large power transient that would result from multiple simultaneous wind turbine contactor closures, only one turbine is allowed to actually start at a time. 36 4.3.1 Statistical AC Machine Dispatch Statistical AC machine dispatch criteria serve as the foundation of AC machine dispatch. Their objective is to determine if any system conditions exist under which the AC machine could provide a benefit and to dispatch the AC machine if that is the case. The primary benefit that the AC machine could provide is to allow diesel-off operation under certain conditions. In addition, the AC machine can be used to prevent another diesel from starting under other conditions. Statistical AC Machine Dispatch will dispatch the AC machine if any of these conditions exist. Statistical AC Machine Dispatch will take the AC machine off-line if none of these conditions exist and there is at least one diesel on-line. Statistical AC machine dispatch criteria are evaluated every minute. The process consists of the following steps: 1. Determining whether or not there is sufficient diesel capacity to run the pony motor 2. Determining the appropriate statistical AC machine dispatch mode 3. Determining if the AC machine could be used to allow diesel-off operation 4. Determining whether or not the AC machine could be used to eliminate the need to start another diesel. The following sections describe these steps. 4.3.1.1 Sufficient Diesel Capacity to Run the Pony Motor The first step in Statistical AC Machine Dispatch is to determine whether or not there is sufficient diesel capacity to run the pony motor. If the AC machine is not on-line yet, it must be spun up to speed by the pony motor. The pony motor’s peak power draw is about 10 kW. Therefore, regardless of other system conditions, Statistical AC Machine Dispatch will not initially dispatch the AC machine unless diesel capacity on-line is at least 10 kW greater than Diesel Capacity Required (as determined by Diesel Dispatch). This requirement no longer applies once the AC machine is already on-line. 4.3.1.2 Statistical AC Machine Dispatch Modes The next step in Statistical AC Machine Dispatch is to determine the appropriate dispatch mode. The primary benefit that the AC machine could provide is to allow diesel-off operation if system conditions exist that would otherwise allow diesel-off operation. Principally, there must be sufficient wind power to allow diesel-off operation. Recall from Statistical Diesel Dispatch Diesel-Off that the criteria used to determine if there is sufficient wind power to allow diesel-off operation depend on whether the battery bank/DC machine is projected to be on-line or not. Therefore, the AC machine dispatch modes are based on whether the battery bank/DC machine is allowed to be on-line or not: 37 1. Statistical AC Machine Dispatch DC Not Allowed Mode 2. Statistical AC Machine Dispatch DC Allowed Mode The battery bank/DC machine is only allowed to be on-line if the system is in Auto Mode 3 and the batteries are not being boost charged. 4.3.1.3 Diesel-Off Criteria Once Statistical AC Machine Dispatch has determined the appropriate dispatch mode, the next step is to determine whether or not diesel-off operation would be allowed if the AC machine were on-line. Statistical Diesel Dispatch requires that there be sufficient wind power to shut all the diesels off before diesel-off operation is allowed. Furthermore, Diesel Run Select will not actually shut all the diesels off until all of them have met their minimum run-time. Therefore, Statistical AC Machine Dispatch must ensure both these requirements, as well as sufficient diesel capacity to run the pony motor, before dispatching the AC machine for the purpose of allowing diesel-off operation. The minimum run-time requirement is straightforward. However, two criteria must be met to satisfy the sufficient wind power requirement. Not surprisingly, these are the same criteria used by Statistical Diesel Dispatch to determine whether or not diesel-off operation is allowed. In addition, if the battery bank/DC machine is allowed to be on-line, these are the same criteria used by DC Machine Dispatch to determine whether or not to run the DC machine. Therefore, both components would be dispatched at the same time. 4.3.1.4 Statistical AC Machine Dispatch Additional Considerations Once Statistical AC Machine Dispatch has determined the appropriate dispatch mode and determined if diesel-off operation would be allowed if the AC machine were on-line, the last step is determine if there are any additional conditions under which the AC machine could provide a benefit. These additional considerations are described below. Rotary Converter kVA Support for a Wind Turbine Start A significant amount of current is required to magnetize the AOC wind turbine’s induction generator when the turbine first comes on-line. It is necessary to ensure that there is sufficient kVA capacity on-line to provide this inrush current before allowing a wind turbine to start. The amount of kVA capacity required is defined to be the most recent 1-minute average kVA load plus a user-specified amount of excess kVA capacity required for a wind turbine (WT) start. If the amount of kVA capacity required is less than the total kVA capacity on-line, then additional kVA capacity must be provided, either by the AC machine or another diesel. We would prefer to dispatch the AC machine rather than another diesel. Therefore, if additional kVA capacity is required for a wind turbine start, the AC machine is not already on-line, and there is sufficient diesel capacity to run the pony motor, then Statistical AC Machine Dispatch will dispatch the AC 38 machine. If any of these requirements are not met, or if the system is not in Mode 2 or 3, then a diesel will be dispatched to provide the additional kVA capacity. Rotary Converter kVAR Support The AOC wind turbines are provided with power factor correction capacitors. However, each turbine still consumes up to about 30 kVAR of reactive power. If there are multiple wind turbines on-line, it is possible that the amount of Diesel Capacity Required to meet the kVAR demands of the system may be significantly greater than the amount of Diesel Capacity Required to meet the kW demands of the system. In these cases, the control system dispatches the AC machine rather than another diesel to provide the kVAR support. To do so, it is necessary to dispatch the AC machine before the kVAR load exceeds the kVAR capacities of the diesel generators on-line in order to prevent another diesel from being dispatched first. Therefore, if the most recent 1-minute average kVAR load exceeds 90% of the current diesel kVAR capacity on-line, the AC machine is not already on-line, and there is sufficient diesel capacity to run the pony motor, then Statistical AC Machine Dispatch will dispatch the AC machine.” 4.3.2 Instantaneous AC Machine Dispatch There is only one instantaneous AC machine dispatch mode and one instantaneous AC machine dispatch criterion. Recall that Statistical AC Machine Dispatch checks for sufficient diesel capacity to run the pony motor before dispatching the AC machine. However, an unexpected change in wind power or load could cause there to no longer be sufficient diesel capacity to run the pony motor. Instantaneous AC Machine Dispatch acts only to immediately abort AC machine dispatch if this situation results in a diesel overload before the AC machine is actually on-line. This is accomplished by monitoring instantaneous diesel kW load. If the instantaneous diesel kW load exceeds the diesel capacity on-line and the AC machine is not yet on-line, AC machine dispatch will be aborted. This criterion is evaluated every PLC scan. 4.4 DC Machine Dispatch Because there is only one DC machine, DC Machine Dispatch simply consists of deciding when to run the DC machine and when to shut it off. DC Machine Dispatch is distinct from DC Machine Run Control, which is the procedure followed to bring the DC machine on-line and take it off-line. DC Machine Run Control is performed by the Rotary Converter Control Cabinet PLC. DC Machine Dispatch is enabled if the system is operating in Auto Mode 3 and the “If the amount of diesel capacity required to meet the kVAR demands of the system is significantly greater than the amount of diesel capacity required to meet the kW demands of the system, the current method will dispatch the AC machine to prevent another diesel from being started. However, if another diesel had already been started for other reasons (e.g., to provide kVA support for a wind turbine start), this method will not dispatch the AC machine to allow one of the diesels to shut off. In the latter case, both diesels would remain on due to the kVAR demands. This will be addressed in future versions of the control program. 39 batteries are not being boost charged. It is disabled if the system is operating in Manual Mode, Auto Mode 0, Auto Mode 1, or Auto Mode 2 or if the batteries are being boost charged (the auxiliary battery charger and the DC machine must never be on-line at the same time). DC Machine Dispatch is based on statistical dispatch criteria only. There are no instantaneous DC machine dispatch criteria because there are no power flow conditions that would require the DC machine to be taken off-line immediately. In addition, there is only one DC machine dispatch mode. The DC machine dispatch criteria are evaluated every minute. The objective of DC Machine Dispatch is to determine if system conditions are such that diesel- off operation would be allowed if the DC machine were on-line and to dispatch the DC machine if so. DC Machine Dispatch will dispatch the DC machine if these conditions exist and take the DC machine off-line if they do not. Statistical Diesel Dispatch requires that there be sufficient wind power to shut off all the diesels before diesel-off operation is allowed. Furthermore, Diesel Run Select will not actually shut off all the diesels until all of them have met their minimum run-time. Therefore, Statistical DC Machine Dispatch must ensure both these requirements before dispatching the DC machine. The minimum run-time requirement is straightforward. However, two criteria must be met to satisfy the sufficient wind power requirement. Not surprisingly, these are the same criteria used by Statistical Diesel Dispatch to determine if diesel-off operation is allowed. In addition, these are the same criteria used by AC Machine Dispatch to determine whether or not to run the AC machine if the battery bank/DC machine is allowed to be on-line. Therefore, both components would be dispatched at the same time. 4.5 Auxiliary Battery Charger Dispatch The need for boost charging is discussed in Section 8.5. Because there is only one auxiliary battery charger, Auxiliary Battery Charger Dispatch, or Boost Charge Dispatch, simply consists of deciding when to turn the battery charger on to boost charge the batteries and when to shut it off. Boost Charge Dispatch is distinct from Boost Charge Control, which is the procedure followed to actually bring the battery charger on-line and boost charge the batteries. Boost Charge Control is performed by the RCCC PLC. Auxiliary Battery Charger Dispatch is enabled if the system is operating in Auto Mode 0, Auto Mode 1, or Auto Mode 2, or if the system is operating in Auto Mode 3 and the DC machine is not on-line (the auxiliary battery charger and the DC machine must never be on-line at the same time). It is disabled if the system is operating in Manual Mode or if the system is operating in Mode 3 and the DC machine is on-line. * Boost Charge Dispatch refers to automatic dispatch of the auxiliary battery charger by the WDCP Controller. The operator can also manually request a boost charge. 40 Boost Charge Dispatch is based on both statistical and instantaneous dispatch criteria. Statistical dispatch criteria determine when to bring the auxiliary battery charger on-line. Instantaneous dispatch criteria determine when to take the auxiliary battery charger off-line. The following sections describe these two criteria. 4.5.1 Statistical Boost Charge Dispatch Criteria Statistical boost charge dispatch criteria are evaluated every minute. The process of evaluating statistical boost charge dispatch criteria is referred to as Statistical Boost Charge Dispatch. There is only one statistical boost charge dispatch mode. Statistical Boost Charge Dispatch determines when to begin boost charging. There are three system conditions required before the WDCP Controller will begin an automatic boost charge (in addition to those conditions required to enable Boost Charge Dispatch, i.e., that the system is in Auto Mode 0, 1, 2, or 3 and the DC machine is not on-line): 1. Sufficient diesel capacity on-line to boost charge the batteries at maximum rate 2. No wind turbines are about to start 3. The user-specified number of days since the last complete boost charge has elapsed. The first requirement is that there be sufficient diesel capacity on-line to charge the batteries at the maximum battery charger current. The maximum allowed battery charger current is 26 amps. This would be 7.8 kW at 300 VDC. Therefore, Statistical Boost Charge Dispatch will not initially dispatch the auxiliary battery charger unless diesel capacity on-line is at least 8 kW greater than Diesel Capacity Required (as determined by Diesel Dispatch). The second requirement is that no wind turbines are about to start. This eliminates the risk of imposing the boost charge step load concurrent with the turbine start current inrush. The last requirement is that the user-specified number of days since the last complete boost charge, called the Boost Charge Interval, has elapsed. 4.5.2 Instantaneous Boost Charge Dispatch Criteria Instantaneous boost charge dispatch criteria are evaluated every PLC scan. There is only one instantaneous boost charge dispatch mode. Instantaneous Boost Charge Dispatch determines when to end boost charging. There are two system conditions that will cause Instantaneous Boost Charge Dispatch to take the auxiliary battery charger off-line (in addition to those conditions that cause Boost Charge Dispatch to be disabled, i.e., if the system is in Manual Mode or if the DC machine is on-line): 1. Insufficient diesel capacity available to provide Diesel Capacity Required 2. Boost charge complete. 41 The first condition is if there is insufficient diesel capacity available to provide the Diesel Capacity Required as determined by Diesel Dispatch. Statistical Boost Charge Dispatch checks for sufficient diesel capacity on-line to boost charge at maximum rate before dispatching the auxiliary battery charger. However, this condition is not required to continue boost charging. Once a boost charge has begun, the boost charge kW becomes a primary load that must always be met (as opposed to a secondary load which can be turned off on/off at any time). Therefore, Diesel Dispatch will dispatch sufficient diesel capacity to continue boost charging. Even so, situations may develop in which there is no longer sufficient diesel capacity available to ensure that all the current primary loads are met. In this case, Instantaneous Boost Charge Dispatch will immediately abort the boost charge to reduce the primary load. The second condition is if boost charge has completed. Boost Charge Control will indicate when the batteries are fully charged and it is no longer necessary to continue boost charging. 42 5 Power Flow Management The most critical task of the wind-diesel hybrid power system is to maintain constant voltage and frequency i in all modes of operation. As with any power system, this is accomplished by maintaining a balance of both real and reactive power at all times. The principles of frequency and voltage regulation are explained more fully below. 5.1 Frequency Regulation The entire power system, including all its generators, distribution wiring, and even motors present in the village load, can be thought of as one big electromechanical entity, as shown in Figure 3. Power flows into this system as power from the wind transferred to the wind turbine rotor, mechanical power developed in the diesel engines as a result of combustion, and electric power drawn from the battery. Power flows out of the system to consumer resistive loads, to consumer mechanical loads, to secondary loads, and as various mechanical and electrical losses. At any given moment, if more power is flowing into the system than out of it, the difference will be stored as an increase in kinetic energy of the rotating machines within the system, both generators and motors, that happen to be on-line at that time. The effect of any power imbalance in the system is expressed in the following equation: > Psources — > Psinxs = one = “> Jia where P = active power (kW) K.E.= kinetic energy of system J =moment of inertia of rotating machine @= angular velocity of rotating machine 4 This i increase in kinetic energy is manifested as an increase in rotational speed of the _ ~* “synchronous machines in the system and thus an increase in electrical frequency. The task of frequency regulation is essentially a problem of maintaining an instantaneous balance of the real power flowing into and out of the system. WIND POWER POWER a) ELECTRIC POWER TO RESISTIVE LOADS SYSTEM DIESEL POWER ——s SHAFT POWER TO MECHANICAL LOADS (STORED KINETIC ENERGY) BATTERY POWER seem MSCELLANEOUS LOSSES Figure 3. Real power flows into and out of the power system 43 5.2 Voltage Regulation Analogously, regulating the AC voltage of the power system is a problem of maintaining equilibrium between the source and sinks of reactive power (VARs) in the system. The_ induction generators of the wind turbines, transformers in the distribution system, and induction motors in the consumer load are all reactive power sinks. Power factor correction capacitors on the wind turbines or the distribution system are sources of reactive power. Synchronous generators, both on the diesel gensets and on the rotary converter, can be either sources or sinks, but generally, they are supplying the reactive power demanded by the sinks. Unlike the case of real power, where an imbalance can be absorbed by the system as a change in stored kinetic energy, there is no storage mechanism for “reactive energy.” The reactive power supplied by the sources is inherently equal to the reactive power absorbed by the sinks. This is expressed in the equation below, in which the reactive power flows for each component are expressed as functions of voltage. DY Ones Ao) in YO smns (Vac) =0 where Q=reactive power (kVAR) Vac = AC bus voltage If the reactive power sources are unable to deliver the reactive power demanded by the sinks, the bus voltage will fall such that the equilibrium is maintained. With reactive power, the issue is not so much ensuring that equilibrium is maintained (which is automatic), but that the equilibri rium_occurs. at_the desired voltage level. On a synchronous machine, the function of the voltage regulator is actually to control the generator excitation such that the generator delivers the reactive power demanded by the load at the desired voltage. 5.3 The Power Flow Management Algorithm There are three main power components subject to the direct control of the wind-diesel controller: the rotary converter AC machine, the rotary converter DC machine, and the secondary load controller (which actually consists of multiple distributed load controllers). Each of these devices has several different control modes associated with it. For example, the AC machine can be controlled to achieve any of the following: Match voltage with the AC bus (before synchronization) Share reactive power with the diesel generators Deliver a specified amount of reactive power to the grid Regulate AC bus voltage The power flow management algorithm determines the appropriate control mode for each of these three devices depending on the system state. In the context of Power Flow Control, system state refers not only to the system operating state as defined in Section 3; it also includes additional detail such as the state of charge of the battery and the instantaneous amount of excess wind energy available. The Wales wind-diesel hybrid power system involves multiple diesels and multiple wind turbines. In addition, there is a power converter consisting of two separate rotating machines and a secondary load that is divided into local dump load and remote dump load. Because each of these components may or may not be operating at any given time, there are a great number of possible system operating states. To develop a power flow management algorithm flexible enough to handle all possible operating states, one must identify a minimum set of key state variables that provide sufficient information to determine the appropriate control mode for each device. Our top-level state variable is the diesel status, because it has the greatest effect on how voltage and frequency is regulated. Diesel ON refers to the state where one or more diesel generators is connected to the bus and loaded (i.e., not in load or unload ramp). Conversely, Diesel OFF refers to the state in which all diesel generators are either disconnected from the bus or connected but not fully loaded. 5.3.1 Diesel ON State The stand-alone diesel generator is designed to regulate the voltage and frequency on an isolated power bus. In a multiple diesel configuration equipped with automatic load-sharing controls, the diesels collectively regulate frequency and share both the real and reactive power load in proportion to their respective ratings. Diesel gensets do an excellent job of frequency and voltage control if the real and reactive power load on them remains within their rated capacity and they are not subject to large reverse power transients. In the Diesel ON state, we allow the diesel generator(s) to perform their intended function of frequency and voltage control, and we control In summary, in Diesel ON state: e The diesel generator(s) assume both frequency and voltage control ¢ Power flow to the secondary loads and/or energy storage is controlled to maintain diesel loading within a comfortable range e The rotary converter AC machine is used to assist the diesel generators in meeting the VAR load, as necessary. 5.3.2 Diesel OFF State In the Diesel OFF state, the only synchronous machine operating in the system is the AC machine of the rotary converter. The rotational speed of the rotary converter will establish the grid frequency. As with the diesel generator, the voltage regulator on the rotary converter AC 45 machine controls the field current so as to maintain the desired AC bus voltage. Frequency is controlled by modulating power flow to the secondary load or battery, depending on factors discussed below. 5.3.3 Other System State Variables Diesel status is only the first of the system state variables used in determining the appropriate control mode for the various system components. The others reflect the state of readiness of the other system components and the nature of the instantaneous real power imbalance on the system. They are embodied in the following questions: 1. Is the (rotary converter) AC machine on-line and ready? Just as with the diesel generator, for the AC machine to be available to perform its control function, not only must its contactor be closed, but it must also not be in an unload ramp, preparing to go off-line. 2. Is the DC machine on-line and ready? Similarly, the DC machine is only available for control when its contactor is closed and it is not in a transitional state. 3. Is there instantaneous excess wind power? In the case where there is excess wind power, secondary (or “dump”) load may be used to provide frequency control. As long as there is excess wind power, this works fine, but suppose the wind suddenly drops, resulting in a power deficit. As wind power drops, secondary load will be rapidly removed in an attempt to maintain grid frequency. Once it has all been removed, the ability to control frequency is lost. The system must switch immediately to frequency control by the DC machine. 4. Is the battery at High State of Charge? This question is actually several criteria rolled into one, all designed to detect when the battery state of charge (SOC) reaches a maximum desired operating level: e Is the battery actually at a high state of charge, as determined by amp-hour integration? e Is the DC field current limit of the rotary converter reached? e Is the charging voltage limit of the rotary converter reached? When any of these criteria are met, the battery is considered to be at a high state of charge, which requires that charging current be limited until the state of charge falls back below a certain level. (See Section 8 for more information on battery management.) 46 Note that the state variables presented above are concerned only with whether or not the various system components are on-line and ready at a particular moment in time, not when and why they are brought on-line. The criteria by which individual diesels, wind turbines, and the rotary converter AC and DC machines are turned on and off are the subject of a whole suite of dispatch algorithms, which are covered in Section 4. 5.3.4 Power Flow Management Algorithm Flowchart The power flow management algorithm is presented in flowchart format in Figure 4. Each decision block represents one of the state variables described above. Each branch in the decision tree specifies the control mode of the devices actively participating in voltage and frequency control in the corresponding state. Note that each branch loops back to the beginning of the algorithm, because any of the key state variables can change at any moment. 47 START POWER FLOW CONTROL Loop as RETURN TO} ‘START DL = SECONDARY LOAD CONTROLLER AC = ROTARY CONVERTER AC MACHINE VOLTAGE REGULATOR Yes SIESEL(S) 0} iD LOADED: Yes ; RETURN TO| S~ DL: DIESEL LOAD CONTROL pont eta ‘| AC: VAR SHARING DC = ROTARY CONVERTER DC MACHINE FIELD CONTROL ‘No: i-@ ‘AC: AC BUS VOLTAGE REGULATION aa ae = wos vase to con <> no] oc rcavmcrcone | fr No-»| DL: OFF RETURN TO| S " DL: OFF RETURN TO| DC: DIESEL LOAD CONTROL ‘START x S o>) DG: FREQUENCY CONTROL ‘START Nows| DL: CHARGE RATE LIMITING RETURN TO| DL: CHARGE RATE LIMITING RETURN TO| DC: DIESEL LOAD CONTROL START DC: FREQUENCY CONTROL ‘START DL: | eet ee LOAD CONTROL Yes, DL: HOLD ZERO AVG. DC CURRENT RETURN ro] | eet ee HOLD ZERO CHARGING CURRENT *| Dc: FREQUENCY CONTROL ‘START : Pe Figure 4. Power flow algorithm In the Wales wind-diesel control system, the loop shown in Figure 4 is executed approximately once every 40 milliseconds (ms). A short loop interval is necessary to detect and immediately respond to changes in component status. For example, when the last diesel goes off-line, the rotary converter must step in immediately to control the grid frequency and voltage. If the transition is too slow, unacceptable deviations of either voltage or frequency could result. When a change in state occurs that calls for a change in the control modes of one or more devices, it is important that the mode changes occur seamlessly, without causing discontinuities in power flow, which would be manifested as frequency or voltage transients on the line. This requirement is not expressed in the flowchart, but it is an important part of the design of the various control modes and requires careful application of bumpless transfer techniques. 48 6 Diesel Load Control As discussed in Section 5, bus frequency is controlled by the diesel generator(s) whenever one or more of them is on-line. During diesel-on operation, secondary load (which may be dump load or power to charge the battery) is used to maintain the total diesel load at the diesel load set point if the diesel load would otherwise fall below it. Diesel Load Control is the software module in the Wind-Diesel Control Panel PLC program that continually calculates the amount of secondary load required to maintain the diesel load at the set point. Diesel Load Control uses the PLC’s built-in proportional-integral-derivative (PID) loop feature to perform this calculation. PID control loops have two inputs, the set point and the process value (the quantity that one is trying to control), and one output, the control output. The controller changes the control output in an attempt to minimize the error, which is the difference between the set point and the process value. In this case, the set point represents the desired diesel kW load, the process value is the actual measured diesel kW load, and the control output is the Secondary Load Request. 6.1 The Diesel Load Set Point In most cases, the diesel load set point will be based on the minimum allowed diesel % loading, which is a user-settable parameter based on the engine manufacturer’s recommendations. Cummins recommends for the LTA10 engines that the genset be loaded so as to maintain an exhaust temperature of at least 650°F. Experience in Wales indicates that this is achieved when the genset is operated at 20-25% of rated load. There is another diesel loading criterion that may override the minimum allowed diesel % loading, which is based on diesel plant heat required. There is a certain amount of heat required to keep the plant warm and the non-operating diesel engines at a sufficiently high temperature to provide for rapid starting and synchronization. The plant heat requirement is calculated as the sum of (1) the heat required to keep the diesel engines hot given a plant indoor temperature of 70°F, and (2) the additional heat required to keep the plant at 70°F given the existing outdoor ambient temperature. Plant heat is supplied through a combination of diesel jacket water heat and electric energy sent to the local dump load boiler. There are situations when the wind and the village load are such that the available wind power, in combination with the diesel running at minimum allowed load, is just enough to meet the village load. When the ambient temperature is very low, the heat supplied by the diesel running at minimum allowed load might be less than the plant heat required. In that case, the diesel load set point is increased until the plant heat produced, counting both jacket water heat and local dump load heat, equals the calculated plant heat required. Note that the jacket water heat is estimated as being equal to the electric output of the diesel generator. 49 6.2 Diesel Load Control Modes PID control uses three parameters to define the control function: proportional gain, integral gain, and derivative gain. In industrial PID controllers, the integral and derivative gains are often reformulated as integral and derivative time constants, often referred to as reset and rate, respectively. Tuning the control loop consists of determining the optimal values for each of these three parameters, which will depend on the dynamics of the system. The dynamics of the Wales power system depend on the response characteristics of the grid (the system being controlled) and the response of the device being used to control it. In the case of diesel load control, the system response depends primarily on which large rotating machines are on-line at any given time. The actuator response depends on whether the dump load or the DC machine is being used to provide the secondary load necessary to maintain diesel load. Considering the multiplicity of diesels and wind turbines, it would be overly complex to provide a different control loop tuning for every possible combination of machines on-line. As a reasonable compromise, we have defined four different control modes (sets of control gains) to correspond to the following four system states: 1) Dump Load Diesel Load Control — AC Machine Off 2) Dump Load Diesel Load Control — AC Machine On, DC Machine Off 3) Dump Load Diesel Load Control — AC Machine On, DC Machine On 4) DC Diesel Load Control - AC Machine On, DC Machine On. 50 7 Nondiesel Power Component Control This section deals with the specifics of how each of the nondiesel power components is controlled. These components are the dump loads (electric boilers), the AC machine side of the rotary converter, and the DC machine side of the rotary converter. 7.1 Dump Load Control As discussed in Section 6, Diesel Load Control is strictly a computational block that determines the amount of secondary load required. Whether the Secondary Load Request is met using dump load or by charging the battery is determined by the Power Flow Management algorithm, discussed in Section 5. Diesel Load Control is just one of several possible control modes for the dump load. Dump Load Control refers to the PLC software module that actually controls the local and remote dump loads. Dump Load Control has three aspects: 1) Determine how much total dump load is required. 2) Determine the appropriate allocation between local and remote DL dump load. 3) Issue the actual switching commands to control individual boiler elements. 7.1.1 Determining Dump Load Required: Dump Load Control Modes There are several different modes of Dump Load Control, depending on the control objective at any given time. Each of these uses a different method and/or different criteria for calculating the amount of dump load required moment by moment. The three modes are Diesel Load Control, Frequency Control, and Battery Charge Current Control. The Diesel Load Control mode of Dump Load Control is active when the dump load is being used to satisfy the Secondary Load Request as determined by the Diesel Load Control module. In this mode, Dump Load Required is simply set equal to the Secondary Load Request. This is known as pass-through, because the output from another computational block is simply passed through to the output of this one, without additional modification. The Frequency Control mode of Dump Load Control is active only in diesel-off operation when the DC machine is also off-line, which would normally only be the case if the DC machine were unavailable. In this state, the system is essentially a high-penetration no storage wind-diesel system. This control mode uses a built-in PID loop in the RCCC PLC. The set point of this PID loop is the desired bus frequency. The default value for this set point is 60 Hz. The process value (feedback) for the PID loop is the grid frequency, as measured by the magnetic pickup on the DC machine.” * Though the DC machine is not on-line in this mode, it is part of the rotary converter and therefore still spinning. Because the AC machine is a synchronous generator, its rotational speed is exactly proportional to the grid frequency. 51 The Battery Charge Current Limiting mode of Dump Load Control is active when the DC machine is on-line, the battery is not full, and there is excess wind power available. In this situation, excess power is used to charge the battery, but the charge current must be kept from exceeding the maximum allowed charging current. This control mode uses a built-in PID loop in the RCCC PLC. The process value for this loop is the measured DC armature current. The default set point value is 260A, which represents the 2C; charging rate. If the DC current starts to exceed this value, dump load will be applied so as to regulate the current at the set point value. If excess power drops such that the battery current drops below the set point, any on-line dump load will be removed, but otherwise there is no consequence. The Battery Charge Current Control mode of Dump Load Control is active in diesel-off operation, with the DC machine on-line, when there is excess wind power and the battery is full. This is the same as the Battery Charge Rate Limiting Mode, except that the set point is zero A instead of 260 A. Because the DC machine is busy controlling bus frequency, there will inevitably be current fluctuations in and out of the battery. The dump load control, however, will act to minimize these fluctuations and center them around zero amps, so that the net flow into the battery is zero. 7.1.2 Local/Remote Dump Load Apportionment Once the total dump load required is known, the controller must determine the appropriate split between local and remote dump load. This is done with the following considerations in mind: e Ensuring that the plant has sufficient heat is a higher priority than supplying excess heat to the school. e¢ No more excess power should be sent to the local dump load than is necessary to keep the plant and engines warm. e Fast response and high-resolution dump load control is only obtainable with the local dump load. Therefore, whenever any dump load is required, for either diesel load control or bus frequency control, the local dump load must be kept within a range that allows for controllability. Remote dump load must be adjusted to prevent the local dump load from either going to zero or reaching its maximum capacity. Hitting either extreme could result in short control dropouts or a complete loss of control. Apportioning total dump load between the local and remote dump loads is based on two parameters, the Local Dump Load Target Value and the Local Dump Load Deadband. The Local Dump Load Target Value represents the desired average kW value to be maintained by the local dump load, which, in turn, is the amount of heat that when added to the diesel jacket water heat will equal the total plant heat required. The Local Dump Load Deadband is the width of the power band, centered around the target value, within which the local dump load will be allowed to vary before making a change to the amount of power sent to the remote dump load. The local dump load target value is calculated as follows: 52 Local Dump Load Target Value = Min[Max( 20 kW, Required Dump Load Contribution to Plant Heat), 69 kW] where Required Dump Load Contribution to Plant Heat = (Plant Heat Required) — (1 minute average Total Diesel kW) Note that the Local Dump Load Target Value is constrained to be between 20 and 69 kW. The upper limit is 20 kW less than the capacity of the local dump load. The Local Dump Load Deadband is set at 30 kW, which is slightly larger than the remote dump load element size of 24 kW. This deadband defines the local dump load upper and lower limits, equidistant above and below the target value. Regardless of which dump load control mode is active (see Section 7.1.1), whenever the local dump load power level reaches one of the limits, the control system will attempt to add or remove a remote dump load element as appropriate. As shown in Figure 5, when the local dump load reaches the upper limit, a remote dump load element will be added if it is available, thereby keeping the local dump load within the deadband. Similarly, when the local dump load reaches the lower limit, a remote element will be removed, assuming there is one on to be removed. With random variations in dump load required, this approach will ensure that the average local dump load is approximately equal to the local dump load target value. 80 Local DL Upper Limit Local DL Lower Limit 70 Total DL = = = Local DL 60 Remote DL 50 = 40 30 20 10 Time Figure 5 Interplay between local and remote dump load as total dump load required varies. (Target local dump load = 25 kW.) 53 As remote dump load elements are added and removed, the amount of local dump load commanded is simultaneously adjusted by an amount equal to the remote dump load step size. Thus, the local/remote apportionment process is completely transparent to the overall dump load control, and the total dump load on the system smoothly varies according to the calculated Dump Load Required. 7.1.3. Dump Load Element Switching Both diesel load control and dump load frequency control require a dump load with fast response and good resolution. The local dump load provides fast response using a high-speed communications link to the RCCC PLC and solid-state relays to switch the elements. It provides good resolution with its sizing of the heating elements in a binary progression. The local dump load boiler is equipped with six heating elements of the following nominal sizes: 2.5, 5, 10, 20, 20, and 20 kW. Two of the 20-kW elements are combined to form a 40-kW virtual element and are always switched on or off simultaneously. Thus, the local dump load is organized as a five element binary progression: 2.5, 5, 10, 20, and 40. By switching these elements appropriately, any power level from zero to the full nominal value of 77.5 kW may be obtained with 2.5-kW resolution. The remote dump load does not have the same requirement for fine resolution as the local dump load. Instead of a binary array of elements, it consists of six equal-sized elements, which has several advantages. Using equal-sized boiler elements and equal-sized, solid-state relays to switch them reduces the number of spare parts required. In addition, this approach lends itself to the future integration of additional remote dump loads, in the community water treatment plant, for instance. In that case, the additional dump load would simply appear to the system as an extension of a linear array of heating elements. The criteria for switching remote dump load elements in or out are described in Section 7.1.2. The actual switching of particular elements is based on a FIFO scheme. When an additional element of remote dump load is needed, the next available off element is turned on. When it is necessary to remove an element, the element that has been on the longest is turned off. This approach ensures equal duty on all of the boiler elements and solid-state relays. If an individual element has been disabled by the operator (or automatically by the control system), that element will be skipped over and the next available element turned on instead. * Because the boiler elements differ slightly from their nominal values, and because the system voltage is higher than the voltage at which the elements are rated, the actual step size is approximately 2.875 kW, and the total local dump load capacity is approximately 89 kW. 54 7.2 AC Machine Control AC Machine Control refers to the processes of starting and stopping the AC machine and controlling it to perform a specific task while on-line. Four outputs control the operation of the AC machine: Pony motor drive contactor close Closes the AC contactor to energize the pony motor signal variable-speed drive. Pony motor drive enable signal Causes the variable-speed drive to accelerate the pony motor and match the frequency of the rotary converter to the measured bus frequency. Voltage regulator bias signal Biases the set point of the AC voltage regulator that controls the field current in the AC machine. AC Machine contactor close signal Connects the AC Machine to the plant bus. The following sections describe how these four control outputs are used to control the AC machine during the various phases of its operation. 7.2.1. AC Machine Startup Sequence Once the controller makes the decision to dispatch the AC machine, it runs a startup sequence to bring it on-line. 1. Close the pony motor drive contactor and enable the drive to accelerate the rotary converter to synchronous speed. The pony motor variable-speed drive attempts to match the rotary converter speed/frequency to that of the grid. A frequency transducer on the plant bus is used for the set point of the pony motor speed control loop. A frequency transducer on the magnetic pickup (MPU) signal from the DC machine is used as the speed feedback to this control loop. Note that the pony motor drive only attempts to match speed. It does not ae eae attempt to bring the AC machine voltage in phase with the bus voltage. 2. Adjust the voltage regulator bias signal such that AC machine armature voltage matches the measured bus voltage. 3. When the AC machine voltage and frequency are in the right range, issue the AC contactor close command. 4. When the AC machine frequency, voltage, and phase match the bus within specified tolerances, the sync check relay in the Basler Generator Protective Relay (GPR) actually allows the AC contactor to close. When the AC contactor closes, the pony motor drive contactor opens. 5. Place the AC machine in kVAR Control Mode, with a zero set point. This mode will bias the voltage regulator such that there is no net reactive power flow in or out of the AC machine. In other words, it will be held at unity power factor. Do 7.2.2. On-Line Control Modes of the AC Machine There are three on-line control modes for the AC machine: kVAR Control In this mode, the voltage regulator set point is biased as necessary to maintain a fixed value of kVAR output by the AC machine. It is used only to maintain zero kVAR during the AC machine startup and shutdown sequences. Reactive Power Sharing In this mode, the voltage regulator is biased so that the AC machine shares the total reactive load with any on-line diesels, with each machine sharing in proportion to its kW rating. The total reactive load on the power system consists of the kVAR component of the village load plus the reactive power consumed by the wind turbines. Note that when multiple diesel generators are on-line, they share reactive load automatically using the cross-current compensation method. The AC machine is not part of the cross-current compensation loop. Its kVAR output is actively controlled by the RCCC PLC. Bus Voltage Control This mode is only used when no diesel generators are on-line. In this mode, the AC machine’s voltage regulator is responsible for controlling the plant bus voltage. Its set point is biased only as necessary to compensate for changing load conditions or drift in the voltage regulator. The operator is able to change the plant bus voltage set point from the RCCC touchscreen. Note that this setting only applies in diesel-off mode. 7.2.3 AC Machine Shutdown Sequence Once the controller makes the decision to no longer dispatch the AC machine, it runs a shutdown sequence to take it off-line. 1. Enter the kVAR Control Mode and ramp the set point to zero kVAR. This minimizes the current flow in the AC machine, allowing for a softer opening of the contactor. The current cannot be brought to zero, because a certain amount of real power is required to keep the rotary converter spinning. 2. When the kVAR drops below a specified threshold, open the AC contactor. When the contactor opens, the rotary converter coasts down to a stop. ae 56 7.3. DC Machine Control DC Machine Control refers to the processes of starting and stopping the DC machine and controlling it to perform a specific task while on-line. Four outputs control the operation of the DC machine, as shown in the following table: Field power supply enable signal Energizes the DC field current power supply Load resistor contactor close signal Connects the dummy load resistor to the output of the field current DC power supply Speed control bias signal Biases the set point of the speed control, which in turn, controls the set point of the field current power supply DC Machine contactor close signal Connects the DC Machine to the battery The rotary converter can be thought of as an additional genset. Rather than a diesel engine, the DC machine acts as the prime mover. The AC machine is a synchronous generator. The difference is that the rotary converter is bidirectional, i.e., it can absorb as well as generate AC electric power. Figure 6 shows the means by which the DC machine is controlled. Because the primary function of the DC machine is frequency control in diesel-off mode, a standard Woodward generator speed control is used as the main control component for the DC machine. Normally the output from the speed control would be connected to the fuel system actuator on an engine. In this case, the speed control output adjusts the set point of the DC field current power supply. By controlling the DC machine field current, the DC machine torque is controlled over its full operating range. When producing positive torque, the DC machine is a motor, driving the AC machine and generating AC power, but discharging the battery. When producing negative torque, the DC machine is a generator, charging the battery but causing the AC machine to act as a motor. POWER SUPPLY Dc BIAS CONTROL SIGNAL SIGNAL FIELD MACHINE [4 [——————"] FIELD CURRENT PLC WOODWARD CURRENT he ANALOG SPEED POWER OUTPUT CONTROL SUPPLY LOAD RESISTOR DC CONTACTOR > UI BATTERY Figure 6 Layout of DC machine control 57 Just as the set point of a diesel speed controller is often biased to perform load-sharing with other diesels or base-loading to a utility, the analog bias signal from the PLC to the speed controller allows the speed control to perform control functions other than pure frequency control, such as battery charge regulation. These control modes are explained in Section 7.3.2. 7.3.1 DC Machine Startup Sequence Once the controller makes the decision to dispatch the DC machine, it runs a startup sequence to bring it on-line. Note that this startup sequence is not initiated until the rotary converter is spinning at synchronous speed and the AC machine is on-line. Connect the DC voltage transducer to the battery and measure the battery voltage. Switch the DC voltage transducer to the DC armature. Turn on the DC field current power supply. Connect a load resistor to the field power supply output for several seconds to establish current flow through the silicon control rectifiers (SCRs). This step is necessary because of the very high inductance of the DC field coil. The impedance of the coil is so high that without an additional load across the output of the power supply, current will not build up enough to hold the SCRs on after the initial turn-on pulse. 5. Adjust the speed controller set point bias signal such that the DC armature voltage matches the battery voltage. 6. Close DC contactor when the voltage difference drops below a specified threshold. Enter DC Current Control Mode with a zero set point. This is a standby mode in which the DC machine is connected to the battery, but no current flows in either direction. (The power to keep the rotary converter spinning comes from the AC bus.) oS oe 7.3.2 On-Line Control Modes of the DC Machine There are three on-line control modes for the DC machine: DC Battery Charging Current Control In this control mode, the speed control’s set point is biased to regulate the DC current. This mode is normally only used when ramping the DC current to zero as part of the DC machine shutdown sequence. This mode is also used to control the battery discharge rate during a battery capacity test. DC Power Control In this control mode, the speed control’s set point is biased to control the DC power sent to the battery. This mode is used when Diesel Load Control is in the DC Diesel Load Control Mode (see Section 6.2). 58 Bus Frequency Control Whenever the DC machine is on-line and the system is in diesel-off operation, the DC machine is responsible for controlling the bus frequency at 60 Hz. Because this is the basic function of Woodward speed control, this control mode requires no outer control loop and no speed control set point bias signal. Thus, it is the most straightforward of the DC machine control modes. There are no PID gains in the PLC associated with this control mode. The DC machine bus frequency control dynamics depend solely on the parameters programmed directly in the Woodward speed control. 7.3.3 DC Machine Shutdown Sequence Once the controller makes the decision to no longer dispatch the DC machine, it runs the following shutdown sequence to take it off-line: 1. Enter the DC Current Control Mode and ramp the set point to zero amps. This provides for a soft opening of the DC contactor. 2. When DC current drops below a specified threshold, open the DC contactor. 3. De-energize the field current power supply. 4. When the DC machine goes off-line, the rotary converter reverts to acting as a synchronous condenser. 59 8 Battery Management 8.1 Battery Characteristics The Wales system battery bank consists of 200 Ni-Cd cells, each having a nominal voltage of 1.2 VDC, giving a total nominal battery voltage of 240 VDC. The cells have a Cs (5-hour discharge rate) rated capacity of 130 ampere-hour (Ah). Thus, the nominal energy storage capacity of the battery bank is 240 VDC x 130 Ah = 31.2 kWh. Because the discharge rates will often be higher than the CS rate, and because it is not desirable to operate the battery at either very high or very low states of charge, the actual usable capacity of the battery will be significantly less, perhaps 40-50% of nominal. If the actual usable battery capacity is 40% of 31.2 kWh, or 12.5 kWh, the rotary converter could deliver 50 kW (less converter losses) to the AC bus for approximately 15 minutes. Although the energy storage is of short duration, it can significantly improve the operating performance of the system. The terminal voltage of any battery is a function of the battery temperature, the charging (or discharging) current, and the battery’s state of charge (SOC). It is also a function of the battery’s recent charge history, the path by which it got to its current state. However, the battery management algorithms employed in the wind-diesel control software neglect any charge history effects. The general relation between battery voltage and state of charge during constant current charging is shown in Figure 7. Starting with a completely discharged battery, the voltage initially rises rapidly. As charging continues, the voltage increases slowly and nearly linearly with increasing state of charge. When the battery approaches a high state of charge, the voltage rises rapidly and then plateaus at a high level. Continued charging at this stage results in very little additional stored energy. Most of the input energy is dissipated as battery gassing. The “knee” in the charging curve typically occurs at 80-90% state of charge. Increasing the charging current shifts the knee to the left (to a lower SOC) and upward (to a higher voltage). 60 Ny - © 7 & , = a Increasing Current x _ fo) Cell Voltage = a 1.4 4 1.3 4 0 10 20 30 40 50 60 70 80 90 100 110 120 State of Charge (%) Figure 7 Constant current charging voltage vs. SOC The general relation between battery voltage and the depth of discharge (DOD) during constant current discharging is shown in Figure 8. Starting with a completely charged battery, the voltage initially falls rapidly. As discharging continues, the voltage decreases slowly and nearly linearly with increasing depth of discharge. When the battery approaches a low state of charge, the voltage falls off steeply. A cutoff voltage is typically defined as that at which the battery is effectively empty, even though it has not delivered its rated capacity. The knee in the discharging curve typically occurs at 95% state of charge at the rated discharging current. However, increasing the charging current shifts the knee significantly to the left (to a lower DOD) and downward (to a lower voltage). 61 Cell Voltage —_ increasing cur 0.9 0.8 + 0.7 0 20 40 60 80 100 120 Depth of Discharge (%) Figure 8 Constant current discharging voltage vs. DOD 8.2 State of Charge Tracking Accurately measuring the SOC of a battery is very difficult and generally requires special instrumentation. It is possible, however, to obtain a reasonable estimate of SOC by starting with a known state of charge and integrating all subsequent ampere-hours in and out of the battery. The longer this integration is performed, the more error will accumulate in the SOC estimate, so it is important to regularly reset the amp-hour count to correspond to a known SOC. In this context, known can mean estimated based on some other method besides amp-hour integration. The wind-diesel system controller uses the amp-hour integration method to provide a continuous estimate of battery SOC. Battery SOC is calculated mainly to provide the operator an approximate indication of the SOC of the battery. The estimated SOC is also used to reset certain battery status flags, as will be described in the next section. It is not directly used to define acceptable limits on battery charging. Determining the normal operating limits on battery charging is a dynamic process that will be described in the following sections. 62 8.3 Battery Status Flags The control system defines four battery status flags that indicate when certain points on the battery charge and discharge curves are reached, as described in the following table: Table 5. Definition of Battery Status Flags Corresponds to the knee in the battery-charging curve (see High SOC Figure 7), i.e., the point at which the battery voltage starts to rise rapidly. Battery Full Corresponds to the point, during constant current charging, where the voltage stops increasing. Corresponds to the knee in the battery-discharging curve (see Figure 8), i.e., the point at which the battery voltage starts to Lowe fall rapidly. This occurs when there is approximately 10% usable battery capacity remaining at the present discharge rate. : . . é Battery Empty Corresponds to the point at which there is less than 2% of usable battery capacity left at the present discharge rate. In a wind-diesel hybrid system, the battery current may be continuously changing, even changing from charging to discharging and back again within a short time. Thus, the battery may be continuously operating on a different characteristic charging or discharging voltage versus SOC curve. It is therefore difficult to detect when the battery is on the knee of a curve. As stated earlier, instantaneous terminal voltage for a given battery is a function of battery temperature, current, and SOC. We have previously analyzed the family of charge and discharge curves and come up with formulas approximating voltage at which the battery will reach three of the four points defined above (High SOC, Low SOC, and Battery Empty, for any combination of battery current and temperature.” These formulas take the following form: VA(L,T)=Vy 9 +k, <+k,(7-68), 5 where Vx = voltage corresponding to criteria X, e.g., High SOC Vxo = uncorrected threshold voltage for criterion X * The “Battery Full” criterion is based on charging voltage stabilization and will be explained in the section on boost charging. 63 ky = current correction factor (volt/hour) kr = temperature correction factor (volt/°F) Cs = nominal battery capacity (Ah) I = battery current (A). (Charging is positive, discharging is negative) T = battery temperature (°F) In addition to the voltage criteria, there are secondary criteria for setting the High SOC and Low SOC flags. These are based on the status of the field coil of the DC machine. The battery current, whether charging or discharging, is controlled by controlling the current in the DC machine field coil. Raising this current will increase the battery current; decreasing the field current decreases the battery charging current (or increases the discharging current). Ata high SOC, it requires more field current to achieve the same battery charging current. Conversely, at a low SOC, it requires a lower field current to achieve the same battery discharging current. There are upper and lower limits on DC field current that must be respected. There is also a maximum field coil operating temperature. Thus, there are field coil current and temperature criteria for setting the High SOC and Low SOC flags. These secondary criteria serve as a backup to the primary voltage criteria. In the previous section, it was mentioned that the running estimate of battery SOC based on amp- hour integration must periodically reset based on some other indication of SOC. The battery status flags are used for this purpose. Because the knee in the charging curve tends to occur near 85% state of charge, the SOC register is arbitrarily reset to 85% whenever the High SOC flag is set. The amp-hour integration then proceeds from the new value. Similarly, whenever the Battery Full flag is set, the SOC register is reset to 100%. The battery status flags are used in various power flow and component dispatch decisions. It is therefore important to reset them when they no longer apply. For example, once the battery discharges a certain amount after reaching a high SOC, the High SOC flag is reset. Similarly, once the battery charges a certain amount after the Low SOC flag is set, that flag will be reset. Table 6 summarizes the conditions for setting and resetting the battery status flags. Table 6. Battery Status Flag Setting/Resetting Criteria 64 Battery Status Flag Criteria for Setting Criteria for Resetting High SOC Voattery 2 Vu(LT) Or High field current limit reached Or High field coil temperature limit reached SOC < 75% Battery Full dV/dt = 0 during boost charge Or Calculated SOC reaches 100% not during boost charge SOC < 85% Low SOC Voattery < ViCLT) Or Low field current limit reached SOC 2 [(SOC when Low Limit was set) + 20%] Battery Empty Voattery < V;(LT) SOC 2 [(SOC when Battery Empty was set) + 20%] 8.4 Normal Charging and Discharging The two main operating modes of the DC machine are frequency control (when the diesel is off) and diesel load control (when the diesel is on). In both of these modes, the battery continuously absorbs or delivers as much power as demanded by the DC machine. Battery current varies continually and somewhat randomly. Thus, battery charging and discharging is essentially unregulated. The only exception to this is that the battery charging current is limited to 260A, which is twice the Cs rate. If the battery is called on to absorb so much excess wind power that the charging current starts to exceed this limit, dump load is increased to limit the battery current. 65 The battery management strategy is to operate the battery in a SOC range bounded on the upper end by the knee in the charging curve and on the lower end by the knee in the discharging curve. We believe that this strategy will provide the highest energy storage efficiency and the lowest rate of wear on the battery. Typically, the normal SOC operating range will be about 20 to 85%, but, in fact, the limits will vary depending on the prevailing charging and discharging currents. 8.5 Boost Charging Most batteries require a periodic boost charge, sometimes called equalization charge, to maintain full capacity. Lead-acid batteries must be brought to a state of full charge fairly frequently. Ni- Cd batteries do not require a boost charge nearly as often, but the manufacturer nevertheless recommends that the procedure be performed several times a year. In accordance with the battery management strategy described above, normally the battery goes no higher than approximately 85% SOC. The only time the battery is brought to 100% SOC is during a boost charge. Boost charging is performed using the constant current charging method, which is the only charging method that will bring a Ni-Cd battery to a full state of charge in a reasonable period of time. When a constant charge current is applied, the battery voltage will stabilize at approximately 1.85 volts/cell. During the boost charge, the battery voltage is monitored. When battery voltage stabilization occurs, the boost charge is terminated. 66 9 Fault Detection and Handling The Wales wind-diesel hybrid system is designed to generate power with high reliability. In a complex system, high reliability implies more than minimizing the failure rate of the various system components. It also implies an extensive system self-diagnostic capability, so that when component faults and other abnormal conditions do occur, they can promptly be detected, and the system can respond appropriately. Specifically, such capabilities should include the following: e Ifa particular component fails, the system should have sufficient operational flexibility that the component can be isolated and the system continue operating without it, perhaps in a more basic configuration. Systems with this capability are often referred to as robust or fault tolerant. e The system should provide sufficiently detailed annunciation of faults or other abnormalities so that the system can be repaired rapidly and cost-effectively with a minimum of troubleshooting. e The system should provide early indication of degraded component performance to alert the operator and service personnel of a potential problem or imminent failure. Timely attention to such indications can avert an actual component failure and possible disruption of service. The foundation for these capabilities in the wind-diesel control system is the fault detection and handling scheme. All faults occur in, or are at least associated with, a particular component or subsystem. For each component, therefore, we attempt to identify all possible faults or other abnormal conditions. These are divided into three categories, cautions, warnings, and alarms, which indicate different levels of urgency and initiate different system responses. A caution indicates that a particular monitored quantity is outside of its normal operating range and indicates a need for service and/or adjustment of the affected component. The system does not respond to a caution other than to annunciate the condition on the operator interface. The condition poses no immediate danger, and the component will continue to be available for dispatch. It is up to the operator and/or service personnel to investigate the condition and determine if any special action is warranted. Cautions are annunciated on the operator touchscreen by blue indicator lights. A warning indicates that a particular monitored quantity is outside of acceptable operating limits. Continued operation at such levels could lead to component damage and/or power system failure. The system responds to a warning by performing a controlled shutdown of the affected component. If it is a power-generating component, sufficient replacement generating capacity is brought on-line before the component is taken off-line. The affected component becomes unavailable and remains so until the precipitating condition is removed and the warning cleared. Warnings are annunciated on the operator touch screen by yellow indicator lights. 67 An alarm indicates a severe malfunction that poses an immediate danger to personnel and/or equipment. The system responds by immediately disconnecting and shutting down the affected component without regard for the ability of the remaining components to meet the primary load. An alarm condition is by definition sufficiently serious that it is preferable to risk a power outage than to allow the component to remain on-line. The affected component becomes unavailable and remains so until the precipitating condition is removed and the alarm cleared. Alarms are annunciated on the operator touch screen by red indicator lights. Except for a few special cases, all cautions, warnings, and alarms must be reset by the operator. This ensures that the operator is aware of the occurrence, even if it requires no special service or repair action in response. 68 10 Data Logging and Reporting The Wales system is the first high-penetration, wind-diesel system deployed to power an entire arctic village. Because it is a technology demonstration project and can serve as the basis for replication to many other Alaskan villages, careful monitoring and reporting of system performance is essential. To meet this need, a complete suite of software modules has been created to perform data acquisition, logging, transfer, decoding, analysis, and reporting. The entire data flow, from its acquisition in the PLC to its eventual reduction to summary report is depicted in Figure 9. BASIC -~ Co-Processor \__ Telephone - Modem ¢ a é NWTC (UNIX Server) a +—< Spreadsheet Decoding Program Monthly Report Summary Program Figure 9. System data logging, reducing, and reporting 10.1 Data Acquisition, Logging, and Transfer Data acquisition and logging capability is incorporated into the design of the control system. In an attempt to ensure the integrity of the data, the system was designed to be completely automatic, not requiring any local operator action. The PLC in the Wind-Diesel Control Panel monitors all energy flows, status relays, and alarm relays for all of the major system components. Because we wish to be able to look retroactively at how the system behaved during a given 69 period of operation, the data logging system is designed to preserve a detailed time history of hybrid system operation. To reduce the operating data to a manageable amount, all numerical quantities are stored as 10-minute averages in the PLC. To reduce the memory and computational overhead in the PLC central processing unit (CPU), the tasks of assembling and uploading the performance data files are handled by a BASIC coprocessor module, which occupies a slot in the PLC backplane. At the end of every data averaging interval, a data record consisting of all averaged quantities and a snapshot of all important status and alarm relays is copied from the CPU to the coprocessor module, where the records are assembled into a time- series data file. The data are stored in a hexadecimal (hex) format to conserve memory and reduce transmission time. The coprocessor module, via a serial port connected to a modem, periodically establishes a dial-up connection with a UNIX server at the National Renewable Energy Laboratory in Boulder, Colo., and uploads the accumulated data. Because of the large amount of data collected and the limited amount of memory available in the coprocessor, the upload must be performed once a day. While the PLC is capable of recording, storing, and sending information on every aspect of the wind-diesel system, this makes up only half of the overall performance monitoring process. When this data is uploaded to a server, it still must be decoded, analyzed, and reduced to a form that a person can look at and easily interpret. The software programs to perform the data decoding and analysis had to be developed for this specific application. 10.2 Data Decoding When the hex data are received by the NREL server, it is first converted back into its appropriate format. For instance, some of the hex characters represent text values, such as the date the information was recorded. Other characters represent groupings of Boolean values, the control relays that indicate the status of the various components or alarm conditions. Still others represent actual decimal values, like the wind speed or the battery voltage. The software that performs this decoding operation is written in PERL (Practical Extraction and Report Language), an interpreted high-level programming language commonly used in web site development. PERL was chosen for this task because of its powerful data manipulation abilities, as well as the fact that the program must be able to run in a UNIX environment, as that is where the raw data resides. This program first parses the data into columns, and then reads through it row by row, decoding each piece of information as it goes. Any gaps in the data, i.e. missing rows, are filled in with zeroes and flagged. Although relatively simple in nature, this program is made slightly more complicated by the requirement that it be fully automated. When the coprocessor sends the data each day, it also issues the command to execute the PERL program. This program must then find the appropriate monthly file to append the data to, based on the date the information was recorded, and then save the data to that file as it is decoded. The program must also create a new monthly file and save the data there if it detects a change in month at any point during the process. 70 10.3 Data Analysis and Reporting At the end of a given month, the data decoding process results is a lengthy text file, in spreadsheet readable format, containing an entire month of 10-minute data, decoded into its appropriate format and containing all recorded information regarding the performance of the wind-diesel system. The data is still extremely unwieldy in this form and would not be of very much use to other than the control system design engineers, and even then only to provide a detailed view of system operation at a given point in time, not to give an overview of system performance. The data must be analyzed and reduced to summary charts and tables to be more generally useful. Because this process can be accomplished most easily with a spreadsheet program, Microsoft Excel was chosen to perform this task instead of PERL. Excel also allows the data to be viewed and manipulated on PCs, giving access to a broad audience of project stakeholders. In essence, this program takes the monthly data file and reduces it to a neat, concise monthly report. It analyzes all of the data, calculating various averages, totals, and other summary values, from which it creates and formats various tables and charts, showing various performance trends and relationships. At the end of every month, the monthly datafile must be manually imported into the Excel program for processing. Once the program is run, the resulting monthly report will be converted to Adobe Acrobat PDF files and copied to the project web site, enabling any authorized person to access and review all of the system performance data. REPORT DOCUMENTATION PAGE OMe ND. 0704-0188 Public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information, including Suggestions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, Arlington, VA 22202-4302, and to the Office of Management and Budget, Paperwork Reduction Project (0704-0188), Washington, DC 20503. 1. AGENCY USE ONLY (Leave blank) | 2. REPORT DATE 3. REPORT TYPE AND DATES COVERED. May 2002 Technical Report 4. TITLE AND SUBTITLE Wales, Alaska, High-Penetration Wind-Diesel Hybrid Power System 5: een ae BERS 6. AUTHOR(S) Stephen M. Drouilhet Mari Shirazi 7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGANIZATION REPORT NUMBER 9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSORING/MONITORING National Renewable Energy Laboratory AGENCY REPORT NUMBER 1617 Cole Blvd. Golden, CO 80401-3393 NREL/TP-500-31755 | 11. SUPPLEMENTARY NOTES 12a. DISTRIBUTION/AVAILABILITY STATEMENT 12b. DISTRIBUTION CODE National Technical Information Service U.S. Department of Commerce 5285 Port Royal Road Springfield, VA 22161 13. ABSTRACT (Maximum 200 words) To reduce the cost of rural power generation and the environmental impact of diesel fuel usage, the | Alaska Energy Authority (AEA), Kotzebue Electric Association (KEA, a rural Alaskan utility), and the National Renewable Energy Laboratory (NREL), began a collaboration in late 1995 to implement a high-penetration wind-diesel hybrid power system in a village in northwest Alaska. The project was intended to be both a technology demonstration and a pilot for commercial replication of the system in other Alaskan villages. During the first several years of the project, NREL focused on the design and development of the electronic controls, the system control software, and the ancillary components (power converters, energy storage, electric dump loads, communications links, etc.) that would be required to integrate new wind turbines with the existing diesels in a reliable highly automated system. Meanwhile, AEA and KEA focused on project development activities, including wind resource assessment, site selection and permitting, community relationship building, and logistical planning. Ultimately, the village of Wales, Alaska, was chosen as the project site. Wales is a native Inupiat village of approximately 160 inhabitants, with an average electric load of about 75 kW. 15. NUMBER OF PAGES 16. PRICE CODE | 17. SECURITY CLASSIFICATION 18. SECURITY CLASSIFICATION 19. SECURITY CLASSIFICATION 20. LIMITATION OF ABSTRACT OF REPORT OF THIS PAGE OF ABSTRACT Unclassified Unclassified Unclassified UL 14, SUBJECT TERMS: wind energy; wind-diesel hybrid systems; Wales, Alaska; Alaska Energy Authority; Kotzebue Electric Association NSN 7540-01-280-5500 Standard Form 298 (Rev. 2-89) Prescribed by ANSI Std. 239-18 298-102 Preparing an Existing Diesel Power Plant for a Wind Hybrid Retrofit: Lessons Learned in the Wales, Alaska, Wind-Diesel Hybrid Power Project Stephen Drouilhet, P.E. National Renewable Energy Laboratory 1617 Cole Bivd. Golden, CO, 80401 USA 1 Introduction In countries around the world, there are many communities not served by national or regional electric grids. In many of these communities, power is generated by small diesel power plants that range in size from about 100 kW to several MW. There are thousands, perhaps tens of thousands, of isolated diesel-powered villages worldwide. In the state of Alaska alone, there— are approximately 200 diesel-powered communities. In addition to village power systems, there are hundreds or thousands of diesel plants providing power to a variety of remote commercial and industrial facilities, including mining operations, military bases, resorts, and fish farming and processing operations. There are many reasons why diesel power systems are so widespread. Diesel generators are by far the lowest capital cost electric generation technology in the sub-MW size range. They are a well-established and well-understood technology and there is a worldwide support infrastructure _in.place. When properly operated and maintained they are also very robust and reliable. However, diesels also have major disadvantages. They are noisy and emit signifi cant air pollution. Though relatively cheap on the world market, transportation costs can make diesel fuel very expensive in remote locations. In arctic communities, where fuel may only be delivered once per year, fuel storage costs also are very high, and the risks of major fuel spills greater. Finally, because diesels require frequent oil changes and other service at regular intervals, they have a relatively high maintenance cost per kWh delivered. Wind-diesel hybrid power systems preserve the advantages of diesel generators while mitigating their disadvantages. Wind turbines have a higher cost per installed kW capacity, but zero emissions, zero fuel cost, and lower routine maintenance requirements than diesels. Rural utilities and national energy agencies worldwide are beginning to see the opportunity that wind- diesel hybrids offer to reduce the life-cycle cost and environmental impact of rural electric service. Because an existing diesel plant frequently represents a substantial investment, it often appears more cost-effective to retrofit wind turbines, system controls, and any other required ancillary components to the existing power system rather than build a completely new wind- diesel hybrid system from the ground up. This was true in the case of the Wales, Alaska, High- Penetration Wind-Diesel Project, a technology demonstration project in which the National Renewable Energy Laboratory (NREL) collaborated with the Kotzebue Electric Association (KEA), the Alaska Village Electric Cooperative (AVEC), and the Alaska Energy Authority (AEA). Most of the engineering effort on the Wales project focused on the design and development of the new system components, primarily the main system controller and the energy storage subsystem. Comparatively little attention was paid to the diesel plant itself and to the modifications necessary to successfully integrate it into a fully automated wind-diesel hybrid system. Consequently, many diesel plant shortcomings were overlooked until they manifested themselves in the field during the start-up and commissioning of the wind-diesel hybrid system. The resulting problems revealed that in such a system, the diesel plant must perform to a higher ————— standard of performance than is often expected of the typical village power plant, which is usually designed to be completely manually operated. These higher performance requirements necessitate more rigorously designed diesel plants. Design shortcomings were found in all of the major diesel plant subsystems (engine cooling and fuel systems, generators, controls, switchgear, and distribution system). These shortcomings primarily affected the following areas: ease of retrofit system installation, frequency and voltage stability, time for diesel start-up and synchronization, generator paralleling, load-sharing stability, and plant and engine temperature control. This paper discusses each of the relevant plant design considerations in detail, in hopes that by sharing this experience, system integrators and project planners will give proper attention to diesel plant preparation (or replacement), and future wind-diesel systems will be installed and commissioned more quickly and cost-effectively. 2 Overview of the Wales, Alaska, High-Penetration Wind-Diesel Project The configuration of the Wales wind-diesel system is shown in Figure 1. The system is composed of the existing diesel power plant, two 65-kW wind turbines, an AC/DC rotary power converter, a battery bank, two electric boilers serving as secondary loads, and a Programmable Logic Controller (PLC)-based main system controller. Wind penetration is a term referring to the ratio of the wind power output to the village electric demand. According to the classification scheme used at NREL’s National Wind Technology Center, a high-penetration wind-diesel system is one in which the annual wind energy output of the wind turbines is at least 50% of the annual primary electric demand and in which the system has the capability to provide electric power with no diesels running during periods of sufficient wind power availability. The Wales wind-diesel system is thus a high-penetration system, because the Wales wind turbines are projected to generate approximately the same amount of energy annually as is consumed by the primary village load, and because the power system can operate diesel-off as long as the short-term average wind power exceeds the average load by a small margin. The individual components of the wind-diesel system and their role in its operation are discussed in detail below. 2.1 Diesel Power Plant The pre-existing diesel plant in Wales consists of two Cummins LTA10 and one Allis-Chalmers 3500 diesel gensets, rated as shown in Figure 1. Prior to the implementation of the wind-diesel system, the plant was entirely manually controlled, with the operator deciding when to run the various generators and manually starting, stopping, and synchronizing them to the grid. The gensets are of different ages and origins and, prior to this project, were equipped with various assorted voltage regulators, governors, and actuators. Manual diesel operation.is_incompatible with the effective implementation of -a-high-penetration wind-diesel system. Maximum fuel savings demands that only the most efficient diesel(s) adequate to meet the net load (village load minus available wind power) are run at all times. Furthermore, to capture the additional fuel and maintenance savings made possible by a reduction in diesel run time, the diesels must be shut down completely when there is more than enough wind power to meet the load. Under such an operating regime, the starting and stopping of any particular genset will be more frequent and unpredictable than is feasible with a manually controlled system. For this reason, the first step in system installation was to retrofit all diesels with controls making them capable of automatic starting, stopping, synchronization, and load-sharing. 2.2. Wind Turbines The project has installed two 65-kW wind turbines manufactured by Atlantic Orient Corporation, each of which has a 3-bladed downwind stall-regulated rotor and an induction generator. Because the turbines use induction generators, they rely on excitation from the line to generate power. Lacking any inherent control over voltage or frequency, induction machines generate power at whatever voltage and frequency exists on the line to which they are connected. Voltage and frequency must be controlled elsewhere in the system. Even when generating positive power, the generators draw reactive power (VARs) from the line. This reactive power “must be supplied by other components in the system, either by the diesel generators or by the AC machine of the rotary converter, as described later. 2.3. Energy Storage As discussed above, the main performance objective for the wind-diesel system is to minimize fuel consumption. This can be done by ensuring that only the smallest adequate diesel is run at any given time and that all diesels are off as much of the time as possible. Several researchers have shown that by providing a small amount of energy storage, unnecessary diesel starts (those that occur even when the average wind power plus the diesel capacity already on-line is adequate to meet the average load) can be greatly reduced, with a significant impact both on fuel savings and on total diesel run time. The optimal amount of storage depends on a variety of factors, including the number of wind turbines (a larger number of turbines will give a greater wind power smoothing effect), the variability of the wind, the variability of the village load, the cost of fuel, and the cost of storage. The Wales system is equipped with a battery bank sized to provide enough energy storage to meet about two thirds of the average village load for about 15 minutes. 2.4 Rotary Power Converter The rotary converter serves two purposes in the operation of the power system. It is the interface between the 480 VAC bus and the battery bank. It also supplies some or all of the reactive power demanded by the load and by the wind turbine generators. The converter consists of an AC synchronous generator shaft coupled to a DC motor. In its dual role, the AC machine acts sometimes as a motor/generator, sometimes only as a synchronous condenser. The field excitation on the AC machine is controlled by a standard generator voltage regulator. The field of the DC machine is controlled by the main system controller. Whenever one or more diesel gensets is operating, the system voltage and frequency is maintained by the gensets’ voltage regulator and governor. When no diesels are operating, both frequency and voltage must be regulated by the rotary converter. The system voltage is controlled by the AC machine’s voltage regulator, and the frequency is maintained by controlling the DC machine field current, thereby modulating the power flow to or from the battery bank. In 2.5 Secondary Loads In any high-penetration wind-diesel system, there are times when the wind turbine power output exceeds the load. To maintain system stability, this excess power must be dissipated. As discussed above, to the extent that the battery bank can accept it, excess power will be absorbed by the battery. However, when the battery is full or when the current to the battery would otherwise be excessive, power is dissipated in secondary loads (also referred to as dump loads, even though the energy is not actually wasted). This approach ensures that wind turbines will never have to be shut down due to excess power production, and that every bit of available wind energy will be used in an economically valuable way, either saving diesel fuel or displacing heating fuel. Remotely controlled electric boilers have been installed in the village school and the waste heat loop of the diesel plant, which is used to heat the plant itself and the diesel engines when they are not running. 2.6 System Controller One of the principal technical objectives in this project is to develop a system that is as reliable and robust as possible consistent with the requirements of high wind penetration and maximum fuel savings. As with the rest of the system, the main control system is built up from proven industrial components using conservative design practices. The heart of the control system is a standard industrial PLC controller outfitted with the I/O modules necessary to monitor and control the system. The control system is also equipped with a telephone interface to facilitate remote performance monitoring and fault diagnosis. 3 Requirements for Diesel Plant Automation As stated above, a high-penetration wind-diesel system requires a fully automated diesel plant. Even though the generators may have been originally sized such that the village load could always be met by a single unit, the automated plant should allow unattended parallel operation of any combination of available units. In addition to ensuring stable load-sharing during diesel generator changeovers, this capability also allows the deferral of diesel capacity upgrades as the village load grows. The diesel generators must also behave stably and consistently when paralleled to other generating sources that are part of the system, such as the wind turbines and the energy storage power converter. The ability to rapidly start and synchronize each diesel generator to the plant bus is another fundamental requirement of the automated plant. Diesel-off operation is the objective in a high- penetration system. In order to safely shut down all diesels, however, the supervisory controller must be able to reliably start and bring a diesel on-line on short notice. To minimize the amount of wind power margin required to run diesel-off and/or to gain the maximum benefit from the installed energy storage capacity, it is important to be able to start and synchronize the diesel__ engine in less than a minute, preferably much less. In addition to these performance requirements, there are logistical issues involved in upgrading the plant. Ideally, the plant modifications should be done to minimize the amount of design, fabrication, assembly, wiring, etc., that must be performed on site. Such work is typically more difficult and more expensive to do in the field. Moreover, it is typically disruptive of plant operations, and to the extent that it causes prolonged power interruptions, it will dampen local enthusiasm for the project. 4 Diesel Plant Automation and Integration Issues Encountered in Wales The start-up and debugging phases of the Wales project revealed a variety of ways in which the diesel plant, even after being retrofit with modern electronic diesel generator controls, failed to meet the requirements identified above. This section discusses the deficiencies in each major diesel plant subsystem. 4.1. Generators During early parallel operation of the generators, we observed that whenever Unit 1 was paralleled with either of the other generators, the VAR load was not shared properly. Unit 1 acted as a reactive power sink. The other generator on-line, therefore not only had to meet the village VAR demand, it also had to meet the VAR demand of Unit 1. After determining that all of the voltage regulators and paralleling (i.e., VAR-sharing) modules were wired and functioning properly, we measured the harmonic content of the generator currents. We found a large third harmonic current circulating between the two paralleled generators, which is characteristic of parallel operation of generators of different winding pitch. We had the generator manufacturer trace the generator serial numbers and found that Unit 1 had a 7/9 pitch whereas Units 2 and 3 had 2/3 pitch generators. The plant had been functioning for years under manual contro! with this generator pitch mismatch. The circulating harmonic currents went unnoticed because parallel operation was limited to short periods of transitions between gensets and because stable long-term unattended load-sharing was not a requirement. Even under these circumstances, mismatched generators are undesirable because the circulating currents increase losses and cause an effective reduction of the individual generator ratings. The generator pitch mismatch has major implications for the performance of the plant in automated operation. It is clear that the large harmonic currents are overwhelming the diesel VAR sharing controls. A similar situation is encountered when Unit 1 is paralleled with the rotary converter AC machine, which has a pitch of 2/3, matching Units 2 and 3. Although still somewhat speculative at this point, we believe that the harmonics are also interfering with the proper operation of the load-sharing modules, such that load-sharing is marginally unstable. Plans have been made to replace the Unit 1 generator with a 2/3 pitch generator. We are confident that this will completely solve the VAR-sharing problem and contribute greatly to load- sharing stability. 4.2 Generator Set Controls The original Wales diesel controls did not contain the load-sharing or synchronizing controls necessary for automatic operation. Nor did they provide any means of automatically actuating the generator circuit breakers. Because they were designed for completely manual operation, there was no provision to place the controls in automatic mode. Early in the project, the decision was made to upgrade the existing control panels with the necessary additional controls rather than to replace them with new panels designed specifically to accommodate the wind- hybrid system. The existing controls for all three generators were very densely packed in a cabinet more appropriately sized for one generator than for three. With some difficulty, the new suite of electronic controls (speed control, synchronizer, and load-sharing module) were added to this cabinet, but there was no room left for the additional relaying required to switch between manual and automatic operating modes, nor any space for terminal blocks for the interface to the new hybrid system control panel. Consequently, we had to design a Diesel Controls Interface Panel to fit in the narrow space between the existing generator control cabinet and the Wind-Diesel Control Panel. Several months were lost while waiting for this panel to be fabricated in Anchorage and then installed in the plant. Because of the large number of interconnections with existing equipment, the installation technician spent days working in electrically live cabinets next to operating diesel engines. This kind of work environment is conducive to wiring errors, which may only show up later and require extensive troubleshooting to locate. Those concerned with this installation agreed in retrospect that the more prudent and probably cost- effective approach would have been to replace the existing generator controls with preengineered and pretested automatic diesel control panels. 4.3 Engine Cooling System Excess Back Pressure The Wales diesel plant incorporated a jacket water waste heat recovery system (see Figure 2). The waste heat is used to keep the diesel plant buildings warm and to preheat off-line engines so that they can start rapidly and pick up load without a lengthy warm-up. The coolant discharge line from each engine is plumbed into a common discharge (“hot”) header, and the coolant intake line to each engine comes from a common intake (“cold”) header. The flow from the hot header passes through a water-to-water heat exchanger, which transfers heat from the engine coolant to the hydronic loop that heats the plant facilities. The flow may then either pass through some externally mounted air-cooled radiators to reject excess heat, or it may be returned directly to the engines via the cold header. A proportioning temperature contro! valve regulates the fraction of flow through the radiators and the fraction bypassing the radiators to regulate the temperature of the coolant returning to the engines. The local dump load, an electric boiler equipped with its own circulation pump, was plumbed into this piping system as though it were a fourth diesel engine. In other words, it operates in parallel with the three engines, taking its flow from the intake header and discharging to the discharge header. An interesting phenomenon was observed during start-up testing of the hybrid system. Whenever multiple engines were operating simultaneously, or when the dump load circulation pump was operating in parallel with one or more diesel engines, there was a tendency for the engines to overheat. This was particularly true of engine #2, which was the smallest and had the weakest water pump. The effect of multiple pumps operating in parallel, all discharging into the same coolant loop, was to cause the pressure drop across that loop to rise above levels normally seen with a single engine operating. The water pumps in the individual engines were not always strong enough to ensure that there was sufficient flow through a particular engine. This problem had never previously manifested itself, because the manually operated plant only had one engine operating for any extended length of time, and this one engine’s water pump never had to compete with another pump trying to force flow through the same circuit. We believe the basic layout of the cooling system to be acceptable, so the solution being pursued currently is to replace various components in the cooling loop with less restrictive versions to reduce the loop’s total resistance. This will ensure that any engine’s water pump is adequate to provide adequate coolant flow regardless of the operation of any other engine or the dump load pump. Wasted Heat Under manual diesel plant operation, with a diesel generator carrying the full village load, there is always more than enough diesel waste heat to keep the plant and the nonoperating engines warm. Consequently, some excess heat must always be dissipated by the radiators. With the wind system, however, this is not necessarily the case. In moderate winds, the wind turbines may be supplying most of the village load, and the diesel generator will be operating at low power, in which case there is little waste heat available. In higher winds, the diesels may be shut off completely, in which case there is no diesel waste heat. Because the plant and engines must nevertheless be kept warm, wind-generated electricity, via the plant dump load, must be used to make up for the loss of diesel waste heat. Whereas thermal energy conservation in the plant was previously not a concern, it now becomes very important. In Wales, the diesel plant is an uninsulated and drafty building, and it requires a lot of energy to heat. This, combined with the fact that the temperature regulation valve that controls how much flow goes to the radiators was not working properly, meant that in early operation of the wind- diesel system, a lot of wind-generated electricity was being lost out the radiators. This meant that much more wind energy was being used to keep the plant warm than should have been required. Such heat loss causes a severe negative impact on the system economics, because wind energy used to keep the plant warm is not available to serve either the primary village electric load or the secondary load installed in the school heating system. Both of these loads are revenue generating, whereas the plant dump load is not. Steps are currently being taken to improve the plant insulation and to ensure that the heat recovery system is working properly, such that no coolant flow goes to the radiators unless there is truly an excess of diesel waste heat. 4.4 Fuel Systems Parallel operation of the diesel generators revealed problems with the existing fuel system as well. As with the cooling system, the fuel supply and return lines from the three engines share common supply and return headers coming from the day tank. Though still somewhat speculative, it appears that when multiple engines are running simultaneously, the combined fuel flow causes the back pressure in the fuel return lines to rise above an acceptable level for proper fuel injector operation. This high fuel back pressure may be contributing to poor engine speed control. Another problem with the layout of the original fuel system is that the routing of the fuel lines between the day tank and the engines appears to contribute to the formation of air or vapor pockets in the lines when the engine is off. The current layout also may make it difficult to purge these air pockets from the piping. Again, while this behavior has not been positively confirmed, it appears that the fuel piping layout is causing fuel flow anomalies that only disappear after a fairly long warm-up time, making rapid synchronization difficult. To address these problems, separate fuel supply and return lines for each engine will be run from the day tank, and these new lines will be routed in such a way as to minimize air/vapor entrapment. 4.5 Starter Battery Systems When the Wales plant was a manual operation, diesel starting events were very infrequent. Since either of the larger engines was capable of meeting the peak village load, one of them was typically operated continuously for 10 days, at the end of which the load would be transferred to the other one and the first one serviced. The duty cycle on the engine starting batteries was therefore minimal. The battery charger did not need to be capable of rapid recharging. Because all starting operations were manual and therefore attended, if an engine was particularly difficult to start and required numerous retries, the operator could rest the battery as long as necessary between crank cycles. With the automated system, however, optimal diesel dispatch requires more frequent diesel starts and stops. In addition, in the case of a hard-starting engine, it is important that the battery and charger be robust enough to start the engine within a programmed number of crank cycles to avoid an alarm condition and consequent loss of availability of that engine. In Wales, the existing battery charger was of an outdated design that did not have an automatic fast-charging mode. In certain situations, this limitation resulted in unreliable diesel starting. The battery charger is being replaced in conjunction with the other necessary diesel plant upgrades. 5 Project Planning and Implementation The problems described in Section 4 of this paper were all discovered only after the diesel plant began to be operated as part of an automated hybrid system. That is because during the project development, our attention was focused on the hybrid system controls development and not on the detailed impact of the system on diesel plant operation. These particular problems would not all necessarily be present in another village power plant, nor are they the only problems that might occur. The main lesson to be learned from this experience is the importance of conducting a thorough assessment of the existing diesel plant early in the planning stages of a wind hybrid retrofit project. The first step in this assessment is to identify any features of the existing plant design that could potentially interfere with fully automatic operation of the plant. It will not always be possible to predict the exact behavior of individual diesel plant subsystems in operating modes that have never been experienced. Therefore, it is prudent to note all potential problems as well as obvious ones. The following are questions that may help to identify problem areas: e What capabilities do the current diesel controls lack that are necessary for automatic operation (e.g., auto synchronization, load-sharing, VAR-sharing, remote breaker closure, automatic feeder control, etc.) e Are any of the diesel piping systems (fuel, lubricating oil, coolant, exhaust, etc.) designed in such a way that the performance of one diesel is influenced by the simultaneous operation of another? e Are the generators matched in pitch? If not, have steps been taken to eliminate or mitigate circulating harmonic currents? ¢ Does the diesel plant rely in any way on having excess waste heat available? e Do the engines have similar dynamic response? (If not, it will be difficult to achieve good load-sharing performance.) e Are there any actions the operators currently perform (consciously or not) that tend to compensate for inadequacies in any of the diesel plant systems? e What alarm or fault conditions currently occur in the diesel plant? What is their impact on manual plant operation? What would their impact be on automatic plant operation? e Are there any factors present that would compromise the performance of a waste heat recovery system? The second step in the assessment is to determine what diesel plant modifications would be required to rectify the problems or deficiencies identified. There may be multiple engineering solutions to any given problem. All possible approaches should be identified. Lastly, one must determine the true cost of performing the upgrades and modifications to the existing plant. To do this, one must consider not only the costs of parts and labor, but various other costs and risks as well, including: the cost of doing engineering designs that may only apply to this one installation the difficulty of doing fabrication and installation work in the field the risk of design and installation error the loss of revenue and customer good will associated with outages necessitated by diesel plant rework e the risk of delays to the project if the plant requires extensive rework In many cases, these costs and risks may be relatively minor. Even when they are significant, they can often be reduced by good project planning. In some cases, however, it will be more cost-effective to replace major subsystems than to upgrade them. Frequently, for example, it will be more cost-effective to scrap the existing generator controls and start fresh with new automation-ready diesel control panels. In cases of very old plants or plants that have been incrementally expanded over the years with lots of dissimilar components and/or with poor documentation, the best approach may be to replace the entire plant with a new one optimized for the wind hybrid system. The gains in system lifetime, speed of installation, ease of maintenance, and overall reliability may more than make up for the increased capital cost. Wind Turbines (Induction, Stall-Regulated) 2X 65 KW = 130 KW ab aL . 1) Diesel #1 a CS 168 kW 1 a J (fifi : 7oKW poe ee ir 75 kW Battery Bank DC MACHINE AC MACHINE 280 YOG, 190A Rotary Converter f 156 kVA Diesel #3 School Heating 3 oe System in Diesel Plant 3 ‘KY wes fc Hydronic Loop Resistance Secondary Load Primary Village Load Heaters Controllers 40-120 kw Figure 1 Wales High-Penetration Wind-Diesel Hybrid Power System no ole a frre #1 L MW WW tO 28" == WwW == A SGe AWWW TEMPERATURE On = ELECTRIC BOILER Figure 2 Wales Diesel Plant Cooling and Heat Recovery System Layout 10 REPORT DOCUMENTATION PAGE OU NO CTO OTBS Public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for rere instructions, searching existing data sources, gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this Collection of information, including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, Arlington, VA 22202-4302, and to the Office of Management and Budget, Paperwork Reduction Project (0704-0188), Washington, DC 20503. 2. REPORT DATE 3. REPORT TYPE AND DATES COVERED August 2001 Conference paper 1. AGENCY USE ONLY (Leave blank) 4. TITLE AND SUBTITLE Preparing an Existing Diesel Power Plant for a Wind Hybrid Retrofit: Lessons Learned in the Wales, Alaska, Wind-Diesel Hybrid Power Project WER11720 5. FUNDING NUMBERS 6. AUTHOR(S) Stephen Drouilhet 7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGANIZATION REPORT NUMBER 9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSORING/MONITORING National Renewable Energy Laboratory AGENCY REPORT NUMBER 1617 Cole Blvd. Golden, CO 80401-3393 NREL/CP-500-30586 11. SUPPLEMENTARY NOTES NREL Technical Monitor: 12a. DISTRIBUTION/AVAILABILITY STATEMENT 12b. DISTRIBUTION CODE National Technical Information Service U.S. Department of Commerce 5285 Port Royal Road Springfield, VA 22161 13. ABSTRACT (Maximum 200 words) This paper describes the wind-diesel hybrid power project, a technology demonstration project conducted by the National Renewable Energy Laboratory, Kotzebue Electric Association, the Alaska Village Electric Cooperative and the Alaska Energy Authority in Wales, Alaska. It discusses each of the relevant plant design considerations in detail, in hopes that system integrators and project planners that read the report will realize the importance of giving proper attention to diesel plant preparation (or replacement), and future wind-diesel systems will be installed and commissioned more quickly and cost effectively. . NUMBER OF PAGES . SUBJECT TERMS wind energy hybrid systems; Wales; Alaska; Kotzebue PRICE CODE . SECURITY CLASSIFICATION 18. SECURITY CLASSIFICATION | 19. SECURITY CLASSIFICATION . LIMITATION OF ABSTRACT OF REPORT OF THIS PAGE OF ABSTRACT Unclassified Unclassified Unclassified UL NSN 7540-01-280-5500 Standard Form 298 (Rev. 2-89) Prescribed by ANSI Std. Z39-18 298-102 September 2001 * NREL/CP-500-30668 Characterizing the Effects of High Wind Penetration on a Small Isolated Grid in Arctic Alaska Gordon Randall, and Rana Vilhauer Global Energy Concepts, LLC Craig Thompson Thompson Engineering Company Presented at AWEA's WINDPOWER 2001 Conference Washington, D.C. June 4 — June 7, 2001 « DNREL uy oy National Renewable Energy Laboratory 1617 Cole Boulevard Golden, Colorado 80401-3393 NREL is a U.S. Department of Energy Laboratory Operated by Midwest Research Institute e Battelle e Bechtel Contract No. DE-AC36-99-GO10337 NOTICE The submitted manuscript has been offered by an employee of the Midwest Research Institute (MRI), a contractor of the US Government under Contract No. DE-AC36-99GO10337. Accordingly, the US Government and MRI retain a nonexclusive royalty-free license to publish or reproduce the published form of this contribution, or allow others to do so, for US Government purposes. This report was prepared as an account of work sponsored by an agency of the United States government. Neither the United States government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States government or any agency thereof. Available electronically at http:/Mwww.doe.gov/bridge Available for a processing fee to U.S. Department of Energy and its contractors, in paper, from: U.S. Department of Energy Office of Scientific and Technical Information P.O. Box 62 Oak Ridge, TN 37831-0062 phone: 865.576.8401 fax: 865.576.5728 email: reports@adonis.osti.gov Available for sale to the public, in paper, from: U.S. Department of Commerce National Technical Information Service 5285 Port Royal Road Springfield, VA 22161 phone: 800.553.6847 fax: 703.605.6900 email: orders@ntis.fedworld.gov online ordering: http://www.ntis.gov/ordering.htm oO la? Printed on paper containing at least 50% wastepaper, including 20% postconsumer waste CHARACTERIZING THE EFFECTS OF HIGH WIND PENETRATION ON A SMALL ISOLATED GRID IN ARCTIC ALASKA Gordon Randall and Rana Vilhauer Global Energy Concepts, LLC 5729 Lakeview Dr. NE, Suite 100 Kirkland, WA 98033 USA grandall@globalenergyconcepts.com rvilhauer@globalenergyconcepts.com Craig Thompson Thompson Engineering Company 721 Sesame Street, Suite 2B Anchorage, AK 99503 teco@gci.net Abstract Utilities have historically assumed that wind penetration levels of more than 25-30% would result in system instability. Higher levels of wind penetration were expected to cause serious stability and reliability problems in both the generation and distribution systems. Because wind penetration levels in the United States have generally been much lower, little data have been available to determine if such problems actually occur. This paper examines the operating characteristics of the wind-diesel system in Kotzebue, Alaska, operated by Kotzebue Electric Association (KEA). KEA began incorporating wind power into its 100% diesel generating system in 1997 with three 66 kW wind turbines. In 1999, KEA added another seven 66 kW turbines, resulting in the current wind capacity of 660 kW. KEA is in the process of expanding its wind project again and ultimately expects to operate 2-3 MW of wind capacity. With a peak load of approximately 4 MW and a minimum load of approximately 1.6 MW, the wind penetration is significant. KEA is currently experiencing greater than 35% wind penetration, sometimes for several consecutive hours. This paper discusses the observed wind penetration at KEA and evaluates the effects of wind penetration on power quality on the KEA grid. Introduction The KEA wind power plant is a 0.66 MW facility of small commercial-scale wind turbines. The project consists of 10 AOC 15/50 66 kW fixed-speed, stall-controlled wind turbines manufactured by Atlantic Orient Corporation (AOC) of Norwich, Vermont. The AOC 15/50 is a three-bladed, downwind turbine with a 15-m (49-ft) rotor diameter installed on 24.4-m (80-ft) lattice towers on piling foundations, resulting in a hub height of approximately 26.5 m (87 ft). The KEA wind power project joined the U.S. Department of Energy/Electric Power Research Institute (DOE-EPRI) Wind Turbine Verification Program (TVP) as an associate project in 1997. Additional information about the KEA wind project performance and operating experience is reported by EPRI[1]. KEA’s project site is located on the tip of the Baldwin Peninsula approximately 42 km (26 mi) north of the Arctic Circle on the northwest coast of Alaska near the town of Kotzebue. With a population of approximately 3,000 residents, Kotzebue is the largest community in northwestern Alaska and serves as the economic, governmental, medical, communication, and transportation hub for the 11 communities in the Northwest Arctic Borough, an area roughly the size of Indiana. Figure 1 shows the location of Kotzebue on the Alaska state map. Arctic Circle - regs aoe tition Pacific Ocean FIGURE 1: KOTZEBUE LOCATION MAP The only source of generation for the KEA power grid is an 11 MW diesel generating plant consisting of six diesel generators. Normally the diesel plant has only one or two generators operating, with the remaining generators providing redundancy. Typical loads on the grid range between 2 and 3 MW, with a peak load of approximately 4 MW and a minimum load of approximately 1.6 MW. The load varies with the time of day and with climatic conditions. Methodology and Data Used For the purpose of this paper, data from August 21 through September 20, 2000 were evaluated. This period was selected because total system load can be somewhat lower during the warmer summer months. The TVP reporting period for the KEA project ends on the 20" of the month, so the period evaluated corresponds to one monthly period. Data for the wind project were generated using the Second Wind Advanced Distributed Monitoring System (ADMS), a commercial supervisory control and data acquisition system (SCADA) that KEA uses to manage and operate the wind project. Parameters measured by the SCADA include turbine production and performance, meteorological data, and a variety of power quality measurements collected by the Second Wind Phaser® power transducers located in each turbine. The Phaser has also been used by TVP to make power quality measurements at other distributed wind projects, as reported by Green [2]. KEA meter readings indicating total system load and production by the diesel generating facility were also available. Wind penetration estimates were calculated by dividing the measured power output from the wind facility by the total KEA system load. KEA Grid Load and Observed Penetration Levels Figure 2 presents the diurnal pattern of grid loads during the period evaluated. The solid line indicates the average load for each hour over the month. Minimum and maximum values within the month are indicated with bars off of this line. The approximate capacity of the wind farm (i.e., approximately 660 kW) is also indicated. Total System Load, August 21 - September 20, 2000 3500 3000 Approximate Capacity of Wind Farm 1000 | 12AM 1AM 2AM 3AM 4AM 5AM 6AM 7AM 8AM QAM 10AM11AM12PM 1PM 2PM 3PM 4PM 5PM 6PM 7PM 8PM 9PM 10PM11PM Time of Day FIGURE 2: DIURNAL DISTRIBUTION OF KEA LOAD As shown on this figure, grid loads are lowest during early morning hours, usually falling below 2 MW from approximately 1:00 a.m. to 6:00 am. The minimum load observed during the month was approximately 1.67 MW. Loads are higher throughout the day, although loads did not exceed 3 MW during the month. The minimum system load occurred at approximately 4:00 a.m. on August 29, 2000. The highest overall wind penetration of approximately 35% was also measured at this time. Figure 3 presents the wind speed and temperature measured over the time around this event, as well as the calculated wind penetration values. August 29 was an unusually warm morning in Kotzebue, with temperatures exceeding 13 degrees Celsius during the overnight hours. These temperatures were over 3 degrees warmer than measured values for the same time period on other days during the month. In addition, the winds were moderate to strong throughout the early morning hours, exceeding 15 m/s at 4:00 a.m. This combination of high winds and low demand on the KEA grid resulted in the unusually high penetration values. 18 45% — Air Temperature —Wind Speed 16 _/\ Wind Penetration |++ 40% 14 \ A 35% Dan \_ Jed \ V 10% Penetration Air Temperature (degrees C)/Wind Speed (m/s) 2 5% 0 0% 12:00 AM 1:00 AM 2:00 AM 3:00 AM 4:00 AM 5:00 AM 6:00 AM 7:00 AM 8:00 AM Time of Day, August 29, 2000 FIGURE 3: AIR TEMPERAURE, WIND SPEED, AND WIND PENETRATION, AUGUST 29, 2000 In addition to the morning of August 29, penetration values in excess of 25% were observed during the early morning hours for several days following the 29". A summary of the observed penetration values is presented in Table 1. Despite a few periods of high penetration, the overall average penetration for the month was approximately 5.6%. This average is highly influenced by the few significantly higher values. The median penetration for the month was 2.0%. Penetration values of less than 1% were calculated for approximately 46% of the month; penetration values of less than 10% were calculated for approximately 77% of the month. Overall, penetration varied only slightly with the time of day. During the early morning hours (from midnight to 8:00 a.m.), the median penetration was 0.71%, which is somewhat below the overall median value. This reflects the generally lower wind speeds during this period, with the exception of August 29 and the days immediately thereafter. However, the early morning average penetration was 6.0%, reflecting the lower energy demand during this period. The opposite trend was seen during evening hours, with a higher 3.1 % median penetration due to higher winds but a lower 5.3 % average penetration because of higher energy demand. Overall wind penetration values are presented graphically in Figure 4. TABLE 1: SUMMARY OF WIND PENETRATION VALUES Evening Early Morning Day (5 p.m. - arameter 12 a.m. - 8 a.m.)} (8 a.m. - 5 p.m. 12 a.m. aximum penetration 35.3% verage penetration 5.3% |Median penetration 3.0% ime with less than 1% enetration ime with less than 10% enetration 76.9% 80.4% ime with less than 20% enetration 93.4% 95.0% 40% el — Night (12 AM - 8 AM) — Day (8 AM - 5 PM) TT — Evening (5 PM - 12 AM) —— Overall 30% 25% iy 20% 15% Penetration 10% 5% 0% 7 + T { 10 20 30 40 50 60 70 80 90 100 Percentile FIGURE 4: WIND PENETRATION PERCENTILES Power Quality at High Penetration A variety of power quality parameters are measured by the Phasers and recorded in the SCADA system or can be calculated from the measured values. These parameters include (among others) line voltage, voltage imbalance, total demand distortion, and frequency deviation. The following section presents an overview of how each of these parameters varied as the wind penetration increased. Power quality measurements described in this section were recorded by the Phaser at Turbine 8. This turbine was used in power performance testing conducted at the site, and the Phaser recorded a wider range of parameters. than those at the other turbines. The measurements at this turbine are believed to be representative of the rest of the wind farm. For the purpose of this analysis, data were used only when the wind facility was on-line and producing at least 200 kW of power. Below this output level, it was assumed that any irregularities in power quality measurements would be caused by sources other than the wind farm. Voltage Figure 5 presents a scatter plot of 10-minute average voltages compared to the wind penetration values. The nominal voltage for the grid is 480 V. As shown, the measured voltage exceeded the nominal value by up to approximately 20 V during the time period evaluated. However, no relationship can be seen between voltage and wind penetration. At the highest penetration levels, voltage was closer to nominal than at some lower penetration values. Consequently, it appears that any effect on line voltage caused by high penetration is dwarfed by effects external to the wind farm, and possibly by effects caused by the wind farm independent of the penetration level. 500 ——— — Average Voltage (V) 482 0% 5% 10% 15% 20% 25% 30% 35% 40% Wind Penetration FIGURE 5: AVERAGE VOLTAGE VS. WIND PENETRATION Voltage Imbalance Figure 6 presents a scatter plot of 10-minute average voltage imbalance compared to the wind penetration values. There is relatively little scatter in the measured values, with a range between about 18.2 V and 19.6 V. No relationship can be seen between wind penetration and voltage imbalance. Total Demand Distortion Figure 7 presents a scatter plot of 10-minute average total demand distortion compared to the wind penetration values. There appears to be a slight inverse relationship between wind penetration and total demand distortion; however, the relationship is not strong and may be more related to other factors. Regardless, no adverse effect on total demand distortion is observed as wind penetration increases. Frequency Deviation Figure 8 presents a scatter plot of 10-minute average frequency deviation compared to the wind penetration values. As shown, the frequency deviation appears to remain relatively constant as wind penetration levels increase. Voltage Imbalance (V) Total Demand Distortion (%) 19.6 19.4 19.2 + 19 + 18.8 + 18.6 + 18.4 18.2 0% 5% 10% 15% 20% 25% 30% 35% 40% Wind Penetration FIGURE 6: VOLTAGE IMBALANCE VS. WIND PENETRATION (aan a — 0% 5% 10% 15% 20% 25% 30% 35% 40% Wind Penetration FIGURE 7: TOTAL DEMAND DISTORTION VS. WIND PENETRATION 9B tty 120 © 100 80 ° 3 8 . 60 es coe 40 Frequency Deviation (mHz) -60 0% 5% 10% 15% 20% 25% 30% 35% 40% Wind Penetration FIGURE 8: FREQUENCY DEVIATION VS. WIND PENETRATION Conclusions Based on measurements collected between August 21 and September 20, 2000, there is no apparent adverse effect on power quality on the KEA grid as wind penetration increases. During this period, wind penetration reached a maximum level of approximately 35%. It is unlikely that wind penetration will significantly exceed 35% at KEA with the current wind turbine capacity, as the highest observed penetration levels occurred during time periods with a combination of high winds and low system load. KEA plans on expanding the wind farm in the near future. Current plans for expansion include addition of two AOC 15/50 turbines during the summer of 2001. With these additional turbines, wind penetration could reach maximum levels of approximately 45%. Eventually, KEA plans to increase wind generation capacity to a total of 2 to 3 MW. References 1. Kotzebue Electric Association Wind Power Project First-Year Operating Experience: 1999-2000, U.S. Department of Energy - EPRI Wind Turbine Verification Program, EPRI 1000957, December 2000. 2. Green, J., VandenBosche, J., Lettenmaier, T., Randall, G., Wind, T. Power Quality of Distributed Wind Projects in the Turbine Verification Program. WindPower 2001 Proceedings, AWEA, Washington, DC, June 2001. September 2001 * NREL/CP-500-30412 Power Quality Issues ina Hybrid Power System Preprint E. Muljadi and H.E. McKenna To be presented at the IEEE—IAS 2001 Conference Chicago, Illinois September 30, 2001—October 4, 2001 ¥ > NRE National onlin Energy Laboratory 1617 Cole Boulevard Golden, Colorado 80401-3393 NREL is a U.S. Department of Energy Laboratory Operated by Midwest Research Institute e Battelle e Bechtel Contract No. DE-AC36-99-GO10337 NOTICE The submitted manuscript has been offered by an employee of the Midwest Research Institute (MRI), a contractor of the US Government under Contract No. DE-AC36-99GO10337. Accordingly, the US Government and MRI retain a nonexclusive royalty-free license to publish or reproduce the published form of this contribution, or allow others to do so, for US Government purposes. This report was prepared as an account of work sponsored by an agency of the United States government. Neither the United States government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States government or any agency thereof. Available electronically at http://www.osti.gov/bridge Available for a processing fee to U.S. Department of Energy and its contractors, in paper, from: U.S. Department of Energy Office of Scientific and Technical Information P.O. Box 62 Oak Ridge, TN 37831-0062 phone: 865.576.8401 fax: 865.576.5728 email: reports@adonis.osti.gov Available for sale to the public, in paper, from: U.S. Department of Commerce National Technical Information Service 5285 Port Royal Road Springfield, VA 22161 phone: 800.553.6847 fax: 703.605.6900 email: orders@ntis.fedworld.gov online ordering: http:/www.ntis.qgov/ordering.htm ie te Printed on paper containing at least 50% wastepaper, including 20% postconsumer waste POWER QUALITY ISSUES IN AHYBRID POWER SYSTEM E. Muljadi * H.E. McKenna National Wind Technology Center National Renewable Energy Laboratory 1617 Cole Boulevard, Golden, CO 80401 (303) 384-6904 ABSTRACT We analyzed a power system network, which consisted of two types of power generation: wind turbine genera- tion and diesel generation. The power quality and the interaction of diesel generation, the wind turbine, and the local load were the subjects of investigation. From an energy-production point of view, it is desir- able to have as much wind energy production as possible in order to reduce both the fuel consumption of the diesel engines and the level of pollution. From the customer’s point of view, it is desirable to have good power quality at the receiving end. The purpose of this paper is to show the impact of the wind power plant on the entire system. Also, we discuss how the startup of the wind turbine and the transient con- dition during load changes affect voltage and frequency in the system. Key words: wind turbine, diesel generator, hybrid power system, power quality, renewable energy |. INTRODUCTION Since ancient times, wind turbine technology has been used to improve the quality of life for many people. Peo- ple have used wind turbines to pump water, mill grain, and do many other things [1, 2, 3, 4]. In today’s world, wind turbines are used for similar purposes (i.e., water or oil pumping, battery charging, or utility generation). One important aspect of wind turbine applications, especially in an industrial environment, is that wind turbines generate electricity without creating pollution. In addition, the generation of electricity using wind turbines is well suited for isolated places with no connections to the outside grid [2, 3, 4]. This paper analyzes a hybrid power system consisting of wind turbines, diesel generators, and a local load. The results and the conclusion of the analysis are applicable to any similar hybrid power systems. Il. SYSTEM CONFIGURATION A physical diagram of the system that was analyzed is shown in Figure 1. The wind turbine is operated with an eduard_muljadi@nrel.gov http://www.nrel.gov induction generator with a capacity from 40 kW to 225 kW. In low wind speeds, the generator is operated at a lower generator speed (900 rpm, rated at 40 kW) and at high wind speeds the generator speed is 1,200 rpm with a rated capacity of 225 kW. Only a fixed-speed wind tur- bine is discussed; variable-speed wind turbine generation [5] will not be covered. The diesel engines have different rated capacities. They are operated in parallel to supply the load. The lo- cal loads are mostly residential and light loads. Other loads include water pumps, compressors, and other heavy equipment. The transient condition of a heavy load is represented by an 80kW water pump. The system has two types of generation: the diesel generator and the wind turbine generator (Figure 1). The diesel generator provides smooth generation, whereas the output power of wind turbine generation depends on wind velocity. The wind velocity is reflected in the power generation. For example, if the wind changes very Leta) Residential/Light (C= Industrial/Heavy Load oo: Diesel Generator ® Wind Turbine Generator Sensitive Load Figure 1. Physical diagram of the analyzed power smoothly, the output power of the wind turbine will also change very smoothly. On the other hand, wind turbu- lence causes the output power to fluctuate The load can be divided into three different types. The first is the residential type of load for lighting, heating, or powering small appliances. The nature of this load is very tolerant of disturbance (immune to low-quality power, such as fluctuating voltage and frequency). The second type of load is the industrial or heavy load, such as compressors, water pumps, and intermittent large loads. This type of load, although insensitive to power quality, may cause voltage and frequency fluctuations on a weak grid. Another type of load is the sensitive load. This load may consist of radar, medical equipment, transceivers, or receivers. Ill. COMPONENTS OF POWER SYSTEM The system under discussion in this paper consists of three major subsystems: diesel generator, wind turbine generator, heavy (industrial) loads. In the power system network, the balance of active power and reactive power must be maintained. Thus, the diesel genset must be able to keep the power balanced when the wind turbine or lo- cal load varies. While important, details of the dynamic model for electric machines will not be covered here. Many good textbooks are available on this subject [6, 7]. A. Diesel Generator From an electrical system point of view, a diesel gen- erator can be represented as a prime mover and genera- tor. Ideally, the prime mover has the capability to supply any power demand up to rated power at constant fre- quency, and the synchronous generator connected to it must be able to keep the voltage constant at any load condition. The block diagram of the diesel generator is illustrated in Figure 2. The diesel engine keeps the frequency con- stant by maintaining the rotor speed constant via its gov- ernor. The synchronous generator must control its output voltage by controlling the excitation current. Thus, the diesel generating system as a unit must be able to control its frequency and its output voltage. The ability of the diesel_ge enerator to respond to frequency Cianses is af fected by the inertia of the diesel genset, the sensitivity of the_governor,.and the power-capability of the diesel _en- gine. The ability of the-s u trol its voltage is affected by t eld eld winding | time con time constant, the availability of the DC | power to er to supply_the fie ield winding, and the response of voltage control regulation. The diesel engine must be capable of following the variation of loads and wind generation. The size of fre- quency variation indicates how well the diesel and its Diesel Engine exciter wt f Governor k@q- ve to the rest of power system. Figure 2. Diesel generator control block diagram governor maintains the balance of active power in the system. And the size of voltage variation indicates how well the genset and its voltage regulator maintains the balance of reactive power through its excitation. Under transient conditions, the frequency and the voltage will not be absolutely constant due to constantly changing wind speeds and load variations. B. Wind Turbine Generator The power and the torque generated by the wind tur- bine are as follows: P=0.5pAC,V° () r=— @) Q. S p = density of air A = swept area of the blade C, = performance coefficient V = wind speed T mechanical torque (at low speed shaft) P = output power of the turbine @,= rotor speed of the wind turbine low-speed shaft. ‘A typical C, curve is shown in Figure 3. This charac- teristic defines C, as a function of the tip-speed ratio (TSR) given by equation 3: ii} o,R TSR = — (3) where R is the radius of the wind turbine rotor. 0.5 3 O4 i = 0.3 8 e 0.2 5 0.1 é oO 0 5 10 15 Tip Speed Ratio (TSR) Figure 3. A typical C, versus tip-speed ratio curve Since this is a fixed-pitch turbine, the wind turbine must rely on the blade-stall condition to limit the output power when the winds are at high speed. During a stall condition, the wind turbine operates at a very low effi- ciency (C, is very low). The tip-speed ratio for a fixed- speed wind turbine varies with wind speed (Figure 4). Note that although the rotor speed of an induction gen- erator varies with wind speed, the speed range is within a 1% to 2% slip. On the other hand, the wind-speed varia- tion may range from 5 m/s to 25 m/s; thus, from the wind turbine point of view, the induction generator operates at a relatively “fixed speed” compared with the range of wind-speed variation. At high wind speeds, the tip-speed ratio is expected to be low (to the left side of Cymax in Figure 3) as the wind speed increases at a relatively “fixed” rotor rpm. The term “fixed speed” is commonly used to describe a wind turbine mechanically coupled to a constant-frequency induction machine. In reality, the rotor speed of wind turbine varies with the slip (0-2%) as the wind speed varies. @ and TSR versus Wind Speed 16 0.5 14 4 42 \ L 0.4 40 Sh | 0.3 % 8}—f 8 F 6 / l o2 + 0.1 2+ 4 = 0 0 0 5 10 15 20 25 Wind Speed (m‘ TSR ind Speed (ms) =a Figure 4. C, and TSR as a function of wind speed for a fixed-speed turbine Pow er and dPYdV versus Wind Speed 250 I 40 200 LN 7 0 F = / 2 £ =~ 150 7 1 = o x £ 100 ov & 50 10% af | -29 & — -30 0 10 20 30 Wind Speed (ns) “== = = Powe | Pidv Figure 5. Power curve and rate of power change of a typical wind turbine As shown in Figure 4, the performance coefficient (C,), which represents the efficiency of the wind turbine, varies as the wind speed increases. As the wind speed increases, the TSR decreases and the C, increases until it reaches the maximum C, at TSR = 7.8 (which corre- sponds to about 8 m/s for this wind turbine). As the wind speed continues to increase, the performance coefficient C, declines. This process makes the wind turbine self- regulate its output power by operating at lower efficiency at high wind speeds. A typical power curve and the rate of power change of a wind turbine are presented in Figure 5. The rate of change of the output power versus wind speed occurs at maximum C, which is about 8 m/s. In the high-wind- speed region, the rate of change is diminished when the wind turbine goes into stall mode (lower C,). This is good news for the power system environment. Thus, wind turbulence reflected in the power system network weakens as the wind turbine operates in high wind speeds (stall region). Power fluctuation is not linearly propor- tional to the cube of wind speed, especially in the high- wind-speed region. The inertia of the turbine rotor (including the blades), together with the shape of the C,-TSR curve, behaves like a low-pass filter that damps out the high-frequency power fluctuation in the power network. This low-pass filter screens out the wind-speed fluctuation in the high-wind- speed region so that the resulting output power is smoother than the wind-speed fluctuation, especially in high wind speeds. C. Induction Machines Most electric machines used as the prime mover in in- dustry are induction motors. Two applications of induc- tion machines in the power system network fall within the scope of this study. One application is as a generator on a wind turbine and the other is as a motor driving large pumps and compressors. By its nature, an induction ma- chine is an inductive load. Either as a motor or as a gen- Source | Motor . ap? jXs} Is Rs j Xis j Xr »— W099 ry qi et Figure 6. Equivalent circuit of an induction ma- chine connected to power system ; Imaginary part of Y (mho) Real part of Y (mho) Figure 7. Current locus of an induction machine from the start-up as a motor to operation as generator erator, this machine absorbs reactive power. The reac- tive power absorbed by the induction machine comes from the line to which it is connected. In hybrid power, the reactive power comes from the synchronous genera- tor of the diesel genset. In.a wind turbine generator, a fixed capacitor is usually installed to supply some of the reactive power needed by the induction generator. The equivalent circuit of an ifi- duction machine connected to a power system is given in Figure 6. The power system is represented by the infinite bus E, and reactance X, which represents the line imped- ance. Figure 7 shows the admittance of a typical induction machine as the operating slip varies from slip = 1 (mo- toring) to slip = -1 (generating). The admittance diagram illustrates a variable load connected to a voltage source. It illustrates the current path as the operating condition changes. If the induction machine is connected to an in- finite bus, the admittance diagram shows how the induc- tion machine current (the magnitude and phase angle) varies as the operation changes from start-up as a motor to operation as a generator. A large inductive current is drawn during start-up (slip = 1), and it shrinks as the slip moves closer to the synchronous speed where the current is minimum. As the induction machine enters the gener- ating region (slip <0), the current increases. In the entire region of operation (both as a motor and as a generator), the current always operates in the inductive (lagging) mode. The stator current changes in magnitude and phase angle (power factor) from start-up to generating rated power. leading lagging Figure 8. Terminal voltage changes as the phase angle of the load current is converted from leading to lagging for the same current amplitude and power factor ‘ motoring generating Figure 9. Terminal voltage changes as the load changes from motoring to generating D. Voltage variation due to load changes Between the load and an infinite bus in a power net- work, impedance represents the line impedance, the transformer impedance, and other elements. Although an infinite bus can maintain its voltage, the voltage at the terminal of the load varies due to variation in the drop in voltage across the line impedance. The voltage drop across the impedance depends on the size of the current and the power factor of the load. The terminal voltage varies (Vs) as the load current is Bw oS Terminal voltage Vs (volts) Y S S 000 1100 1200 1300 1400 rotor speed (rpm) +++ Vs (terminal voltage) ---- Es (infinite bus voltage) Figure 10. Per-phase terminal voltage variation as the generator speed varies (synchronous speed is 1,200 rpm) changed from leading to lagging for the same current magnitude and the same power factor (Figure 8). The terminal voltage (Vs) of an induction machine drops (with respect to an infinite bus Es) as the induction ma- chine is operated at the same power factor (both lagging) and the same current magnitude (Figure 9). In Figure 10, per-phase terminal voltage Vs (line-to- neutral) variation is shown as the induction machine is operated from start-up as a motor and then as it is oper- ated as a generator. During start-up, voltage drops sig- nificantly at the terminal voltage of the induction ma- chine. The voltage drop across line impedance is caused by the current surge during start-up. In addition, the phase angle of the stator current is very large and lagging (Figure 7). A poor power factor, lagging, and a large current surge combine to create a voltage dip at the ter-_ ‘inal of the induction machine during start-up. Thus, a short-duration start-up is preferable to a prolonged one. One way of reducing the voltage drop during start-up is to reduce the starting current. This is accomplished by using variable-frequency drives (as in a variable-speed wind turbine) or a_soft-start device. Although the dura- tion of start-up is prolonged when using the soft-start, the voltage drop across the line impedance is reduced sig- nificantly. The voltage dip occurs only at the time of very high slip (i.e., during start-up). In the operating slip region (slip = + 2-3%), the voltage drop is very small. IV. DYNAMIC ANALYSIS In this section, the power system we studied is simu- lated using a package program\ RPMSim, developed at the National Renewable Energy Laboratory [4]. The case studies look at different aspects of major power system components in the power network. The major component investigated is the diesel power. A diesel generator must respond properly to changes in the power balance in the power network. The power balance in the power network is maintained by the diesel control. Case Study I: Wind Turbine Start-up During start-up, the in-rush current entering the induc- tion machine operating as a generator may cause a volt- age dip on the power system network. Many wind tur- bine designers let the wind turbine self-start by using wind energy to bring the generator rpm up to speed and then connecting the generator when it is near its synchro- nous speed. Thus, the motoring interval is very short. Single-speed and two-speed wind turbines A single-speed wind turbine is equipped with only one generator. One advantage is that many off-the-shelf, sin- gle-speed induction machines are available from many manufacturers. A single-shaft, input-output gearbox is needed to transfer the aerodynamic power from wind to the generator. Two-speed wind turbines can be imple- mented by using two induction machines with different speed ratings or by using one induction machine with two different windings with a different set of poles wound on the same stator frame. When two induction machines are used, the wind turbine needs to use a gearbox with a sin- gle-shaft input connected to the low-speed shaft and two output shafts connected to the two generators. On the other hand, with a single-induction machine with two sets of windings, the gearbox needs to be connected to a sin- gle shaft only. The dual winding must be a custom- wound induction generator. From an energy point of view, the two-speed wind turbine can yield more energy than a single-speed wind turbine because the wind tur- bine can be operated at low rpm when the wind speed is low and operated in higher rpm when the wind speed is high. The simulation is started with zero initial output voltage and zero initial speed of the diesel generator. Thus, the observation is started after t= 1.5 s when both the voltage and the frequency reach their rated values. The rated output of the diesel genset is set to 750 kW and the line-to-line voltage reference is 480 V. > Hybrid Project - ~O-Vs at the WTG terminal > €Vs at the Diesel Genset terminal Lie > 10 2 ae? & 3 os 3 s ates 6 > JS + + > 13°) 2) oP 2 6 OS 8 aS 6 eS Ts Time (sec) Figure 11. Per-phase voltage variations at the terminal of the wind turbine and at the terminal of the diesel generator Figure 11 shows the voltage variation at the terminal of the wind turbine and at the terminal of the synchronous generator as the wind turbine is started and load is ap- plied. The wind turbine is self-started (i.e., the turbine starts as the wind speed increases). The wind turbine is connected to the power network as the generator speed reaches 1,050 rpm (about 83% of synchronous speed) at = 2s. The start-up electrical transient is over within less than 0.5 s because of the self-start capability of the wind turbine. The voltage dips at the diesel genset terminal drop about 13% before the exciter of the synchronous generator recovers the voltage to normal. As expected, when starting (motoring) the wind turbine generator, the voltage at the terminal of the wind turbine generator (Vs) Hybrid Project Frequency and WTG Speed 1.050 -~©-Diesel output Frequency (per unit) €WTG speed (per unit) 1.025 1.000 :975 .950 925 Diesel Freq and WTG speed (in p.u.) .900 Time (sec) Figure 12. The speed of wind turbine and the output frequency of the diesel generator is lower than the voltage at the terminal of the synchro- nous generator because of the voltage drop across the line impedance between the diesel genset and the wind turbine. Figure 12 shows the frequency variation as the wind turbine is started at t= 2s. The frequency dips about 5% before it recovers to the normal frequency. Also shown in Figure 12 is the trace of the wind turbine speed. Once the wind turbine speed passes the synchronous speed (above 1.0 per unit or p.u.), the induction generator starts generating. As seen on the chart, the rotor speed varies as the wind changes. The output frequency of the diesel genset also varies because of imperfections of the diesel governor in keeping the frequency constant. A two-speed wind turbine that uses a smaller genera- tor to start the wind turbine will show a less severe de- viation in voltage and frequency. This is because the starting current of the smaller generator is much smaller than the starting current for a large generator. Sequential start-up of wind turbines Most wind farms have a number of turbines. Some turbines are installed in a straight line and spaced at least three blade diameters apart with respect to one another. If the wind comes in a perpendicular direction to this line of turbines, the wind will reach the turbines at the same instant. Thus, all the turbines in the line may start at the same time, creating an accumulation of surge current drawn from the power system. In this case, the voltage dip is amplified. On the other hand, if the turbine is not perpendicular to the line, there will be a delay between the start-up of one turbine and that of the next turbine. The voltage and frequency dips recover before the next turbine starts. A signal can be added to each individual turbine that tells the turbine to start up during the short period of time between the start-up of one turbine and that of another. With a sequential start-up, instantaneous start-up can be avoided and severe voltage dips can be avoided. Case Study Il: Water Pump Start-up The prime mover of the water pump and compressor are the same. Both of them are driven by the induction motor. The start-up of a large induction motor can sig- nificantly affect the voltage dip and frequency dip on the power system. Direct on-line start for a water pump and compressor have the same effect on the power system as the wind turbine start-up. The difference is that in a wind turbine, the wind helps the generator reach its synchronous speed faster, whereas in a water pump or compressor, the pump or compressors do not. In a two-speed wind turbine, the smaller induction machine is used during start-up; thus, the voltage and frequency dip is not the same as if the larger induction machine is used during start-up. In Figure 11, the voltage dips are due to the start-up of the water pump at t= 6s. The voltage reduction is much smaller than the one caused by wind turbine start-up (about 4.3%). This is obvious because the size of the water pump is only 80 kW (35% of the size of the wind turbine). The voltage at the terminal of the wind turbine is also affected by the voltage drop at the terminal of the synchronous generator. In Figure 12, the frequency variation during the start-up of the water pump (at t = 6 s) is barely noticeable. A wind turbine is started up when the generator speed is very close to the rated speed. Thus the transient time during the start-up of the wind turbine is shorter than the transient time during the start-up of the water pump. The real power of the diesel, wind turbine, water pump, and the local load is shown in Figure 13, which also shows how the interaction among the generators and the loads takes place. The sign convention for real power and reactive power is that a positive sign means that a particular piece of equipment is producing real power or > Hybrid Project Real Power (kW) 1500 > ~S-Diesel Genset >€Wind Turbine » 1000). ;-@-Water Pump -é-Local Load > a g 500 é > 3 ob > -500F > Water pump starts Wind turbine starts -1000 [ 1 1 1 1 > 1 2 3 4 5 6 7 8 Time (sec) Figure 13. Real power of the diesel genset, wind turbine, water pump, and local load reactive power and a negative sign means that the equip- ment is absorbing real power or reactive power. Initially, the synchronous generator supplies a constant 400-kW local load when, at t = 2 s, the wind turbine is turned on. The wind turbine is in the motoring mode while its speed is below synchronous speed; therefore, the power (and the reactive power) absorbed by the wind turbine must be supplied by the diesel genset. Once the wind turbine reaches synchronous speed, it starts producing power and contributes to power generation (supplying the 400-kW local load). The diesel power generation drops from 400 kW to about 175 kW because 225 kW of power is con- tributed by the wind turbine. Also, the wind turbine gen- erates relatively smooth power at high wind speed. The water pump is started at t= 6 s. When it is started, the starting power required comes from the power system (diesel genset and wind turbine). Thus, the diesel genset must generate additional power to supply the start-up of the water pump. Once the transient period is over, the water pump absorbs a constant power of 80 kW. The diesel genset and wind turbine share the total load of 480 kW. The contribution from the wind turbine is relatively constant, and the power absorbed by the local load is also constant at 400 kW. Case Study Ill: Diesel - Wind Turbine Interaction A diesel generator consists of a diesel engine and a synchronous generator. The diesel engine is responsible for controlling the frequency and keeping it constant through its governor. The synchronous generator is re- sponsible for controlling the voltage via its field winding and voltage controller. Undersized diesel engine The ability of a diesel engine to change speed is its ac- celerating or decelerating power. The diesel accelerates when the diesel power is higher than the electrical output power of the generator (including losses). The diesel de- celerates when the diesel power is lower than the electri- cal output power of the generator (including losses). An oversized diesel engine does not have problems in accel- eration or deceleration, but an undersized diesel engine may create problems, for example during the startup of a wind turbine or large compressor. Figure 14 illustrates a condition in which the diesel is undersized with respect to the load. The genset fre- quency and the terminal voltage of the wind turbine gen- erator are shown on the top graph, and the real power of the diesel, wind turbine, water pump, and local load are shown in the bottom graph. At the start-up, the wind tur- bine uses the smaller, 40-kW generator to motor up the wind turbine and bring the induction machine up to speed. The wind is low; thus, the wind turbine operates at low output power, and the local load is set to 200 kW. The diesel engine has a rated power of 400 kW. Att =2 Hybrid Project a 1.10/-—S-\eat the WTG terminal & 1.05. ><Genset. frequency g 1.00) i 95 zy a 90 3 s s 85h 80) 1 1 1 1 1 1 1 1 0 4 6 8 10 12 14 16 18 20 Time (sec) > Hybrid Project Real Power (kW) > —900rSaDiesel Genset €Wind Turbine >» 1 -4-Water Pump ~@Local Load 200 oe = o 4 -200 > -400h . ‘Wind turbine starts Water pump starts FH Additional toad -600 —. as 0 4 6 8 10 +12. «14 +16 18 20 Time (sec) Figure 14. Voltage, frequency and power to illustrate an undersized diesel genset s, the wind turbine is turned on. As we can see, the volt- age dip and the frequency dip are not very large because the wind turbine is started up using a smaller generator. At t = 10 s, the 80-kW water pump is started up. The start-up time for the water pump takes longer than the start-up time for the wind turbine. This is because the wind turbine is started when the rotor speed is close to the synchronous speed; it also gets some help from the wind. The voltage drop is not significant, but the fre- quency of the diesel drops about 3%. The diesel output power increases to cover the real power needed, whereas the contribution from the wind turbine is insignificant because the wind is low. For a short period of time, the induction generator enters the motoring region between t = 10.8 s andt= 11.3 s. After the condition is restored, at t = 14 s, the additional local load (300 kW) is turned on, thus bringing the total load to 580 kW. The diesel can carry only up to 400 kW, and the contribution from the wind is very small (about 40 kW). The voltage and fre- quency start decreasing. The voltage and frequency sen- sors detect the change. If the frequency drops below 95% and the voltage drops below 90% for an elapsed time of 0.5 s, the controller will drop the additional load (300 kW) and keep the critical load (200 kW) to regain the voltage and frequency. After the load is shed at t = 14.5 s, the frequency and voltage eventually return to normal. When the frequency drops, there is a sudden jump in the wind turbine power contribution due to a sudden increase in the generating slip. Eventually, the genset frequency increases again for a short period of time and the induction generator enters into a motoring condition (between t = 14.5 s and t= 15s). This condi- tion is made worse if the mechanical time constant of the wind turbine rotor (including the blade) is higher than the diesel genset time constant. In other words, the changing of the genset rotor speed is much faster than the changing of the wind turbine rotor speed. The response to the load change is shown by how fast the governor corrects the frequency and how fast the field excitation control of the generator reacts to the voltage changes. Oversized wind turbine An oversized wind turbine presents another problem to the diesel power plant. The diesel engine can present a driving torque only to the synchronous generator it is driving. It can present a small braking torque only when it is driven. The braking torque (if any) is due to engine compression. The synchronous generator generates the power needed by the load in the power network. When the wind turbines operate, the power they generate offsets the diesel power. The balance of power is maintained by the diesel engine. Assuming a constant load, any in- Hybrid Project ~O-Vs at the WTG terminal |. -+Rotor speed (p.u.) wo \ Freq. runaway a a a Vs (pu) and rotor speed (pu...) eo Time (sec) Hybrid Project Real Power (kW) —SDiesel Genset 475s. 2€Wind Turbine ~@-Local Load ju E ¥ » vv ‘Real Rwer, P 3 6 -150 ay Load decreased 275 Wind turbine starts ae -400 mshi) 2 2.5 3 3.51]|]/4 45 5 Time (sec) Figure 15. Voltage, rotor speed, and power of an oversized wind turbine crease in the wind power output means a reduction in the diesel power ouput. When the wind power output ex- ceeds the power required by the load, the synchronous generator of the diesel genset becomes a synchronous motor that tends to accelerate the rotor speed of the die- sel engine. Thus, the excess energy from the wind power tries to drive the diesel engine. Because the diesel engine has only a small braking capability due to engine com- pression, the frequency control can be lost when the extra power generated by the wind turbine is sufficiently high. In Figure 15, the diesel generator has a rated power of 400 kW, the local load is initially set to 280 kW at t = 4 s, and the local load is set to 100 kW. When the diesel is started, there is only a local load of 280 kW. Then, the wind turbine is started at t= 2 s with a 225-kW induction machine. Although the diesel genset is rated at only 400 kW and the wind turbine is started with a 225-kW induc- tion machine, the effect of wind turbine start-up on the power system is very mild, mostly because the induction machine current is limited by a soft start. As shown in Figure 13, the same wind turbine (225-kW) draws a starting power of 300 kW. After the installation of the soft start (Figure 15), the power surge during start-up drops to about 100 kW. Soft start is a device that limits starting current during start-up. It consists of a pair of back-to-back thyristors installed in a series with each phase of the motor winding. Because the firing angle of current can be adjusted by controlling the firing angle of the thyristors. Variable-frequency drive (VFD) is seldom used to start a wind turbine for economic reasons. For a variable-speed wind turbine employing VFD (not the subject of this paper), the start-up capability is integrated into the system. Soft start is very useful to reduce the current surge during start-up of large induction machines; thus, the voltage dip and frequency dip can be reduced. During start-up, the resulting voltage dip at the wind turbine terminal is insignificant in comparison to direct online start-up (Figure 13). When the wind turbine is started, the voltage at the wind turbine drops to 90%. In comparison, without a soft start, even with a larger diesel, the voltage at the wind turbine terminal drops more than 30% (Figure 11). After the wind turbine enters generating mode (at about t = 2.5 s), the local load (280 kW) is shared be- tween the diesel genset (55 kW) and the wind turbine (225 kW). The voltage and frequency are maintained constant. The diesel genset generates only a small per- centage of its rated load (about 13%). This indicates a significant contribution to fuel savings from wind energy. At t= 4 s, the local load is reduced from 280 kW to 100 kW, while the wind speed stays the same. As a re- sult, the wind turbine tries to supply 225 kW, while the only load available is 100 kW. As a result, the synchro- nous generator of the diesel genset turns into a motor (negative power), the governor loses its speed control, and frequency run-away occurs. This is an example of the wind turbine being oversized in comparison to the local load. In such a case, a dump load (water heater, water pump, battery charger, etc.) is usually deployed to keep the diesel genset generating, thus preventing it from motoring. Minimum power generation of the diesel gen- set is usually preset (for example, 15%-40% of the rated load). If the generated power of the diesel genset is less than the preset value, the dump load should be deployed. The dump load must be sized so that the diesel genset will always generate power above its minimum set point. The dump loads are normally noncritical loads used to store excess electrical energy into another form of en- ergy, such as heat (water or space heater), electric charge (battery charging), or potential energy (water pump). V. CONCLUSIONS This paper starts with an overview of the components of the power system under investigation. The operating characteristics of the components are described as they relate to power quality in the power network. Steady- state predicted effects of wind power fluctuations on the power system are presented. ;—_ The following factors contribute to voltage fluctuation: e Voltage drop across the line impedance caused by | large current surges | e Current surges developed at the start-up of an in- duction machine operating from fixed frequency (large induction machines), which produce reactive current surges ¢ Current surges, which can be reduced as follows: - For wind turbines with a dual-winding generator, use smaller-sized induction machines during start-up, and connect the generator when the rpm starts rotating close to the synchronous speed of the generator. Or, use a soft start to control the stator current. - For other large loads (water pumps or compres- | sors), use a soft-start device to limit the current surges, use VFDs if affordable, and use sequen- tial start-up when appropriate. e Slow response of the exciter as the synchronous generator side reactive power mismatch ¢ Uncoordinated capacitor switching. | The following factors contribute to frequency aes tion: e Large, real power surges during start-up of the in- duction motor and wind power fluctuations } ¢ Slow response of the governor control at the diesel side | Frequency dip when the diesel engine is loaded by a | sudden, large load or the diesel engine is under- | rated with respect to the load ¢ Frequency run-away when the wind turbine produces | more power than needed; in this case, the diesel en- | gine loses control of its rotor speed | ¢ Change in genset frequency that may affect the op- | eration of the induction generator (e.g., during fre- quency fluctuation, the induction generator may J enter the motoring region for a very short period of time). Power fluctuation in the high-wind-speed region is very mild in comparison to the wind fluctuation. Thus, the wind turbine acts as a low-pass filter to the wind- speed input. Many technical solutions can be implemented to rem- edy the shortcomings presented in this paper. However, in any power generation, the economic implications of the solutions have to be considered very carefully. ACKNOWLEDGMENTS The authors acknowledge the strong support of the National Renewable Energy Laboratory and U.S. De- partment of Energy during the completion of this study. This project was supported under contract number DE- AC36-98GO10337. REFERENCES 1] E. Muljadi, L. Flowers, J. Green, and M. Bergey, "Electrical Design of Wind-Electric Water Pumping," ASME Journal of Solar Energy Engineering, Nov. 1996, Vol. 118, pp. 246-252. 2] J.T.G. Pierik and M. De Bonte, Quasi Steady State Simulation of Autonomous Wind Diesel Systems (Status Report), Report No. ECN-85-091, Netherlands Energy Research Foundation, Petten, May 1985. 3] A.J. Tsitsovits and L.L. Freris, “Dynamics of an Iso- lated Power System Supplied from Diesel and Wind,” Proc. IEEE, 130, Part A, No. 9, pp. 587-595, 1983. 4] J.T. Bialasiewicz, E. Muljadi, G. Nix, and S. Drouil- het, “RPM-SIM simulator: A comparison of simulated versus recorded data, ”Proceedings of WINDPOWER ’98,” Bakersfield, California, pp. 423-432, 1998. 5] E. Muljadi and C.P. Butterfield, "Pitch-Controlled Variable-Speed Wind Turbine Generation," Transactions of the IEEE-Industry Applications Society, Jan./Feb. 2001, Vol. 37, No. 1, pp. 240-246. 6] P.C. Krause and C.H. Thomas, “Simulation of Sym- metrical Induction Machinery,” JEEE Trans Power Ap- paratus and Systems, Vol. 84, pp. 1038-1053. 7| C.M. Ong, Dynamic Simulation of Electric Machin- ery Using Matlab/Simulink, Prentice Hall PTR, 1998. REPORT DOCUMENTATION PAGE Onl NO. G7080188 Public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other as} of this Collection of information, including Suggestions for reducing this burden, to Washington leader Services, Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, Arlington, VA 22202-4302, and to the Office of Management and Budget, Paperwork Reduction Project (0704-0188), Washington, DC 20503. 4. AGENCY USE ONLY (Leave blank) | 2. REPORT DATE 3. REPORT TYPE AND DATES COVERED September 2001 Conference Paper . TITLE AND SUBTITLE : , : 5. FUNDING NUMBERS Power Quality Issues in a Hybrid Power System . AUTHOR(S) WER13220 E. Muljadi, H.E. McKenna . PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGANIZATION REPORT NUMBER ). SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSORING/MONITORING National Renewable Energy Laboratory AGENCY REPORT NUMBER 1617 Cole Blvd. Golden, CO 80401-3393 NREL/CP-500-30412 11. SUPPLEMENTARY NOTES 12a. DISTRIBUTION/AVAILABILITY STATEMENT 12b. DISTRIBUTION CODE National Technical Information Service U.S. Department of Commerce 5285 Port Royal Road Springfield, VA 22161 13. ABSTRACT (Maximum 200 words) We analyzed a power system network, which consisted of two types of power generation: wind turbine generation and diesel generation. The power quality and the interaction of diesel generation, the wind turbine, and the local load were the subjects of investigation. From an energy-production point of view, it is desirable to have as much wind energy production as possible in order to save fuel consumption of the diesel engines and to reduce the level of pollution. From the customer point of view, it is desirable to have good power quality at the receiving end. The purpose of this paper is to show the impact of wind power plant in the entire system. Also, we discuss how the startup of the wind turbine and the transient condition during load changes affects the voltage and frequency in the system. 44. SUBJECT TERMS 15. NUMBER OF PAGES wind turbine; diesel generator; hybrid power; power system; power quality; renewable energy; energy production; wind energy 16. PRICE CODE . SECURITY CLASSIFICATION 18. SECURITY CLASSIFICATION 19. SECURITY CLASSIFICATION 20. LIMITATION OF ABSTRACT OF REPORT OF THIS PAGE OF ABSTRACT Unclassified Unclassified Unclassified UL NSN 7540-01-280-5500 Standard Form 298 (Rev. 2-89) Prescribed by ANSI Std. 239-18 298-102 NREL/CP-500-24681 * UC Category: 1213 Modular Simulation of a Hybrid Power System with Diesel and Wind Turbine Generation J. T. Bialasiewicz, E. Muljadi, S. Drouilhet, G. Nix National Wind Technology Center Presented at Windpower '98 Bakersfield, CA April 27-May 1, 1998 fe. ens — oll NREL National Renewable Energy Laboratory 1617 Cole Boulevard Golden, Colorado 80401-3393 A national laboratory of the U.S. Department of Energy Managed by Midwest Research Institute for the U.S. Department of Energy under contract No. DE-AC36-83CH10093 Work performed under task number WE802230 May 1998 NOTICE This report was prepared as an account of work sponsored by an agency of the United States government. Neither the United States government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercialproduct, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States government or any agency thereof. The views and opinions of authord expressed herein do not necessarily state or reflect those of the United States government or any agency thereof. Available to DOE and DOE contractors from: Office of Scientific and Technical Information (OSTI) P.O. Box 62 Oak Ridge, TN 37831 Prices available by calling (423) 576-8401 Available to the public from: National Technical Information Service (NTIS) U.S. Department of Commerce 5285 Port Royal Road Springfield, VA 22161 (703) 487-4650 ay, ue Printed on paper containing at least 50% wastepaper, including 20% postconsumer waste MODULAR SIMULATION OF A HYBRID POWER SYSTEM WITH DIESEL AND WIND TURBINE GENERATION J.T. Bialasiewicz, E. Muljadi, S. Drouilhet, G. Nix National Wind Technology Center National Renewable Energy Laboratory 1617 Cole Boulevard Golden, CO 80401 U.S.A. ABSTRACT In this paper, we present the modular simulation tool we developed to help study the system dynamics for wind-diesel power systems. The principal modules of the simulator, which include a diesel generator, a wind turbine generator, a rotary converter with a battery, a village load and a dump load, are described. With a case study, we demonstrate how the designer benefits from easily understanding the effects of system modifications. Using this tool, a designer can easily develop control strategies to balance the system power flows under different generation/load conditions. INTRODUCTION The advantage of hybrid power systems is the combination of the continuously available diesel power and locally available, pollution-free wind energy. With the hybrid power system, the annual diesel fuel consumption can be reduced and, at the same time, the level of pollution can be minimized. A proper control strategy has to be developed to take full advantage of the wind energy during the periods of time it is available and to minimize diesel fuel consumption. Therefore, a proper control system has to be designed, subject to the specific constraints for a particular application. It has to maintain power quality, measured by the quality of electrical performance, i.e., both the voltage and the frequency have to be properly controlled. This results in a need for a simulation study of each new system to confirm that a control strategy results in desired system performance. Using the VisSim visual environment, we developed the modular simulation system to facilitate an application-specific and low-cost study of the system dynamics for wind-diesel hybrid power systems [1]. The simulation study can help in the development of control strategies to balance the system power flows under different generation/load conditions. Using the typical modules provided, it is easy to set-up a particular system configuration. In this paper, we present the principal modules of the simulator and, using a case study of a hybrid system, we demonstrate some of the benefits that result from easily understanding the effects of the designer’s modifications to these complex dynamic systems. In these systems, the voltage and the frequency are controlled by the diesel generator. The wind speed varies with time and so does the village load. Therefore, we regard the diesel generator as a controlled energy source, whereas the wind is an uncontrolled energy source and the village load is an uncontrolled energy sink. The difference between the power consumed by the village load and the power generated by the wind turbine is balanced by the diesel generator. On occasion, the wind speed can be very high [2], resulting in energy generation that exceeds the energy demand of the village load. Under such circumstances, the power from the diesel becomes very low and the wind may try to drive the diesel engine. Should the wind turbine override the diesel, the frequency control could be lost and the system would enter the instability region. To avoid this condition, the dump load is controlled so that the power generated by the diesel will always be higher than a minimum value. In addition, the energy surplus can be saved for future use by utilizing the rotary converter/battery assembly. Therefore, the dump load should be regarded as a controlled energy sink. When the battery is being discharged the rotary converter/battery assembly should be regarded as a controlled energy source. On the other hand, when the battery is being charged, it should be regarded as a controlled energy sink. By properly choosing the sequence of events programmed for our case study, we demonstrate that all operation aspects of the hybrid power system, briefly discussed above, can be easily taken into consideration. The single-line diagram of a simulated system, discussed as a case study, which involves all modules available in the simulator, is shown in Fig.1. We must include the point of common coupling (PCC) module as a node at which all power sources and power sinks are connected in every simulation diagram. The other principal modules shown in Fig.1 are the diesel generator (DG), the AC wind turbine (WT) with the induction generator and the wind speed time series as the input, the rotary converter (RC) with the battery bank (BB), the village load (VL) and the dump load (DL). In addition, R+jX represents the transmission line impedance. In all electrical simulations, we use the d-q axis convention and synchronous reference frame. In particular, in the PCC module, the q-axis and d-axis components v,,,; and Vqs; of the line voltage V, are defined. Fig.1 A typical hybrid power system The electric machine models we used and modified in the development of the simulator can be found in many references [3, 4]. The wind turbine model can be derived from many papers in wind energy [5, 6]. BASIC CHARACTERISTICS OF THE PRINCIPAL MODULES OF THE SIMULATOR Diesel Generator Module The top-view block diagram of this module is presented in Fig.2. Notice that the user can easily set the voltage set point V,,., and the frequency (speed) set point f. at required values. The frequency is controlled via the diesel torque and the line voltage is controlled through the field current of the synchronous generator. According to this block diagram, the speed control block generates a proper fuel/air ratio (represented by the variable %vgz,) for the diesel engine, so that the engine can generate the proper torque for the synchronous generator, as a result of which the power requirement can be satisfied and, simultaneously, the system frequency can be kept close to the required value. Notice also that the line voltage V; is controlled by the voltage regulator through the proper adjustment of the field current of the synchronous generator. torque kW SYNCHRONOUS GENERATOR KVAR % fuel field AC terminal frequency current voltage set point 60 (fet VOLTAGE REGULATOR AC terminal voltage voltage set point [266 >| Vs ref} > + Fig.2 Block diagram of the diesel generator module: principal functional blocks and their interconnections Figure 3 represents a simple simulation diagram of the diesel engine. Its POWER={(% vez) characteristic is represented by the minimum value of the % vez, (the dead zone) and the slope of the straight line. The user defines this characteristic to approximate the diesel engine involved. One can see from Fig.3 that the rated power for the simulated diesel engine is 200 kW. Using the generated power and angular velocity Gem rad/sec, the torque Tyiese: is generated assuming the first-order dynamics, whose time constant represents the diesel modeled. The actual power generated by the diesel, represented by the variable Peen, is compared with the minimum diesel power (chosen in Fig.3 to be 25% of the diesel rated power or 50 kW in our simulation). The difference is used to control the number of active dump load elements (connected in parallel) needed to keep the generated power P,., above the minimum diesel power. The number of active dump load elements is represented in this simulation diagram by the variable DL gir. This variable is used to control the dump load. %_FUEL S| 1 torque * et *EOWER p / , Voor | Easiest ——>— Cc 2000 +f Gen_m. rad/seq 5 Diesel slope * 1 minimum value of the %FUEL +(Peenl-¥ L,| 103 HEH} ADL switch} 100 n engine seed 100%_FUEL x ‘ [Gen_m_rad/sed =} Gamadiel = Minimum Diesel Load Fig.3 Simulation diagram of the diesel engine block The torque equation 1 Gen = — J (Taser ~ Tren ~ Boo Ger m_rad/sec ~ "y gen SD jdt , m_rad/sec is simulated to obtain the angular velocity of the diesel generator Gentm_paasec. In this equation, the torque Ten represents the generated electrical power Prgen_p (includes generator power losses), i.e., rT P, Egen_D gen > Genny, pat/sec Jsp is the moment of inertia, and Bsp is the viscous friction coefficient of the synchronous generator. The speed of the diesel engine is controlled by the variable %uvg1, generated by the governor (represented in the simulation by the PI controller), so that the relative frequency error 1- f/f, is driven to zero, where the frequency f is determined by the relation @ = 27f . In this relation, the angular frequency @ is found as corresponding to the angular velocity Gentm rad/sec - The field current iy of the synchronous generator is calculated according to the equation ae : ip => JU, - Ry at, it s. where V; is the voltage applied to the field winding. It is controlled by the line voltage error V, ,., —V, through a PI controller. AC Wind Turbine Module Figure 4 represents the principal functional blocks of the AC wind turbine module with their interconnections and all inputs and outputs clearly shown and labeled. One also can see that the reactive power has two components: one contributed by the induction generator and the other one contributed by the power factor correction capacitors block. Clicking with the mouse on any of the blocks shown in Fig.4, the user obtains its lower level expansion. We briefly discuss below the principal components of this module, i.e., the induction machine and the wind turbine. wind IG_kW speed WIND torque torque INDUCTION }—?>— +——————| TURBINE }—>—— el IG_KV. kVAR V_wind ROTOR GEARBOX| + wind ceneratog AS) POWER FACTOR PFI_KVAR CORRECTION > CAPACITORS Fig.4 Block diagrams of the AC wind turbine module Induction generator The mechanical input to the induction machine comes from the wind turbine via the gearbox. It should be noted that there is neither voltage control nor speed control in this machine. At the beginning, when the wind speed increases above the cut-in speed, the induction machine operates as a motor, drawing a large starting current and absorbing the real power. As the motor speed comes closer to synchronous speed, the stator current decreases and eventually reaches the value of the rated current. As the rotor speed increases above synchronous speed, the induction motor becomes an induction generator and starts to generate power. At the start-up of the induction machine there is an inrush current that must be sustained by the power system. The time during which the change of the rotor speed, from the value of zero to the rated speed, takes place is also influenced by the kinetic energy supplied to the wind turbine. The diesel generator must be able to provide the energy required by the induction generator and the wind turbine during the start-up operation without disturbing the power system stability. Wind turbine model The wind turbine is modeled based on the aerodynamic model provided by the wind turbine characteristic, described by the following equation defining the aerodynamic power Piero, generated by the rotor: Pry = OSPAC,V?, where V is the wind speed, C, is the performance coefficient, A is the turbine rotor area swept and p is the air density. For a particular wind turbine C, is given as a function of the tip-speed ratio TSR, defined as @,R mr TSR = where @, is the angular speed of the rotor with the radius R. This function assumes a maximum for a certain value of TSR. From the wind speed input V and the rotor speed @,, the operating point of the wind turbine can be determined. When the induction machine is generating, which corresponds to its operation in the vicinity of synchronous speed, rotor speed is relatively constant. Rotary Converter Module with the Battery The block diagram of the rotary converter, shown in Fig.5, represents its principal functional blocks with their interconnections and all inputs and outputs clearly shown and labeled. As shown in this figure, the rotary converter simulation diagram consists of the following principal blocks: e DC motor e Synchronous generator e DC machine field controller e Voltage regulator. battery torque voltage kW —,> pe ( SYNCHRONOUS] + MOTOR GENERATOR: {= KVAR field field AC terminal current current voltage Vs DC MACHINE FIELD CURRENT FIELD CONTROLLER REGULATOR. AC terminal voltage Vs Fig.5 Block diagram of the rotary converter module. The DC motor, when the battery is being discharged, generates a positive torque Tpc, driving the synchronous generator, and the DC power from the battery is converted into AC power supplied to the grid. Alternatively, when the battery is being charged, the DC motor works as a generator, i.e., it develops a negative torque Tpc or is driven by the synchronous generator which, due to this condition, is forced to operate as a synchronous motor. One of these two modes of operation is determined by the sign of the battery reference power Pgar rer With negative sign corresponding to charging the battery. In the simulation, the sign and the magnitude of Pg,r_,r can either be chosen by the user or can be used to help with the system energy balance. We can control the DC machine by controlling the armature voltage/current or by controlling the field voltage/current. While the former mode of control requires a large-size power converter, the latter requires a smaller-size power converter. Thus, the latter mode of control is used in our application. The field current of the DC motor i; pc is generated so that the battery power Paar = Vaart oc follows the battery reference power Pp47 re, Where vgsr is the battery voltage applied to the armature circuit of the motor and Jpc is the armature current. The emf Ep, generated by the synchronous generator driven by the DC motor, has the phase angle y= J rc —@)dt where @z~ is the angular frequency generated by the rotary converter and @ is the angular frequency generated by the synchronous generator driven by the diesel engine. Note that the difference @p¢ —@ is non-zero only in the transient and is used here to establish the angle y which, in steady-state, remains constant as a result of the equality @,,. = @ forced by the diesel generator. [_/ +1 K£_DCK steady-state component Kfc dynamic correction component PID controller correcting the value of Kf_DC »{0.0001 *LP_BAT ref I >) »[2e-005}-9[178 r + +(P_BAT] ae) {10000 -»{ 0.0001 1/sk {Ki mal peal if_DC qb 3) *L4fspfl »[if, Dc-—> M > +{REDC}Y, » ifDC LfDC UX Fig.6 Simulation diagram of the DC machine field controller. In Fig.6, the simulation diagram of the DC machine field controller is shown. It illustrates how the field current i pc is generated to drive the error Pyyr _,.¢ — Pg4r to zero. First of all, let us note that in this simulation diagram a new variable Vp. = Vg, is used. The field current i; pc is controlled by the DC/DC conversion coefficient Ky pc . In the upper part of Fig.6, this coefficient is shown as consisting of the steady-state component K; and the dynamic or transient correction component Ky, i.e, Kypc= Ky+ Kye - Let us reiterate that Ky represents the steady-state value of Ky pc, obtained under two assumptions: Mpc =@ and Pyyr = BAT _ref * Therefore, the necessary dynamic correction component K;, is generated as the output of the PID controller whose input is the difference P,4, — Pg4r yy. Due to the PID action, he coefficient K,. will, at the beginning of the transient, be greater than 1. Therefore, in order to make sure that Ky pc remains between 0 and 1, the sum K;+ K;,. is passed through the limiter (as shown in Fig.6). Dump Load Module The dump load is a set of parallel load resistances. Each parallel resistance can be turned on or off upon command. In a real application, the dump load may take the form of an electric heater to heat air or water. The main purpose of the dump load is to keep the diesel-generated power above a user-prescribed fraction of its rated power. It is introduced to avoid overpowering the diesel by the wind turbine but, under special circumstances, it also can be used to control the frequency. Any of these two control strategies dynamically determines the number of the dump load elements to be connected in parallel. For diesel power control, this number is represented by the variable DLswitch, determined in the simulation diagram in Fig.3. For the frequency control strategy, the calculation of this number is directly performed in the dump load module, which also contains a user-operated switch for choosing a particular strategy. A SIMULATION CASE STUDY OF THE DYNAMICS OF A HYBRID POWER SYSTEM The power system we chose to simulate consists of the following principal modules: e Diesel generator with rated power of 200 kW and minimum load of 50 kW e AC wind turbine driven by the wind given by a file of the wind speed time series e Village load of 50 kW at the power factor pf=0.75, switched to 80 kW at 11s e Rotary converter with a preprogrammed battery reference power ¢ Dump load with the diesel power control strategy selected. The results of the simulation are shown in Fig.7. We use the convention that the power generated is positive and the power consumed is negative. Notice that at 0 the diesel generator starts up. At approximately t=4s, the line voltage V, reaches its reference value of 266V, the relative frequency f/f, is close to 1, and both the real diesel power generated and the village power consumed are approximately 50 kW, as required. The first action of the diesel power control strategy appears in the transient period (before this steady state is reached) is seen as an additional load on the diesel when its generated power is below 50 kW. At f=5s, the battery reference power of the rotary converter is switched from the level of 0 to the level of 20 kW. We note that the response represented by the power Pyar is slow. This is a consequence of the field current control of the DC machine in the rotary converter. This additional power generation is immediately followed by its consumption by an increasing dump load, so that the diesel generation does not drop below the minimum value of 50 kW. This load is taken off immediately when at f=7.5s the induction machine starts to motor the wind turbine. In this phase of the simulation, the power is provided by both the diesel generator and the rotary converter. At 10s, the power consumption of the induction machine, reaching synchronism, rapidly drops and eventually, at 11s, it starts to generate. The power released and then generated by the AC wind turbine is now consumed by the village load, increased at 11s to the value of 80 kW, and by the rotary converter, which at f=12s starts to charge the battery due to the preprogrammed drop of the battery reference power from the value of 20 kW to the value of -10 kW. One can also notice small corrections of the power consumed, performed in this phase of the simulation by the diesel power control strategy. The influence of the sudden changes in the power generated and consumed in the system can be noticed in the line voltage V, and the relative frequency //f, transients, but their quality remains satisfactory. Jomod ‘poads putm Jo soovly, /SIy sjuouoduiod snorea 10J Aouanboy pur o8eq[OA ‘ (Pp) (y) S coheed N + ie o = 5 an \, a Z a, . . 5 mms Gog Fa 4 a. —~ 3 523 peg ge 3 e z tos o a as p= g . a . @ 2 3 ._ = S = z a c=) + %& © Se eo s a s & o Nn & - c z a £ 9 ao $ a oa EOS ra} ~sa 2 2B go 5 He § & a & 3 = a o a a ee fee q Q g 2 £ = oe & a @ ch a I Lhe 9 81 CONCLUSION The simulation results of a hybrid power system with diesel and wind power generation, presented in this paper, demonstrate that the modular simulation system, developed using the visual programming environment, constitutes a very useful tool for analysis and design of such systems. The modular simulation system has been developed to e Facilitate an application-specific and low-cost performance study of wind-diesel hybrid power systems (both mechanical and electrical components are simulated) e Analyze both static and dynamic performance e Help in the development of control strategies e Simulate different wind speed profiles and different village load profiles. The system has the following capabilities/characteristics: e Modular and multilevel structure provided by VisSim visual environment e Clear and easy-to-understand system presentation e Setting-up a particular desired configuration is within a click of the mouse © Modifications are easy to make e Effects of system modifications can be immediately examined. ACKNOWLEDGMENTS We wish to thank Neil Kelley for providing the wind data sets used in this work. We wish to acknowledge our management at the National Renewable Energy Laboratory and the U.S. Department of Energy (DOE) for encouraging us and approving the time and tools we needed for this project. DOE supported this work under contract number DE-AC36-83CH10093. References 'Visual Solutions, Inc., “VisSim User's Guide” Professional VisSim Version 3.0, 1997, Westford, Massachussetts. *Muljadi, E.; Buhl, Jr., M.; Butterfield, C.P., “Effect of Turbulence on Power Generation for Variable Speed Wind Turbines,” Presented at the ASME Wind Energy Symposium, Reno, NV, Jan. 6-9, 1997. 3Ong, Chee-Mun; “Dynamic Simulation of Electric Machinery,” Prentice Hall, 1998. ‘Kraus, P.C.; Thomas, C.J., “Simulation of Symmetrical Induction Machinery,” IEEE Trans. On Power Apparatus and Systems, Vol. PAS 84, No. 11, Nov, 1965, pp. 1038-1053. ‘Johnson, G.L., Wind Energy Systems, Englewood Cliffs, NJ: Prentice Hall, 1985. ‘Drouilhet, S.; Muljadi, E.; Holz, R.; Gevorgian, V. “Optimizing Small Wind Turbine Performance in Battery Charging Applications,” Presented at Windpower '95 Conference, Washington, D.C., March 27- 30, 1995. NREL/CP-440-22108 * UC Category: 1213 * DE97000083 An Analysis of the Performance Benefits of Short-Term Energy Storage in Wind-Diesel Hybrid Power Systems Mariko Shirazi Stephen Drouilhet Prepared for 1997 ASME Wind Energy Symposium Reno, Nevada January 6-9, 1997 te. @n*s we — | <> NREL National Renewable Energy Laboratory 1617 Cole Boulevard Golden, Colorado 80401-3393 A national laboratory of the U.S. Department of Energy Managed by Midwest Research Institute for the U.S. Department of Energy under contract No. DE-AC36-83CH10093 Work performed under task number WE712360 Revised April 1997 NOTICE This report was prepared as an account of work sponsored by an agency of the United States government. Neither the United States government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States government or any agency thereof. Available to DOE and DOE contractors from: Office of Scientific and Technical Information (OST!) P.O. Box 62 Oak Ridge, TN 37831 Prices available by calling 423-576-8401 Available to the public from: National Technical Information Service (NTIS) U.S. Department of Commerce 5285 Port Royal Road Springfield, VA 22161 703-605-6000 or 800-553-6847 or DOE Information Bridge http://www.doe.gov/bridge/home.html ay ue Printed on paper containing at least 50% wastepaper, including 10% postconsumer waste AN ANALYSIS OF THE PERFORMANCE BENEFITS OF SHORT-TERM ENERGY STORAGE IN WIND-DIESEL HYBRID POWER SYSTEMS Mariko Shirazi Stephen Drouilhet National Wind Technology Center National Renewable Energy Laboratory Golden, Colorado, USA Abstract A variety of prototype high penetration wind-diesel hybrid power systems have been implemented with different amounts of energy storage. They range from systems with no energy storage to those with many hours worth of energy storage. There has been little consensus among wind-diesel system developers as to the appropriate role and amount of energy storage in such systems. Some researchers advocate providing only enough storage capacity to supply power during the time it takes the diesel genset to start. Others install large battery banks to allow the diesel(s) to operate at full load and/or to time-shift the availability of wind-generated electricity to match the demand. Prior studies indicate that for high penetration wind-diesel systems, short-term energy storage provides the largest operational and economic benefit. This study uses data collected in Deering, Alaska, a small diesel-powered village, and the hybrid systems modeling software Hybrid2 to determine the optimum amount of short-term storage for a particular high penetration wind-diesel system. These findings were then generalized by determining how wind penetration, turbulence intensity, and load variability affect the value of short term energy storage as measured in terms of fuel savings, total diesel run time, and the number of diesel Starts. Introduction The main performance objective of a wind-diesel hybrid power system is to maximize fuel savings relative to a diesel-only system. The role of energy storage in accomplishing this goal has been addressed in previous studies." In a system without storage, the diesels must either be run continuously or switched on and off to meet the instantaneous net load, defined as the consumer load minus the available wind power. Fuel savings will be small for the continuous diesel case, and the diesel start/stop frequency will be high for the intermittent diesel case,'54689 While the diesel start/stop frequency can be reduced by imposing a minimum run time on the diesels, this has the effect of decreasing the fuel savings."“* In addition, since diesel generator sets cannot be started and This paper is declared a work of the U.S. Government brought on-line instantaneously, fuel savings is further limited in no-storage systems by the need to maintain a spinning reserve (additional on-line diesel capacity) to meet net load peaks, due to wind and load fluctuations. It has been shown that the introduction of even a small amount of energy storage increases the fuel savings while significantly reducing the number of diesel starts."**"* Our purpose is to further investigate the value of energy storage as measured in terms of fuel savings, total diesel run-time, and the number of diesel starts relative to a no- storage system. We restricted our attention to short-term storage, i.e., storage that is used to cover peaks in the net load due to stochastic wind and load variations, not to time-shift the wind resource to match the diurnal load pattern. This study attempts to quantify the benefit of short-term storage in a particular high penetration wind- diesel system and then to generalize the findings by determining how wind penetration, turbulence intensity, and load variability affect this value of storage. It was our intent both to corroborate and to extend the prior studies cited above. Although our results are not specific to batteries as the storage medium, in this study we have assumed battery storage, since it is currently the only field-proven industrial storage technology with sufficient capacity and power delivery capability. The study site is Deering, Alaska, a small diesel-powered village of approximately 160 inhabitants, where Kotzebue Electric Association, in partnership with the National Renewable Energy Laboratory and the Alaska Department of Community and Regional Affairs, is planning to install a high penetration wind-diesel demonstration project. The system was modeled using wind speed and load data collected from Deering and NREL’s hybrid system simulation model, Hybrid2. Hybrid2 is a computer software tool which can predict the long-term performance, including fuel use, diesel run- time, and diesel starts, of hybrid power systems under user-specified renewable resource and load conditions."™"* In addition, we were able to use Hybrid2 to perform a sensitivity analysis using different levels of wind penetration, turbulence intensity, and load variability, allowing the results to be applied to power and is not subject to the copyright protection in the United States. systems at different sites and of different size and wind penetration levels than the Deering system. This paper presents the methodology and results of the Hybrid2 analysis. , Methodology We selected fuel use, diesel run-time, and diesel starts as the criteria by which to judge the value of energy storage, all of which are simulation results provided by Hybrid2. Hybrid2 was thus a good tool for evaluating the benefit of various amounts of energy storage for the Deering wind- diesel system. After determining the benefit of energy storage under the conditions applying in Deering, we used Hybrid2 to determine the sensitivity of these results to varying levels of wind penetration, turbulence intensity, and load variability. The performance of a hybrid system depends on wind penetration, wind power variability, and load variability. Wind penetration, as used here, is the ratio of the generated wind power to the primary system load. Most often, we are referring to average wind penetration, for example, the annual wind energy generated divided by the annual electric demand. In our analysis, we have expressed wind power variability as turbulence intensity, which is a property of the local wind. Turbulence intensity is defined as the standard deviation of the wind speed divided by the mean over a given averaging Wind Turbines Rotary - I Tr Converter interval. The term turbulence intensity is normally used in the context of short averaging intervals (up to several minutes). At the time scales we are interested in here (several minutes to half an hour), it may be more correct to speak of the coefficient of variation of the wind speed. The mathematical definition is the same, and in this paper we use the term turbulence intensity merely to be consistent with the terminology of Hybrid2, our principal modeling tool. The actual relation between turbulence intensity and wind power variability depends on the specific model of wind turbine and the number of wind turbines used. Load variability is defined as the standard deviation of the load divided by the mean over a given averaging interval. System Configuration The existing Deering diesel power system consists of four diesel gensets rated at 60 kW, 113 kW, 125 kW, and 135 kW. The planned Deering wind-diesel hybrid power system (see Figure 1) consists of the smaller three diesel gensets, three 65-kW wind turbines, a rotary power converter, a battery bank, an 180-kW optional resistive heating load (“dump” load) and associated power controllers, and a main system controller. The village load varies from around 30 to 130 kW. Over the data collection period (Jan 26 - July 14, 1996), the average village load was 53 kW while the expected average wind power (from Hybrid2 results) was 42 kW, giving an average wind penetration of 80%. Battery Bank me [7} BE kK SCR Controller Waste Heat } System Resistance Heater 60 kW Diesel m wi 113 kW Diesel au tt 125 kW Diesel Village Distribution System Figure 1. Layout of the Deering Wind-Diesel System Choice of Simulation Time Step Hybrid2 is a combined _ probabilistic/time-series simulation model which uses a time-series to predict long-term performance and applies statistical analysis to predict short-term behavior within each time step." To run a simulation in Hybrid2, the user must input time- series of wind speed and load data, as well as specify the complete power system and control (dispatch) strategy. The system dispatch strategy determines the interaction between the storage and the diesel generators and how each will be used to supply the load. For this study, the selection of an appropriate dispatch strategy and the simulation time step turned out to be non-trivial. In the Deering system, batteries will be used for “peak shaving.” Diesels will be dispatched as necessary to meet the average net load (based on, for example, a 15-minute moving average of the net load). Power will only be drawn from the battery to eliminate the need to bring a(nother) diesel on-line to meet a short-term increase in net load that exceeds the diesel capacity already on-line. In Hybrid2, this operating strategy is effectively modeled if one specifies the “Multiple Diesel Load Following” dispatch strategy, which dispatches only enough diesel capacity to cover the average net load for each time step. Hybrid2 uses a statistical algorithm, based on the user- specified standard deviation of wind and load during each time step, to determine the maximum net load during each time step. If, according to Hybrid2's battery model, there is not enough available energy stored in the battery to meet the peaks within that time step, then additional diesel capacity is run for that time step. In this Hybrid2 dispatch strategy, the battery is only discharged to cover any (probabilistically determined) transient peaks above the rated power of the on-line diesels within each time step. Consequently, this method will only model battery charge and discharge events which are shorter in duration than the time step. If the simulation time step is small, e.g., one minute, then the batteries will only be used to cover net load peaks smaller than one minute, and enough diesel capacity will be dispatched to cover the minute-average load. In this case, Hybrid2 will underestimate the battery usage, and Overestimate the diesel cycling frequency. In order to use Hybrid2 to model longer storage discharge events, we were obliged to use a longer simulation time step. Conversely, since Hybrid2 cannot model diesel state transitions within a time step, it requires that the minimum diesel run-time be greater than or equal to the simulation time step. However, the longer the diesel minimum run-time, the lower the fuel savings. Thus, we were forced to strike a balance between a time step long enough to allow the batteries to cover all net load peaks within our range of interest, and a time step short enough to allow for a useful minimum run time. As a compromise, we selected a time step of 30 minutes. The specified time step and diesel dispatch strategy ensure that there is enough diesel spinning reserve to meet the 30-minute average load. Thus, on average, battery discharges (to cover net load peaks) would be limited to 15 minutes in duration. Consequently, our method (using the 30-minute time step) cannot accurately evaluate the performance of a system in which the battery storage is used to cover net load peaks of more than 15 minutes duration. This is not a major shortcoming, however, because, as will be seen, the performance gain due to the addition of energy storage decreases rapidly beyond around 10-minutes nominal storage capacity (10 minutes at average system load). Model Inputs The data inputs for Hybrid2 are time-series of wind speed and load data from Deering for Jan 26 - July 14. The wind speed was logged over this time interval as one- minute averages and standard deviations. One-minute load data was only available for June, so the June data was scaled for the rest of the months using monthly load averages from Deering. The one-minute data was converted to 30-minute data for use in Hybrid2. The nominal turbulence intensity of the 30-minute wind speed data was calculated as 0.12. Since standard deviation was not included in the load data, a constant load variability of 0.10 was specified. Both of these values, as well as the average wind penetration, were scaled up and down for each of the test cases in the sensitivity analysis to determine each parameter’s effect on the value of energy storage. The Deering system and all of the modified systems were run with various amounts of storage. The storage size is indicated by its nominal energy capacity in kWh, which is its rated amp-hour capacity times the nominal battery bank voltage. Storage size is also expressed as the amount of time that the nominal energy capacity, if fully available, could cover the average system load. These values are nominal battery sizes presented for purposes of comparison only and do not necessarily represent the amount of energy storage actually available to the system at any given time, which is dependent on charge history and discharge rate. In a no-storage system, the diesels would need to be dispatched to cover the instantaneous net load (load minus wind power). The Hybrid2 code “knows” what the future maximum net load will be (calculated from time- series data and statistics within each time step) and therefore can dispatch the minimum amount of diesel capacity necessary to ensure that the load can always be met. Real wind-diesel systems cannot predict the future, and so a system without storage must maintain enough spinning reserve to cover all possible sudden net load peaks due to village load peaks and/or wind power drops. Therefore, the fuel savings in a real system will be less than as predicted by Hybrid2 for the no-storage case. To approximate a real system in Hybrid2, we have added a fixed 20 kW offset to the maximum net load. The zero offset cases were included to show the theoretical fuel savings possible with a hypothetical no-storage system capable of accurate short-term load and wind prediction. Finally, along with the “Multiple Diese! Load Following” dispatch strategy and 30-minute minimum diesel run time, we specified the minimum battery % State of Charge (below which the batteries will not be allowed to discharge) at 20%. In addition, the minimum allowed power of all diesels was specified at 20% of rated power. Results The results for the Deering case are shown in Figures 2 through 4. We ran simulations varying the storage capacity from no storage to 65.8 kWh nominal, equivalent to a nominal 74 minutes of energy storage at average load. Fuel consumption, diesel run-time, and diesel starts all decrease sharply, relative to the no-storage case, with increasing storage, up to a storage equivalent of approximately 10 minutes at average load. A nominal 10-minutes equivalent worth of storage reduces the fuel use by 18%, the diesel run-time by 19%, and the number of diesel starts by 44% (from an average of 7.6 starts/day to 4.2 starts/day) compared to the no-storage case. In this case, there does not appear to be any benefit to increasing the amount of storage beyond a nominal 18 minutes at average load. In order to determine the extent to which the amount of battery capacity that Hybrid2 shows to be sufficient is artificially limited by step-size, we ran a second set of simulations using 60-minute time steps. The 60-minute time step results did show slight improvement in performance from the nominal 18- to 37-minute storage sizes, however the difference is small. In addition, the shape of the 60-minute curve is almost identical to the shape of the 30-minute curve: the “knee” of the curve, where the majority of the benefits of storage have been realized, occurs at approximately 10-minutes nominal storage capacity for both time step results. While the use of longer and longer time steps will continue to show slight performance improvement for the larger storage cases, the rate of improvement becomes much smaller. Thus we conclude that there will in fact be little economic benefit to further increasing the storage size, considering the high cost of batteries. The optimum amount of storage for the Deering system appears to be 9-14 kWh nominal or 10-15 minutes nominal at average load. Figure 4 shows that a small amount of energy storage greatly reduces the amount of “dumped” energy, i.c., the amount of wind (or diesel) energy generated (here expressed as a percentage of the village demand) in excess of the village demand. While it is intended in Deering to use all excess wind and diesel energy in space and water heating applications, thereby saving heating fuel, the value of this energy is not as high as that of the energy that goes to meet the village electric load. For a given level of wind penetration, the lower the excess energy the better the economics of the wind-diesel system. Sensitivity Analysis The value of short-term storage is apparent for the Deering specific case, but the question is, how does this value change as the wind penetration, turbulence intensity, and load variability of the system change? A sensitivity analysis on these variables enables us to generalize the results of the Deering analysis to other sites and other wind-diesel systems with more or less wind penetration. Wind Penetration Figures 5 through 7 show the effect of wind penetration on fuel use, diesel run-time, and diesel starts. The fuel use trends are similar to those shown by Beyer et al., namely that even a small amount of storage has a strong effect on fuel savings relative to the no-storage cases, and that increasing the amount makes relatively little difference’. However, we are still faced with the question: to what extent is the lack of difference in performance between the storage sizes larger than 18-nominal minutes equivalent a reflection of our time step and operating strategy? Again, we ran a second set of simulations with 60-minute data. The results showed only marginal improvement in performance from an 18-nominal minute storage to a 37- minute storage. We conclude that for all these systems, the majority of the benefits of storage are realized by our strategy of using it only to cover peaks up to 15 minutes duration. Therefore the results in Figures 5 through 7 will still approximate actual trends concerning the value (or lack thereof) of increasing storage. The value of short-term storage, in terms of fuel savings and diesel run time, increases as the wind penetration increases, because there will be an increasing amount of time that the available wind power exceeds the load. At 50% wind penetration the systems with storage have approximately 20% greater fuel savings and 20% fewer diesel run-time hours than the no-storage, 20-kW offset case. Beyond 50% wind penetration, these benefits increase only slightly. The diesel cycling trends are similar to those shown by Beyer, et al... Because the wind-hybrid diesels are approaching continuous operation at the lowest wind penetrations, the number of diesel starts approaches the diesels-only number of starts (2.4 starts/day). The number of diesel starts for all the hybrid cases increases sharply as wind penetration increases and then levels off at about 80% wind penetration. The inclusion of storage significantly mitigates this increase in diesel starts, so that above 80% wind penetration, diesel starts per day are reduced by 50% relative to the no-storage, 20-kW offset case. Turbulence Intensity Figures 8 through 10 show the effect of turbulence intensity on the value of storage. It has been shown that whereas the fuel use relative to the diesels-only case climbs as the wind turbulence intensity (and hence the wind power variability) increases, the greater the energy storage the less the impact of wind variability’. At high turbulence intensities (high wind power variability), there is apparent benefit to increasing the nominal storage size well beyond 15 minutes, even though the simulation is subject to the same 15-minute discharge limitation discussed earlier. This is to be expected, because under conditions of high wind power variability, there will be high magnitude net load peaks. The effective battery capacity drops with increasing discharge current. For the same delivered energy, the higher the battery power required to meet the net load peaks, the larger the battery must be. The trend for diesel run-time is similar. Since most sites will have a turbulence intensity in the range from 0.1 to 0.2, it is evident in Figure 9 that some amount of energy Storage is necessary to preserve the diesel-run-time- reducing benefit that addition of wind power offers. The trend for diesel starts is similar but more pronounced, Anytime the storage is unable to cover a transient peak in the net load, another diesel must be Started. The diesel may not need to be run for a long time, but it must be started. In addition, the performance of the smaller storage sizes not only decreases with increasing turbulence intensity, but also begins to approach or even fall below the performance of the no- storage cases. This is because the no-storage case diesels are dispatched according to the maximum net load (plus offset), but the storage-case diesels are dispatched according to the average net load with the storage covering any transient peaks (above the rated capacity of the on-line diesels) if possible. Thus there will be situations in a high turbulence intensity, small storage case, where depending on the size of the transient peak, a diesel may be required one time step but not the next, potentially leading to a higher number of diesel starts than with the no-storage case, in which the diesel in question would have been running continuously. In any case, Figure 11 shows that as wind turbulence increases, energy storage is increasingly important in limiting the frequency of diesel starts. Finally, we must point out a potentially misleading aspect of our turbulence intensity analysis. Note that at high turbulence intensity, it appears that no-storage cases can have greater diesel run times and consume more fuel than if there were no wind power at all. This is due to the algorithm Hybrid2 uses to estimate the maximum net load in a particular time step. Diesels are dispatched to meet the maximum expected net load, which is the average net load plus the expected net load variation. The latter is determined by statistically combining the expected wind power variation (based on turbulence intensity) with the expected village load variation. In cases of very high turbulence intensity, and thus high wind power variation, this calculation leads to high values of net load variation, which can result in maximum net loads that actually exceed the maximum expected village load. This causes Hybrid2 to dispatch more diesel capacity than it would for a diesel-only system. If the peak net load is higher than the village load, then there are moments when the wind turbines are drawing power from rather than delivering power to the bus. This can in fact occur with certain wind turbines in gusty wind conditions. The results presented in Figures 8 through 10 would therefore be accurate if the turbines were allowed to motor at significant power levels for short periods of time, because in that case, the maximum diesel load would indeed be greater than in the diesel-only case. Most wind turbines, however, are designed to preclude large motoring currents. With such turbines, we would expect the fuel use and diesel run-time curves to level off below the diesel-only values as turbulence intensity increased. Load Variability Figures 11 through 13 show the effect of load variability on the value of storage. Load variability is different than the other parameters we investigated in that the diesels-only performance itself changes as the load variability changes which means that as the load variability increases, the fuel consumption in all cases (storage and no-storage hybrid as well as diesel- only cases) increases, so that the performance of the hybrid cases relative to the diesels-only case may be the same or even increase. This is the case for fuel use, where the no-storage cases show only a slight decrease in performance from 0.1 to 0.3 load variability, while the storage cases actually show a slight increase in performance. The trend for both fuel use and diesel run-time is for the value of storage to increase as the load variability increases; however, the actual amount of storage only makes a small difference. At low load variability (less than 0.1) the benefit of all storage cases above no-storage cases is essentially constant (17% reduction in fuel use and diesel run-time), because at low load variability, the variability of the net load is dominated by wind variability. Beyond 0.1 load variability, the storage cases begin to differentiate slightly, with the larger storage sizes showing slightly increasing benefit over the smaller storage sizes and all storage cases showing greater benefit Telative to the no-storage cases. Load variability most significantly impacts the number of diesel starts, The basic trend is similar to that for turbulence intensity in that as the load variability increases, the storage cases begin to separate, with the performance of the smaller storage cases approaching that of the no-storage cases. For load variability, however, the number of starts for the diesels-only case also begins to approach that of the no-storage cases, and exceeds the number of starts for all storage cases at any load variability above 0.15. This is another manifestation of the energy storage’s beneficial effect of eliminating diesel Starts that occur only to handle short-term peaks in the net ~ load. Conclusions We evaluated the effect of various amounts of energy storage on the operating performance of the wind-diesel system planned for Deering, Alaska, and found that a modicum of energy storage, 10-15 minutes nominal capacity at average load, greatly reduced diesel fuel consumption, diesel run-time, and diesel starts, relative to the no-storage case. When modeled with three 65-kW wind turbines, using actual measured wind and load data, the fuel savings by the no-storage hybrid system, relative to the diesel-only case, were about 21%. The savings increased to about 37% with the addition of a nominal 15-kWh battery. We also examined three factors that significantly effect the benefit of, and need for, energy storage. These are wind penetration level, wind power variability (expressed as turbulence intensity), and load variability. The benefits, relative to a no-storage case, of including energy storage increase to varying extents as each of these factors increase. At wind penetration levels below 25%, energy storage contributes very little, but even a small amount of storage contributes a great deal in most high penetration systems. However, there is not significant benefit to adding larger amounts of storage except in cases of high wind power variability. In a very steady wind, (e.g. trade wind) the benefit of energy storage will be much less than in a wind regime with higher turbulence intensity. A large enough storage can effectively eliminate a hybrid system’s reduction in performance as turbulence intensity increases. The same trend is observed for load variability, but to a lesser degree. At low levels of load variability, however, the benefit of energy storage is somewhat insensitive to the actual value of load variability, since at those low levels, the variability of the net load is likely to be dominated by the wind power variability anyway. In much larger wind-diesel systems than our reference case, say 1- to 2-MW average load, the potential performance gains from energy storage may be reduced, since both the short-term load variability and the wind power variability may be less than with a smaller system. The wind power variability would be less if a much larger number of similarly sized machines were used. On the other hand, it is just as likely that a small number of larger wind turbines would be used. A determination of the need for energy storage in large wind-diesel systems must be based on a similar analysis based on the actual system architecture and local wind and load conditions. Note on Energy Storage System Design The authors wish to stress that the results presented here are not sufficient in themselves to properly design a battery energy storage system, which consists not only of a battery but a power conversion system to interface it to the AC power bus. A variety of factors must be considered in such a design, including the real and reactive power demands, the conversion efficiencies of the power converter, and the actual pattern of charging and discharging that will be experienced by the battery. Knowing the actual charge and discharge profile is also essential to predicting the life of a given battery bank in a particular application. Hybrid2 is designed to be used with a relatively long simulation time step (typically in the range of 15 to 60 minutes) and is therefore unable to accurately model the actual high-rate short-duration charge and discharge events experienced by short-term energy storage. To overcome this limitation, we have developed a simple wind power surplus/deficit analysis program that determines the actual magnitude, duration, and frequency of occurrence of battery charge and discharge events in a hybrid power system with energy storage. This program uses one-minute average wind and load data as input. Examples of this analysis will be presented in a separate paper on battery life prediction. References 1. Beyer, H.G., Degner, T., Gabler, H., “Operational Behaviour of Wind-Diesel Systems Incorporating Short-Term Storage: An Analysis via Simulation Calculations,” Solar Energy, Vol. 54, , No. 6, pp. 429-439, 1995. 2. Scotney, A., Infield, D.G., “Wind-Diesel Systems for Developing Countries,” published in Technology and Developing Countries, Frank Cass, London, 1995. 3. Beyer, H.G., Degner, T., Gabler, H., Stubbe, G., Cheng-Xu, W.., “Effect of Wind Field Properties on the Fuel Saving Potential of Wind-Diesel Systems,” Proceedings European Wind Energy Conference, Madrid, 1996, pp. 675-679. 4. Beyer, H.G., Degner, T., “Accessing the Maximum Fuel Savings Obtainable in Simple Wind-Diesel Systems,” Proceedings European Union Wind Energy Conference and Exhibition, Goteborg, Sweden, 1996. 5. Slack, G., Sexon, B., Collins, R., Dunn, P., Lipman, N., Musgrove, P., “Wind Energy Systems with Battery Storage and Diesel Back-up for Isolated Communities,” Proceedings of the Second British Wind Energy Association Workshop, 1980, pp. 134-142. 6. Freris, L.L., Attwood, R., Bleijs, J.A.M., Infield, D.G., Jenkins, N., Lipman, N.H., Tsitsovits, A., “An Autonomous Power System Supplied from Wind and Diesel,” Proceedings of the European Wind Energy Conference, Hamburg, 1984, pp. 669-673. 7. Contaxis, G.C., Kabouris, J., Chadjivassiliadis, J., “Optimum Operation of an Autonomous Energy 10. 11. 12. 13. 14. 15. 16. System,” Proceedings of the European Wind Energy Conference, Rome, 1986, pp. 305-310. Lipman, N.H., De Bonte, J.A.N., Lundsager, P., “An Overview of Wind/Diesel Integration: Operating Strategies and Economic Prospects,” Proceedings of the European Wind Energy Conference, Rome, 1986, pp. 75-92. Bullock, A.M., Musgrove, P.J., “The Effects of Turbulence Spectra and Load Profiles on the Operation of a Modelled 60 kW Wind/Diesel System,” Wind Energy Conversion 1987, Proceedings of the Ninth British Wind Energy Association Wind Energy Conference, Edinburgh, 1987. Infield, D.G., Bleijs, J.A.M., Coonick, A., Bass, J.H., White, J.T., Harrap, M.J., Lipman, N.H., Freris, L.L., “A Wind/Diesel System Operating with Flywheel Storage,” Proceedings of the European Community Wind Energy Conference, Herning, Denmark, 1988, pp. 367-372. Skarstein, O., Uhlen, K., “Design Considerations with Respect to Long-Term Diesel Saving in Wind/Diesel Plants,” Wind Engineering, Vol. 13, No. 2, 1989, pp. 72-87. Manwell, J.F., McGowan, J.G., Jeffries, W., “Experimental Data from the Block Island Wind/Diesel Project,” Wind Engineering, Vol. 13, No. 3, 1989, pp. 111-131. Infield, D.G., “An Assessment of Flywheel Energy Storage as Applied to Wind/Diesel Systems,” Wind Engineering, Vol. 14, No., 2, 1990, pp. 47-60. Toftevaag, T., Uhlen, K., Skarstein, O., “Wind/Diesel/Battery Systems - The Effect of System Parameter Variations on Long-Term Fuel Savings and Operation,” European Wind Energy Conference, Amsterdam, 1991, pp. 505-513. Baring-Gould, E. Ian, The Hybrid System Simulation Model, Version 1.0, User Manual, National Renewable Energy Laboratory, Golden, CO, June 1996, NREL/TP-440-21272. Manwell, J.F., Rogers, A., Hayman, G., Avelar, C.T., McGowan, J.G., (University of Massachusetts), DRAFT Theory Manual for Hybrid2, The Hybrid System Simulation Model, National Renewable Energy Laboratory, Golden, CO, June 1996, NREL/TP-440-21182. 2 ® —?@—fuel use ——run time die Starts 9 a 2 N a P pp pp “eo WYN w& Relative Diesel Starts of diesels-only case) o @ e bars Wind Penetration = 0.80 Turbulence Intensity = 0.12 1.7 Load Variability = 0.10 Relative Fuel Use and Diesel Run Time (fraction ° & _ io e a 0 20 40 60 80 Nominal Storage Size (minutes at average load) Figure 2. Relative Fuel Use, Diesel Run-Time, and Diesel Starts vs. Storage Size 8 —@— run time 21.5 —@- starts a _ i =~ 2y On O©adsa = an 18.5 os a anoaannn @ a Average Diesel Starts (starts/day) Wind Penetration = 0.80 Turbulence Intensity = 0.12 Load Variability = 0.10 Average Diesel Run -Time (hours/day) _ o a 17.5 oe N @ Nomindl Storage Size (Pinutes at sveoge load) 7 Figure 3. Absolute Diesel Run-Time and Diesel Starts vs. Storage Size 0.43 0.42 ? 0.41 Ef os S 0.36 Wind Penetration = 0.80 0.35 Turbulence Intensity = 0.12 Load Variability = 0.10 0 20 40 60 80 Nominal Storage Size (minutes at average load) Figure 4. "Dump" Energy vs. Storage Size - S © only case) ° ao Relative Fuel Use (fraction of diesels —@— No Storage, 20 kW Offset —f— No Storage, No Offset ~~~ 8.2 kWh (9 minutes) —<— 16.4 KW (18 minutes) —H— 32.9 kWh (37 minutes) —@— 65.8 kWh (74 minutes) 0.7 0.6 Turbulence Intensity = 0.12 Load Variability = 0.10 05 0 05 1 15 2 25 3 3.5 Average Wind Penetration Figure 5. Fuel Use vs. Wind Penetration 32 —#—Dieseis Only Zz 30 i No Storage, 20 KW Offset 3 28 ~~~ No Storage, No Offset S 26 —%—8.2 kWh (9 minutes) = a —¥— 16.4 kW (18 minutes) § —@— 32.9 kWh (37 minutes) m 22 —+—65.8 kWh (74 minutes) 3 20 18 3 3 6 Turbulence Intensity = 0.12 < 14 Load Variability = 0.10 12 0 05 1 15 2 25 3 3.5 Average Wind Penetration Figure 6. Diesel Run-Time vs. Wind Penetration 13 —@— Diesels Only i No Storage, 20 kW Offset " —-#e~ No Storage, No Offset —<— 8.2 kWh (9 minutes) 9 —— 16.4 KW (18 minutes) —@— 32.9 kWh (37 minutes) —+— 65.8 kWh (74 minutes) Average Diesel Starts (starts/day’ N 0 0.5 1 1.5 2 25 3 Average Wind Penetration Figure 7. Diesel Starts vs. Wind Penetration 3.5 Turbulence intensity = 0.12 Load Variability = 0.10 Relative Fuel Use (fraction of diesels ~—@— No Storage, 20 kW Offset 1 —— No Storage, No Offset ~~~ 8.2 KWh (9 minutes) 0.9 —<— 16.4 KW (18 minutes) q —¥%— 32.9 kWh (37 minutes) z 08 —@— 65.8 kWh (74 minutes) 07 06 Wind Penetration = 0.80 Load Variability = 0.10 S a 0 0.1 0.2 0.3 0.4 Average Turbulence Intensity Figure 8. Fuel Use vs. Turbulence Intensity a —— Diesels Only g 30 —# No Storage, 20 kW Offset 2 28 -~#e-~ No Storage, No Offset 2 26 —%—8.2 kWh (9 minutes) i 2 —#— 16.4 kW (18 minutes) & 22 —@— 32.9 kWh (37 minutes) i 20 —+— 65.8 kWh (74 minutes) 3 ie $ Wind Penetration = 0.80 = 14 Load Variability = 0.10 12 0 0.1 0.2 0.3 0.4 Average Turbulence Intensity Figure 9. Diesel Run-Time vs. Turbulence Intensity 13 —?— Diesels Only =z —— No Storage, 20 kW Offset : 1 te No Storage, No Offset 29 —%—8.2 kWh (9 minutes) 2 —#— 16.4 kW (18 minutes) & 7 —@—32.9 kWh (37 minutes) 3 —+— 65.8 kWh (74 minutes) a5 2 $3 Wind Penetration = 0.80 Load Variability = 0.10 0 0.1 02 0.3 0.4 Average Turbulence Intensity Figure 10. Diesel Starts vs. Turbulence Intensity 10 Relative Fuel Use (fraction of diesels 0.1 0.15 02 0.25 Load Variability Figure 11. Fuel Use vs. Load Variability —#—No Storage, 20 KW Offset —l— No Storage, No Offset ~~~ 8.2 kWh (9 minutes) —>— 16.4 KW (18 minutes) —#*— 32.9 kWh (37 minutes) —@®— 65.8 kWh (74 minutes) Wind Penetration = 0.80 Turbulence Intensity = 0.12 888K aSRR Average Diesel Run-Time (hrs/day a — a =~ wD 0 0.05 0.1 0.15 Load Variability Figure 12. Diesel Run-Time vs. Load Variability 02 0.25 03 —?— Diesels Only —#— No Storage, 20 kW Offset —~te-~ No Storage, No Offset —>—8.2 kWh (9 minutes) —#— 16.4 KW (18 minutes) —@— 32.9 kWh (37 minutes) —+— 65.8 kWh (74 minutes) Wind Penetration = 0.80 Turbulence Intensity = 0.12 a Average Diesel Starts (starts/day a ~~ 0 0.05 0.1 0.15 Load Variability 0.2 0.25 0.3 Figure 13. Diesel Starts vs. Load Variability —? Diesels Only —i— No Storage, 20 kW Offset ~~ No Storage, No Offset —%— 8.2 kWh (9 minutes) —%— 16.4 kW (18 minutes) —@— 32.9 kWh (37 minutes) —+— 65.8 kWh (74 minutes) Wind Penetration = 0.80 Turbulence Intensity = 0.12 11 January 2001 * NREL/CP-500-29174 RPM-Sim: A Comparison of Simulated Versus Recorded Data Preprint J.T. Biatasiewicz University of Colorado at Denver E. Muljadi, G. Nix, and S. Drouilhet National Renewable Energy Laboratory To be presented at the 39" American Institute of Aeronautics and Astronautics (AIAA) Aerospace Sciences Meeting Reno, Nevada January 8-11, 2001 fe omy - > MEL * National Renewable Energy Laboratory 1617 Cole Boulevard Golden, Colorado 80401-3393 NREL is a U.S. Department of Energy Laboratory Operated by Midwest Research Institute e Battelle e Bechtel Contract No. DE-AC36-99-GO10337 NOTICE The submitted manuscript has been offered by an employee of the Midwest Research Institute (MRI), a contractor of the US Government under Contract No. DE-AC36-99G010337. Accordingly, the US Government and MRI retain a nonexclusive royalty-free license to publish or reproduce the published form of this contribution, or allow others to do so, for US Government purposes. This report was prepared as an account of work sponsored by an agency of the United States government. Neither the United States government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States government or any agency thereof. Available electronically at http://www.doe.gov/bridge Available for a processing fee to U.S. Department of Energy and its contractors, in paper, from: U.S. Department of Energy Office of Scientific and Technical Information P.O. Box 62 Oak Ridge, TN 37831-0062 phone: 865.576.8401 fax: 865.576.5728 email: reports@adonis.osti.gov Available for sale to the public, in paper, from: U.S. Department of Commerce National Technical Information Service 5285 Port Royal Road Springfield, VA 22161 phone: 800.553.6847 fax: 703.605.6900 email: orders@ntis.fedworld.gov online ordering: http://www.ntis.gov/ordering.htm 7 fae Printed on paper containing at least 50% wastepaper, including 20% postconsumer waste RPM-Sim SIMULATOR: A COMPARISON OF SIMULATED VERSUS RECORDED DATA Jan T. Biatasiewicz Department of Electrical Engineering University of Colorado at Denver Denver, Colorado Eduard Muljadi Senior Engineer R. Gerald Nix Manager Geothermal Program Stephen Drouilhet Senior Engineer National Wind Technology Center National Renewable Energy Laboratory Golden, Colorado ABSTRACT This paper compares simulated versus recorded data for the RPM-SIM simulator, developed at the National Re- newable Energy Laboratory’s National Wind Technol- ogy Center. The simulator was used to study the system dynamics of a wind/diesel hybrid power system. We also provide information on newly developed simulator modules that will be released. The simulator performed extremely well, demonstrating flexibility in making modifications and including specialized modules re- quired for problem solving. We also outline several possible applications for this tool. INTRODUCTION Hybrid power systems combine continuously available diesel power with pollution-free wind and/or solar en- ergy. Another advantage of these systems is that annual diesel fuel consumption can be reduced, thus minimiz- ing pollution levels. However, to take full advantage of the wind and/or solar energy during periods of maximum availability, a proper control system has to be designed, subject to the constraints for a particular application. It has to maintain power quality measured by electrical performance (i.e., both the voltage and the frequency have to be properly controlled). Thus, each new system must be simulated to confirm that a par- This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States. ticular control strategy results in desired system per- formance. Using the VisSim™ visual environment, we developed the modular simulation system RPM-SIM'® to facilitate a low-cost application-specific study of the system dy- namics of wind-solar-diesel hybrid power systems. The simulation can aid developing system control strategies to balance the power flows under different generation and load conditions. Using the typical modules pro- vided, it is easy to set up a particular system configura- tion. Some simulation studies require modifications of the existing modules and/or inclusion of specialized modules*. In the simulation study presented in this paper, we use the dump load module, which we modi- fied to represent full physical model of the Wales Con- trol System Dump Load Dispatch used at the Hybrid Power Test Bed (HPTB) located at the National Re- newable Energy Laboratory. Many researchers recognized the need for a tool that would facilitate analysis and design of hybrid power systems. An interesting study of modeling efforts of wind/diesel systems was performed by Jeffries.° Among those who have developed dynamic models of wind/diesel systems (in chronological order) are Tsitso- vits and Freris,° Pierik and De Bonte,’ Papadopulos et al.,® Uhlen and Skarstain,? Manwell et al,,'° Papadopulos 1 VisSim is the trademark of Visual Solutions. et al.,'! Lundsager et al.,’? Binder," Jeffries et al.'* and Ladakakos et al.'® Of the various simulation tools, our simulator includes probably the largest selection of modules and control strategies. For example, the user can model systems that contain a diesel generator and an inverter con- nected to a battery bank or to a photovoltaic (PV) array; the inverter can work in a slave mode or in a master mode when the diesel generator is idling. In such a system, the user can easily program the master/slave switching strategy. It is also possible to simulate sys- tems without diesel generators (such as battery charging systems) with one or more wind turbine generators, in which the frequency and voltage are controlled by the inverter operating in the master mode. The RPM-SIM is the first dynamic hybrid system with a symbolic graphical user interface. Figure | presents principal modules of the RPM-SIM included in a single-line diagram of a hybrid power system. All elements of the simulated system are con- nected to one module, called the point of common cou- pling (PCC). The other principal modules shown in Figure 1 are the diesel generator (DG), the alternating current (AC) wind turbine (WT) with the induction generator and the wind speed time series as the input, the rotary converter (RC) with the battery bank (BB), the inverter with the PV array, the village load (VL), and.the dump load (DL). R+jX represents the transmis- sion-line impedance and PFC represents the power- factor-correcting capacitors. In all electrical simulations, we use the d-q axis con- vention and synchronous reference frame. In the PCC module, the q-axis and d-axis components v,.; and vgs) of the line voltage V, are defined. In electric machine and power system analysis, it is common to use the transformation from three phase quantities a, b, and c into the d-q axis. This transformation, known as Park’s transformation, was pioneered by Park'® and Stanley.'” In 1965, Krause and Thomas'® generalized the d-q transformation for arbitrary reference frame. The Park’s transformation has the unique property of elimi- nating all time-varying inductances from the voltage equations. It also is used by Ong’? for dynamic simu- lation of electric machinary. For all modules included in the simulator, we assumed (for both real and reactive power) the following general power sign convention: the power absorbed is displayed as negative and the power generated is displayed as positive. A village load (as an inductive load) always absorbs both the real and the reactive power and both powers are always negative. The real power of a diesel generator always remains positive and its reactive power may be positive or negative. The real power of the wind turbine generator is negative during motoring and is positive during generation. Its reactive power is always absorbed, i.e., it is displayed as a negative quantity. This power convention makes easy the inter- pretation of the simulation results. In particular, it makes apparent the interpretation of the instantaneous power balance. In the sections that follow we first describe the modules of the simulator. We discuss some modifications done to account for unmeasured variables and/or to include the specifics of the simulated system. Next we present simulation results versus recorded data. Finally we summarize the results and present the potential applica- tions of the simulator. SIMULATOR MODULES Diesel generator module The diesel generator module, presented in Figure 2, includes models of the diesel engine and the synchro- nous generator, the engine speed control block, and the voltage regulator. The engine speed control block gen- erates the fuel/air ratio, represented by the %pug, vari- able, to keep the frequency constant. The voltage regulator determines the field current of the synchro- nous generator necessary to keep the voltage constant nder varying load conditions. The voltage set point V, re and the frequency set point /;, can be easily set at required values by the user using the dialog boxes in the simulation diagram at the level shown in Figure 2. With the addition of the first-order dynamics, we model the diesel engine according to the static relationship between the fuel/air ratio input and the generated power output. The aforementioned static characteristic is rep- resented as a straight line with the dead zone.’ The slope of the straight line part, which represents diesel capacity, and the dead zone, which represents the minimum value of the fuel/air ratio, are to be chosen by the user to approximate the engine involved. In addi- tion, the user can set the minimum diesel power Piiesel min AS a required percentage of the rated value. This variable is used by the dump load controller, which maintains this minimum diesel load. Thus, the dump load is active only when the diesel power is lower than the minimum diesel power. Figure 3 presents the principle of the voltage and fre- quency control implemented by the RPM-SIM model of the diesel generator. AC wind turbine module The AC wind turbine module simulates two-step con- version of wind power to electrical power. In the first step, wind power is converted to mechanical power represented by the torque developed by the wind tur- bine rotor and transmitted through the gearbox to the induction generator. In the second step, electrical power is obtained from the induction generator connected to the line. The wind speed is represented in time series. Figure 4 shows the principal functional blocks of the AC Wind Turbine module, together with their intercon- nections and all inputs and outputs. The reactive power has two components, one absorbed by the induction generator and the other contributed by the power factor correction capacitor block. The user can obtain its lower level expansion by clicking with the mouse on any of the blocks shown in Figure 4. Dump load module The dump load module is composed of parallel resistive loads. The principle purpose of the dump load is to keep the diesel-generated power above a_ user- prescribed fraction of its rated power. It can also (un- der special circumstances) be used to control the fre- quency. So, we have the following control strategies available: © Diesel power control strategy. e@ Frequency control strategy. Either control strategy dynamically determines the number of the dump load elements to be connected in parallel. Village load module The village load (VL) module generates the i,y and igy components of the load current. The user declares the rated real power consumption P, and the power factor pf and has a choice between fixed load (for which he/she provides the constant values of P,, and pfi)and the load profile (for which he/she provides the time series of P, and pf). The user makes his/her choice by clicking the button in the parameter module of the VL. The RPM-SiM format of a village load profile is shown in Figure 5. Rotary converter/battery bank assembly Figure 6 shows, that the rotary converter/battery bank assembly consists of a battery bank and two machines: (1) a DC machine and (2) a synchronous machine. Fig- ure 6 shows the principal functional modules of the rotary converter, together with their connections and all inputs and outputs. The rotary converter/battery bank assembly can be set up to operate in the synchronous condenser mode, i.e., to provide or absorb reactive power. This is accom- plished by setting to zero the battery reference power and consequently maintaining zero shaft torque and zero real power output. Other RPM-Sim_modules to be included in the second release of the simulator In this section, we briefly present the RPM-SIM mod- ules, which have not been included in the simulation performed to present the model performance versus recorded data. These include the Inverter Module and the PV Array Module. The inverters can work in one of two modes, i.e., the master mode or the slave mode. In the master mode, the inverter controls the system’s frequency and volt- age. The power exchange is determined by the system’s power balance. In the slave mode, the real and reactive power required to be generated or to be absorbed are specified by the user. The transfer from slave mode to master mode is deter- mined based on the control strategy designed by the hybrid power plant designer or operator. This is suit- able, for example, when the inverter uses a sufficient battery storage site, and during nighttime operation when the inverter’s battery carries the load and the die- sel generator is turned off. Power generated by occa- sional wind during the night will be stored in the bat- tery. During the day, when the load is close to rated power, the diesel is turned on and the inverter is oper- ated in the slave mode to charge the battery and to sup- port occasional peak loads. We developed a model of the inverter, which makes possible all specified options of operation. We tested this model under variable load conditions, switching between the slave mode, in which the system’s voltage and frequency are controlled by the diesel generator, and the master mode, in which the diesel generator is disconnected and the voltage and frequency control is taken over by the inverter. In addition, we tested the inverter’s operation in conjunction with the battery and the PV array. The PV arrays are commercially available in modules. The PV modules are used to build an array and their I-V characteristics are considered as I-V characteristics of the elementary PV array unit. In our model, we in- troduced a single solar cell as this elementary unit. Consequently, when setting up the simulation with commercial PV arrays, the user must declare N,., as a number of modules in one row or connected in series, and Npay as a number of module rows connected in par- allel. The representation of a cell for varying insolation and temperature is accomplished by the scaling coefficients K, and K;, which are functions of the insolation and temperature specified by the files K_v.map and K_i.map, respectively. These functions are represented at several points in a two-dimensional table. The values of K, and K; for temperature and isolation values not included in the table are determined by interpolation and extrapolation. The I-V characteristics for a number of the commercial PV modules under adjustable environmental parameters such as temperature, beam irradiance, diffuse irradiance, wind speed, site altitude, sun elevation, angle of inci- dence and others along with manufacturer’s parameters of the module can be obtained using Sandia National Laboratories’ I[VTracer Program. These data (for a particular solar module involved in the simulated sys- tem) can be used to generate the files K_v.map and K_imap. In addition, we developed the PV Array-Inverter As- sembly, which converts the direct current (DC) power generated by the PV array to AC power. COMPARISON OF SIMULATED VERSUS RECORDED DATA In this study, we use the data recorded from the Hybrid Power System Test Bed (HPTB) located at the National Renewable Energy Laboratory. The power system in- cluded: the diesel generator, the AOC wind turbine, the dump load (DL), and the village load (VL). The ar- rangement of the test bed is illustrated in Figure 7. The recorded time series sets include the real and reactive power of each device, the line voltage, and the fre- quency. The data were recorded over 10-second inter- vals with a sampling period of 0.001second. Because the wind speed was not measured, we could only com- pare simulated versus recorded data using two simula- tion runs: (a) Generating the simulated data for the system with the standard models of the diesel generator and the village load, and the models of the other system modules modified to conform to the recorded power files. (b) Generating the simulated data for the same system as in (a), but with the full physical model of the Wales Control System Dump Load Dispatch used at the HPTB when the recorded data was acquired. We performed simulations for five sets of data for the same system run under different conditions. However, we discuss the results for only one of these sets and two simulation runs: using a simplified and a full physical model of the dump load. The wind turbine recorded data consisted of the real and reactive power files, Pwr and Qy7, respectively (i.e., the wind speed was not measured). Consequently, we had to modify our wind turbine generator model. Using these data and the reference voltage, we calcu- lated the equivalent resistance and inductance. Then, using the real voltage generated in the simulation, we calculated the equivalent d and q current components contributed by the wind turbine generator at the point of common coupling (PCC). In addition, using these cur- rents and the q and d components of the voltage gener- ated in the simulation, we calculated the wind turbine real and reactive power. This simulation approach is shown in the block diagram in Figure 8. In the first run of the simulation, we included the modi- fied dump load block, for which we calculated the power absorbed using the recorded power file and the voltage generated in the simulated system. For the second simulation run, we used the same data but did not use the dump load power file. Instead, we included the model of the real dump load block with the Wales Control System Dump Load Dispatch used at the HPTB. The dump load at the HPTB has 20 elements of 10 kW each. The Wales Control System Programmable Logic Controller (PLC) dispatches dump load elements in order to maintain a minimum load for the diesel generator. The algorithm uses a modified Proportional+Integral+Derivative (PID) loop to deter- mine the dump load kW required. It subtracts the cur- rent value of the dump load to determine the delta dump load power required, and divides this value by the dump load step size and rounds to determine the number of dump load elements to add or remove.' The recorded village load data consisted of the real and reactive power files P, and Q,, respectively. Therefore, the corresponding power factor pf was first determined and then the standard option of the village load profile data was used. The simulated power, voltage, and frequency traces closely follow those recorded after 2 seconds approxi- mately, which is the start-up time of the diesel genera- tor in the simulation. This is the time needed for the voltage level at the synchronous generator to reach the reference. At time zero, the initial condition of the field current is zero. We only present the representative re- sults. In both cases (see Figures 9 and 10), the dark line is used for simulated traces and the light line is used for recorded traces. The results of comparison run (a) (with the standard diesel generator model) are represented in Figure 9 by the recorded and simulated traces for one of the data sets. We also combined as one variable the real power consumed by the village and the dump load. Considering a slight oscillatory power imbalance in the recorded data (shown in Figure 9) and the smoothing involved in the measuring system, there is a very good agreement (within 2%-5%) between the recorded and simulated traces. The results for the comparison run (b) (with the physi- cal model of the dump load) and the same recorded data are shown in Figure 10. We separated the power traces for the village load and the dump load. We do not again illustrate the power imbalance of the data re- corded. A good agreement (within 2-5%) can also be observed in this case. CONCLUSIONS In this paper, we compared RPM-Sim Simulator data versus the data recorded at the Hybrid Power System Test Bed (HPTB) at the National Wind Technology Center, NREL. We developed the modular simulation system in order to e Study applications and cost-effective performance of wind-diesel hybrid power systems. (Both me- chanical and electrical components are simulated.) e Analyze both static and dynamic performance. ¢ Develop control strategies. ¢ Simulate different wind speed and village load profiles. The system has the following capabilities and/or characteristics: e Modular and multilevel structure is provided by the VisSim visual environment. e System presentation is clear and easy to under- stand. ¢ Customized configuration setup is within a click of the mouse. © Modifications are easy to make. e Effects of system modifications can be immedi- ately examined. This simulation tool can be used for: ¢ Control strategy simulation A proper control strategy must be developed to take full advantage of the available wind energy and to minimize diesel fuel consumption, while maintaining desired system performance. e Inclusion of constraints To implement this control strategy, a control system must be designed subject to the con- straints for a particular application. These in- clude the power generation limitations of the die- sel generator, wind turbine generator, and battery bank/rotary converter assembly, excitation time constants, and dump load parameters. e Checking stability of the power system under time-varying conditions To properly control the system’s voltage and fre- quency, the time-varying power genera- tion/consumption conditions of the system must be considered. The levels of changes that drive the system into instability should be determined. The three factors to be considered are: (1) Wind speed: High winds may drive the diesel engine and cause loss of frequency control and instability. (2) Village load (represented by the real power and the power factor): This includes events such as start-up of induction motor load, start-up of large heating load, loss of load, and sudden change of power factor. (3) Minimum diesel load: To maintain system stability diesel generation must be kept at the required minimum level by a proper control strategy of the dump load. These are just a few of the possible uses of the RPM-SIM. However, because of its flexibility and modularity, it can, if necessary, be easily extended to meet any need that might emerge in the design of an autonomous power system with renewable energy sources. Thus, if some specialized modules for a par- ticular simulation are needed, the user can easily in- clude them. The RPM-SIM simulator can be obtained by contacting NREL. ACKNOWLEDGEMENTS We wish to thank Dr. Susan Childs of Atlantic Orient Corporation, Dr. William Q. Jeffries of Yankee Envi- ronmental Systems, Inc., and Dr. Vahan Gevorgian of NREL for their valuable comments and suggestions. We also wish to extend our thanks to the anonymous reviewers who helped in improvement of the presenta- tion of our work. The recorded data were provided by Ian Baring-Gould of NREL. This work was supported by the U.S. Department of Energy. REFERENCES 1 J. T. Biatasiewicz, E. Muljadi, R.G. Nix and S. Drouilhet, Renewable Energy Power System Modular Simulator RPM-SIM User’s Guide, NREL Technical Report No. NREL/TP-500-25951, Octo- ber 1999. 2 J.T. Biatasiewicz, E. Muljadi, S. Drouilhett and G. Nix, Hybrid Power System with Diesel and Wind Turbine Generation, Proceedings of the 1998 American Control Conference, vol. 3, pp. 1705- 1709, Philadelphia, PA, 1998. 3 J. T. Biatasiewicz, E. Muljadi, S. Drouilhet and G. Nix, Modular Simulation of a Hybrid Power System with Diesel and Wind Turbine Generation, Proc. Windpower ’98, Bakersfield, CA, 1998. 4 E. Muljadi, Nix, G. and Biatasiewicz, J. T., Analysis of the Dynamics of a Wind Turbine Water-Pumping System, Proc. IEEE PES 2000 Summer Meeting, Seattle, WA, 2000. 5 W. Q. Jeffries, Analysis and Modeling of Wind/Diesel Systems Without Storage, Ph.D. The- sis, Department of Mechanical Engineering, Uni- versity of Massachusetts, 1994. 6 A. J. Tsitsovits and Freris, L.L., Dynamics of an Isolated Power System Supplied from Diesel and Wind, Proc. IEEE, 130, Part A, No. 9, pp. 587-595, 1983. 7 10 11 12 13 J. T. G. Pierik, and De Bonte, Quasi Steady State Simulation of Autonomous Wind Diesel Systems (Status Report), Report No. ECN-85-091, Nether- lands Energy Research Foundation, Petten, May 1985. M. P. Papadopoulos, et al., Penetration of Wind Turbines in Islands with Diesel Power Stations, Proc. EWEC 1988, pp. 512-517, 1988. K. Uhlen and O. Skarstein, A Short Term Dynamic Simulation Model for Wind/Diesel Systems, Proc. 10 BWEA Conference, pp. 239-242, 1988. J. F. Manwell, et al., Developments in Experimental Simulation of Wind/Diesel Systems, Proc. EWEC 1989, pp.759-763, 1989. M. P. Papadopoulos, et al., Simulation of the Paral- lel Operation of Diesel, Wind Turbines and Photo- voltaic Systems, Proc. EWEC 1991, pp. 495-499, 1991. P. Lundsager, et al., The JODYMOD Dynamic Wind Diesel Simulation Program Part I: Description of the Model and its Validation, Proc. BWEA/RAL Workshop on Wind Energy Penetration into Weak Electricity Networks, Abingdon, UK, pp.133-149, 1993. H. Binder, The JODYMOD Dynamic Wind Diesel Simulation Program Part II: Presentation of the Pro- gram: Demonstration and Examples, Proc. BWEA/RAL Workshop on Wind Energy Penetra- tion into Weak Electricity Networks, Abingdon, UK, pp. 150-160, 1993. 14 W. Q. Jeffries, McGowan, J.G. and J.F. Manwell, Development of a Dynamic Model for No Storage Wind/Diesel Systems, Wind Engineering, Vol. 20, No.1, pp. 27-38, 1996. 15 P. D. Ladakakos, et al., Development of a Simula- tion Model for Investigating the Dynamic Operation of Autonomous Wind-Diesel Systems, Proc. EWEC, 1997. 16 R. H. Park, Two-Reaction Theory of Synchronous Machines-Generalized Method of Analysis-Part I, AIEE Trans., Vol. 48, pp. 716-727, 1929. 17 H. C. Stanley, An Analysis of Induction Motor, AIEE Trans., Vol. 57 (Supplement), pp. 751-755, 1938. 18 P. C. Krause, and Thomas C. H., Simulation of Symmetrical Induction Machinery, IEEE Trans Power Apparatus and Systems, Vol. 84, pp. 1038- 1053. 19 C.-M., Ong, Dynamic Simulation of Electric Ma- chinery Using Matlab/Simulink, Prentice Hall PTR, 1998. INVERTER Figure 1. Principal modules of the RPM-SIM included in a single-line diagram of a hybrid power system. % fuel frequency set point field current 60 > fb 4+ ENGINE wind speed +— > _ TURBINE wind Vi Start_up time in the induction SPEED CONTROL Dp voltage ~ 266 set point L—) Vs _ref voltage > SYNCHRONOUS GENERATOR VOLTAGE REGULATOR AC terminal AC terminal voltage + Figure 2. Block diagram of the DG module. DIESEL ENGINE DIESEL DYNAMICS GOVERNOR Ge FREQUENCY CONTROL torque -—>| GENERATOR iy C field current VOLTAGE CONTROL Figure 3. The principle of voltage and frequency control. IG_kW kw WIND torque +—> ROTOR GEARBOX torque INDUCTION ——> p eee ny generator block to be declared by the user. CAPACITORS PFI_kVAR IG_kKVAR +, T_wind GENERATOR a) kVAR POWER FACTOR CORRECTION Figure 4. Simulation diagram of the wind turbine generator. C:AVISSIM\W_D_SIM\power.dat > P_v2 Dia ra VL: REAL POWER kW 0.001 =e 1 >40h C:\VISSIM\W_D_SIM\pf.dat > pa | a power.dat 0 . 2 : i 0.0000000e+000 2.0000000e+004 Di eee ee erie 0.9999999e+000 2.0000000e+004 TEES) 1,0000000e+000 5.0000000e+004 TE 2.4999999e+000 5.0000000e+004 2.5000000e+000 2.0000000e+004 2.0000000e+001 2.0000000e+004 le Pages a pf.dat 0.0000000e+000 0.5000000e+000 o i 5 a 5 0.9999999e+000 0.5000000e+000 nite GO) 1,0000000e+000 1.0000000e+000 2.4999999e+000 1.0000000e+000 2.5000000e+000 0.7500000e+000 2.0000000e+001 0.7500000e+000 Figure 5. An example of a village load profile represented graphically and by ASCII data files power.dat and pf.dat. battery voltage torque kW ———————_> nd pee _— De SYNCHRONOUS | MOTOR asides kKVAR field field AC terminal current current voltage Vs DC MACHINE VOLTAGE FIELD CONTROLLER REGULATOR AC terminal voltage Vs Figure 6. Principal functional blocks of the rotary converter. DE Loe control Figure 7. Connection diagram for the system tested at the HPTB. kw »P_wT }—————_ AC /-—>— WIND dt kVAR Figure 8. Symbolic representation of the wind turbine generator model with real and reactive power file as the input. AOC kW SIMULATED & RECORDED VL + DL SIMULATED & RECORDED Paes isk On Betas se t9 Oe) 7 ee Bh ED. Time (sec) DGkW SIMULATED & RECORDED Time (sec) ol kW BALANCE SIMULATED & RECORDED 10 Time (sec) > 0 > 2 > -4 -6 Wi ORI AD HAE NA PS SeeG TE BD Time (sec) Time (sec) # eee SIMULATED & RECORDED 00 VOLT AGE SIMULATED & RECORDED : + 61.0 490 + 60.5 + 60.0 480 59.5 4 470 59.0 58.5 460 y 0 2 4 6 8 10 8% 0 2 4 6 8 Figure 9. Simulation results for the run (a): the dark line is used for traces simulated and the light line is used for traces recorded. 10 AOC kW SIMULATED & RECORDED 5 DL SIMULATED & RECORDED 480 4-20 70 -40 454 > 50 -60 40 > > 30 30 20 100 WO Tee ae a sees Te OTe || WO CS ete Orn ners sk LO. Time (sec) Time (sec) DGkW SIMULATED & RECORDED i VL SIMULATED & RECORDED > Re FF iok -15 > -20 - lo 4 4 He 8 2 4 6 8 10 Time (sec) Time (sec) FREQUENCY SIMULATED & RECORDED VOLTAGE SIMULATED & RECORDED #615 61.0 4 60.5 Figure 10. Simulation results for the run (b): the dark line is used for traces simulated and the light line is used for traces recorded. 11 REPORT DOCUMENTATION PAGE a as Public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information, including Suggestions for seducing this burden, to Washington Headquarters Services, Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, Arlington, VA 22202-4302, and to the Office of Management and Budget, Paperwork Reduction Project (0704-0188), Washington, DC 20503. 1, AGENCY USE ONLY (Leave blank) | 2. REPORT DATE 3. REPORT TYPE AND DATES COVERED January 2001 Conference Paper ” TITLE AND SUBTITLE | RPM-SIM Simulator: A Comparison of Simulated Versus Recorded Data See Ne . AUTHOR(S) WER13010 J. Biatasiewicz, E. Muljadi, R. G. Nix, and S. Drouilhet 7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGANIZATION National Renewable Energy Laboratory REPORT NUMBER 1617 Cole Blvd. NREL/CP-500-29174 Golden, CO 80401-3393 9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSORING/MONITORING AGENCY REPORT NUMBER 11. SUPPLEMENTARY NOTES 12a. DISTRIBUTION/AVAILABILITY STATEMENT 12b. DISTRIBUTION CODE National Technical Information Service U.S. Department of Commerce 5285 Port Royal Road Springfield, VA 22161 13. ABSTRACT (Maximum 200 words) This paper compares simulated versus recorded data for the RPM-SIM simulator, developed at the National Renewable Energy Laboratory's National Wind Technology Center. The simulator was used to study the system dynamics of a wind/diesel hybrid power system. We also provide information on newly developed simulator modules that will be released. The simulator performed extremely well, demonstrating flexibility in making modifications and including specialized modules required for problem solving. We also outline several possible applications for this tool. . NUMBER OF PAGES 14. SUBJECT TERMS wind energy, simulator, wind/diesel hybrid systesm . PRICE CODE 17. SECURITY CLASSIFICATION 18. SECURITY CLASSIFICATION 19. SECURITY CLASSIFICATION OF REPORT OF THIS PAGE OF ABSTRACT Unclassified Unclassified Unclassified . LIMITATION OF ABSTRACT UL NSN 7540-01-280-5500 Standard Form 298 (Rev. 2-89) Prescribed by ANSI Std. 239-18 298-102 Nes > aT Sy NY) September 1999 * NREL/CP-500-27114 Wind-Diesel Hybrid Systems for Russia’s Northern Territories V. Gevorgian and K. Touryan National Renewable Energy Laboratory P. Bezrukikh Ministry of Fuel and Energy of Russian Federation P. Bezrukikh Jr. and V. Karghiev Intersolarcenter Presented at Windpower ‘99 Burlington, Vermont June 20-23, 1999 f° NREL a National ae Energy Laboratory 1617 Cole Boulevard Golden, Colorado 80401-3393 NREL is a U.S. Department of Energy Laboratory Operated by Midwest Research Institute e Battelle e Bechtel Contract No. DE-AC36-98-GO10337 NOTICE This repomans prepared as an account of work sponsored by an agency of the United States government. Neither the United States government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States government or any agency thereof. Available to DOE and DOE contractors from: Office of Scientific and Technical Information (OST1) P.O. Box 62 Oak Ridge, TN 37831 Prices available by calling 423-576-8401 Available to the public from: National Technical Information Service (NTIS) U.S. Department of Commerce 5285 Port Royal Road Springfield, VA 22161 703-605-6000 or 800-553-6847 or DOE Information Bridge http://www.doe.gov/bridge/home.htm! Pha ae Printed on paper containing at least 50% wastepaper, including 20% postconsumer waste Wind-Diesel Hybrid Systems for Russia's Northern Territories Vahan Gevorgian, Kenell Touryan National Renewable Energy Laboratory 1617 Cole Blvd. Golden, CO 80401 USA Pavel Bezrukikh Ministry of Fuel and Energy of Russian Federation 7 Kitaigorodsky pr. Moscow, 103074, Russia Pavel Bezrukikh Jr., Vladimir Karghiev Intersolarcenter 2, 1* Veshnyakovski proezd Moscow, Russia, 109456 ABSTRACT This paper will summarize the DOE/ Russian Ministry of Fuel and Energy (MF&E ) activities in Russia’s Northern Territories in the field of hybrid wind-diesel power systems over the last three years (1997— 1999). The National Renewable Energy Laboratory (NREL) supplied technical assistance to the project, including resource assessment, system design, site identification, training and system monitoring. As a result, several wind-diesel systems have been installed and are operating in the Arkhangelsk/Murmansk regions and in Chukotka. NREL designed and provided sets of data acquisition equipment to monitor several of the first pilot wind-diesel systems. NREL's computer simulation models are being used for performance data analysis and optimizing of future system configurations. INTRODUCTION The Russian Ministry of Fuel and Energy (MF&E) together with U.S. Department of Energy (DOE) and U.S. Agency for International Development (USAID), have joined efforts to bring efficient, competitive, off-grid, renewable energy based power solutions to the Russia's Northern Territories. Currently, most of Russia's Northern inhabitants, approximately 10-15 million people, live without access to the central electricity supply grid. Diesel/gasoline power stations serve as the main source of energy supply on the Arctic coast of Russia. More than 60,000 people are involved in the fuel supply network for Russia's Northern Territories. Thus, substantial funds are spent each year to deliver fuel for the diesel power stations. In addition, the short Arctic summers, insufficient funding, and institutional problems threaten the stability of fuel transportation and supply. The objectives of the DOE/MF&E project are to demonstrate energy and infrastructure cost savings in off-grid wind-diesel applications; to facilitate commercialization of wind technologies; and to establish a technically and economically viable model from which an alternative loan program can be designed by the Russian government. BACKGROUND The current Russian-American cooperation in renewable energy was established in September 1993, during the Joint Committee on Economical and Technical Cooperation (JCTC) meeting between Prime Minister Chernomyrdin and Vice President Gore. The agreement was reached to join efforts in examining options for Russia’s energy future (thermal/nuclear power, power transmission, energy efficiency, renewable energy and financing). It was followed by a Memorandum of Cooperation signed in October 1993 by Russian Minister of Fuel and Energy Yuri Shafrannik and DOE Secretary of Energy Hazel O’Leary. The Memorandum sets the framework for joint activities in energy efficiency and renewable energy to provide more economical, cleaner energy solutions that yield employment and economic benefits for both countries. As a result of the above activities, the Joint Coordinating Committee (JCC) in energy efficiency and renewable energy was established during the meeting between Russian deputy minister Bushuev and DOE Assistant Secretary Christine Ervin in October 1994. NORTHERN RUSSIA’S WIND RESOURCE The maps shown in Figure 1 identify the areas of renewable energy resource availability for Russia [1]. The centralized grid covers only the European part of the country and Southern Siberia. The rest of the country has access to local grids, or has no grids at all. The wind resource is available in the North and Far East. The highlighted area on the “Wind Energy” map shows the areas with annual average wind speed of 6 m/s and higher. The wind appears to be the only renewable energy resource for the North. The other renewable energy resources (hydro, biomass, and solar) are not available in these regions. Traditional Energy Sources: RUSSIA Oil, Natural Gas, Coal — 73% Hydro Power — 18 Renewable Energy Resources (ifthe deur) Wind Energy i bth > r fo Figure 1: Renewable Energy Resources of Russia Russia’s Northern territories include the European North (Murmansk and Arkhangelsk regions) and the Asian North (Yamal, Taymir, Yakutia and Chukotka regions). These territories, stretched along Russia’s Arctic coast for about 7000 km, have enormous wind energy potential. The other areas with large wind potential are located in the Russian Far East (Kamchatka, Magadan, Khabarovsk and Vladivostok regions). The Arctic coast and the Far East of Russia appear to be promising for wind energy development with average annual wind speeds of 6-7 m/s or higher. The fluctuation between the winter maximum and the summer minimum rates is about 30%—40% for the coastal areas. The intensity of the wind resource is less for lower inland where biomass is considered the best alternative energy source. There are two more prospective wind energy development regions in Southern Russia—the coastal areas of the Azov and Caspian seas. These regions are beyond the scope of work for the current project. However, some studies are being conducted there by the Russian Ministry of Energy and Fuel in order to include them in future activities. Through the U.S. Department of Energy's (DOE) Initiative for Proliferation Prevention (IPP) program for Newly Independent States (NIS), a program has been established to help facilitate and accelerate the large-scale use of wind energy technologies in areas of Russia that are of interest to the U.S. wind energy industry. A key component of this program is a wind resource assessment project that will reveal the level of wind resource present in specific areas of Russia with the primary focus on the Northern Russian Territories. The major product of this assessment project will be the development of detailed wind resource maps which will use advanced Geographic Information Systems (GIS) technology and sophisticated mapping techniques. New computing techniques will use existing data sets obtained from sources in the United States and Russia. The resultant maps and accompanying summaries of the wind resource characteristics will facilitate the rapid identification of candidate locations for additional wind monitoring and consideration of wind energy development. U.S./RUSSIAN JOINT COLLABORATION FOR HYBRID WIND-DIESEL SYSTEMS The first step of the joint collaboration for renewable energy under the JCC framework was establishing two research centers by the Russian MF&E. The first one was the Intersolarcenter (ISC) established at the Russian Rural Electrification Institute (VIESKh) as a sister organization to NREL for joint R&D activities. The second—Federal Center of Small Scale and Unconventional Energy (FC) was established under the Russian MF&E for renewable energy systems certification and deployment, renewable energy policy, and regulation development. Researchers and engineers from both centers work in close contact with NREL. In 1997, as a part of the agreement, USAID purchased and shipped to Russia 40 wind turbines with batteries and inverters to be installed in Northern Russia. The wind turbines were manufactured by Bergey Windpower Company (BWC) and included thirty 7.5-kW Excel and ten 1.5-kW generators as well as Trojan batteries and solid-state Trace Engineering power inverters. This amounted to a nominal total of 315 kW of wind generating capacity. The list of equipment shipped to Russia is given in Table 1. The Russian MF&E selected twenty-one sites in the Murmansk, Arkhangelsk and Chukotka regions for the installation of pilot wind-diesel hybrid systems. The candidate project sites were selected on the basis of (1) the infrastructure necessary to maintain the systems, (2) the wind resource, (3) the fuel price and availability at the site, and (4) a variety of applications to serve as pilot projects. System design and training in installation and maintenance were provided by a U.S. team from NREL and BWC. The Russian team was responsible for systems installation, local personnel training, monitoring, and performance analysis of the installed systems. The Federal Center was assigned responsibility for systems installation and operation, and the Intersolarcenter role is mainly in systems monitoring and analytical studies. NREL provided four complete sets of monitoring equipment for the projects. Table 1: List of Equipment Provided by the U.S. Government Description Cost per unit, $ uantity per unit Total delivered Total cost, $ BWC Excel WTG 18,770 1 30 563,100 Tower 8,370 1 30 251,100 Trojan L-16 batteries 5,388 24 720 161,640 Trace SW 4548E inverter 7,395 2 60 221,850 Wind turbine accessories 5,874 30 176,220 Sub total 1,373910 BWC 1500 WTG 5,210 1 10 52,100 Tower 3,050 1 10 30,500 Trojan T-105 batteries 715 8 80 62,000 Trace DR 2424E inverter 1,225 1 10 12,250 Wind turbine accessories 3,813 10 38,130 Sub total 194,980 Grand Total 40 1,568890 HYBRID SYTEMS ANALYSIS The local grids serve only a small portion of remote sites in the Northern Territories. A large number of fishing villages, meteorological stations, lighthouses, and frontier outposts in the Northern territories are supplied with electrical power from small 4-8 kW diesel generators. Diesel fuel prices range from $0.40 to $1.40/liter, and the demand for electricity often exceeds the supply of available fuel. Table 2 shows 12 sites in the Murmansk and Arkhangelsk regions that the Russian Ministry of Fuel and Energy selected for pilot project installation. Table 2: Potential Hybrid Power System Sites in Northern Russia No | Site Long. | Lat. Diesel Peak Load, | Average Fuel cost, Generator, | kW wind speed | $ per liter kW. (m/s) Murmansk region 1__| Tzip-Navolok (met. station) 33°10’ _| 69°70" | 8 5.5 TA 0.47 2__| Kharlov Island (met. station) 37°40’_| 68°80’ | 8 5.5 9.2 0.47 3__| Sviatoy Nos (met. station) 39°45’ _| 68°08" | 8 5.5 8.3 0.47 4__| Vaida Guba (met.station) 33°10’ _| 69°70" | 8 5.5 6.9 0.47 5__| Svyatonosski Lighthouse 39°45’ _| 68°08’ | 30 20 8.3 0.4 6 Tuvagubski Lighthouse 33°10’ _| 69°00’ | 8 5 4.9 0.47 Arghangelsk region 7__| Morzovetz (met. station) 42°50’ _| 66°70’ | 8 5.5 14 0.47 8 | Kanin Nos (met. station) 43°30’ _| 67°80’ | 8 55 8.1 0.47 9__| Intzy village 40°70" _| 65°90’ | 16 16 5:2 0.40 10_| Krasnoe village 53°60’ _| 67°70’ | 8 8 3.9 0.40 11_| Miada village 42°00’ _| 66°30’ | 100,60,20 | 112 6.3 0.40 12_| Megra village 41°6’ 66°10’ | 60, 20 58 6.3 0.40 The fuel cost for the selected sites is within a range of $0.40-$0.50 per liter, which includes transportation costs. Fuel is much cheaper in the Murmansk and Arkhangelsk regions compared to the rest of Russia’s Northern territories for a number of reasons. All of these sites are geographically closer to the major industrial centers of Murmansk and Arkhangelsk. They are also located in the areas were the Arctic sea never freezes because of warm Atlantic streams that reach the tip of the Cola peninsula. NREL’s task was to select the optimum configuration for each site based on the site information provided by the Russians. The limited amount of equipment was considered as well. The analysis was done using the Hybrid2 simulation code, developed at NREL [2]. After multiple runs of Hybrid2, configuration recommendations were developed, similar to ones shown in Table 3. The optimum number of wind turbines and other equipment for each site that gives the maximum economy of diesel fuel was recommended. Table 2 shows the simulation results for 12 sites in the Arkhangelsk and Murmansk regions. The estimated percentage of fuel saved varies from site to site, depending on the local wind resource, and the type of equipment supplied. Table 3: Recommended Configurations No | Site Number | Number Number | Number Number Number | Estimated of BWC | of BWC | of of of SW | of | DR | Fuel 1500 Excel T-105 L-16 inverters inverters | saving, % wind wind batteries | batteries turbines _| turbines Murmansk Region 1 | Tzip-Navolok 1 16 1 84.2 2_| Kharlovy Island 3 24 3) 78.2 3 | Sviatoy Nos 3 24 3 72.6 4 | Vaida Guba 1 24 1 81.0 5 | Sviatonoski 3 72 6 84.0 lighthouse 6 | Tuvagubski 1 32 1 67.9 Lighthouse Arkhangelsk region 1 { 7 | Mortzovetz 1 24 1 85.0 8 | Kanin Nos 3 24 3 74.0 9 [Intzy 3 ma | 3 | 79.5 10 | Krasnoe 4 96 3 70.5 11 _| Maida 4 96 6 26.0 12 | Megra 4 96 3 51.0 PERFORMANCE MONITORING Another major contribution that NREL is making in this project is to help the Russians monitor the hybrid systems. NREL has provided four data acquisition systems (DAS) that allow monitoring of all basic performance parameters of the hybrid system. The DAS are based on the equipment manufactured by Campbell Scientific and Ohio Semitronics, both U.S. companies. The monitoring equipment includes dataloggers, wind speed and direction sensors, ambient and battery temperature sensors, and various AC and DC current/voltage/power sensors. The purposes for using monitoring systems are Determine component and system efficiencies Verify proper system functioning Provide system trouble shooting Detect and analyze significant village load changes Calculate actual cost of utilized energy Validate models Provide information to improve systems to be installed in later stages of project implementation. The Russian-developed monitoring system will be tested at the demonstration wind hybrid site in Istra, near Moscow. This site is being established with NREL assistance as a training and demo center for Russian technician and engineers who will be involved in implementation of hybrid technologies in Russia. PILOT PROJECTS According to the Federal Center, a total of twelve wind turbines have already been installed in the Murmansk and Arkhangelsk regions and Chukotka as of July 1999. Two of these turbines are used in a hybrid system installed in the village of Krasnoe located in the Arkhanglesk region. The map of the region is shown in Figure 2 [3]. This site has been visited several times by NREL staff. The system was installed in 1997 by BWC team. It is the only system so far that has been equipped with a DAS. Figure 2: Map of Arkhangelsk region, Russia The village of Krasnoe is typical of communities in the Northern Territories. It consists of about 30 households and an 8-kW Russian-made diesel generator was the only power source for the village. Krasnoe is an agricultural and fishing community with a very low average household income. The diesel generator was used to provide electrical power to the village for about four hours a day (6:00—10:00 P..). The rest of the time the village did not have electric power. In recent years, due to the deepening economic crisis in Russia, villagers could afford even less diesel fuel. Thus, the power supply for the village became extremely irregular. The hybrid patent installed in Krasnoe is supposed to provide power to the village on a regular basis. The diagram of the hybrid system installed in Krasnoe is shown in Figure 3. The system consists of two 7.5-kW Bergey Excel wind turbines on an 18-m tower, connected to a 48 VDC, 200 Ah total capacity battery bank made up of Trojan L-16 batteries. Three 4.5-kW single-phase Trace Engineering SW4845 inverters (220 V, 50 Hz) are connected in parallel to the battery bank on the DC side. On the AC side, the inverters are combined in a way so that they provide three-phase power to the village loads. The system was set up as a switched configuration. Thus no parallel operation of inverters and diesel generator is possible. This was done because the diesel generator is too old and has a poor voltage/frequency control system. The inverters often refuse to recognize the diesel generator as a reliable voltage source and will stop operation. In addition, the diesel generator does not have an automatic start-up option, which is typical for other sites in Northern Russia. This does not allow full use of the inverters’ capability to turn the diesel generator on and off automatically depending on battery voltage. As a result, system performance will suffer because of the human factor: it depends on operator judgment when to turn the diesel on or off. This will decrease the overall operational efficiency of the system. The hybrid system in Krasnoe is shown in Figure 4. The site is located on Nikolski Island, about 15 miles north of Arkhangelsk. The prevailing wind direction is from Northwest and has an average annual wind speed of about 4 m/s. The Russians plan to install one more 7.5-kW wind turbine at the site to increase system’s energy production. A photo of the control room with the Trace Engineering inverters and other BWC equipment is shown in Figure 5. Figure 5: Control room The hybrid system in Krasnoe is equipped with one of the DAS provided by NREL. The DAS components are shown in Figure 6 and include Wind speed, direction and ambient temperature DC output currents for each individual wind turbine Battery bank voltage, current and temperature DC input currents for each individual inverter Phase AC power and power factor for each phase. 10 kW WTG EHO ‘Transformer LOAD Bank 4avpc Te ves-10 mn @ bc Center 3k Transformer Wing Speed/Directon Measuring Set aaah oes ee ee & © Figure 6: Diagram of the hybrid system in Krasnoe List of DAS components is shown in Table 4. Three similar sets were shipped to Russia to be installed on other sites. Total cost of the equipment in Table 4 is about $9,000. Table 4: List of DAS components (by manufacturer No. Description Qty Campbell Scientific, Inc 1 | Datalogger CR10X-2M-XT with accessories 1 2_| AM416 — 16 -XT channel. Multiplexer 1 3 | PS12LA 12 V regulated power supply 1 4 | 108-L Temperature Probe 2 5_| 2:1 voltage divider modules 8 Qhio Semitronics, Inc. 6 | 12973 Current transformer (100:5) 3 7 | PC20-002E-22 Watt/VAR/PF transducer (4-20 mA output) 3 8 | CTL-200T Current transducers 5 9 | CTL-400T Current transducer 1 10 | CTA101X5-48 Signal conditioner (0-5 V output) 6 11_| VT7-004X5-48 Voltage transducer (0-5 V output) 1 NRG systems |__12 | NRG #40 Maximum Anemometer 2 13_| NRG #200P Wind Direction Vane, 10K 1 The other systems were installed in the village Bolshie Kozli (Arkangelsk region), Sosonovka (Murmansk region), Mirni (Cheliabinsk region), and in Chukotka. NREL is studying and evaluating the information of these systems provided by the Russians that is to be presented in future publications. Figure 7 shows the wind speed and load data measured at the site in June of 1999 by NREL staff during their last visit to Russia. The measurements were taken over about 4.5 hours of operation. The wind speed was measured on both wind turbines and the AC power to the load was measured in all three phases. The first thing that can be noticed from the power profile is the extreme imbalance between phases. This kind of load profile is typical for small villages in Northern Territories. 10 WIND SPEED (mm/s) AC POWER (kW) —— AC Power, Phase A seeees AC Power, Phase B ~=~= AC Power, Phase C Figure 7: Wind speed and load data CONCLUSIONS AND LESSONS LEARNED The following main conclusions can be drawn after installation of 12 wind turbines and based on accumulated project experience: © Russian diesels are old and often have not been properly maintained. They usually don’t have an operational automatic start- up circuitry. So, automatic operation with inverters is not possible. ¢ No problems have been recorded so far with the Bergey wind turbines. All 12 of them have survived through two consecutive winters with no breakdown or malfunction. © Qualified technicians and engineers are available almost everywhere in Russia, resulting in proper operational conditions and maintenance of installed hybrid systems. ¢ Direct contact with local authorities helps to speed up the project implementation schedules significantly. Better fuel usage records are necessary to conduct economic analysis. Valid records or data on fuel consumption do not exist for most of the sites. The option of using some fuel metering equipment as part of the DAS is being considered. FUTURE PLANS The following activities were included in the next five-year hybrid systems cooperation plan between DOE and the Russian Ministry of Fuel and Energy: Complete installation of the remaining systems in Northern territories Complete testing of the Russian-made monitoring equipment prior to installation at the remaining sites NREL will continue providing technical assistance The Russian Ministry of Fuel and Energy will attempt to secure a $300 million loan to retrofit 900 sites in Northern Russia using similar systems as the pilot sites above. These new sites may also require the installation of controls on existing diesel generators. Complete development of wind resource atlas for Northern Russia which is currently being developed by the Intersolarcenter under an NREL subcontract. ACKNOWLEDGEMENTS The authors of this paper would like to thank all those who have been instrumental in these pilot projects: from Russia, Sergey Mikhaylov of FC, and Dr. Askar Pinov of Intersolarcenter. From the hybrid power systems industry: Mike Bergey, Peter Hubner and Ken Craig. We would also like to thank all of the members of the Village Power Team at the National Renewable Energy Laboratory who have assisted in this project, especially Dennis Barley. This project is funded by the U.S. AID and DOE. REFERENCES ile Sokolsky A.K.; Kharitonov V.L.; Mikhaylov S.A. “Windpower Installations in the Arkhangelsk Region.” Renewable Energy Bulletin, No. 2, 1998. pp. 30-33. Baring-Gould, E.1.; Green, H.J.; van Dijk, V.A.P.; Manwell, J.F. “Hybrid2—The Hybrid Power System Simulation Model.” Windpower ’96 Proceedings, Denver, CO, June 23-27, 1999. Washington, DC: American Wind Energy Association. pp. 497-506. Arkhanglesk region map. Russian Federal Mapping Service. Geographical maps. Moscow, 1995 10 August 1999 * NREL/CP-500-26927 Feasibility of Hybrid Retrofits to Off-Grid Diesel Power Plants in the Philippines C.D. Barley and L.T. Flowers National Renewable Energy Laboratory P.J. Benavidez, R.L. Abergas, and R.B. Barruela Strategic Power Utilities Group, National Power Corporation Quezon City (Manila), Philippines Nee Sv) > a Y a) es eae panties! aga es pee Setar Reies Presented at Windpower ‘99 Burlington, Vermont June 20-23, 1999 GAs — National eeaiiinabee Energy Laboratory 1617 Cole Boulevard Golden, Colorado 80401-3393 NREL is a U.S. Department of Energy Laboratory Operated by Midwest Research Institute ¢ Battelle e Bechtel Contract No. DE-AC36-99-GO10337 NOTICE The submitted manuscript has been offered by an employee of the Midwest Research Institute (MRI), a contractor of the US Government under Contract No. DE-AC36-99GO10337. Accordingly, the US Government and MRI retain a nonexclusive royalty-free license to publish or reproduce the published form of this contribution, or allow others to do so, for US Government purposes. This report was prepared as an account of work sponsored by an agency of the United States government. Neither the United States government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States government or any agency thereof. Available electronically at http://www.osti.gov/bridge Available for a processing fee to U.S. Department of Energy and its contractors, in paper, from: U.S. Department of Energy Office of Scientific and Technical Information P.O. Box 62 Oak Ridge, TN 37831-0062 phone: 865.576.8401 fax: 865.576.5728 email: reports@adonis.osti.gov Available for sale to the public, in paper, from: U.S. Department of Commerce National Technical Information Service 5285 Port Royal Road Springfield, VA 22161 phone: 800.553.6847 fax: 703.605.6900 email: orders@ntis.fedworld.gov online ordering: http://www.ntis.gov/ordering.htm fae Printed on paper containing at least 50% wastepaper, including 20% postconsumer waste FEASIBILITY OF HYBRID RETROFITS TO OFF-GRID DIESEL POWER PLANTS IN THE PHILIPPINES* C. Dennis Barley? and Lawrence T. Flowers National Renewable Energy Laboratory Golden, CO, U.S.A. Pio J. Benavidez, Rafael L. Abergas and Rene B. Barruela Strategic Power Utilities Group National Power Corporation Quezon City (Manila), Philippines Introduction The Strategic Power Utilities Group (SPUG) of the National Power Corporation (NPC) in the Philippines owns and operates about 100 power plants, mostly fueled by diesel, ranging in energy production from about 15 kilowatt-hours (kWh)/day to 106,000 kWh/day. Reducing the consumption of diesel fuel in these plants, along with the associated financial losses, is a priority for SPUG. The purpose of this study is to estimate the potential fuel and cost savings that might be achieved by retrofitting hybrid power systems to these existing diesel plants. As used in this report, the term "hybrid system" refers to any combination of wind turbine generators (WTGs), photovoltaic (PV) modules, lead-acid batteries, and an AC/DC power converter (either an electronic inverter or a rotary converter), in addition to the existing diesel gensets. The resources available for this study did not permit a detailed design analysis for each of the plants. Instead, the following five-step process was used: 1. Tabulate some important characteristics of all the plants. 2. Group the plants into categories (six classes) with similar characteristics. 3. For each class of system, identify one plant that is representative of the class. 4. For each representative plant, perform a moderately detailed prefeasibility analysis of design options. 5. Summarize and interpret the results. The analysis of each representative plant involved the use of time-series computer simulation models to estimate the fuel usage, maintenance expenses, and cash flow resulting from various designs, and to search the domain of possible designs for the one leading to the lowest life-cycle cost. Cost items that would be unaffected by the retrofit, such as operator salaries and the capital cost of existing equipment, were not included in the analysis. Thus, the results are reported as levelized cost of energy (COE) savings: the difference between the cost of the existing diesel-only system and that of an optimized hybrid system, _ hh Ati AOA ts — a — een expressed in_ of U.S. dollars per kWh (US$/kWh) of energy p i This analysis is one phase of a study entitled "Analysis of Renewable Energy Retrofit Options to Existing Diesel Mini-Grids," funded by the Asia-Pacific Economic Cooperation (APEC) and the U.S. Department of Energy (DOE), and performed jointly by NPC, the U.S. National Renewable Energy Laboratory (NREL), and Sustainable Energy Solutions in New York, New York (Morris et al. 1998). A more detailed version of this paper is included in that report. ' Condensed from a longer version in eons et al. (1998). ? Currently an independent consultant in Boulder, Colorado. 1 Features of Existing Systems In order to get an overview of the existing plants, certain features of all the plants were tabulated. These features include: ¢ Hours per day of electrical service provided by the plant * Average energy production in kWh/day * Delivered fuel price at the plant, in Philippine pesos (PHP) per liter (1997 average prices) « Use of a dump load to preclude the operation of the diesel generators at a loading of less than 40% of rated power. Loads. The distribution of plant operating hours is shown in Figure 1. 43% of the plants provide 6-hour service, 36% provide 24-hour service, and 21% provide a variety of other service periods. Based on this Hours Histogram distribution, 6-hour and 24-hour service are identified as the most significant Mommnen a groupings in terms of number of 6 hr | plants. ° | 35 To gain further perspective on the 2) loads, the daily energy production of the various plants was correlated to the hours of service. All of the 24-hour plants are grouped as Classes 1 through 4, depending on the daily energy production. Most of the 6-hour plants produce less than 500 ss Sar ca Pat a PI S kWh/day; these are grouped as Se ee Classes 5 and 6 according to whether or not a dump load is used. In the Class 6 systems, a dump load is used Figure 1. Distribution of plant operating hours to prevent diesels from operating at less than 40% of their rated power; such low-load operation increases maintenance requirements. In the Class 5 systems, diesels may be permitted to operate at less than 40% of rated power; the increased maintenance costs resulting from this practice are not known and therefore are not included in this study. A summary of the six load classes identified in this manner is shown in Table 1. For each class, a representative example of a plant belonging to that class is also identified. These are the plants that were analyzed. The daily load profiles at these six representative plants are shown in Figures 2 and 3. These profiles are assumed to be constant throughout the year. (Significant seasonal variations in the load are not expected in this tropical climate.) Of the 99 plants owned by SPUG, 69 are encompassed by these six classes. The analysis of the six classes leads to some general conclusions that shed light on the other plants as well. 25 20 15 Number of Plants 10 Plant Operating Hours Table 1. Summary of Load Classes 10,000-20,000 Number Example of Plants Plant 10 400-800 Romblon 5,000-10,000 (No dump load) 200-400 6 (Dump load) Typical Load Profiles 1-Tablas 3-Cuyo 2000 a 8 1000 Plant Load, kW 500 16 20 24 Hour Figure 2. Load profiles for Class 1, 2, 3 Fuel Prices. The distribution of the delivered fuel prices at all the SPUG plants is shown in Figure 4, based on averages for the year 1997. Two significant groupings in the distribution are the range 4.0-8.5 PHP/L, with an average value of 6.5 PHP/L, and the range 8.5-12.0 PHP/L, with an average value of 9.9 PHP/L. At the 1997 average exchange rate of 29.5 PHP/US$, the average values of these two groupings are equivalent to US$0.22/L and US$0.34/L, respectively. One approach to the analysis would be to include the fuel price in the grouping of the plants into classes. However, it is believed that the fuel prices will Typical Load Profiles 150 4-Kabugao 5-Palanan 6 -Cagancillo Hour Figure 3. Load profiles for Class 4, 5, 6 Fuel Price Histogram 20 $.22/1 Frequency ore Fuel Price, PHP/Ii ee > MP Wwe e Figure 4. Distribution of delivered fuel prices at SPUG power plants fluctuate in future years due to fluctuations in the world market price of fuel, the removal or reduction of subsidies affecting the fuel price incurred by NPC, privatization, and other factors. So that the results may be reinterpreted as fuel prices change, the fuel price is treated as a variable in the analysis. Values of US$0.22/L and US$0.34/L are used to represent the two main groupings in the distribution, as well as US$0.46/L to represent the possibility of increased fuel prices in future years. There is a trend that the largest plants have low fuel prices; however, fuel prices of about US$.30/L or more are seen at some plants producing as much as 20,000 kWh/day. Resource Assessment Wind. The role of wind turbines in an optimized hybrid design depends heavily on the wind resource in the vicinity of plant site. A wind atlas of the Philippines was recently completed at NREL (Elliott et al., in progress). However, at the time of this study, these results were not yet available. Therefore, the magnitude of the wind resource is treated as a variable in this analysis, so that the results may be interpreted as more information about the wind resource becomes available. Although this approach leads to conclusions that are less specific, it has the advantage that the results may be applied to a wider range of applications. In a previous study (Barley et al. 1998), a one-year set of hourly wind speed data was obtained from Basco, in the northernmost province of Batanes. The annual average of this data set (adjusted for the estimated long- term average) is 6.13 m/s at a height of 12 m. Although some sites in the Philippines with a greater wind resource than this have been identified, it is generally true that the wind is strongest in the north. In order to represent a reasonable range of wind speeds in the analysis, the data from Basco were scaled to annual averages of 4.5 m/s, 5.5 m/s, and 6.5 m/s. This range serves to illustrate the sensitivity of wind power cost- effectiveness to the wind speed. The power densities corresponding to these three wind speeds are 98 W/m?, 179 W/m?, and 295 W/m, respectively. The seasonal profile of the wind resource, based on this data set, is shown in Figure 5. Of course, actual weather patterns vary throughout the country; therefore, hourly data measured at actual project sites will be used in further analysis of recommended retrofit projects. 5 ———$—$<$<$_______.. ee ee oe Solar. Site-specific estimates of annual average global insolation data for the Philippines were obtained from Bonjoc et al. (1985). The average value for all the sites is 4.01 kWh/m? /day. In addition, daily values of global insolation were obtained from the World Radiation Data Centre Web site (undated reference) for one site: Science Garden, in Manila. The long-term average of this data set is 4.50 kWh/m?/day. For the analysis, hourly values were extrapolated from one year of daily values at Science Garden. (The most complete year of data, 1981, was scaled down from 4.86 kWh/m?/day to the long-term average of 4.50 kWh/m?/day.) Although Bonjoc et al. indicate some sites with a higher estimated solar —_ Figure 5. Seasonal profiles of the wind and solar resource, the value 4.50 kWh/m?/day is above the resources average for all the sites. The approach in this analysis is to use a liberal estimate of the solar resource and a conservative estimate of the cost of PV modules. Then, if PV does not appear to be a cost- effective component in the optimized retrofit designs (which is in fact the result that is obtained), cases of lower resource or higher module costs are ruled out as well. If it is subsequently determined that a higher insolation or lower module cost would apply at a particular site, the analysis should be repeated for such a case. Wind Power, Avg kW/AOC 15/50 Insolation, Tilted, kWh/m*2/day Month The seasonal profile of this data set is shown in Figure 5, for comparison with the wind profile. Recalling that the seasonal load profile is assumed to be flat, the wind and solar resources are somewhat complementary. For example, the solar resource is strong during February through May, when the wind is relatively weak. However, the months of June and October are notably weak in combined resources. Hybrid Designs The system hardware configuration considered in this study includes wind turbines, batteries, and a rotary power converter, in addition to the existing diesels. Photovoltaics is also considered, although none of the optimized designs includes a PV component. A dump load component is used to stabilize the AC wind turbines. The rotary converter component, with higher available capacities, is replaced with an electronic inverter, with a higher conversion efficiency, in the smaller systems. The rotary converters are also less expensive per volt-ampere of capacity in the size range of the larger systems. In all cases, the wind turbines are connected to the AC bus, and all diesels are permitted to shut off when the power from other sources is adequate to meet the load. When the diesels are off, the reactive power requirements (volt-amperes reactive, or VARs) of the wind turbines and the village load are met by the power conversion device (inverter or rotary converter). The inclusion of a PV array and of batteries are options in the optimization procedure, except in the 6-hour plants (Classes 5 and 6), where the use of batteries is assumed. This type of hybrid system — AC wind turbines, with shutoff of all diesels allowed — is a relatively new technology with little field experience. In fact, some of the designs indicated in this analysis feature larger wind turbines than have previously been used in this configuration. Therefore, adequate technical support for such a venture would be especially important, and a recommended approach would be a turn-key installation by an experienced system integrator. Experience with this type of system will be gained in Wales, Alaska (Drouilhet, undated reference). Component Costs Installed costs of the various components featured in the hybrid designs are estimated as shown in Table 2. The installed cost of the large wind turbines is difficult to estimate because of the requirement of a crane for installation and the difficulty of moving a crane to the various Philippine islands and hilltop sites. In order to indicate the sensitivity of the results to this cost item, two different cost estimates were used, identified in Table 2 and in the results as sub-cases A and B. For a more accurate analysis in the design stage, this cost should be ascertained more definitively. A commercial installer of PV systems in the Philippines has.estimated the installed cost of a PV array as about US$7.00/watt. Allowing for some economy of scale in a larger system, this cost is estimated as US$6.00/watt in this analysis. Battery life was computed on a case-by-case basis within the models, which are discussed below. The hybrid system architectures assumed in this study feature wind turbines connected to the AC power bus and diesel gensets which may be shut off. In order to maintain a constant electrical frequency in such a system, a dump load is used. The dump load dissipates any power in excess of the load to avoid an increase in wind turbine speed and electrical frequency, and it is controlled using high-speed switching to maintain a stable frequency as the wind speed and the load vary. Such a dump load may have economic value, as in water heating, water pumping, or ice making. However, in order to vary the dump load quickly enough and in small enough increments to maintain a steady frequency, the use of a dump load component, featuring a bank of successively larger resistors (for at least a portion of the power dissipation) is assumed. The cost of this _ component, with associated controls, is assumed to be US$200/kW (Drouilhet 1998). The size of the dump — load component is taken as the rated power of one wind turbine (or two wind turbines, in Class 3), under the — assumption that any additional excess power could be utilized in a productive load or eliminated by shutting down some of the wind turbines. The cost of existing diesel equipment is not included this analysis, because it would not be altered by a hybrid retrofit. However, the addition of more sophisticated diesel controls would be required, to enable automatic starting and stopping, synchronization, and load sharing in a hybrid system. The cost of these controls is assumed to be US$15,000 per diesel genset (Drouilhet 1998). In addition, combined maintenance-and- overhaul costs are estimated aS US$7.007run-hour for each diesel genset. This estimate is based in part on NPC maintenance records. It may be somewhat unrealistic to apply a Maintenance cost that does not depend on the size of the machine. Thus, a more detailed accounting of maintenance costs should be performed in a subsequent analysis of recommended retrofit projects. Table 2. Component Costs Assumed in the Analysis Component Assumed Price Bergey Excel $24,000 ea. 10 kW, w/ 24-m tower, + 5% Inst.; Maint. 10%/10 yr WTG AOC 15/50 $75,000 ea. 50 kW, w/ 25-m tower, + 5% Inst.; Maint. 10%/10 yr $600,000 ea. | 550 kW, w/ 40m tower, installed; Maint. G Zond ZAOFS (A) $50,000/10 yr WTG Zond $825,000 ea. | 550 kW, w/ 40m tower, installed; Maint. ZAOFS (B) $50,000/10 yr PV Generic $6,000/kW $110 ea. + 5% Inst., Maint. 5%/yr $1,000/kVA + 5% installation $100/kVAR < 200 kVAR $60/kVAR > 200 kVAR $20,000 + added to synchronous condenser cost $200/kW According to a SPUG economist, low-interest loans at an annual rate of 2.7% may be available for projects such as these. However, this analysis of least-cost designs is based on the standard interest rate of 15% for SPUG projects, so that the resulting designs will be most conducive to replication. It is assumed that income tax deductions do not apply to any of the expenses. Additional financial parameters are listed in Table 35 Battery Trojan L-16 Diesel controls | Generic Dump load Generic Electronic ~ AES inverter Synchronous condenser Small, generic Synchronous condenser Large, generic Generic Rotary converter Financial Parameters Table 3. Financial Parameters Used in the Analysis a Annual discount rate General inflation rate Annual rate of increase in battery replacement expense Down-payment fraction of first cost Fractional salvage value at end of equipment life Models Two hybrid system computer models are used in this analysis. Both models are time-series simulation programs that were developed at NREL. HOMER (Hybrid Optimization Model for Electric Renewables) (Lilienthal 1995) is simplified to run quickly, facilitating iterative computations to search the range of possible designs for the one leading to the lowest life-cycle cost. Hybrid2 (Baring-Gould et al. 1996) is a more sophisticated model which is used to verify the HOMER results and more accurately determine the fuel usage, maintenance requirements, and cash flow associated with the recommended design. In this analysis, both models are run for a period of one year using a one-hour time step. Results Results for Classes 1-4. The results of the analysis for Classes 1-4 are summarized in Figure 6. (Curve markings correspond to the system classes listed in Table 1.) The curves in this graph are based on visual interpolations and extrapolations of more detailed results, which are not included here due to the space limitation (Morris et al. 1998). For these 24-hour power plants, it is shown that hybrid retrofits could save both fuel and money, relative to the existing diesel-only systems, at wind speeds of about 5.5 m/s and higher and fuel prices above about US$0.20/L to US$0.25/L. For the cases studied, cost savings as high as about US$0.12/kWh and fuel savings as high as about 65% are predicted. Also, the amount of Wind Spnea battery capacity recommended for the least-cost designs tends to increase as the size of the plant decreases and Figure 6. Approximate conditions needed as the wind speed and fuel price increase. Some of the for wind/diesel retrofit feasibility with 24 designs feature no storage component. Because the hour/day operation © & 5 £ an = 3 z © 8 *Significant revisions of the HOMER model are not yet reported in the literature. 7 latter is less well-proven than the simpler wind/diesel technology, especially for the larger systems, a more detailed subsequent analysis of recommended designs should include a consideration of the trade-off between economics and simplicity for systems with and without energy storage. In the example Class 4 system that was analyzed, adding one smaller diesel genset to the existing array has the effect of reducing the COE by about US$.04/kWh to US$.06/kWh in a diesel-only system, depending on the fuel price. This modified diesel array serves as the base case against which the hybrid designs are evaluated. Results for Classes 5 and 6. For the 6-hour plants (Classes 5 and 6), the analysis indicates that hybrid retrofits are not cost-effective over the same ranges of wind speed and fuel cost. However, in each case, the addition of one smaller diesel genset yields considerable cost savings. For Class 5. Adding a 90-kW diesel genset reduces the COE by: Fuel COE Price Savings US$/L US$/kWh 0.22 0.032 0.34 0.050 0.46 0.068 (Existing maintenance costs incurred due to running diesels below 40% of rated power are unknown, and thus are not included in these results.) For Class 6. Adding a 70-kW diesel genset reduces the COE by: Fuel COE Price Savings US$/L US$/kWh 0.22 0.093 0.34 0.144 0.46 0.195 It is a general rule that hybrid retrofits are less likely to be economical in part-day diesel systems than in 24- hour systems, because: ¢ For the same average load, more of the energy from renewable sources needs to be stored. A larger power converter is needed to meet peak load. More energy is lost in power conversion and energy storage. The cycle life of the batteries is exhausted more quickly. * For the same average load, the diesel-only base case features less diesel run time, so there is less money to be saved on maintenance by shutting off diesels. Conclusions This analysis shows that wind retrofits to the existing isolated power plants in the Philippines are most likely to be cost-effective for the plants providing 24-hour service, for wind speeds of approximately 5.5 m/s and greater, and for fuel prices above about US$0.20 to US$0.25/L, with some trade-off between the minimum wind speed and the minimum fuel price. Photovoltaics is not likely to be cost-effective in this application for a solar resource of 4.5 kWh/m?/day or less and an installed module cost of US$6/watt or more. COE savings for the recommended designs range from about US$0.02 to US$0.12/kWh, with associated fuel savings ranging from about 30% to 65%. In the smaller systems studied, including the 6-hour plants and the smallest class of 24-hour plants, adding a smaller diesel to the existing equipment could save between US$0.04 and US$0.20 /kWh. The primary trade-off determining the cost-effectiveness of hybrid retrofits is the cost of the hybrid equipment vs. that of the diesel fuel and maintenance saved. Because the hybrid equipment considered in this study would be imported, and because most of the diesel fuel used by SPUG is also imported, changes in the currency exchange rate would not be expected to have much effect on these results, with the exception that the diesel maintenance costs would shift relative to the other expenses. Deferrable loads, also known as productive dump loads ( such as water heating, water pumping, and ice making) can enhance the economic advantage of a hybrid power system by utilizing the excess wind energy that is occasionally available. This possibility was not considered in this study; however, it should be included in any subsequent analyses. One of the objectives for this work was to develop a systematic approach for evaluating options for retrofitting diesel mini-grids with hybrid renewable technologies. The five-step approach outlined on the first page, in conjunction with the tandem use of the two computer models (HOMER and Hybrid2), proved to be an efficient and effective approach to developing an overview of the retrofit opportunities. Assembling the basic information about all of the power plants in Step 1 would not have been possible without the personal contact and cooperation of a SPUG engineer (co-author Rafael Abergas). This alone took about a week of two analysts’ time. More specific recommendations would result from the study if the wind resource assessment had been completed in advance. Since the NREL wind resource mapping of the Philippines has been completed, specific candidate sites have been identified through the use of the wind maps, existing mini-grid maps, topographic maps, and aerial photography. Anemometers will be installed at those sites to collect data for a more accurate analysis. Some of the diesel power plants treated separately in this study are actually interconnected in a grid system. Subsequent analysis will involve treating the plants collectively. This will require a determination of the power capacity and the condition of these grids. Because the results are not very sensitive to the size of the plant, grouping the plants is not expected to make much difference in the results, except in cases where the interconnected grid passes through a higher wind resource than an individual plant grid. Because the results of this analysis are presented with wind speed and fuel price shown as variables (Figure 6), the question may arise as to whether or not the same results would apply in another country. The aspects of this study that are specific to the Philippines are: * The assumed flat seasonal profile of the load, based on the tropical climate + The seasonal profiles of the wind and solar resources, shown in Figure 5 + The trade wind climate (in the northern part of the country, where the wind speed data were recorded), associated with a rather consistent wind resource * The assumed level of solar radiation: 4.5 kWh/m?/day (Because PV was ruled out in this study, only higher levels than this, on the tilted array plane, would contradict the assumption.) * The component costs, shown in Table 2 * The financial assumptions, shown in Table 3. Acknowledgments Primary funding for this study was provided by the U.S. Department of Energy and the Asia-Pacific Economic Cooperation. In addition, DOE's Energy Research Undergraduate Laboratory Fellowship program funded the participation of Matthew Meares, who performed the simulation analyses. Tony Jimenez, at NREL, assisted with use of the HOMER and Hybrid2 models. Stephen Drouilhet, also at NREL, provided information about auxiliary component requirements and costs. Silverio Navarro, of Solar Electric Co., Inc., in the Philippines, and Ellen Morris, of Sustainable Energy Solutions, in New York, were helpful in obtaining solar radiation data. References Baring-Gould, E.I.; Green, H.J.; Manwell, J.F.; Van Dijk, V. (1996). "Hybrid2—The Hybrid Power System Simulation Software." American Wind Energy Association Windpower Conference Proceedings; June 23-27, 1996; Denver, Colorado. NREL/TP-440-21506. Golden, CO: National Renewable Energy Laboratory; pp. 497- 506. See also http://www.ecs.umass.edu/mie/labs/rerl/hy2/intro.htm. Barley, C.D.; Nufiez, C.S.; Magpoc, G.B. Jr.; Abergas, R.L. (March 1998). "Proposal for Hybrid Power Systems in the Province of Batanes, Philippines." Golden, CO: National Renewable Energy Laboratory; Quezon City, Philippines: National Power Corporation. Bonjoc, M.C.; Buan, R.D.; Leano, V. (1985). "The Profile of Solar Insolation in the Philippines." Bureau of Energy Development, Nonconventional Resources Division, the Philippines. Drouilhet, S. (1998). Private communication. National Renewable Energy Laboratory, Golden, Colorado. Drouilhet, S. "A High-Penetration Wind-Diesel Hybrid Power System Pilot Project in Northwest Alaska." Golden, CO: National Renewable Energy Laboratory. (Unpublished NREL paper) Elliott, D.; Schwartz, M.; George, R.; Haymes, S.; Heilmiller, D.; Scott, G.; McCarthy, E. "Wind Energy Resource Atlas of the Philippines" (in progress). TP-500-26129. Golden, CO: National Renewable Energy Laboratory. Lilienthal, P. (March 27-30, 1995). "Economic Analysis of Renewable Technology Options for Village Electric Applications." Presented at the 1995 Windpower Conference, Washington, D.C. — Morris, E. et al. (Oct. 1998). "Analysis of Renewable Energy Retrofit Options to Existing Diesel Mini-Grids." ARES Energy Working Group, APEC Pub. No. 98-RE-01.6. E2» apecseceara » sg 5 | devin load /Pub bP “World Radiation Data Centre Internet Web site, http://wrdc-mgo.nrel.gov/, » ana for the World Meteorological Organization by the Russian Federal Service for Hydrometeorology and Environmental Monitoring, A.!. Voeikov Main Geophysical Observatory, St. Petersburg, Russia. 10 dee 2, Fr July1999 © NREL/CP-500-26827 Power Flow Management in a High Penetration Wind-Diesel Hybrid Power System with Short-Term Energy Storage S.M. Drouilhet National Wind Technology Center Windpower ’99 June 20-23, 1999 Burlington, Vermont fe. *s, No om il . National Renewable Energy Laboratory 1617 Cole Boulevard Golden, Colorado 80401-3393 NREL is a U.S. Department of Energy Laboratory Operated by Midwest Research Institute « Battelle « Bechtel Contract No. DE-AC36-98-GO10337 NOTICE This report was prepared as an account of work sponsored by an agency of the United States government. Neither the United States government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States government or any agency thereof. Available to DOE and DOE contractors from: Office of Scientific and Technical Information (OST1) P.O. Box 62 Oak Ridge, TN 37831 Prices available by calling 423-576-8401 Available to the public from: National Technical Information Service (NTIS) U.S. Department of Commerce 5285 Port Royal Road Springfield, VA 22161 703-605-6000 or 800-553-6847 or DOE Information Bridge http://www.doe.gov/bridge/home.htm! ae ae Printed on paper containing at least 50% wastepaper, including 20% postconsumer waste POWER FLOW MANAGEMENT IN A HIGH PENETRATION WIND-DIESEL HYBRID POWER SYSTEM WITH SHORT-TERM ENERGY STORAGE Steve Drouilhet, PE National Wind Technology Center National Renewable Energy Laboratory 1617 Cole Blvd. Golden, Colorado 80401 USA INTRODUCTION In the last several years, interest in medium to large scale (100 kW to multi-MW) wind-diesel hybrid power systems for rural electrification has grown enormously among energy officials and utility planners in the developing countries, multilateral lending institutions, and utilities serving the far northern areas of the developed countries. There are many indications that there is a large potential market for such systems, and though there are an increasing number of demonstration projects, a true market for such systems has yet to emerge. Consequently, an industry capable of serving this market is still only in its infancy. Only a small fraction of researchers and engineers working in the wind power industry, which is relatively small itself, are involved in hybrid systems for off-grid applications. There is therefore relatively little information available on the technical issues involved in implementing a wind-diesel power system. It is tempting to view the addition of wind turbines to a diesel mini-grid as a straightforward task, only slightly more complicated than a conventional grid-connected installation, requiring only a few ancillary components at a relatively modest cost. While this is true for low penetration wind-diesel systems, where the wind turbine output averages no more than about 15% of the load, high penetration systems, in which the average wind power generated can approach or even exceed the average load, require much more sophisticated controllers and more extensive components in addition to the wind turbines. This additional cost and complexity can often be justified by the much greater fuel savings (and associated environmental benefits) and reduced diesel operating time made possible by high wind penetration. This paper is intended as an introduction to some of the control challenges faced by developers of high penetration wind-diesel systems, with a focus on the management of power flows in order to achieve precise regulation of frequency and voltage in the face of rapidly varying wind power input and load conditions. The control algorithms presented herein are being implemented in the National Renewable Energy Laboratory (NREL) high penetration wind-diesel system controller that will be installed in the village of Wales, Alaska, in early 2000. BACKGROUND Since 1995, the National Wind Technology Center (NWTC) at NREL has been researching wind-diesel hybrid power systems. Areas of study have included optimal diesel dispatch strategies, the value of energy storage, village mini-grid optimization, power converter efficiency, system stability, and long-term performance and economic modeling. In 1997, the Hybrid Power Test Bed was constructed at the NWTC to facilitate the development and testing of both experimental and commercial hybrid power systems. One of the ways NREL supports the development and growth of the renewables-based hybrid power systems industry is to provide technical assistance to pilot and demonstration projects, both domestically and internationally. Since 1995, NREL has been engaged in a collaborative project with the Alaska state energy office and Kotzebue Electric Association, a rural Alaskan electric cooperative, to design, test, install, and monitor a high penetration wind-diesel hybrid power system in Wales, Alaska, a small village on the northwest coast on the Bering Strait. NREL’s role in this project has been the following: e Identify a hybrid system architecture well suited to implementation in small northern villages e Design and build the non-wind turbine hardware components of the system (system controller, secondary load controllers, energy storage subsystem) Develop the control software necessary to operate the system stably and reliably Fully test and debug the control system at its test facility in Boulder, Colorado Provide training in system operation and maintenance. The control system, the secondary load controllers, and the energy storage subsystem have all been built and installed at the Hybrid Power Test Bed, where the control software is currently undergoing test and refinement. Installation of the system in the village of Wales is scheduled to begin in July 1999. SYSTEM DESIGN OBJECTIVES Because of NREL’s substantial technical involvement in the project and the availability of its hybrid power test facility, this project represented an opportunity to develop and test a system to meet design and performance objectives beyond what could be met by then-available commercial hybrid power systems. Some of the specific objectives that guided the development of the system are the following: 1. One requirement of the system architecture was that it be designed to “wrap around” an existing village power plant. Unlike many remote regions of developing countries, nearly 100% of Alaska’s rural villages are already electrified. These village power systems represent a considerable investment in diesel generation equipment. It is important in these cases to use as much of the existing diesel genset and controls equipment as possible. 2. Second, the system should have sufficient penetration to achieve at least a 50% reduction in the diesel fuel consumed for electric power generation. 3. To maximize the potential fuel savings and to reduce diesel maintenance expenses, the system should allow the diesel generators to be shut off as much of the time as possible. Prior analysis at NREL had suggested that this objective required that the system include a small amount of high power density energy storage. 4. To further maximize the return on investment, the system needs to ensure that 100% of the wind turbines’ energy output serves a productive load. In other words, wind power in excess of what was needed to meet the primary village electric demand must be diverted to another application having economic value. In Wales, the only other significant energy demand is for space heating. Because the major heat loads exist at several distributed points in the village, a distributed secondary load arrangement was required. The architecture chosen to meet these objectives is shown, in simplified form, in Figure 1. The existing diesel power station consists of three 480 VAC, three phase generator sets with ratings of 75, 142, and 148 kW. The retrofit package consists of two AOC 65 kW wind turbines, an AC-DC rotary converter, a 240 VDC 130 Ah nickel-cadmium (Ni-Cd) battery, and several secondary load controllers to control power to electric boilers located at several points in the village. These secondary loads will be used to displace heating fuel that would otherwise be burned in existing boilers. Figure 1 shows only the power components, not the variety of control hardware required to make the system operate. Figure 2 depicts the control architecture of the system. Control signals pass among the various control components either via hardwired analog and discrete control lines, or via high speed serial networks, as shown. Batter 240 Wind Turbines (Induction, Stall-Regulated) 2X65 KW= 130 KW A 1 KY Tit Bank » 130 Ah ——— DC MACHINE 0 m1 ‘AC MACHINE. Rotary Converter 156 KVA Diesel #1 142 kW School Heating Diesel #2 75 kW Diesel #3 148 kW Xt System i Diesel Plant Xb Load lers Hydronic Loop ; Resistance Secondai Heaters Con' os be Primary Village Load 40-120 kW FIGURE 1. WIND-DIESEL SYSTEM ARCHITECTURE FOR WALES, ALASKA EXISTING DIESEL GENSET DIESEL PLANT CONTROL PANEL HYDRONCLOOP (DUMP LOAD CONTROLLER: ANALOG DISCRETE SGVALS CONTROL DISCRETE SGViS CONTROL SIGNALS WAD TORRE Sooeoe WIND-DIESEL Wwe----------- > CONROUER = |*4 HYBRIDPOWERSYSTEM — <- -S-REMEMEK____.) Rotary CONVERTER ' CONTROL CABINET 1 MAIN CONTROL PANEL rood \ | AOC 15/50 ' eos ms | RS-485 VONETWORK = PHONE CONNECTION | FOR REVOTE SCADALINK y ‘SCHOOL HYDRONC LOOP DUMP LOAD CONTROLLER FIGURE 2. WIND-DIESEL SYSTEM COMMUNICATION AND CONTROL DIAGRAM THE PRIMARY TASKS OF A POWER GENERATION SYSTEM An automated wind-diesel hybrid power system controller is called upon to do a wide variety of tasks. These include such things as (1) automatic dispatch of the diesel generators to ensure proper loading and good operating efficiency, (2) operator notification of any warning or alarm conditions, (3) performance data logging to facilitate troubleshooting and maintenance, and (4) management of the secondary loads to ensure that excess power is directed where it is most needed. Fundamentally, however, the most critical tasks of the system are to provide good frequency and voltage regulation. Unless the system can provide good power quality, as measured primarily by frequency and voltage stability, it is not viable. WIND POWER » POWER = ELECTRIC POWER TO RESISTIVE LOADS SYSTEM DIESELPOWER [ (STORED =) SHAFT POWER TO MECHANICAL LOADS KINETIC ENERGY) BATTERY POWER c= —— MSCELLANEOUS LOSSES FIGURE 3 POWER FLOWS INTO AND OUT OF THE HYBRID POWER SYSTEM Frequency Regulation The entire power system, including all its generators, distribution wiring, and even motors present in the village load, can be thought of as one big electromechanical entity, as shown in Figure 3. Power flows into this system as power from the wind transferred to the wind turbine rotor, mechanical power developed in the diesel engines as a result of combustion, and electric power drawn from the battery. Power flows out of the system to consumer resistive loads, to consumer mechanical loads, to secondary loads, and as various mechanical and electrical losses. At any given moment, if more power is flowing into the system than out of it, the difference will be stored as an increase in kinetic energy of the rotating machines within the system, both generators and motors, that happen to be on-line at that time. The effect of any power imbalance in the system is expressed in Equation 1. > Psources — >) Psinxs = wae = “yu a (1) ee where P = active power (kW) K.E. = kinetic energy of system J = moment of inertia of rotating machine @ = angular velocity of rotating machine This increase in kinetic energy is manifested as an increase in rotational speed of the synchronous machines in the system and thus an increase in electrical frequency. The task of frequency regulation is essentially a problem of maintaining an instantaneous balance of the real power flowing into and out of the system.’ ' This relationship of power imbalance to frequency change only applies to power systems in which the frequency is determined by the rotational speed of one or more synchronous machines in the system. In systems governed by a Voltage Regulation Analogously, regulating the AC voltage of the power system is a problem of maintaining an equilibrium between the source and sinks of reactive power (VARs) in the system. The induction generators of the wind turbines, transformers in the distribution system, and induction motors in the consumer load are all reactive power sinks. Power factor correction capacitors on the wind turbines or the distribution system are sources of reactive power. Synchronous generators, both on the diesel gensets and on the rotary converter, can either be sources or sinks, but generally they are supplying the reactive power demanded by the sinks. Unlike the case-of real power, where an imbalance can be absorbed by the system as a change in stored kinetic energy, there is no storage mechanism for “reactive energy”, which only actually exists as a mathematical construct. The reactive power supplied by the sources is inherently equal to the reactive power absorbed by the sinks. This is expressed in Equation 2, in which the reactive power flows for each component are expressed as functions of voltage. YX QsourcesV40) — YX Osinas V0) = 0 (2) where Q = reactive power (kVAR) Vac = AC bus voltage If the reactive power sources are unable to deliver the reactive power demanded by the sinks, the bus voltage will fall such that the equilibrium is maintained. With reactive power, the issue is not so much ensuring that equilibrium is maintained (which is automatic), but that the equilibrium occurs at the desired voltage level. On a synchronous machine, the function of the voltage regulator is actually to control the generator excitation such that the generator delivers the reactive power demanded by the load at the desired voltage. THE OPERATING STATES OF THE WIND DIESEL SYSTEM There are three devices subject to the direct control of the wind-diesel controller: the rotary converter AC machine, the rotary converter DC machine, and the secondary load controller (which actually consists of multiple distributed load controllers). Each of these devices has several different control modes associated with it. For example, the AC machine can be controlled to achieve any of the following: Match voltage with the AC bus (prior to synchronization) Share reactive power with the diesel generators Deliver a specified amount of reactive power to the grid Regulate AC bus voltage. The power flow management algorithm determines the appropriate control mode for each of these three devices depending on the operating state of the power system. The Wales wind-diesel hybrid power system involves multiple diesels and multiple wind turbines. In addition there is a power converter consisting of two separate rotating machines and a secondary load that is divided into “local dump load” and “remote dump load”. Because each of these components may or voltage source inverter, the frequency is typically set by a crystal oscillator and does not vary. However, a similar situation exists in that any power imbalance then typically shows up as an increase or decrease in voltage on the AC and/or DC side of the inverter. The problem then becomes one of voltage control rather than frequency control. may not be operating at any given time, there are a great number of possible system operating states. To develop a power flow management algorithm flexible enough to handle all possible operating states, one must identify a minimum set of key state variables that provide sufficient information to determine the appropriate control mode for each device. Our top level state variable is the diesel status, because it has the biggest impact on how voltage and frequency is regulated. “Diesel ON” refers to the state where one or more diesel generators is connected to the bus and loaded (i.e. not in load or unload ramp). Conversely, “Diesel OFF” refers to the state in which all diesel generators are either disconnected from the bus or connected but not fully loaded. Diesel ON State The stand-alone diesel generator is designed to regulate the voltage and frequency on an isolated power bus. In a multiple diesel configuration equipped with automatic load sharing controls, the diesels collectively regulate frequency and share both the real and reactive power load in proportion to their respective ratings. Diesel gensets do an excellent job of frequency and voltage control provided that the real and reactive power load on them remains within their rated capacity and they are not subject to large reverse power transients. In the Diesel ON state, we allow the diesel generator(s) to perform their intended function of frequency and voltage control, and we control the rotary converter and/or secondary loads to maintain the diesel loading in a comfortable range. In summary, in Diesel ON state, e The diesel generator(s) assume both frequency and voltage control ¢ Power flow to the secondary loads and/or energy storage is controlled to maintain diesel loading within a comfortable range e The rotary converter ac machine is used to assist the diesel generators in meeting the var load, as necessary. Diesel OFF State In the Diesel OFF state, the only synchronous machine left on the system is the AC machine of the rotary converter. The rotational speed of the rotary converter will establish the grid frequency. As with the diesel generator, the voltage regulator on the rotary converter AC machine controls the field current so as to maintain the desired AC bus voltage. Frequency is controlled by modulating power flow to the secondary load or battery, depending on factors to be discussed below. System Sub-States Diesel status is only the first of the system state variables that are used in determining the appropriate control mode for the various system components. The others reflect the state of readiness of the other system components and the nature of the instantaneous real power imbalance on the system. They are embodied in the following questions: 1. Is the (rotary converter) AC machine on line and ready? Just as with the diesel generator, for the AC machine to be available to perform its control function, not only must its contactor be closed, but it must also not be in an unload ramp, preparing to go off- line. 2. Is the DC machine on line and ready? Similarly, the DC machine is only available for control when its contactor is closed and it is not in a transitional state. 3. Is there instantaneous excess wind power? In the case where there is excess wind power, secondary (or “dump”) load may be used to provide frequency control. As long as there is excess wind power, this works fine, but suppose the wind suddenly drops, resulting in a power deficit. As wind power drops, secondary load will be rapidly removed in an attempt to maintain grid frequency. Once it has all been removed, the ability to control frequency is lost. The system must switch immediately to frequency control by the DC machine. 4. Is the battery “full”? This question refers to whether the present level of current into the battery can be sustained. It is actually several questions rolled into one. With a “yes” answer to any one of them, the battery is considered “full”. e Is the battery at a high state of charge (i.e., actually full)? e Is the DC charging current limit of the rotary converter reached? e Is the charging voltage limit of the rotary converter reached? Note that the state variables presented above are concerned only with whether the various system components are on line and ready at a particular moment in time, not when and why they are brought on line. The criteria by which individual diesels, wind turbines, and the rotary converter AC and DC machines are turned on and off are the subject of a whole suite of dispatch algorithms not covered in this paper. THE POWER FLOW MANAGEMENT ALGORITHM The power flow management algorithm is presented in flow chart format in Figure 4. Each decision block represents one of the state variables described above. Each branch in the decision tree specifies the control mode of the devices actively participating in voltage and frequency control in the corresponding state. Note that each branch loops back to the beginning of the algorithm, since any of the key state variables can change at any moment. START POWER DL = SECONDARY LOAD CONTROLLER ‘AC = ROTARY CONVERTER AC MACHINE VOLTAGE REGULATOR DC = ROTARY CONVERTER DC MACHINE FIELD CONTROL No: DL: DIESELLOAD fener) _ CONTRO (ra) ee See ‘AC: AC BUS VOLTAGE REGULATION ore 04: eesexvoanconraoy | _—fReTuRW TO No-+ ou Frequency contno, |_—+[RERRNTO aay & Yes XCESS DL: OFF RETURN TO DL: OFF RETURN TO WIND No-> |-_—+| No DL: ——+4 POWER? DC: DIESEL LOAD CONTROL START = DC: FREQUENCY CONTROL ‘START Yes Yes DL: OFF RETURN TO DL: CHARGE RATE LIMITING RETURN TO| o~ DC: DIESEL LOAD CONTROL [——*|__ START <a> No—> 0c: FREQUENCY CONTROL ‘START Yes DL: DIESEL LOAD CONTROL rR ——* DG: HOLD ZERO CHARGING CURRENT Yes. DL: FREQUENCY CONTROL RETURN TO ——* DC: HOLD ZERO CHARGING CURRENT START FIGIIRE 4 POWFR FIOW MANAGEMENT AT GORITHM PERFORMANCE REQUIREMENTS OF THE CONTROL SYSTEM In the Wales wind-diesel control system, the loop shown in Figure 4 is executed approximately once every 20 milliseconds (ms). A short loop interval is necessary in order to detect and immediately respond to changes in component status. For example, when the last diesel goes off line, the rotary converter must step in immediately to control the grid frequency and voltage. If the transition is too slow, unacceptable deviations of either voltage or frequency could result. When a change in state occurs that calls for a change in the control modes of one or more devices, it is important that the mode changes occur seamlessly, without causing discontinuities in power flow, which would be manifested as frequency or voltage transients on the line. This requirement is not expressed in the flow chart, but it is an important part of the design of the various control modes and requires careful application of bumpless transfer techniques. CONTROL CHALLENGES OF A HIGH-PENETRATION WIND-DIESEL SYSTEM WITH ENERGY STORAGE Compared to conventional power generation systems, where short-term load variations are typically small and the principal power source is dispatchable on demand, high penetration wind-diesel power systems are challenging to implement. The wind power input to the system is stochastic in nature and highly variable, particularly at gusty sites. At the Wales project, there will be times when the wind power exceeds 200% of the village load, with short-term variations as large as the load itself. There is also the fact that on small isolated power grids, single loads tend to represent a larger percentage of the total load. Starting even a small industrial motor, for example, could have a perceptible impact on the system. The variability in the wind and the variability in the load combine to yield rapid and high amplitude fluctuations in the net load, which is the difference between the primary load and the instantaneous wind power. The net load represents the power that must be supplied by the diesel generator(s) and/or energy storage, or if it is negative, must be absorbed by secondary loads to maintain the real power balance discussed above. Low system inertia is another factor that contributes to the challenge of providing tight frequency regulation in wind-diesel hybrids. The rotating mass contained in a typical isolated wind-diesel hybrid system is disproportionately smaller than that of large utility-scale power systems. Whereas the time constant in a utility system for the frequency to respond to a change in load is measured in seconds or even minutes, it is measured in tenths of a second for the hybrid system. The actual PID control loops used to control power flow in the secondary loads and in the rotary converter must provide very fast response. Because of the requirements for speed and automatic control mode switching, AC-based wind- diesel hybrid systems require active computer control systems to provide stable operation and good power quality. CONCLUSIONS The Wales, Alaska, project will demonstrate the feasibility of retrofitting an existing village diesel plant to create a high-penetration wind-diesel system that achieves a large reduction in diesel fuel consumption and run time and that uses all available wind energy in a productive manner. The Wales system consists of nine active power system components (three diesel gensets, two wind turbines, two secondary load controllers, an AC synchronous generator, and a DC motor). The variable status of each of these power components gives rise to many possible system operating states. The control system must respond rapidly to changes in system operating state and smoothly transition among a variety of control modes. The controller must regulate both real power to provide stable frequency and reactive power to provide stable voltage. Implementing such systems requires not only controls expertise but also a detailed knowledge of the individual system components and how they interact. For example, some wind turbines have a large inrush current on synchronization. Some turbines have no inrush and can control their own power factor. The power converter interface to the energy storage must be designed to operate in a way that is compatible with the wind turbine requirements. As another example, some secondary loads have a very fast response (e.g., electric resistance heaters) and some a slow response (e.g., water pumps, ice makers). > In AC-based systems, the wind turbine generators are typically of the induction type and connected to the local AC distribution system. These are in contrast to small DC-based hybrid systems where wind turbines, and often photovoltaic panels, are connected to a DC bus, in parallel with a battery. A hybrid system controller algorithm would need to take these differing characteristics into account in order to achieve acceptable frequency control. Because of the considerable impact of individual component characteristics on overall system operation, it appears that in the near term, the design of hybrid power control systems will be fairly specific to a particular power system architecture. Only after considerably more high penetration wind hybrid systems operating experience has been obtained, and the many control issues posed by various generation, load, and storage devices are better understood, will it be possible to design generic hybrid system controllers capable of adapting to a wide variety of components and system architectures. Regarding the cost of wind hybrid power systems, it is misleading to think of the nongeneration hybrid system components (system controller, power converters, energy storage, secondary loads) as mere accessories to either the wind turbines or the diesel power plant. These components typically represent 25%-50% of the equipment cost of the system, not including the cost of the diesel plant and distribution system, which is pre-existing in many cases. Wind hybrid project promoters, developers, and potential customers often underestimate these costs when considering a project. 10 ~ ASeE = — oS leit ie SS) | Da ed July 1999 * NREL/SR-500-24663 Performance and Economics of a Wind-Diesel Hybrid Energy System: Naval Air Landing Field, San Clemente Island, California Ed McKenna National Renewable Energy Laboratory Golden, Colorado Timothy L. Olsen Tim Olsen Consulting Denver, Colorado NREL National iihowsbts Energy Laboratory 1617 Cole Boulevard Golden, Colorado 80401-3393 NREL is a U.S. Department of Energy Laboratory Operated by Midwest Research Institute e Battelle e Bechtel Contract No. DE-AC36-98-GO10337 July1999 * NREL/SR-500-24663 Performance and Economics of a Wind-Diesel Hybrid Energy System: Naval Air Landing Field, San Clemente Island, California DOE Strategic Environmental Development Research Program Ed McKenna National Renewable Energy Laboratory Golden, Colorado Timothy L. Olsen Tim Olsen Consulting Denver. Colorado NREL Technical Monitor: Ed McKenna Prepared under Subcontract No. CAK-6-15387-01 ote oa, « »NR=EL + National Renewable Energy Laboratory 1617 Cole Boulevard Golden, Colorado 80401-3393 NREL is a U.S. Department of Energy Laboratory Operated by Midwest Research Institute e Battelle « Bechtel Contract No. DE-AC36-98-GO10337 NOTICE This report was prepared as an account of work sponsored by an agency of the United States government. Neither the United States government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States government or any agency thereof. Available to DOE and DOE contractors from: Office of Scientific and Technical Information (OSTI) P.O. Box 62 Oak Ridge, TN 37831 Prices available by calling 423-576-8401 Available to the public from: National Technical Information Service (NTIS) U.S. Department of Commerce 5285 Port Royal Road Springfield, VA 22161 703-605-6000 or 800-553-6847 or DOE Information Bridge http://www.doe.gov/bridge/home.html| ae fa Printed on paper containing at least 50% wastepaper, including 20% postconsumer waste CONTENTS Page BOTG WOT sees eccxesccseseesucassusuvauestsvnsncsovseoucenstssevsasessseyestesteavssetivewsssiessosoussssstesssnsedsdssrcnsteeses Acknowledgments.... i EXCCUtive SUMMaLy i. cssssccccscsssacsresncsvaveesonesvrvssousvessvarvisevervaceresrsenasssesenrserearnvenenioess 120) TtPODUGHON. ..........5sc..sseusecesetensnssodsoaecese tan aaisessaivesesssesessvesasusstusaresietsest sonessesessateasesees 1 2.0 Background ........ccececcscssssescsecssesesescseesesesssesesnsesecsescscssscsssssesesesecscasseeeeesasseeaacasssseasees 1 2.1 San Clemente Island o......c.cceccesessessesesesesesesesescseseseeeeecsesesesesseseseseessseeeneeeeeeees 2.2 Naval Installation Mission. ve 2.3 Energy Demand ...........ccecesesceseseseesessescsescscsescsesesescseeeecsssesesesssssseseeesscseseeeeeeees Dif Diesel Mery SY Stet ces sezexern ca coeaswors emacs sre svscsessususravevee pssenrucweususnsususesesvovnss 2.5 Wind Energy Site Description .. 2.6. Wind Energy System sccscccsccscncscsescsscvscossveswasssssvescvasesveastavewauataveavesstwscavetawsteess 3.0 The Wind Resource .........ccccccccssseseseseseeeseeseseseesesesesesesseseesssscsssseseeeesasasseeseasseseeasanasaees 15 3.1 Wind Data Collection and Analysis. 1S 3.2 Historical Wind Data................ 1S 3.3 Current Wind Data... cccccecceesessssesesctecseseesescseescsesscsesscsscsesesaesseeessecassecatees 19 4.0 Wind-Diesel Hybrid Energy System............ccccscssssessssssssesesesesesssssseseseseseneseseaeseseenes 29 4.1 Diesel System Operation ... .29 4.2 Wind Turbine Operation........0.00ccce 30 4.3 Wind Energy vs Load Profile Correlation . 31 4.4 Energy System Operating Resullts..................ssssssssesessesescesessssseesessseesesesssssees 32 4.5 Hybrid System Spreadsheet Model sic. ssisesescvssses sessescesesivcasscssocessvsosscoevestvese 32 5.0 Cost Analysis .......cccccccceesceesseseeseseescssesesecsecsesescsesecsesesecscsecaeeesaeeesseesesesseesseeeseeasseceesaees ISD. MTOM ORY sas ccsccscsssesesseveceseesassecoeseasessssinsescsuedeusessescveseucecesteesesorseesutesseces dovse 5.2 Diesel System Costs. 5.3 Wind Energy System Costs 5.4 Wind-Diesel Hybrid System Operational Savings...........c.ccceceeceeeseeeeeeeeeeeee 37 5-5) SEMSILIVAty: ANALYSIS ressesestcscecexsescaxeveseuessesesasveuseseersvisestavsessesrevescecs\aueeouiestoeiessn 39 6:0) (COMCHISIONS ic. i<ci.c:.ccass canes ceressescacateccacst acs acstvestststsetscssnonedsesosunadstbuebescavedsstevessteerersess 40 RREPCTEN CES. 5c:esensecseeqevecseseocesexesaseceateessvosessacooseisosovaresasssefeiscuntosessesssuesesesesesisnsedessee? 42 Appendices .......ccccccesesesesesesesensesesesesescescsesesssescsescsesesescseseseseuesescsesenececseeeeseaeaeneneeeeeees 43 A. Hybrid System Model and Economic Summary Tables B. SCI 1998-9 Power Plant Status and Production Reports C. Independent Paper on Wind-Diesel Hybrid Energy System Design and Operation D. Wind-Diesel System Operational Guidelines LIST OF TABLES Page SCI System Demand Statistics .......c..cccccccccescesesseseesesceseescssesecsceesecsecscacseeseeeceteseeees 5 Diesel / Generator Power Rating and Fuel Consumption.............cccesseeeeeeeeeeeees 9 Source Years for Composite Wind Data Set .........cceeeeesesseeeseseeeeeseeeeeseeeeeeees 19 Summary of Current SCI Meteorological Data .............c.ccececeseseeseeeeeereseeeneneeeeeees 19 1998 Hybrid Systems vs Baseline: Spreadsheet Model Results «0.0.00... cece 34 Initial Capital Costs for One Additional 225 kW Wind Turbine... 36 Economic Sensitivity to Wind Speed Variations..............cceceseesesesseseseeseseeeeneeeeneee 39 ii 10. il. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. LIST OF FIGURES Page SCI Location Map......ccccecsesesccssescsecsseseesscsessescseeecscsecscsecscseseeeceeceeeessecseeecseeeaeeaes 3 SCI Wind Farm Location Map .........:ccccsesescsesssscscscsssseesseecscecseeeeecsssesesscseseseceeacaes 4 SCT oad Frequency Distribution izecerececsssscseconctsssossasnsnscossetarssassasasatvotcpenssavesseccnestd 6 SCL Annual Average Diutnal Load iscissccsssesasicscsscsevastsasnateccasssdensusenqivicnsassisstsostrereies 6 SCI Annual Energy Production Record .......:.cccsessesesesseseseseseseneneessseeescseeseeseeenes 7 SCI Annual Fuel Consumption Record ..0......:cccecesseesceeeseeeeseescseeseseesesseeecseeeeseeeees 7 SCI Energy Production vs Fuel Consumption ..........c.ccccccsscseseseseseeeeseeeseeseeseaeseseees 8 SCI Naval Facilities and SERDP Wind Farm Site... ccccceseseseseseteeeeeeeneees 11 Arial View Of SCI Wind -F atin oesssrasccssesesavccecrsescsexscsssonsescvecssssvastevssevssesesssoreevsiaras 12 SCI SERDP-Funded NM 225-30 (225 kW) Wind Turbines «0.0.00... ceeeeseseeeteees 14 SCI Historical Wind Speeds .........cccsceeeeeseeeeessesesesssescseseseeenscscseseeecssseeerssscseseeees 16 Annual Average Diurnal, Jacobs Site, 1983-400... eceseseseseeeesescsereeseseseeseeeeeeeeees 17 Annual Average Diurnal, Jacobs Site, 1984-5 .......c.ceeeecseseeseseseseeeeeseeeeeeneneneeesees 18 SCI Wind Speed Frequency Distribution ...........ccceeceseseeeeeeceeneeeeeeeeeteeneeeeaeeeees 20 SCI Monthly Averaged Wind Speed, 1995 oe ccceeseeseseeseeeeseeseseeseeseseeseeseees 21 SCI Monthly Averaged Wind Speed, 1996 .......cccceeceseseseeeesesessseesesssseeeessseseeses 21 SCI Monthly Averaged Wind Speed, 1997 o....ceeeceseeseseseeseeeseseseesesesenerstseseseeeees 22 SCI Monthly Averaged Wind Speed, 1998 oo... eeeesesescseeeeeeseneeeeetsesetetseseeeteees 22 SCI Daily Averaged Wind Speed, 1998 oo... cceececeseeseseteseeeeeeseseeteeseeteteeseeeereeees 23 SCI Monthly Averaged Temperature, 1998 .0.....ccccesccsseseseseescsesesestsesesseesnesseeees 23 SCI Monthly Averaged Pressure, 1998.........ccssescssssssesesescseseseeseseeeeetecscseeeeeeeeeates 24 SCI Monthly Averaged Density, 1998 ssscscsscccsessrcesnesssssurecssvessesesssascossnesrucstsuvsnensess 24 iii 23. 24. 25. 26. 2: 28. 29. 30. 3h. 32. 33. 34. SCI Monthly Averaged Wind Power Density, 1998.........cecseseseeesseteeeeeseeeeeeeee 25 SCI Annual Average Diurnal Wind Speed ..........c.ceccceceeeseseceseseeeeeeeeteneeeeeseeeeeeeees 26 SCI Wind: Rose} Percent Time ssacccsccsnssiscsnsscssosascmnassurvcrscsvssvasesversecsrsscsvssssvesesvsten 26 SCI Wind Rose: Average Wind Speed (11/Ss)..........cccceseesesssesesesesseeeeeseeeseeeeeeee 27 SCI Wind Rose: Average Wind Speed (kmots).........:.:ecssesessseeseseseseeteseseeeseeeseseees 27 SCI Wind Rose: Time Weighted Average Wind Speed (11/s)............:cseeeeereees 28 SCI Wind Rose: Time Weighted Average Wind Speed (knots)............::ccceeee 28 Power Curve, NM 225-30 (225 kW) Wind Turbine..............cccccceeeeeeeeeeeeeeeeees 31 SCI Diurnal Load and Wind Speed Overlay............:.ccccccseseseseseeeeseneseseseseeeseeeeeenens 32 System COE vs Number of Wind Turbines. ...........c.c.ccceseseseeseseeseeetsesesesesenseseees 38 Payback Period vs Number of Wind Turbines ..............ccccccssssseeseseseseseeeseseseeseeees 38 Internal Rate of Return vs Number of Wind Turbines ..............:c:cceseseseessseseseeeeeeee 39 iv FOREWORD This report was prepared as an account of work for others funding contract, sponsored by the Department of Defense (DoD) Strategic Environmental Research and Development Program (SERDP) under Department of Energy (DOE) Contract # DE-AC02-83CH10093. This report provides an overview of the wind resource, economics and operation of the recently installed wind turbines in conjunction with the with diesel power for the Naval Air Landing Field (NALF), San Clemente Island (SCD, California Project. The purpose of the SERDP and Federal Energy Management Program (FEMP) funded installation is to use wind power, a form of renewable energy, to decrease the Navy’s dependency on fossil fuel at San Clemente Island. Wind-powered electrical generation would allow a reduction in the current use of diesel-powered generators on the island. The primary goal of the SCI wind power system is to operate with the existing diesel power plant and provide equivalent or better power quality and system reliability than the existing diesel system. The wind system is intended to reduce, as far as possible, the use of diesel fuel and the inherent generation of nitrogen-oxide emissions and other pollutants. The first two NM 225/30 225kW wind turbines were installed and started shake-down operations on February 5, 1998. This report describes the initial operational data gathered from the February through January 1999, as well as the SCI wind resource and initial cost of energy provided by the wind turbines on SCI. In support of this objective, several years of data on the wind resources of San Clemente Island was collected and compared to historical data. The wind resource data was used as input to economic and feasibility studies for a wind-diesel hybrid installation for San Clemente Island. Timothy L. Olsen, an engineering consultant, was contracted by National Renewable Energy Laboratory (NREL) to assist with data reduction analysis, research historical wind resource data, perform wind-diesel hybrid analysis, and assist in the generation of this report. ACKNOWLEDGMENTS The authors wish to acknowledge major contributions to the success of this project by the following people: Ed Cannon of National Renewable Energy Laboratory (NREL), the Strategic Environmental Research and Development Program (SERDP) project manager, and Neil Kelley of NREL, the meteorological consultant. Bob Keller and crew of Mountain Valley Energy, who installed, commissioned, and helped operate the meteorological towers and instrumentation. Extensive typing, editing, and graph building was provided by Roni Olsen of Highline Editions. Scott Davis, the Utilities Supervisor at San Clemente Island (SCI), advised on aspects of SCI facilities, costs, operations, and detailed breakdowns for the SCI diesel plant operations and costs. Scott also manages daily wind energy project activities. More detailed diesel cost information was provided by Valre Kehler and Mike Stevens of Valley Detroit Diesel Allison. Additional support was provided by Brian Cable, the SCI project manager of Naval Facilities Engineering Service Center (NFESC); Ken Nicoll, past Division Director of Public Works Center (PWC); Diana Bendle, Assistant Operations PWC. Tom Brule, Division Director at PWC San Diego, and Joyce Sengpaseuth, the electrical engineer at PWC San Diego made the site available for study and arranged travel to and from SCI. Scott Miller of NFESC assisted with meteorological instrumentation, and the rest of the SCI Navy support staff assisted with this project in countless ways. And yes, one other that kept the SCI Wind Energy dream alive — Norm Groth, USN COGEN-SD Director. Each of these people deserves special thanks for their role in bringing this project together. vi EXECUTIVE SUMMARY In 1991, Congress authorized the Strategic Environmental Research and Development Program (SERDP) to help Department of Defense (DoD) meet their environmental obligations. The SERDP efforts included the use of alternative energies to reduce emissions. The long-term objectives of the U.S. Navy for San Clemente Island (SCI) are to install about 8 MW of wind capacity and to develop a pumped-hydroelectric storage system, using the ocean as the lower reservoir. SCI’s electrical system is powered with diesel generators, using wind energy to reduce the overall diesel system operating costs. To accomplish this mission, National Renewable Energy Laboratory (NREL), with the aid of Naval Facilities Engineering Service Center (NFESC), was charged with collecting wind resource data, and then providing the wind turbine generation system installations. The first two turbine installations were funded through DOE/SERDP, the third is being funded by the Department of Energy (DOE) Federal Energy Management Program (FEMP) funds, and the fourth will be funded through DoD Navy funds. This report summarizes the results of those tasks and the operational data learned to date. The 1995 through 1998 wind resource at the designated SCI Wind Turbine Site has an annual average wind speed of 6.1 m/s (11.8 knots) as measured by NREL and NFESC on a 42.7 m (140 foot) meteorological tower. Data were collected between August 1995 and January 1999 with several sections missing throughout. This work presents a study of the operation of a wind-diesel hybrid system using two wind turbines, along with predictions for an expansion up to 4 wind turbines. The study shows that wind energy can be cost effective in this application. As the third and fourth wind turbines are added, further savings are expected as the power plant can then run with fewer or smaller diesel engines. Further additions would start to see diminishing savings. Additional wind turbine installations may be limited at SCI, as the island has 7000 protected archeological sites and other Navy facilities have priority. Higher wind potential is available at the southeast end of the island, but that region is used for the bombing range and is off limits. Using two 225 kW wind turbines, the wind energy COE of $0.142/kWh helps reduce the wind- diesel hybrid system COE from the baseline $0.476/kWh to $0.461/kWh. This reduces system COE by 3.2%. The payback period is 6.5 years, the internal rate of return 14.4%. The four-turbine case had a wind energy COE of $0.139/kWh and a hybrid system COE of $0.447/kWh, saving 6.1%. The payback period is 6.3 years, the internal rate of return 14.8%. The COE for this case is relatively insensitive to annual average wind speed, varying 2.6% for a 17.5% change in wind speed. But the payback period is quite sensitive to wind speed, varying 28% to 63% for a 17.5% change in wind speed. As a preliminary review, this study used 1-hour average wind and load data for the hybrid system modeling to develop a general sense of economic tradeoffs. Dynamic load management should be addressed using load and wind data at shorter intervals (1 minute or less) to study system dynamics. vii 1.0 INTRODUCTION This report outlines and summarizes the local wind resource and evaluates the costs and benefits of supplementing the current diesel-powered energy system with wind turbines at the Naval Auxiliary Landing Field, San Clemente Island, California. This renewable electrical power generation provides a reduction of emissions from the diesel power plant. Specifically, the project began with two operational 225 kW wind turbines, and construction has begun on a third turbine, which should be installed and online by July, 1999. In Section 2.0 the San Clemente Island (SCI) site, naval operations, and current energy system are described, as are the data collection and analysis procedures. Section 3.0 presents the wind resource data and its analysis results, including historical wind speed averages, recent annual records, diurnal wind speeds, and annual wind roses. Sections 4.0 and 5.0 present the conceptual design and cost analysis of a hybrid wind and diesel energy system on SCI, with conclusions following in Section 6. Appendix A presents summary pages of the hybrid system spreadsheet model. Appendix B contains actual system operating data. Appendix C presents the results of a preliminary load and fuel analysis. Appendix D presents Wind-Diesel System Operational Guidelines developed by NREL and the RMH Group (Lakewood, Colorado). 2.0 BACKGROUND 2.1 San Clemente Island Installation Setting SCI is one of the Navy’s largest real estate assets and is among its most unique installations. SCI is the southernmost of the eight Channel Islands located off the southern California coast, lying approximately 89 km (55 mi) southwest of Long Beach and 135 km (84 mi) northwest of San Diego. The next nearest land-mass to SCI is Santa Catalina Island, lying approximately 40 km (25 mi) away between SCI and the mainland. SCI’s geographical center is 32° 54’N, 118° 29°W. The Island is approximately 34 km (21 mi) in length, with a land area of about 148 km? (57 mi’, or 35,540 acres), making it one of the larger Channel Islands shown in Figure 1. The rugged southern third of the island has an average width of about 6.4 km (4 mi), with the remainder tapering to 1.6 km 1(mi) across at the flatter and lower north end. SCI is considered the most biologically and historically distinctive coastal island owned by the United States. Because the island supports unique natural, cultural, and anthropologic resources as well as a variety of activities for Naval operations and training. The island, generally treeless, is relatively flat on top and drops off sharply on the east side with a more gradual slope to the ocean on the west side. The interior terrain is a rolling mesa, with little vegetation, mostly coarse grasses and few large shrubs. Its highest point, Vista View Point, is 592 m (1,943 feet) at the southwestern portion of SCI. The San Clemente Island wind turbine site, Figure 2, is located along Telemetry Road in the island’s north-central portion (33° 59’ N, 118° 53’ W). Prevailing winds on SCI are from the west and northwest and are moderate and steady most of the year. The average wind speed at the wind turbine site is 6.1 m/s (11.8 knots) and seasonal variation is small. Climate SCI’s climate is distinctly maritime, with cool summers and mild winters. Except for fog and overcast conditions and generally cooler year-round temperatures, the weather is similar to that of the southern California mainland coastal region [1]. Temperature One of the outstanding features of SCI’s climate is the narrow temperature range, with mean winter temperature just —12.2°C (10F) lower than the mean summer temperature. Mean annual temperature at the lower elevations is about 15.6°C (60°F), 16.6°C (62°F) at the higher elevations. Temperatures above 32.2°C (90°F) are rare, but occasionally when Santa Ana wind conditions occur between August and October, temperatures of 32.2°C (90°F ) and even 37.8°C (100°F) have been recorded. No temperatures below freezing have ever been recorded at the location of the air field station, but at the higher elevations such as Mt Thirst such temperatures appear to occur according to the Navy based public utility crews. Humidity High relative humidity is experienced throughout the year with an annual average of 78 percent. The exception occurs during Santa Ana conditions when the relative humidity is generally less than 25 percent [1]. Winds Gale force winds are common at higher elevations during the winter, but are infrequent elsewhere on the Island. Average wind speeds measured at the airfield are less than ten knots. The airfield sits at a low elevation next to a rise in the land, which can deflect or shelter the wind. Precipitation Annual precipitation averages just five to eight inches, with the majority falling between November and April, and the driest period being June to September. Snowfall has been reported at the highest elevations on the Island, Mt. Thirst and Mt. Vista, but in minimal amounts. Occasionally small hail accompanies the passage of strong storm fronts [1]. 2.2 Naval Installation Mission The Naval Auxiliary Landing Field (NALF), SCI serves a variety of weapons research, development, testing and evaluation activities, and a number of military training functions as well. It is used primarily by several major Naval tenant commands, but is also used by research divisions of government agencies and private companies working on government contracts. The Island is administered by the Commanding Officer, Naval Air Station North Island. SCTI’s relative isolation, restricted airspace, variable topography, adjacent deep seas and clear water conditions permit a great deal of flexibility in accommodating specific testing and training programs [1]. Figure 1: SCI Location Map Source: San Clemente Island Site Manual “a wenvus wip | ’ ; . Farmt60m) | I , = : a I} i i | t ‘ Meteoralogical Monitoring al 4 ‘ a Tower (Existing) e 4 c , o} \ &\| \ Proposed 3\) q Underground =|} § Utility Line || ; ° ; P g ‘ Praposed Vi : \ ‘ Staging Area \\ § ' \ ‘ ; Proposed See: ' i ; Substation —— ! : i { Northern Wind Farm ae sa | ae eee \ Study Area Bound ypOse orthern Wind Fa \ ‘ . Parco \ Study Area Boundary m\ Giver o \ 5 | north 2 <\\ i o \ S 3 2\ 2. | x &, S|) & es aaa 2 Bldg. 60256 oe ° 30 180 m June 21, 1995 Source: NFESC and SwDtv San Clemente island SERDP Wind Farm Location of SERDP Wind Farm within Northern Wind Farm Study Area Figure 2: SCI Wind Farm Location Map Source: USN NFESC and SWDIV Environmental Assessment Report 2.3 Energy Demand Energy production information follows in Table 1. The current (1998) average hourly electrical demand at SCI is 846 kWh; the hourly average peak is 1350 kW. The SCI electrical power system supplied 7.42 million kWh in 1998, up from 6.15 million kWh in 1996. Table 1: SCI System Demand Statistics San Nicolas Year 1996 1997 1998 ___ Island 1995 Peak daily demand (kWh) 22,400 25,900 26,788 N/A Low daily demand (kWh) 12,250 14,000 13,650 N/A Average hourly demand (kWh) 711 785 846 771 Average dailydemand (kWh) 17,075 18,838 20,320 19,275 Average monthly demand (kWh) 512,269 565,146 618,080 586,281 Peak monthly demand (kWh) 547,100 724,150 708,438 N/A Low monthly demand (kWh) 478,000 523,250 578,025 N/A Annual energy production (kWh) 6,147,230 6,781,750 7,416,959 6,753,000 Annual energy from diesel (kWh) 6,147,230 6,781,750 6,631,021 6,753,000 Annual fuel consumption _ (liter) 1,983,120 2,033,260 2,074,240 1,996,584 Annual fuel consumption (gal) 523,942 537,190 547,958 527,499 Energy / Fuel ratio (kWh/1) 3.10 3.34 3.58 3.38 Energy / Fuel ratio (kWh/gal) 11.7 12.6 13.54 12.8 Demand growth (%) - 10.3 9.4 ~ Source: SCI Public Works Center The load frequency distribution in Figure 3 shows predominant operation between 700 kW and 1000 kW, with peaks up to 1400 kW. Annual diurnal loads are shown in Figure 4. These figures are based on the composite data set using original 1998 loads data. Included with the data set are daily energy production and monthly fuel consumption. Annual records of monthly energy production and fuel consumption are shown in Figures 5 and 6, and the relationship between the two in Figure 7. San Clemente Island, 1998 900 800 700 600 + 400 + Hours per Year 300 + 200 + 100 + 04 400 475 550 625 700 775 850 925 1000 1075 1150 1225 1300 1375 1450 1525 1600 Load (kW) Figure 3: SCI Load Frequency Distribution San Clemente Island, 1998 Hourly Averages [+ x* von fet x [ Tit tax tt tx = I —e—mean = = = mean-std a 800 4 mean+std S 800 yee x x ex cl x min 400 x max 200 0 To s RS s es se Time of Day Figure 4: SCI Annual Average Diurnal Load Energy (kWh) 720,000 700,000 680,000 San Clemente Island, Feb 98 - Jan 99, Monthly Totals 660,000 640,000 620,000 600,000 - 580,000 560,000 + 540,000 + 520,000 Feb-98 4 4 + Jun-98 Aug-98 Month Figure 5: SCI Annual Energy Production Record Fuel (gal) 60,000 San Clemente Island, Feb 98 - Jan 99, Monthly Totals 50,000 40,000 30,000 20,000 Apr-98 Aug-98 Oct-98 Month Dec-98 Figure 6: SCI Annual Fuel Consumption Record San Clemente Island, Feb 98 - Jan 99, Monthly Totals 750,000 700,000 +* 7 650,000 o a = 600,000 +? e+ — 550,000 > © 500,000 FS i 450,000 400,000 350,000 300,000 30,000 35,000 40,000 45,000 50,000 55,000 60,000 Fuel (gal) Figure 7: SCI Energy Production vs Fuel Consumption 2.4 Diesel Energy System The existing SCI power plant consists of four diesel generators. The diesel generator plant is located in a sheltered cove about 3.2 km (2 mi) from the hill where the wind turbines are located. Power for the island grid is generated by diesel generators at 4,160 V and stepped up through two 2,000 KVA transformers to 12,470 VAC, three-phase, three-wire (Delta) for distribution on the island. The existing power plant and island power grid has the following electrical characteristics: Grid Voltage: 12.470 kV Frequency: 60 Hz +/- 1.0 Hz (1998 PIE-NREL Data Record) Power Factor: 0.8 — 0.95 lagging Average Load 846 kW (1998 SCI Data) Maximum Load 1350 kW (1998 SCI Data) Minimum Load 500kW (1998 SCI Data) eoeee#eee Diesel Generator Sets: Electrical power at SCI is presently supplied by four Navy-owned, 3- phase, 4160 V, diesel driven electric generators that are operated by the Public Works Center located at San Diego. The diesel plant on the island was rebuilt in 1994 as Building 60137. The SCI operating data for 1998 shows an average diesel fuel consumption rate of 236.8 I/h (62.6 gph), and average energy conversion rate of 3.2 kWh/I (12.1 kWh/gal, from Table 1). The engines’ specific diesel fuel rates are shown in Table 2. Table 2: Diesel / Generator Power Rating and Fuel Consumption Power Fuel Usage Manufacturer & Model Rating Full Load Diesel Generator kW Vhr (gal/hr) 1. EMD 8-645E1 KATO (A258730000, 720 rpm) 500 144 (38) 2. EMD 8-645E1 KATO (A258730000, 720 rpm) 500 144 (38) 3. EMD 12-645E4 KATO (A257780001, 720 rpm) 1,200 329 (87) 4. EMD 12-645E1 KATO (A258710000, 720 rpm) 750 216 (57) Unfortunately, there is little real data for the units operating at no load. The only information available is from EMD Power Products for its similar 12-cylinder, turbo-charged units during EMD factory testing. On units similar to 12-645E4, they recorded 57 liters per hour (15 gal per hour) at rated speed, no load. Based on this information, the hybrid system analysis presented later in this report uses a no-load consumption value of 17% of full load consumption. The other units probably would have no-load rates of 25% of full load, but the hybrid system model uses 17% because the turbo-charged unit is on the majority of the time. In the current operating protocol, online diesel capacity typically exceeds average demand by a substantial margin to ensure enough capacity to cover excursions and to avoid too frequent switching between diesels. The resulting excess margin causes the diesels to run below their ratings most of the time, resulting in lower energy conversion efficiency. A tighter margin would allow more efficient operation but the more frequent switching would cause faster wear on the diesels and more work for the operators. Fuel Supply System: Petroleum products are delivered to SCI by regularly scheduled barge and unloaded at Wilson Cove. Diesel fuel (DL-2) is delivered by barge to the fuel tank farm at the north end of Wilson Cove. Barges dispensing fuel tie up to a buoy and pump fuel directly into above-ground storage tanks. The DL-2 fuel is stored in the 37,854 liter (10,000 gal) above-ground tank located to the north of the power plant, Building 60137. Fuel is continuously circulated and centrifuged in this tank. Upon demand, fuel is automatically diverted from returning to the main storage tank and sent to day tanks located just outside the power plant instead. From the day tanks (one for each engine), fuel flows by gravity to each operating engine’s fuel pump. A 757 liter (200 gal) lubrication oil tank is located within Building 60137. Oil is added to each running engine via pumps or centrifuge. Each engine has a direct pipe connection to the lube oil centrifuge, and oil can be gravity-fed at this point or pumped in with pump 7. The plant is also provided with a waste-oil collection system. This system consists of one 1514 liter (400 gal) holding tank. The tank and pumps are located immediately outside the power plant and are equipped with secondary containment and interconnecting piping. Balance of Plant: The plant is operated 24 hours per day. Operators observe equipment operation, make hourly log entries, and start and stop the generators as required. The control room has been recently upgraded and is enclosed by sound-reducing insulation and double doors leading to the engine room. The station auxiliary equipment includes two150 kVA, three-phase, 4160-120/208-V station service transformer, a 120/208-V distribution panel board, a 20-battery 125-V DC station battery bank, and two 2,000-kVA, three-phase, grounded-wye-delta-connected grounding transformers. There is one grounding transformer for each bus in the switchgear to provide a neutral for single- phase, 2400 V-loads. The power plant switchgear, installed in 1994, has two buses with a vacuum circuit-breaker tie. The circuit-breaker tie will trip automatically in the event of a fault on either bus. In addition to the 4160-V generators, local emergency generators provide back-up power for critical loads. The power is generated at utilization voltage (120/208V or 480V) and is applied to the load through manual or automatic transfer switches. Distribution: Electricity is distributed throughout the island by three 12.4-kVA, 4160-V feeders. Feeder # D-Line serves most of the southern end of the island. Feeder # C-Line serves the north- central area of the island, including personnel living facilities, administration and recreational facilities, and the public works buildings. Feeder # A-Line serves the air terminal and associated hangars and maintenance facilities, and loads in the northwest part of the island. The SCI Air Field portion of the distribution (feeder # A-Line from pad A- 57-7) is completely underground. Feeders # C-Line and # D-Line use mostly overhead lines, consisting of wood poles supporting bare copper conductors. 2.5 Wind Energy Site Description The wind energy site is located along Telemetry Road in its north-central portion (33° 59 N, 118° 53 W), as shown in Figures 8 and 9. The highest elevation at this location is 222 m (730 ft). The SERDP site is surrounded by, and dominated by, open, undeveloped habitats. The older wind farm that consisted of six Jacobs 20-kW wind turbines is presently shut-down. It will be removed in 1999, It was constructed in 1987, and is positioned approximately 400 m (1312 ft) to the north of the current wind site at a lower elevation. Telemetry Road runs East-West through the wind site and a Navy building is located roughly 300 m (985 ft) to the south. There are no trees or other wind obstructions on the site, just light vegetation including grasses and cacti. Several low water tanks and buildings, including the Power Plant Island Utilities, are located to the north and on lower elevations from the wind turbine site. The nearness of the power plant minimizes power line distances to the wind energy site. This site has moderate winds throughout the year. Although more optimal wind site locations exist on the southern section of the island at higher elevations, this particular site was selected because it does not interfere with radar, communications, or other naval operations. It is close to the San 10 Clemente Island power plant (approximately 4.3 km or 2 mi), and it does not pose environmental or anthropologic constraints or interfere with Naval Weapons Testing Operations. Figure 8: SCI Naval Facilities and SERDP Wind Farm Site Source: USN NFESC Aerial Photograph 11 Figure 9: Aerial View of SCI Wind Farm Source: USN NFESC Aerial Photograph 2.6 Wind Energy System The wind turbine electrical power generation facility is composed of two NEG Micon Model 700- 225/40 wind turbines, now called the NM 225/30 model following the merger of Nordtank Energy Group and Micon. Shown in Figure 10, this wind turbine has a rotor area of approximately 700 square meters and a rated output of 225 kW. (The new "30" designation in the model name of the turbine reflects the approximate diameter of the rotor measured in meters, instead of the lower generator rating). The wind turbines start producing power at approximately 4m/s (9 mph) and continue producing power up to 25 m/s (56 mph). Turbine #1 is located at 32° 59.147 N by 118° 33.127 W, Turbine #2 is located at 32° 59.211 N by 118° 33.072 W and Turbine #3 will be located nearby, all at 223 m (730 ft) altitude. The third 225-kW wind turbine is under construction as this report is being written. 12 Nacelle: The base frame is designed as a self supporting, integrated welded steel plate construction which also supports the main shaft bearing, gearbox, generator, yaw system, rotor, etc. The integrated construction is hot-dip galvanized and makes up the bottom half of the nacelle cover. The upper half is made from lighter, hot-dip galvanized steel plate. Yaw System: The yaw system applies forced yaw by electrical gear drive over a cogged ball bearing ring with friction brake system. Rotor: The rotor consists of three blades manufactured by LM, type LM 13.4, fastened to a hub. The blade diameter is 29.6 m (97.1 ft) with a swept area of 688 m? (7407 ft’). The height to the blade tip in straight upright vertical position is 44.8 m (147.0 ft). Tower: The tower height is 28.7 m ( 94.2 ft), with a hub height at the center of the rotor of 30.0 m (98.4 ft). The tower weight is approximately 12,000 kg (26,455 Ibm), and has three layers of zinc silicate for protection from the island’s marine environment. (An alternative standard tower height for this turbine model provides a hub height of 36.0 m (118.1 ft), and is recommended for sites with a wind shear coefficient higher than 0.1.) Wind Turbine Control: The main wind turbine control panel is located inside the tower bottom, protected against weather and unauthorized access. Its function is to provide automatic cut-in of the generator to the SCI electrical grid and fault detection and wind turbine protection. This control panel has easy access to operate and control the wind turbine. This controller has displays with fault indicators to secure quick fault finding in case of a turbine stop condition. If the SCI grid fails, and then is brought back on line the wind turbines can be automatically re-started. These wind turbine controllers are under the supervision of the main wind turbine control computer located at the SCI PWC diesel power plant. Electrical: The MICON wind turbine is equipped with phase compensation which improves the power factor to 0.96 lagging. Over-voltage protection in case of lighting is provided in the control system. Soft cut-in is also provided by thyristors that limit the in-rush current to 1.3 times normal current. 3 Figure 10: SCI SERDP-Funded NM 225-30 (225 kW) Wind Turbines Source: USN NFESC Photograph 14 3.0 THE WIND RESOURCE 3.1 Wind Data Collection and Analysis In July 1994, the National Renewable Energy Laboratory (NREL) entered into a cooperative agreement with the Naval Facilities Engineering Service Center (NFESC) to collect one full year of high quality wind energy resource data at San Clemente Island (SCI) old Jacobs wind turbine facility, (Tower #6) at 18.3-m (60-ft) height. Three additional UNR-ROHN 43-m (140-ft) towers were installed by NREL crews at SCI sites Met2: 32° 59.236N by 118° 33.209W (at the present 450 kW wind turbine site), Met3: 32° 58.630N by 118° 33.977W (approximately 1 mile south of Met2), and Met4: 33° 01.248N by 118° 33.041W (Lemon Tank Reservoir). We examined the Met2 data in detail, and reviewed historical summary data to describe long-term wind characteristics. The new data were collected through a full wind-energy meteorological sensor system including two anemometers, two wind vanes, a temperature probe, and a barometric pressure sensor. The anemometers were mounted 24.4-m (80-ft) and 42.7-m (140-ft) high at the new wind energy site on tower Met2. Data collection began in August 1995, and continued through January 1999. All data was sampled at 1 Hz and then stored as 10-minute and 24-hour averages. The 10-minute average data was used for this report. Annual records of the 10-minute average wind speed, and the monthly records use daily averages. An annual record is derived for air density using p=p/(R*T) where p is density, p is pressure, T is temperature, and R = 0.286 kJ/(kg*K) for air. Then wind power density is derived using P/A=0.5*p*V> where P is power, A is area, and V is wind speed. Using hourly average data, the diurnal wind speeds are created by computing an average for each hour of the day over all days in the period. Wind direction data is difficult to present because the most common directions do not necessarily have the strongest winds. Therefore, this report includes three types of wind roses: percent time at each direction, average wind speed at each direction, and time-weighted average wind speed at each direction. 3.2 Historical Wind Data This section begins with a review of 19 years of wind-speed data (1960-1978) at SCI station number 93117, compiled by Pacific Northwest Laboratories and archived by the National Climatic Data Center [2]. Historical annual average wind speeds follow in Figure 11. 15 The airfield began its operations in 1960 and the historical 19-year anemometer locations changed several times for this collection of historical wind data, and used different sensors, mountings, heights, exposures, and possibly drifting calibrations. Readings on the historical data were made 24 times a day after the first 3-years, which were read 5 to 11 times a day. The heights varied from 5.2 m to 7.9 m, so each year’s data were adjusted to the wind turbine hub height [30.0 m (98.4 ft)] using the 1/7 power law. These low measurement heights are very susceptible to the effects of obstructions. The average 19-year wind speed at SCI adjusted to the 30.0 m (98.4 ft) height is 4.0 m/s (7.8 knots) based on annual averages of hourly data, and the average of the annual standard deviations is 2.6 m/s (5.1 knots). The standard deviation of the annual averages is 0.7 m/s (1.3 knots), giving a variability of 0.7 / 4.0 = 0.175, or 17.5%. Although confidence in the average wind speed is low, this variability implies that the annual average wind speed will fall within +/- 53% (3 standard deviations) 99% of the time, assuming these values are normally distributed. Some bias toward lower wind speed measurements is expected because of low heights, proximity to buildings and other obstructions, and possible binding (bearing or shaft roughness) of older anemometers. The airfield’s altitude is 55.5 m (182ft). The ASR-8 Radar hill with an east-west ridge peaking at 160 m (524 ft) south of the airfield is approximately 2,300 m (7,500 ft) away. Although the ridge does not shadow the prevailing north to northwest winds, it can deflect them upward and cause lower measurements below. Winds from the northeast to southwest are sheltered. Because these factors are not tractable, no attempt is made to adjust the data to account for them. However, the averages found here are not used for the hybrid system modeling later in this report, but the interannual variability of 17.5% is used for the sensitivity analysis. Annual Averages, Scaled to 30.0 m (98.4 ft) 10 : +18 8 + 16 _ | = n 2 7 74s £ 6 ti2 & 3 Ss 3 54 +10 @ a o no a 3 44 — 8 ij Ts £ 3 +6 £ = = 2H « 44 +2 04 1. 1 1 1 1. 4 1 1 1 1 1 : h 1 1 1. . LQ 60 62 64 66 68 70 72 74 76 78 | Year Figure 11: SCI Historical Wind Speeds 16 The following figures, 12 and 13, show some average diurnals from 1983-4 and 1984-5 collected at the old Jacobs site [3]. The data was collected with a MAXIM type 40 anemometer at 32 ft elevation and a Second Wind Datalogger Model A1-2002-4K. The average wind speed for both years at this site was 6.1 m/s (13.6 mph) at 9.8 m (32 ft) height, which would indicate a speed of 7.2 m/s (16.0 mph) at 30.5 m (100 ft) using the 1/7 power law. Wind Speed (mph) at Year 0000 0200 0400 0600 0800 1000 1200 1400 1600 1800 2000 2200 1983 0200 0400 0600 0800 1000 1200 1400 1600 1800 2000 2200 2400 Aug 11.6 11.5 10.5 9.4 8.9 10.0 13.5 18.0 19.8 19.6 17.6 13.4 13.7 10.9 Sep 11.5 10.6 10.3 9.1 9.8 11.1 13.6 16.5 18.3 17.9 15.5 12.3 13.0 9.2 Oct 11.9 98 98 9.5 8.3 8.1 10.0 12.9 15.1 14.5 143 124 114 7.0 Nov 13.6 12.9 12.5 12.0 11.6 12.9 13.4 15.1 16.4 16.1 15.6 15.1 13.9 4.8 Dec 13.5 13.8 12.9 10.6 10.8 12.1 14.4 14.6 13.9 14.3 13.4 12.9 13.1 4.0 1984 Jan 10.5 10.8 10.3 9.7 85 9.5 9.8 11.9 15.1 15.9 15.1 12.1 116 7.4 Feb 14.6 13.4 11.6 11.6 11.8 13.0 13.8 15.6 17.6 18.1 169 15.5 145 65 Mar 13.3 12.1 9.8 9.5 11.1 12.3 14.4 17.0 20.4 20.6 18.9 15.9 14.6 11.1 Apr 17.4 14.9 13.6 13.0 13.9 16.8 20.5 24.1 24.4 22.8 20.8 18.9 184 11.4 May 12.3 11.5 10.5 9.1 10.3 11.4 14.0 17.8 18.4 19.2 16.8 14.8 13.8 10.1 Jun 11.1 94 89 85 86 9.9 11.9 15.9 19.0 18.6 17.0 13.5 12.7 10.5 Jul__ 13.9 12.5 12.4 10.9 9.9 11.4 17.4 20.8 19.0 17.6 17.8 16.8 15.0 10.9 San Clemente Island, California Wind Speed (mph) 2@ eS p B °o °o °o o oo > °o > Pp 4 8 8 10 12 14 16 18 20 22 24 1983-1984 Time of Day Averages Figure 12: Annual Average Diurnal, Jacobs Site, 1983-4 17 Wind Speed (mph) at Avg Diurnal Year 0000 0200 0400 0600 0800 1000 1200 1400 1600 1800 2000 2200 WS_ Swing 1984 0200 0400 0600 0800 1000 1200 1400 1600 1800 2000 2200 2400 (mph) (mph) Aug 11.4 10.5 9.6 8.6 8.2 10.0 13.2 17.2 19.0 18.2 15.9 12.9 12.9 108 Sep 12.5 12.2 11.4 11.2 11.2 11.9 14.4 17.1 18.5 18.4 17.5 15.6 143 7.3 Oct 12.8 12.5 11.4 11.2 12.2 13.8 14.5 17.1 19.4 18.9 15.5 126 143 82 Nov 12.1 12.2 11.5 11.2 11.2 12.2 13.8 14.2 14.9 13.6 13.1 13.1 12.8 3.7 Dec 10.8 9.9 10.5 10.9 10.8 11.2 10.8 11.1 13.1 13.1 12.6 11.2 11.3 3.2 1985 Feb 10.5 10.1 10.1 9.6 9.9 10.8 11.5 13.1 13.0 12.9 122 11.9 11.3 3.5 Mar 14.4 13.8 12.9 12.4 12.9 14.2 16.8 18.5 19.6 18.6 16.6 15.1 15.5 7.2 Apr 13.6 11.4 11.1 10.5 10.8 12.4 16.0 17.9 19.4 18.2 15.9 14.2 14.3 8.9 May 12.4 11.8 10.5 9.6 10.8 10.8 13.6 18.4 20.2 18.9 16.2 14.2 14.0 10.6 Jun 10.2 96 89 88 9.0 9.8 12.5 14.4 16.9 17.0 13.5 11.1 11.8 8.2 Jul_ 13.1 11.6 10.2 10.1 10.1 11.5 14.5 18.1 20.8 21.9 18.4 15.5 14.7 11.8 San Clemente Island, California Wind Speed (mph) & > 10 12 14 18 18 20 22 24 1984-1985 Time of Day Averages Figure 13: Annual Average Diurnal, Jacobs Site, 1984-5 18 3.3 Current Wind Data Data were collected between August 1995 and January 1999 at the 43.6-m (140-ft) meteorological tower number 2 at the designated SCI Wind Turbine Site. The wind speed data was collected at 43.6-m (140-ft) height and temperature and pressure at 3-m (10-ft). The data collection rate was about 75%, with several gaps spread throughout the data sets. This low collection rate is attributed to a lack of available staff for checking, downloading data, and maintaining the site data acquisition systems. In addition, the data shows error rates of 5 to 10%. Because no year has a full data set, a composite 10-minute data set was created to use for generating a wind speed histogram, annual diurnal, wind rose, and hourly data for hybrid system modeling. Using 1998 as the baseline data set, both missing data and bad data segments were filled in with good data segments from the other years as itemized in Table 3. Table 3: Source Years for Composite Wind Data Set Julian Source Day Year 1-31 1999 32-36 1996 37-238 1998 239-273 1995 274-284 1998 285-287 1995 288-365 1998 Statistical analysis of the last 3-years of daily meteorological data yielded the results shown in Table 4, and a full wind speed distribution is presented in Figure 14. Subsequent hybrid system modeling used the composite data set adjusted to wind turbine hub height. The 10-minute data set was not used here because of the amount of manual processing required to remove bad data segments. However, the 10-minute data would indicate somewhat higher standard deviations of 3.3 m/s and lower minima of 0.0 m/s (both affected by bad data), with maxima reaching 25.9 m/s. Table 4: Summary of Current SCI Meteorological Data Standard Channel Units Average Deviation Minimum _ Maximum Wind Speed, 1996 m/s 6.1 Past} 1.8 16.9 Wind Speed, 1997 m/s 535 2.6 1.4 18.0 Wind Speed, 1998 m/s 6.6 7ATL 23 17.0 Wind Speed, composite m/s 6.4 ZT 2.3 17.0 Wind Speed, composite knots 12.4 5.2 4.5 33.0 Ambient Temp, 1998 oC 14.4 DES 8.2 22.8 Ambient Pressure, 1998 mbar 990 333 981 1002 Air Density, 1998 kg/m? 1.20 0.01 1.17 1.24 Power Density, 1998 W/m? 267 400 7.6 2990 19 San Clemente Island, Site 2, 42.7m (140 ft) 1998, Histogram 700 600 « 500 4 3 = 400 | o Qa Y 300 s ° = 200 | 100 | 0+ 012 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Wind Speed (m/s) Figure 14: SCI Wind Speed Frequency Distribution Annual records using monthly averages have been plotted for wind speed, ambient temperature, ambient pressure, air density, and power density. The source data were derived from NREL testing on SCI at 42.7-m (140-ft) on tower Met2 for the August 1995 through January 1999 period. Missing and bad data segments were removed from the daily and monthly-averaged data for these records. Wind-speed records for 1995 through 1999 appear in Figures 15-19, and records for other meteorological parameters in 1998 are shown in Figures 20-23. Wind speeds are fairly consistent at this site; no months stand out as significantly higher or lower between the 3-years examined. The wind speed range generally falls between 5 and 8 m/s. Temperature and pressure cycle gently with more warmth and lower pressure in the summer, causing slightly lower summer densities. Power density looks like an exaggeration of wind speed, as expected from its cubic relationship. 20 San Clemente Island, Site 2, 42.7 m (140 ft) 1995, Monthly Summary of Daily Averages 20 18 35 16 —— 7 = 7° = —e— mean 14 E £ x ox = mean-std 3 ° loo 8 4 meantstd 10 ® . 2. 2 x min a 8 ¥ 411459 3 4 3 x max £ 6 x £ = 7 40S knots 4 24 J ‘ 75 0 + + + + + + + + + + + 0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Month Figure 15: SCI Monthly Averaged Wind Speed, 1995 San Clemente Island, Site 2, 42.7 m (140 ft) 1996, Monthly Summary of Daily Averages 20 18 35 x 16 +x 130 => P Z 30 8 —¢— mean £ 7a x < x 2 x = mean-std 3 * 3 eel i x ‘aer 3 4 meantstd a & 8 i a 4 x a ; t ie S x min eee Pre Bor mex = ° : Se +10 5 knots 4 x k 2+ x } * ; i } ° 0 + + + + + + + + + + + 0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Month Figure 16: SCI Monthly Averaged Wind Speed, 1996 21 San Clemente Island, Site 2, 42.7 m (140 ft) 1997, Monthly Summary of Daily Averages 20 18 * 35 16 D4 x x 7 30 s —*—mean £ EZ 425 & = mean-std t—* x 3 io i . 7 | 20 g 4 meantstd a a A a x min ef ea ee es Eee 3 x max £ AN = : ™ +10 5 knots s 2+j & t 5 X } ; x x t ; 5 0 + + + + + + + + + + + 0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Month Figure 17: SCI Monthly Averaged Wind Speed, 1997 San Clemente Island, Site 2, 42.7 m (140 ft) 1998, Monthly Summary of Daily Averages 20 18 35 16 x t a 14 * ¥ 7 30 3 —e— mean £ a ¥i26x = mean-std Ss 3 i y Gas (EE t eo 8 a meairstd BD 5 K ee ee ee iain z Tite i 3 x max £6 +——4— £ = + i +10 knots ® * 2+ etn ci Rls 0 +—+ + + + + 0 Month Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Figure 18: SCI Monthly Averaged Wind Speed, 1998 22 San Clemente Island, Site 2, 24.4 m & 42.7 m 1998, Daily Averages | | 20 | 18 — — - a 35 > 16 4 - 0 ‘e144 e = $252 3 12 3 £ 10 7 20 @ n aS 3 8 41459 | ¢ 3 | | £ 6 £ | s 10 5 | 44 24 +5 ot - 0 oownoeoewnon#sdwn 8 mwondondoewndeoexwvn#ene¢gwnwvws?se - OFT ORBDAN oDODOr NTNYORKR DOH YO 3 oO Ferrer Ff FTF FAA KSA A SS SS © & oO Julian Date Figure 19: SCI Daily Averaged Wind Speed, 1998 Temperature (deg C) SR RGDS oN F&F D OO Figure 20 San Clemente Island, Site 2, 42.7 m (140 ft) 1998, Monthly Summary of Daily Averages Jan Feb Mar Jul Aug Sep Oct Nov Dec Apr Month May = Jun : SCI Monthly Averaged Temperature, 1998 23 Figure 21: SCI Monthly Averaged Pressure, 1998 1998, Monthly Summary of Daily Averages San Clemente Island, Site 2, 42.7 m (140 ft) San Clemente Island, Site 2, 42.7 m (140 ft) 1998, Monthly Summary of Daily Averages 1000 996 | ——___— —— —— a 5 990 4 - ~ — iS oa 5 985 — — no n © so | — - — a 975 + 970 + + + + + + + + + Jan Feb Mar Apr May — Jun Jul Aug Sep Oct Nov Dec Month 1.30 1.28 — —— — 1.26 ——— - —_— 1.24 —_—____—_ — — — 1.20 + 1.18 1.16 4 1.22 +— —_—. —__— ——__— Density (kg/m‘43) 1.14 4AZ 1.10 Jan Feb Mar Apr May — Jun Jul Aug Sep Oct Month Figure 22: SCI Monthly Averaged Density, 1998 24 San Clemente Island, Site 2, 42.7 m (140 ft) 1998, Monthly Summary of Daily Averages 500 450 — ——— ——————————————E———————————— 400 350 300 250 +— 200 +— 150 +— 100 + 50 + b Mar Jul Aug Sep Oct Nov Dec Apr May — Jun Power Density (W/m“2) Jan Fe Month Figure 23: SCI Monthly Averaged Wind Power Density, 1998 The annual average diurnal given in Figure 24 shows a stable pattern, with wind speeds falling between 5 and 7.5 m/s. They are slightly lower through night and morning, and slightly higher through the afternoon and evening. The diurnal is derived from the 1-year composite hourly data set described earlier. Each hour is averaged through the whole year; any specific day could be quite different. For reference, the column labeled “0000” refers to the first hour of the day: 0000 to 0100. The wind roses shown in Figures 25-29 also use the composite hourly data set. They indicate prevailing winds from the west and west by northwest, with somewhat stronger average wind speeds in these directions as well as in the northwest, southwest, and south. 25 San Clemente Island, Site 2, 42.7m (140 ft) 1998, Hourly Averages 25 x + 50 3 20 * OK KOK x 3 o x x x o aa 15 xx tf Bx Rf 40 8c O's +30 7 2 oS 10 Tx £ p20 = = 54 i} 10 5 04 0 © © © © © © S S sS ss S ¥ & xv Co Time of Day —e— mean = mean-std 4 meantstd x min max Figure 24: SCI Annual Average Diurnal Wind Speed True Source Direction, 1998-Composite San Clemente Island, 42.7 m (140 ft), Met2 Figure 25: SCI Wind Rose: Percent Time 26 True Source Direction, 1998-Composite San Clemente Island, 42.7 m (140 ft), Met2 Figure 26: SCI Wind Rose: Average Wind Speed (m/s) True Source Direction, 1998-Composite San Clemente Island, 42.7 m (140 ft), Met2 Figure 27: SCI Wind Rose: Average Wind Speed (knots) 27 True Source Direction, 1998-Composite San Clemente Island, 42.7 m (140 ft), Met2 Figure 28: SCI Wind Rose: Time Weighted Average Wind Speed (m/s) True Source Direction, 1998-Composite San Clemente Island, 42.7 m (140 ft), Met2 Figure 29: SCI Wind Rose: Time Weighted Average Wind Speed (knots) 28 4.0 WIND-DIESEL HYBRID ENERGY SYSTEM The San Clemente Island (SCI) hybrid-energy system, consisting of combined wind and diesel generators is proving to be economically and environmentally advantageous for SCI. A study of such a system was conducted using a spreadsheet program to compare the cost of power generation for the diesel-only baseline to several wind-diesel hybrid cases. The hybrid cases were compared to determine the most cost-effective number of wind turbines. The wind-diesel hybrid system is relatively simple. Two commercially available wind turbines (total capacity of 450-kW) are combined with the existing 2950-kW diesel generation, a third wind turbine (total wind capacity of 675-kW) will be online July 1, 1999, and a fourth is scheduled for 2001 (total wind capacity of 900-kW). With a demand peak of 1350-kW, no more than 1700-kW of diesel should be necessary at any time. Therefore, wind penetration of “on-line” capacity with two wind turbines is 450/1700 = 26%, and four wind turbines is 900/1700 = 53%. Based on instantaneous power, wind penetration can range from 0 % when there is no wind to 180% when peak wind power of 900-kW is combined with a minimum load of 500-kW. Several assumptions were made regarding the power that can be generated by the wind. First, at least 200-kW always must be generated by the existing diesel generators, even if there is excess wind capacity. Second, it is assumed that only the necessary number of turbines will generate power at any given time, with the remaining turbines idled. 4.1 Diesel System Operation As mentioned in Section 2.3, there are four diesel generator sets, two generators rated by the Navy at 500-kW, one at 750-kW, and one at 1200-kW. One generator of each size is included in the hybrid-system model. The generator fuel/energy curves were given in Section 2.3. Typically, only one diesel is run at a time, unless the island’s electrical demands require more than 1000-kW. Then two diesels are generally on line to provide the electrical capacity required. The latest SCI power demand for 1998 ranges from a minimum of 500-kW to a maximum of 1350-kW. The fuel needed (with no wind-energy input) is calculated based on minimizing the number and rating of operating diesel generators. Configuring the diesels to produce 500, 750, 1200, or 1700-kW can meet the power demand. The diesel generators follow the load automatically through speed and frequency monitoring. The diesels have no specific selection priority, but there are other constraints. At least one diesel must be on line at all times to ensure reliable capacity and system stability; the present minimum operating load for the diesels are set at 200 kW, or 40% rated power for the smallest unit. In the future, this may be set to 100 kW. Also, according to SCI power system operating data from 1996 to 1998, the manual operating scheme tends to favor running the large engine for long periods of time, as estimated from the Navy-supplied data. That reduces the number of starts. An optimized operating scheme alone 29 could provide significant fuel savings, but it would require many more diesel starts and some form of automated system control. For this study, the spreadsheet model follows the actual manual operating scheme for all cases, wind and baseline. The hybrid wind-diesel systems likely would show greater savings over the baseline with such an optimizer. 4.2 Wind Turbine Operation The wind generation system modeled uses between 1 and 4 commercial wind turbines rated at 225- kW each. The sea level power curve for this turbine is shown in Figure 30. A fifth-order polynomial was fit to the curve for use in the spreadsheet model. No density correction was made to the power curve, as the present San Clemente Island wind site is only 700-ft to 750-ft above sea level. The wind turbines can be curtailed (shut down) as necessary when excess wind energy is available. The net annual energy production (AEP) can be computed by multiplying the power production level by the number of hours for each wind speed level and summing the results. If Pj is power and N; is number of hours at each wind speed, then: AEP = sum (P; * N;), i=0.0, 0.5, 1.0, ... 100.0 m/s. The actual AEP is often lower because of various system losses. Assessment of the wind site showed that there are not any significant obstructions to the prevailing wind flow. Also, there is room for additional wind turbines without interference, so array losses should be mitigated with proper siting. Other sources of loss could include 1 - 5% availability loss for operation and maintenance, up to 5% for blade soiling losses, up to 2% for turbulence losses, and up to 3% for control, grid, and collection system losses. Using 97% availability, the combination of these sources is significant, having a possible net loss of 11.5%. However, the first year of operation for the two SCI wind turbines has generated power curves that match the manufacturer’s published curve. Therefore, no losses have been applied in the energy system model for this report. 30 Power Curve Data, 225 kW Wind Turbine 250 T = 200 - = $ 150 ° a = Data o 5 — Poly. (Data) o 2 Ww 50 4 10.0004x® + 0.038x‘ -|1.296x° + 18.602x? -|89.524x + 136.5 0 0 5 10 15 20 25 Wind Speed (m/s) Figure 30: Power Curve, NM 225-30 (225 kW) Wind Turbine 4.3 Wind Energy vs Load Profile Correlation The hybrid system model used hourly load data derived by combining hourly diesel production data with wind production based on hourly wind speed data. The evidence of the power production statistics in Section 2.3 indicates the loads at SCI grew 10% per year from 1996 through 1998. Several new buildings and facilities would indicate a more substantial load increase in 1999. The load frequency distribution was shown earlier in Figure 3. Short-term load variability is defined as 0.135 based on the following rationale: the average load, 846 kW, gives 1 sigma (one standard deviation) = 114 kW and 3 sigma = 342:kW. These values would require very large operating margins. However, the change in demand from one hour to the next is lower using the derived SCI hourly demand data, with an average of 48 kW and a peak of 287 kW. These fluctuations coincide with operating experience, which has demonstrated about 50 kW normal fluctuation and an occasional 100-kW to 300-kW demand step or spike. The hybrid spreadsheet model accommodates this fluctuation by reserving a 300-kW margin of diesel capacity above the net demand for each 12-hour period. The hourly wind speed averages from 1998 composite data set was used in the hybrid-system model. Both wind and load data were arranged to a calendar year to assure proper synchronization. The wind frequency distribution was presented earlier in Figure 14. Annual diurnal wind speed and load are overlaid in Figure 31. Both the diurnal load and wind speed are relatively steady, 31 with the diurnal wind speed cycling between 5.4 and 7.7 m/s, giving a somewhat neutral correlation. San Clemente Island, Hourly Averages 1000 10 900 HMM ye Se 9 L x — Ky seul JU aoe xf 800 f=x—»—* 8 700 a = at OE 600 6 = 3 o> 500 5 |e 6 no S 400 | ML jaed 4 3 300} —= Wind Speed 3 5 200 2 100 | 44 0 1. J. 1. J L 1 a. 1. 1. i abe i J. 1 1 1 1 1 1 1 1 1 0 0000 0400 0800 1200 1600 2000 Time of Day Figure 31: SCI Diurnal Load and Wind Speed Overlay 4.4 Energy System Operating Results Using the first year’s operating data with two 225 kW wind turbines (February 1998 through January 1999), the diesel energy production and fuel consumption fell by 11.2%. More detail is given in the monthly power plant production logs in Appendix B. The wind turbines were not fully utilized during this time because of initial startup and adjustment activities and subsequent grid problems and construction activities. 4.5 Hybrid System Spreadsheet Model The hybrid system model uses the existing wind-diesel system plus new wind generation; the load data are from SCI 1998 operating data and the wind data are from the NREL/SCI 1996 through 1999 measurements. The spreadsheet model starts by calculating a diesel rating that covers the load demand with sufficient margin to ensure a minimum diesel run time of 12 hours and capable of handling 400-kW excursions. Diesel consumption for a diesel-only baseline system is calculated next, based on demand and efficiency. Finally, the number of diesel starts and run time are computed. 32 The wind-diesel hybrid section follows by calculating the power produced by a single wind turbine for each hour of the year. Then it calculates the optimal wind power usage by choosing the greatest number of turbines to operate without exceeding demand, and while maintaining at least 200 kW of diesel energy online. This wind power, when subtracted from the demand, reduces the amount of power required from the diesel generators. Only in very low or very high winds are the diesel power demand unchanged. Diesel fuel consumption is then calculated from this net demand and fed into the fuel savings over the baseline system. Four different cases of the wind-diesel hybrid system were examined. The results are summarized in Table 5. In the first case, just one 225-kW wind turbine was added to the baseline diesel system, the second case two 225-kW turbines, and so on, up to four 225-kW wind turbines. The minimum and maximum net loads (demand minus wind power) are 0 kW (loss of grid) and 1472 kW for all cases. The number of diesel starts is determined by incrementing a counter every time the diesel demand changes. The diesel run time is 8760 hours (1 diesel all year), plus the number of hours at 1700 kW capacity (2 diesels on). The diesel-only case required 100 starts and 15,491 engine-hours of total run time for the year. Four 225 kW wind turbines reduce diesel energy production by 23.1% and fuel consumption by 17.0%. Fuel savings fall below energy savings because the high wind variability necessitates greater diesel capacity running at somewhat less efficient conditions. However, these fuel savings could be improved significantly if the diesel usage was optimized, but at the cost of starting and stopping the engines much more frequently. 33 Table 5: 1998 Hybrid Systems vs Baseline: Spreadsheet Model Results Hybrid Results Baseline 1 Wind 2 Wind 3 Wind 4 Wind Diesel Turbine Turbines Turbines Turbines Parameter Units Only 225 kW. 225 kW 225 kW 225 kW Average WS, lyr m/s 6.1 6.1 6.1 6.1 6.1 Average Load, lyr kW 870 870 870 870 870 Run Duration hour 8,760 8,760 8,760 8,760 8,760 Avg Net Diesel Load (1), kW 870 817 763 711 669 Energy Demand, lyr MWh 7618 7618 7618 7618 7618 Diesel Energy, lyr MWh 7618 7153 6687 6232 5856 Wind Energy, Used MWh 0 466 931 1386 1762 Wind Energy, Unused MWh 0 0 0 10 100 Diesel Energy % 100.0 93.9 87.8 81.8 76.9 Wind Energy % 0.0 6.1 12.2 18.2 23.1 Wind Energy Incremental Turbine % 0.0 6.1 6.1 6.0 49 Wind System Capacity Factor % (2) n/a 23.6 23.6 23.4 22.3 Wind Sys Inctl Turbine Cap Fac% = n/a 23.6 23.6 23.1 19.1 Fuel Usage kltr 2689 2568 2447 2329 2232 Fuel Usage % of base 0.0 95.5 91.0 86.6 83.0 Fuel Saving kltr 0 121 242 360 457 Wind Energy COE $/kWh n/a 0.152 0.142 0.138 0.139 Effective Wind COE (3), $/kWh n/a 0.361 0. 353 0.350 0.351 Hybrid System COE $/kWh 0.476 0.469 0.461 0.453 0.447 System COE Saving $/kWh 0.000 0.007 0.015 0.023 0.029 System COE Saving % of base 0.0 1.5 3.2 4.8 6.1 Payback Period year n/a 7.00 6.49 6.28 6.32 Internal Rate of Return, % n/a 13.1 14.4 14.9 14.8 Notes: (1) “Net Diesel Load” means net power required from the diesels, or system load minus useable wind power. (2) Wind System Capacity Factor = Wind Energy [MWh] / (#turbines*rating[0.225MW]*8760[h)). (3) Effective Wind Cost of Energy (COE) = (hybrid COE*energy demand - baseline COE*diesel energy) / wind energy. (4) All other values derived from spreadsheet model results, Appendix A. (5) For efficiency, this table includes some of the economic results developed and discussed in the next section. 34 5.0 COST ANALYSIS 5.1 Methodology After estimating 1998 operating costs for the four cases of the wind-diesel hybrid system and for the baseline diesel-only system, the resulting levelized costs of energy (COE) were compared. Also, payback periods were computed for the four cases of hybrid system investment. COE is derived using COE = NPV * CRFI/ AEP, where NPV is the total net present value of all system costs, CRFI is the capital recovery factor for system income, and AEP is annual energy production (system load). A simple payback period is calculated by dividing the total initial capital cost by the annual savings from system operation, which includes the difference in fuel, overhaul, and operations and maintenance (O&M) costs between the wind-diesel hybrid and baseline systems [4]. Economic assumptions included 2% general inflation, 2% fuel inflation, 6.9% discount rate, 20 year system life, and 100% down payment on new investment. Although new wind turbines will start with a 20-year life, the existing diesel systems have been in service for several years and have limited lives of their own. This is covered by a fund for major diesel overhauls. It was further assumed that no additional labor would be required to operate the wind-diesel hybrid plant beyond that already-assigned to operate the existing diesel power plant. 5.2 Diesel System Costs The diesel system operating costs are derived partly from San Nicolas Island cost data because the San Clemente Island information is incomplete. Fuel costs are based on various memos, email and verbal conversations with the PWC. Since a full breakdown of SCI power system costs was not available, we resorted to the rate SCI charges its customers: $0.390 / kWh, which gives $2,971,205 for 7,618,475 kWh. The inherent assumption is that this rate reflects true and total life-cycle costs for the SCI power system without profit, since its customers are other Navy entities and their subcontractors. We suspect the true diesel system costs are lower, but don’t have any other basis to work with at this time. The fuel price also is known, at $0.206/liter ($0.78/gal), and adding transportation-and other hidden costs bring the total fuel cost up to $0:264/liter ($1.00/gal). That translates into $0.082 / kWh-for fuel using the baseline fuel and energy totals for 1998. The remaining amount of $0.308 / kWh is included in the O&M item in the economic analysis spreadsheet, but it must cover O&M, diesel overhauls, and eventual replacement. However, some of these costs are fixed and part variable. We will assume they split half-and-half, based on experience with similar facilities [5]. Therefore the variable part is $0.154 / kWh, and the fixed part is 0.5 * (0.308*7,618,475) = $1,173,245. 35 5.3 Wind Energy System Costs Wind-diesel hybrid system costs include the baseline costs as given above, plus new costs associated with the wind turbines and interconnect and control equipment. The interconnect and control equipment are included with the wind turbine balance of station (BOS) costs, along with foundations, installation, spare parts inventory, site surveying and preparation, O&M facilities and equipment, permits and licenses, project management and engineering, and construction insurance and contingency. Initial capital costs including BOS are detailed in Table 6. If multiple turbines were installed at one time, per-unit turbine price and BOS costs would drop considerably, but we have not applied this adjustment. Also, it may be possible to further reduce installation and operation costs by utilizing Navy heavy equipment (such as a crane). Each 225kW wind turbine costs $220,000, plus $25,000 shipping from Denmark to Los Angeles then to SCI. An additional $450,510 is required to cover BOS costs. Thus, the total capital cost required for four wind turbines is $2,742,040, when installed one at a time. If all four turbines were installed together, the overall cost would be about $2,500,000. Overhaul costs are fixed at an annual $1000 per wind turbine, regardless of turbine usage. Actual wind turbine O&M costs of $0.005/kWh are doubled to $0.01/kWh to account for the small system size and the extra burden SCI represents with its remote setting. As implied by its units, this O&M cost is variable, or fully dependent on wind turbine usage. These amounts are based on a working system using the 225 kW wind turbine. Table 6: Initial Capital Costs for One Additional 225 kW Wind Turbine Mainland Extra Cost for Total SCI Item Cost SCI Access Cost 1998 225kW Wind Turbine $220,000 $25,000 $245,000 Turbine installation on SCI 75,000 67,180 142,180 225kW Service Parts 3,000 400 3,400 Turbine Maintenance / Warranty 2,000 2,730 4,730 Turbine Siting 3,100 2,000 5,100 Turbine Foundation 90,000 41,000 131,000 Project Engineering 25,000 5,500 30,500 Trenching & Land Improvements 5,000 8000 13,000 Electrical Infrastructure 80,000 23,000 103,000 Electrical Maintenance / Warranty 4.000 3,600 7,600 Total 507,100 178,410 $685,510 If multiple turbines were installed together, a discount should be available on the wind turbines, and significant economies of scale would be possible with the BOS costs. Based on bids for multiple turbine projects at SCL, the initial capital cost for one wind turbine should drop by $50,000 each for two turbines, $75,000 each for three, and $100,000 each for four. These 36 reductions are used in the economic model. However, the wind farm at SCI began with two units and will have a third and a fourth added at different times. 5.4 Wind-Diesel Hybrid System Operational Savings Once all of the engineering and cost data were ready, an economic assessment was performed according to the procedure used in Hunter [4]. Figure 32 shows the resulting COE decreasing as the number of wind turbines increases. The trend will eventually reverse and start to increase with number of turbines, each additional wind turbine would be less efficiently utilized because of the growing wind energy penetration and lack of system storage. For this same reason, Figures 33 and 34 show the payback period dropping and the internal rate of return rising from one to three turbines and then leveling and reversing thereafter. The complete economic tables can be found in Appendix A. These results are provided for those who need data points to check their own simulations. Copies of the spreadsheets used here can be obtained from the authors. The $1,271,000 capital investment in the two-turbine hybrid system was easily offset by savings in fuel and operating costs for diesel generation of $196,000 annually, giving a 6.49 year simple payback period, 14.4% internal rate of return, $0.142/kWh wind COE, and dropping the system COE from $0.476/kWh to $0.461/kWh. This would give net savings of $0.015/kWh, or $114,000 in 1998. Using the EPRI TAG approach gives a wind energy COE = (ICC*FCR)/AEP + O&M = (635510*.102) / 466000 + 0.01 = $0.149/kWh. With a capital investment of $2,342,000, four 225 kW wind turbines have annual operating savings of $370,000, and would give a 6.32 year simple payback period, 14.8% internal rate of return, $0.139/kWh wind COE, and $0.447/kWh system COE, with net savings of $0.029/kWh, or $223,000 in 1998. 37 COE vs Number of Turbines 0.50 0.469 0.461 COE 0.447 0.476 0.453 0.45 + 0.40 + 0.35 + 0.30 + 0.25 + 0.20 + +- + + 0 1 2 3 Number of Turbines Figure 32: System COE vs Number of Wind Turbines Payback vs Number of Turbines ” a 6.28 6.32 © o 2 x o © 2 > © o 1 2 3 4 Number of Turbines Figure 33: Payback Period vs Number of Wind Turbines 38 IRR vs Number of Turbines 0.16 9 0.15 + oe 14.8% 0.15 + 14.4% 0.14 + 0.14 + 13.1% 0.13 + 0.12 + 1 Internal Rate of Return 2 3 4 Number of Turbines Figure 34: Internal Rate of Return vs Number of Wind Turbines 5.5 Sensitivity Analysis To check the sensitivity of the results to variations in average wind speed from year to year, the two-turbine case was run with the wind speeds adjusted upward and downward by 17.5%, which is the interannual variability (one standard deviation) found in the historical wind measurements. The results are shown in Table 7. With the wind speed 17.5% lower than the NREL measurement year, COE and payback period rose by 2.6% and 63%. With the wind speed 17.5% higher, COE and payback period dropped by 2.6% and 28%. Table 7: Economic Sensitivity to Wind Speed Variations Spreadsheet Model for 2 Turbines Diesel Cost of Payback Internal Saving Energy Period Rate of Case Wind Speed (kltr) ($/kWh) (years) _ Return (%) minus 17.5% 5.0 m/s 149 0.473 10.6 7.0 baseline 6.1 m/s 242 0.461 6.5 14.4 plus 17.5% 7.2 m/s 336 0.449 4.7 21.0 39 The wind energy COE of $0.15/kWh is much higher than the typical range of $0.04/kWh to $0.09/kWh. Several factors contribute to this discrepancy, including costs associated with island access, severe Naval construction requirements, and moderately low wind speeds at this site. To gain some understanding of the COE sensitivity to these issues, we ran an economic case for two turbines without the severe Naval and island access requirements. This case uses mainland costs with those for installation, foundation, and electrical infrastructure reduced by half, plus a $30,000 discount for each of two turbines. The net initial capital cost of $354,600 per turbine generates a wind energy COE of $0.086/kWh, a hybrid system COE of $0.454/kWh, a payback period of 3.6 years, and a 27.4% internal rate of return. Using the EPRI TAG approach gives COE = (ICC*FCR)/AEP + O&M = (354600*.102) / 466000 + 0.01 = $0.088/kWh. 6.0 CONCLUSIONS San Clemente Island (SCI) has a moderate wind resource, with an annual average wind speed of 6.1 m/s (11.6 knots) as measured by the National Renewable Energy Laboratory and the Naval Facilities Engineering Service Center at the Met2 tower wind site location from the 1995 through 1999 data collection period. Recognizing this, the recently constructed wind-diesel hybrid energy system was modeled to examine its performance and economics, as well as the merits of adding more wind energy generation. Using generally conservative assumptions (unfavorable to wind energy) in the model, the hybrid system displayed favorable operation and economics. Using two 225 kW wind turbines, the wind energy COE of $0.142/kWh helps reduce the wind- diesel hybrid system COE from the baseline $0.476/kWh to $0.461/kWh. This reduces system COE by 3.2%. The payback period is 6.5 years, the internal rate of return 14.4%. The four-turbine case had a wind energy COE of $0.139/kWh and a hybrid system COE of $0.447/kWh, saving 6.1%. The payback period is 6.3 years, the internal rate of return 14.8%. The costs of energy (COE) for this case is relatively insensitive to annual average wind speed, varying 2.6% for a 17.5% change in wind speed. But the payback period is quite sensitive to wind speed, varying 28% to 63% for a 17.5% change in wind speed. Different economic assumptions, such as higher and lower inflation, do not appear to have much impact on the results. Because cost and savings components are well distributed, there does not appear to be a dominant factor affecting the economic results. Factors that could affect the results include the actual capital and installation costs of the wind equipment, diesel fuel costs, and diesel system O&M and overhaul costs. This work presented a study of the SCI wind-diesel hybrid system using two wind turbines with an option for two more. For the operating and economic conditions examined, it appears wind energy is cost effective in this application. We believe these conditions are realistic but regret the lack of complete cost data on the existing diesel system. Certainly many alternatives to these cases merit consideration. For instance, it appears that the wind penetration could be increased, thus producing further, yet diminishing, savings. 40 Moreover, excess electrical energy should not be curtailed or wasted on dump loads; rather, it should be used for beneficial purposes, provided those purposes make economic sense. Within the SCI electrical grid, such benefits may be realized by using excess wind energy for deferrable loads such as the SCI reverse osmosis water system, water heating, or space heating. The Navy is planning for a reverse osmosis water system with its fourth wind turbine in 2001. As a preliminary review, this study used 1-hour average wind and load data for the hybrid system modeling to develop a general sense of economic tradeoffs. Dynamic load management should be addressed using load and wind data at shorter intervals (1 minute or less) to study system dynamics. 41 REFERENCES Le Marriott, Ken, et al (1983). Operational / Land Use Compatibility Study for NALF San Clemente Island, KTU+A under Southwest Division Contract # N68711-91-C-0035 for USN NAS North Island Staff Civil Engineers. Elliott, D.L., et al. (1987). Wind Energy Resource Atlas of the United States, DOE/CH10093-4, Pacific Northwest Laboratory. Pal, D. and Casteel, F. (1986). Wind Resource Assessment for Naval Auxiliary Landing Field, San Clemente Island, California, TM M-73-86-10, Navy Energy and Natural Resources R&D Office, Naval Facilities Engineering Command. Hunter, R. and Elliott, G. (1994). Wind Diesel Systems, Cambridge University Press. Olsen, T and McKenna, E (1996). Hybrid Energy System Cost Analysis: San Nicolas Island, California, NREL/TP-440-21120. Miller, Alan (1996). The Evaluation of the Wind Resource at San Nicolas Island, report to NREL, Problem Solvers, International, Golden, Colorado. 42 APPENDIX A: Hybrid System Model and Economic Summary Tables This appendix contains the hybrid system model and economic summary tables used to develop the economic conclusions reached in this report. 43 BOO Or yonsenennnneenertee enereennsnimeenctneneetsmenecnen | Aduoted Turbine Alowod Number wh Diesel Percent of Litres ot stats Diesel «Wind Ss Wind Wind ot Net Digsel Percentot Litres of tartan © Divs! — tres. = Speed Date Time Demand = Rating Rating Ofoscl_ counter muntima © Speod ~— Power © Demand Turbines Demind Rating Rating Diesel © counter. mntime = saved «6.1 me (or) (Kw) (ew) Used Consumed (houre) (rvs) (KW) on) nw) (ew) Used — Consumed (hours) Average 0e07 «1631.7 = SBR 808.0 61 831 0807 08 8165 1448.9 OS = 2a3e 19.7844 6.119481 Standard Deviation 1140 296.9 % 31 60.8 114.0 04 107.0 280.0 O1 96.6 18.80889 3.10047 Maximum 1819.1 1700.0 6% 404.3 216 2204 1s1e1 10 1280.8 1700.0 os 420g 60.48818 21.62711 Minimum 5142 12000 32% 107.1 03 00 «S142 00 488.8 1200.0 os 197.4 © 0.300872 Total 7018478 2688873 196 14872 408647 5808476 7182820 2608122 em 19121120781 408847 Wind Enorgy Used 405847 Wind Energy Available 0 Wind Energy Curtaliad Maximum numbor of wind titbinee: 2 0001 On ye anne nner nent wind hypiide-——~ ee on es Singo Maximum Adueled Turbine lowed Number stare = Diesel «= Wind «Wind Wind ot Dieso! Percent of Litres of Net Diesel Percent of Litree ot starts = Diesel ~—siitres. «= Speed Date Time Demand «Rating «= Rating —Deeel counter muntime = Speed = Power Demand Turbines Demand Rating Rating ional + counter nuntime «saved «6.1mm on ew) ) eed Consumed (hours) (ra) (ew) (ew) “o) qe) eed Consumed (hours) Average 800.7 1531.7 8% 308.9 81 = 53.1887 17 7684 1908.1 0S © 270.4 27.5888 6.119481 ‘Standard Deviation 114.0 298.9 % «37.0 31 409 1140 07 1816 = 2a88 o1 698 31.81318 3.10947 ‘Masten 1519.1 1700.0 60% «= 484.3. 218 2204 1919.1 2.0 12778 1700.0 08 = 421.7 119 2162711 ‘Minimum 514.2 1200.0 82% 197.1 03 00 3142 00 261.1 780.0 02 «184.8 0 0.900872 Tot 7018476 2088879 196 14572 405547 5860476 0087381 2447371 298 12144 241803 931095 Wind Energy Used 931095 Wind Energy Available © Wind Energy Cuttalled Maximum number of wind tuibines: & Ce >| wind hybtige cee nee _ severance mene 96.4 fh Singe = Maximum Orginal Adueted Turbine Alowed = Number Wind Diesel Percentot Litresot starts Diesel «Wind = Wind Wind ot Net Diese! Porcontof Liresof etarta © loss == siitras Speed Date Time Demand «Rating =—-Rating = Diesel counter mintime = Speed = Power Demand Turbines Demand ating Rating © Dieeel = counter, untime = saved 6.1 mae (ne) (Kew) (ew) Used Consumed (houre) (nwa) aw) ow) (RA) (Kw) Used Consumed (hours) Average 060.7 16917 88% «808.0 61 $8.1 660.7 25° 7114 1985.1 05 = 205.8 41 6.119481 Standard Deviation 114.0 296.9 8% = 97.8 31 609 114.0 14 1720-247. O1 48.8 46.75209 3.10047 Maxum 1700.0 BBB. 216 © 2204 = 1919.1 80 1277.8 1700.0 08 = 421.7 178.4945 21.62751 Minion 12000 82% 03 00 = $142 00 © ©=201.0 780.0 o1 110.1 © 0.300672 Total 198 14872 405547 5800476 6232205 2320909 282 959564 1988271 Wind Energy Used 1908642 Wind Energy Available 10871 Wind Energy Ourtaited Maximum number of wind wibines: 4 Oe ea | wind hybrid Sing — Madmum Original Adusted Turbine Alowed = Number Wind Diesel Percent of Lines ot tarts = Olea! «= Wind = Wind «Wind of Not Diesel Percent of Litres of tate © Dies! —litres. «Speed Date Time = Demand Rating = ating Dieses! countor. runtime = Speed = Power © Demand Turbines Demand Rating Rang lesa! counter nintme = saved «8.1 ma om (Kw) (mw) Used Consumed (hours) (rwe) “wy (Kw) (Kw) ny Ueed Consumed (hours) Average 0607 1531.7 88% = 3008.9 61 = 53.1 088,7 33 0088S | 1928.3 04 = 284.8 82 6.110881 ‘Standard Deviation 114.0 236.3 8% 609 114.0 14 2017 250.2 04 $5.0 57.20081 3.10047 1519.1 1700.0 60% 484.3 218 © 2204 1319.1 40 12778 — 1700.0 08 = 42.7 297.9827 21.62711 5142 12000 © SMe 807.1 os oo 3142 00 ©2002 ©7500 O1 1161 © 0.900872 Total 7618476 2088873 19814672 405547 §868476 5856374 2231828 312 «11987487045 1762102 Wind Energy Used 1862180 Wind Energy Available 100087 Wind Energy Curtalled SCI_Diewel_projate, 4/1499 ECONOMIC ANALYSIS Input Values System load (kWh/y) Diesel energy (kWh/y) Wind energy (kWh/y) Diesel fuel usage, no wind (\/yr) Diesel fuel usage, with wind (I/yr) Diesel fuel cost ($/I) Diesel ops cost, variable ($/kWh) Diesel ops cost, fixed ($/y) Wind ICC ($) Wind O&M cost ($/kWh) System life, (yrs) General inflation Fuel inflation Discount rate Interest Term of loan, (yrs) Calculated Values for Both Systems Capital cost Initial payment on system Loan Annual payment NPV of annual payment Fuel cost per annum NPV of fuel costs Overhaul cost per annum NPV of overhaul costs O&M costs per annum NPV of O&M costs Total annual costs Total system NPV, TNPV Annual savings Levelized cost of energy, COE Payback period, years Internal rate of return, IRR, (x) Site: Turbine: Quantity: 1 SL 7,618,476 7,152,929 465,547 FL 2,688,873 FL 2,568,122 FC 0.264 Ov 0.154 OF 1,173,245 wc 685,510 wo 0.01 L 20 i 2.0% e 2.0% d 6.9% b 10.0% N 10 C =ICC+BOS Ad Al=C-Ad Ap = Al * CRFP Apnpv = Ap*PWFP Af =FL* FC Afnpv = Af * PWFF Ao Aonpv = Ao * PWFO Am Amnpv = Am*PWFO At = Ap+Af+Ao+Am = Ad+sum(NPVs) Sv = dsl At - hbd At = TNPV*CRFI/SL Diesel Only oooo°o 709,863 8,995,402 0 0 2,346,490 29,734,802 3,056,353 38,730,204 0.476 [(1+x)*L-1])/[x*(1+x)4L] - P = San Clemente Island, CA, 6.1 m/s avg 225 kW, Commercial Economic Factors Present worth factor of fuel costs, PWFF, a=(1+e)/(1+d) Present worth factor of O&M costs, PWFO, a=(1+i)/(1+d) Present worth factor of interest payments, PWFP, a=1/(1+b) Capital recovery factor for system income, CRFI, a=1/(1+d) Capital recovery factor for interest payments, CRFP, a=1/(1+b) Hybrid System Diesel Part 0 0 0 0 0 677,984 8,591,439 0 0 2,274,796 28,826,290 2,952,780 37,417,728 0.490 0.000 a variable 0.95416277 0.95416277 0.9354537 a variable 0.9354537 0.90909091 Hybrid System Wind Part 685,510 685,510 0 0 0 0 0 1,000 12,672 4,655 58,994 5,655 757,176 0.152 ovariable = _Y(a,n 20 = 12.67203 20 12.67203 10 7.05616 n variable X(a.n) 20 0.09366054 10 0.16274539 Hybrid System Total 685,510 685,510 0 0 0 677,984 8,591,439 1,000 12,672 2,279,451 28,885,284 2,958,436 38,174,905 97,917 0.469 7.00 13.1% (NPV = net present value; ICC = initial capitol cost; BOS = balance of station = 26% ICC; O&M = operations and maintenance) Appendix_A.xls, 9/14/99 ECONOMIC ANALYSIS Input Values System load (kWh/y) Diesel energy (kWh/y) Wind energy (kWh/y) Diesel fuel usage, no wind (\/yr) Diesel fuel usage, with wind (I/yr) Diesel fuel cost ($/|) Diesel ops cost, variable ($/kWh) Diesel ops cost, fixed ($/y) Wind ICC ($) Wind O&M cost ($/kWh) System life, (yrs) General inflation Fuel inflation Discount rate Interest Term of loan, (yrs) Calcul Val Capital cost Initial payment on system Loan Annual payment NPV of annual payment Fuel cost per annum NPV of fuel costs Overhaul cost per annum NPV of overhaul costs O&M costs per annum NPV of O&M costs Total annual costs Total system NPV, TNPV Annual savings Levelized cost of energy, COE Payback period, years Internal rate of return, IRR, (x) for Both Site: stems Turbine: Quantity: 2 SL 7,618,476 7,042,272 576,204 FL 2,688,873 FL 2,539,421 FC 0.264 Ov 0.154 OF 1,173,245 wc 635,510 wo 0.01 L 20 i 2.0% e 2.0% d 6.9% b 10.0% N 10 C = |ICC+BOS Ad Al=C-Ad Ap = Al * CRFP Apnpv = Ap*PWFP. Af = FL* FC Afnpv = Af * PWFF Ao Aonpv = Ao * PWFO Am Amnpv = Am*PWFO At = Apt+Af+Ao+Am = Ad+sum(NPVs) Sv = dsl At - hbd At = TNPV*CRFI/SL Diesel Only oooo°o 709,863 8,995,402 0 0 2,346,490 29,734,802 3,056,353 38,730,204 0.476 [(1+x)*L-1]/[x*(1+x)L] - P = San Clemente Island, CA, 5.0 m/s avg 225 kW, Commercial Economic Factors Present worth factor of fuel costs, PWFF, a=(1+e)/(1+d) Present worth factor of O&M costs, PWFO, a=(1+i)/(1+d) Present worth factor of interest payments, PWFP, a=1/(1+b) Capital recovery factor for system income, CRFI, a=1/(1+d) Capital recovery factor for interest payments, CRFP, a=1/(1+b) Hybrid System Diesel Part ooooco 670,407 8,495,420 0 0 2,257,755 28,610,344 2,928,162 37,105,764 0.493 0.000 a variable 0.95416277 0.95416277 0.9354537 a variable 0.9354537 0.90909091 Hybrid System Wind Part 1,271,020 1,271,020 0 0 0 0 0 2,000 25,344 5,762 73,017 7,762 1,369,381 0.223 n variable Y(a,n) 20 = =12.67203 20 = =12.67203 10 7.05616 n variable X(a,n) 20 0.09366054 10 0.16274539 Hybrid System Total 1,271,020 1,271,020 0 0 0 670,407 8,495,420 2,000 25,344 2,263,517 28,683,361 2,935,924 38,475,145 120,429 0.473 10.55 7.0% (NPV = net present value; ICC = initial capitol cost; BOS = balance of station = 26% ICC; O&M = operations and maintenance) Appendix_A.xls, 9/14/99 ECONOMIC ANALYSIS Input Values System load (kWh/y) Diesel energy (kWh/y) Wind energy (kWh/y) Diesel fuel usage, no wind (I/yr) Diesel fuel usage, with wind (I/yr) Diesel fuel cost ($/!) Diesel ops cost, variable ($/kWh) Diesel ops cost, fixed ($/y) Wind ICC ($) Wind O&M cost ($/kWh) System life, (yrs) General inflation Fuel inflation Discount rate Interest Term of loan, (yrs) Calculated Values for Both Systems Capital cost Initial payment on system Loan Annual payment NPV of annual payment Fuel cost per annum NPV of fuel costs Overhaul cost per annum NPV of overhaul costs O&M costs per annum NPV of O&M costs Total annual costs Total system NPV, TNPV Annual savings Levelized cost of energy, COE Payback period, years Internal rate of return, IRR, (x) Site: Turbine: Quantity: 2 SL 7,618,476 6,687,381 931,095 FL 2,688,873 FL 2,447,371 FC 0.264 Ov 0.154 OF 1,173,245 wc 635,510 wo 0.01 L 20 i 2.0% e 2.0% d 6.9% b 10.0% N 10 C =|ICC+BOS Ad Al=C-Ad Ap = Al * CRFP Apnpv = Ap*PWFP Af = FL* FC Afnpv = Af * PWFF Ao Aonpv = Ao * PWFO. Am Amnpv = Am*PWFO At = Ap+Af+Ao+Am = Ad+sum(NPVs) Sv = dsl At - hbd At = TNPV*CRFI/SL Diesel Only eooooco 709,863 8,995,402 0 0 2,346,490 29,734,802 3,056,353 38,730,204 0.476 [(14x)SL-1/fx*(14x)AL] - P= San Clemente Island, CA, 6.1 m/s avg 225 kW, Commercial Economic Factors Present worth factor of fuel costs, PWFF, a=(1+e)/(1+d) Present worth factor of O&M costs, PWFO, a=(1+i)/(1+d) Present worth factor of interest payments, PWFP, a=1/(1+b) Capital recovery factor for system income, CRFI, a=1/(1+d) Capital recovery factor for interest payments, CRFP, a=1/(1+b) Hybrid System Diesel Part 0 0 0 0 0 646,106 8,187,475 0 0 2,203,102 27,917,777 2,849,208 36,105,252 0.506 0.000 a variable 0.95416277 0.95416277 0.9354537 a variable 0.9354537 0.90909091 Hybrid System Wind Part 1,271,020 1,271,020 0 0 0 0 0 2,000 25,344 9,311 117,989 11,311 1,414,353 0.142 n variable Y(a,n 20 = 12.67203 20 = =12.67203 10 7.05616 nvariable = X(a,n) 20 0.09366054 10 0.16274539 Hybrid System Total 1,271,020 1,271,020 0 0 0 646,106 8,187,475 2,000 25,344 2,212,413 28,035,766 2,860,519 37,519,605 195,834 0.461 6.49 14.4% (NPV = net present value; ICC = initial capitol cost; BOS = balance of station = 26% ICC; O&M = operations and maintenance) Appendix_A.xis, 9/14/99 ECONOMIC ANALYSIS Input Values System load (kWh/y) Diesel energy (kWh/y) Wind energy (kWh/y) Diesel fuel usage, no wind (I/yr) Diesel fuel usage, with wind (I/yr) Diesel fuel cost ($/I) Diesel ops cost, variable ($/kWh) Diesel ops cost, fixed ($/y) Wind ICC ($) Wind O&M cost ($/kWh) System life, (yrs) General inflation Fuel inflation Discount rate Interest Term of loan, (yrs) Calculated Values for Both Systems Capital cost Initial payment on system Loan Annual payment NPV of annual payment Fuel cost per annum NPV of fuel costs Overhaul cost per annum NPV of overhaul costs O&M costs per annum NPV of O&M costs Total annual costs Total system NPV, TNPV Annual savings Levelized cost of energy, COE Payback period, years Internal rate of return, IRR, (x) Site: Turbine: Quantity: 2 SL 7,618,476 6,323,127 1,295,349 FL 2,688,873 FL 2,352,892 FC 0.264 Ov 0.154 OF 1,173,245 WC 635,510 wo 0.01 L 20 i 2.0% e 2.0% d 6.9% b 10.0% N 10 C =ICC+BOS Ad Al=C -Ad Ap = Al * CRFP. Apnpv = Ap*PWFP. Af =FL* FC Afnpv = Af * PWFF Ao Aonpv = Ao * PWFO Am Amnpv = Am*PWFO At = Apt+AftAo+Am = Ad+sum(NPVs) Sv = dsl At - hbd At = TNPV*CRFI/SL Diesel Only eooooco 709,863 8,995,402 0 0 2,346,490 29,734,802 3,056,353 38,730,204 0.476 [(1+x)*L-1]/[x*(1+x)4L] - P = San Clemente Island, CA, 7.2 m/s avg 225 kW, Commercial Economic Factors Present worth factor of fuel costs, PWFF, a=(1+e)/(1+d) Present worth factor of O&M costs, PWFO, a=(1+i)/(1+d) Present worth factor of interest payments, PWFP, a=1/(1+b) Capital recovery factor for system income, CRFI, a=1/(1+d) Capital recovery factor for interest payments, CRFP, a=1/(1+b) Hybrid System Diesel Part eooooo 621,164 7,871,405 0 0 2,147,007 27,206,938 2,768,170 35,078,343 0.520 0.000 avariable n variable Y(a,n 0.95416277 20 12.67203 0.95416277 20 = 12.67203 0.9354537 10 7.05616 avariable n variable X(a,n) 0.9354537 20 0.09366054 0.90909091 10 0.16274539 Hybrid System Hybrid System Wind Part Total 1,271,020 1,271,020 1,271,020 1,271,020 0 0 0 0 0 0 0 621,164 0 7,871,405 2,000 2,000 25,344 25,344 12,953 2,159,960 164,147 27,371,085 14,953 2,783,124 1,460,511 36,538,854 273,229 0.106 0.449 4.65 21.0% (NPV = net present value; ICC = initial capitol cost; BOS = balance of station = 26% ICC; O&M = operations and maintenance) Appendix_A.xls, 9/14/99 ECONOMIC ANALYSIS Input Values System load (kWh/y) Diesel energy (kWh/y) Wind energy (kWh/y) Diesel fuel usage, no wind (\/yr) Diesel fuel usage, with wind (I/yr) Diesel fuel cost ($/|) Diesel ops cost, variable ($/kWh) Diesel ops cost, fixed ($/y) Wind ICC ($) Wind O&M cost ($/kWh) System life, (yrs) General inflation Fuel inflation Discount rate Interest Term of loan, (yrs) Calculated Values for Both Systems Capital cost Initial payment on system Loan Annual payment NPV of annual payment Fuel cost per annum NPV of fuel costs Overhaul cost per annum NPV of overhaul costs O&M costs per annum NPV of O&M costs Total annual costs Total system NPV, TNPV Annual savings Levelized cost of energy, COE Payback period, years Internal rate of return, IRR, (x) (NPV = net present value; ICC = initial capitol cost; BOS = balance of station = 26% ICC; O&M = operations and maintenance) Appendix_A.xis, 9/14/99 Site: Turbine: Quantity: 3 SL 7,618,476 6,232,205 1,386,271 FL 2,688,873 FL 2,329,309 FC 0.264 Ov 0.154 OF 1,173,245 wc 610,510 wo 0.01 L 20 i 2.0% e 2.0% d 6.9% b 10.0% N 10 C = ICC+BOS Ad Al=C-Ad Ap = Al * CRFP Apnpv = Ap*PWFP Af = FL* FC Afnpv = Af * PWFF Ao Aonpv = Ao * PWFO Am Amnpv = Am*PWFO At = Ap+Af+Ao+Am = Ad+sum(NPVs) Sv = dsl At - hbd At = TNPV*CRFI/SL Diesel Only oooo°o 709,863 8,995,402 0 0 2,346,490 29,734,802 3,056,353 38,730,204 0.476 [(44x)9L-1VPx"(14x)4L] - P= San Clemente Island, CA, 6.1 m/s avg 225 kW, Commercial Economic Factors Present worth factor of fuel costs, PWFF, a=(1+e)/(1+d) Present worth factor of O&M costs, PWFO, a=(1+i)/(1+d) Present worth factor of interest payments, PWFP, a=1/(1+b) Capital recovery factor for system income, CRFI, a=1/(1+d) Capital recovery factor for interest payments, CRFP, a=1/(1+b) Hybrid System Diesel Part 0 0 0 0 0 614,938 7,792,510 0 0 2,133,005 27,029,503 2,747,942 34,822,014 0.523 0.000 a variable 0.95416277 0.95416277 0.9354537 a variable 0.9354537 0.90909091 Hybrid System Wind Part 1,831,530 1,831,530 0 0 0 0 0 3,000 38,016 13,863 175,669 16,863 2,045,215 0.138 n variable n variable Y(a,n 20 12.67203 20 12.67203 10 7.05616 X(a,n) 20 0.09366054 10 0.16274539 Hybrid System Total 1,831,530 1,831,530 0 0 0 614,938 7,792,510 3,000 38,016 2,146,867 27,205,172 2,764,805 36,867,229 291,548 0.453 6.28 14.9% ECONOMIC ANALYSIS Input Values System load (kWh/y) Diesel energy (kWh/y) Wind energy (kWh/y) Diesel fuel usage, no wind (I/yr) Diesel fuel usage, with wind (I/yr) Diesel fuel cost ($/!) Diesel ops cost, variable ($/kWh) Diesel ops cost, fixed ($/y) Wind ICC ($) Wind O&M cost ($/kWh) System life, (yrs) General inflation Fuel inflation Discount rate Interest Term of loan, (yrs) Calculated Values for Both Systems Capital cost Initial payment on system Loan Annual payment NPV of annual payment Fuel cost per annum NPV of fuel costs Overhaul cost per annum NPV of overhaul costs O&M costs per annum NPV of O&M costs Total annual costs Total system NPV, TNPV Annual savings Levelized cost of energy, COE Payback period, years Internal rate of return, IRR, (x) Site: : 225 kW, Commercial Quantity: 4 SL 7,618,476 5,856,374 1,762,102 FL 2,688,873 FL 2,231,828 FC 0.264 Ov 0.154 OF 1,173,245 WC 585,510 Wo 0.01 L 20 i 2.0% e 2.0% d 6.9% b 10.0% N 10 Diesel Only C =ICC+BOS 0 Ad 0 Al=C-Ad 0 Ap = Al * CRFP 0 Apnpv = Ap*PWFP 0 Af = FL* FC 709,863 Afnpv = Af * PWFF 8,995,402 Ao 0 Aonpv = Ao * PWFO 0 Am 2,346,490 Amnpv = Am*PWFO 29,734,802 At = Ap+Af+Ao+Am 3,056,353 = Ad+sum(NPVs) 38,730,204 Sv = dsl At - hbd At = TNPV*CRFI/SL 0.476 P=C/Sv ((14x)4L-1)/bx*(14x)4L] - P= San Clemente Island, CA, 6.1 m/s avg Economic Factors Present worth factor of fuel costs, PWFF, a=(1+e)/(1+d) Present worth factor of O&M costs, PWFO, a=(1+i)/(1+d) Present worth factor of interest payments, PWFP, a=1/(1+b) Capital recovery factor for system income, CRFI, a=1/(1+d) Capital recovery factor for interest payments, CRFP, a=1/(1+b) Hybrid System Diesel Part oooo°o 589,203 7,466,395 0 0 2,075,127 26,296,072 2,664,329 33,762,467 0.540 0.000 avariable nvariable Y(a,n) 0.95416 20 = 12.67203 0.95416 20 = =12.67203 0.93545 10 7.05616 avariable n variable X(a.n) 0.93545 20 0.09366 0.90909 10 0.16275 Hybrid System Hybrid System Wind Part Total 2,342,040 2,342,040 2,342,040 2,342,040 0 0 0 0 0 0 0 589,203 0 7,466,395 4,000 4,000 50,688 50,688 17,621 2,092,748 223,294 26,519,366 21,621 2,685,950 2,616,022 36,378,490 370,403 0.139 0.447 6.32 14.8% (NPV = net present value; ICC = initial capitol cost; BOS = balance of station = 26% ICC; O&M = operations and maintenance) Appendix_A.xls, 9/14/99 ECONOMIC ANALYSIS Input Values System load (kWh/y) Diesel energy (kWh/y) Wind energy (kWh/y) Diesel fuel usage, no wind (I/yr) Diesel fuel usage, with wind (I/yr) Diesel fuel cost ($/l) Diesel ops cost, variable ($/kWh) Diesel ops cost, fixed ($/y) Wind ICC ($) Wind O&M cost ($/kWh) System life, (yrs) General inflation Fuel inflation Discount rate Interest Term of loan, (yrs) Calculated Values for Both Systems Capital cost Initial payment on system Loan Annual payment NPV of annual payment Fuel cost per annum NPV of fuel costs Overhaul cost per annum NPV of overhaul costs O&M costs per annum NPV of O&M costs Total annual costs Total system NPV, TNPV Annual savings Levelized cost of energy, COE Payback period, years Internal rate of return, IRR, (x) Turbine: 225 kW, Commercial 2 SL 7,618,476 6,687,381 931,095 2,688,873 2,447,371 0.264 0.154 1,173,245 354,600 0.01 20 2.0% 2.0% 6.9% 10.0% 10 FL FL FC Ov OF wc Wo L zoao~ Diesel Only C =ICC+BOS Ad Al=C - Ad Ap = Al * CRFP Apnpv = Ap*PWFP Af = FL * FC Afnpv = Af * PWFF Ao Aonpv = Ao * PWFO Am Amnpv = Am*PWFO At = Apt+Af+Ao+Am = Ad+sum(NPVs) Sv = dsl At - hbd At = TNPV*CRFI/SL ooooco°o 709,863 8,995,402 0 0 2,346,490 29,734,802 3,056,353 38,730,204 0.476 [(1+x)9L-1]/fx*(14+x)AL] - P= Fictitious Mainland Site, Non-Naval, 6.1 m/s avg Economic Factors a variable Present worth factor of fuel costs, PWFF, a=(1+e)/(1+d) 0.95416277 Present worth factor of O&M costs, PWFO, a=(1+i)/(1+d) 0.95416277 Present worth factor of interest payments, PWFP, a=1/(1+b) 0.9354537 a variable Capital recovery factor for system income, CRFI, a=1/(1+d) 0.9354537 Capital recovery factor for interest payments, CRFP, a=1/(1+b) 0.90909091 Hybrid System Hybrid System Diesel Part Wind Part 0 709,200 0 709,200 0 0 0 0 0 0 646,106 0 8,187,475 0 0 2,000 0 25,344 2,203,102 9,311 27,917,777 117,989 2,849,208 11,311 36,105,252 852,533 0.506 0.086 0.000 nvariable _Y(a.n 20 12.67203 20 12.67203 10 7.05616 nvariable — X(a,n) 20 0.09366054 10 0.16274539 Hybrid System Total 709,200 709,200 0 0 0 646,106 8,187,475 2,000 25,344 2,212,413 28,035,766 2,860,519 36,957,785 195,834 0.454 3.62 27.4% (NPV = net present value; ICC = initial capitol cost; BOS = balance of station = 26% ICC; O&M = operations and maintenance) Appendix_A.xls, 9/14/99 Graphs Chart 1 COE 0.50 COE vs Number of Turbines 0.45 + 0.40 + 0.35 + 0.30 + 0.20 0.476 2 Number of Turbines Page 1 APPENDIX B: SCI 1998-1999 Power Plant Status and Production Reports SCI 1998-1999 power plant status and production reports displaying the measured wind turbine and diesel electrical energy production contributing to the total San Clemente Island electrical demand. These spreadsheets indicate fuel usage, individual diesel operation and production, and individual wind turbine operation and production. February 1998 - January 1999 42580 38122 57757 510068 45250 37839 44341 520800 43710 46161 44819 531300 40940 53621 52120 496300 41780 49255 46972 495600 45251 31481 29385 542150 48700 23649 21977 584850 45736 23594 22423 555778 48441 20356 19511 563150 51180 32155 27362 607600 55380 29516 33522 645400 54680 28982 30333 634550 563628 414731 430522 6687546 DUCED DIESEL KWH PRODUCED PER GAL OF FUEL % WIND TURBINE KWH PRODUCED TO TOTAL GRID FUEL SAVINGS BY WIND TURBINE OPERATIONS IN GALLONS (AVOIDED GALLONS USED) WT#1 WT#2 Total Total WTs Diesel 38122 57757 95879 510068 37839 44341 82180 520800 46161 44819 90980 531300 53621 52120 105741 496300 49255 46972 96227 495600 31481 29385 60866 542150 23649 21977 45626 584850 23594 22423 46017 555778 20356 19511 39867 563150 32155 27362 59517 607600 29516 33522 63038 645400 28982 30333 59315 634550 This Spreadsheet only shows 11 months of Wind Turbine Production ‘SCI 1998 Production.xls Of which Feb-Mar 98 were initial shakedown - limited production 9/17/99 kWH 120000 100000 80000 60000 40000 20000 San Clemente Island, California Total Electrical Production 1 2 3 4 5 6 7 8 9 10 11 % Feb 98 - Jan 99 San Clemente Island, CA Total Production Levels 700000 600000 500000 400000 To o o Ss 3 Q a = and 300000 200000 100000 6 7 12 Feb 1998 - Jan 1999 "SCI 1998 Production.xls 9/17/99 February 1998 17150 36050 56350 78050 100450 118650 133700 151900 168000 188650 214200 234150 1-Feb| 1600 0} 1600 0 700 2-Feb| 1630 -4| 3230} 68.4 900 3-Feb| 1690 -6| 4920} 291 1800 4-Feb| 1810 0} 6730} 291 2100 5-Feb| 1880 0} 8610} 291 800 6-Feb| 1470 .0} 10080} 291 1200 7-Feb} 1310 0} 11390} 291 200 8-Feb| 1480 -0| 12870} 291 1800 9-Feb| 1260 -0} 14130} 291 1800 10-Feb| 1970 -0} 16100} 291 2000 11-Feb| 1810 17910] 291 700 12-Feb| 1780 19690} 319.2 615 S/S SC|O/OCl|O|oO|C|o g SO[S/O]/S/O/O]/O}O|oO N oe s So wo = Qa o 13-Feb| 1550 21240| 334.2 600 253050 14-Feb| 1420} 0.0} 22660) 334.2 500 270900 15-Feb| 1460} 19.7) 24120) 353.9 2200 287700 308350 325150 344050 365750 379400 393400 406700 423850 437850 454650 474600 491726 510068 16-Feb| 1610} 0.0] 25730} 353.9 400 17-Feb| 1410} 15.4] 27140} 369.3 300 18-Feb| 1540| 0.0} 28680] 369.3 600 19-Feb| 1630} 15.0] 30310} 384.3 630 20-Feb} 1310] 44.0] 31620} 428.3 2200 21-Feb| 1330] _ 0.0] 32950) 428.3 600 22-Feb| 1180] 8.8} 34130} 437.1 600 23-Feb| 1130] 0.0} 35260} 437.1 700 24-Feb| 1340] 13.2| 36600] 450.3 600 25-Feb| 1430} 0.0} 38030| 450.3}. 09:00 26-Feb| 1510] 20.0} 39540| 470.3} 19:00 27-Feb| 1560} 10.0] 41100] 480.3} 07:00 28-Feb| 1480| 16.6] 42580| 496.9| 06:00 SCLOLOLO/O/ O/C] OJ O/O}|O|OC|O MONTHLY TOTAL KWH PRODUCED KWH PRODUCED PER GAL OF FUEL % WIND TURBINE KWH PRODUCED TO TOTAL GRID FUEL SAVINGS BY WIND TURBINE OPERATIONS IN GALLONS (AVOIDED GALLONS USED) 8004 The Wind Turbines first started shakedown operations on February 5, 1998 ‘02-Feb-98 Production.xls 23 Days of Wind Turbine Production 9/17/99 MARCH 1998 0 17500 941 867 867 _ 17500 _17500 1808 1930) 2800 10500 598 1539 5611428 16450 33950 1159 1760! 3-Mar_1460 0 4350 69.8 700 960 92 0 16800 O 1042 2581 1008 2436 16800 50750 2050 1885 4-Mar_1620 0 5970 69.8 _1000 960 91 2100 19600 0 445 3026 421 2857 21700 _72450 866 2256 5-Mar_ 1500 3 7470 100.8 600 1120 92 16800 O 1080 4106 1036 3893 16800 89250 2116 1891 6-Mar_ _ 940 8410 100.8 2400 590 83 0 10500 4941 9047 4858 8751 10500 99750 9799 20299 7-Mar_1350 9760 100.8 700 805 91 015750 834 9881 1027 9778 15750 115500 1861 17611 8-Mar_1310 11070 100.8 2300 855 __91 QO 15750 1144 11025 1062 10840 15750 131250 2206 17956 9-Mar_1540 12610 100.8 550 905 93 5600 _ 10500 126 11151 197 11037 19250 150500 323 1957: 10-Mar_1460 14070 100.8 600 895 _93 19600 400 11551 865 11902 19600 170100 1265 20865 11-Mar 1580 3 15650 130.8 1000 937 92 19600 104 11655 121 12023 19600 189700 225 1982: 12-Mar_ 1530 17180 130.8 1800 935 19600 599 12254 613 12636 19600 209300 1212 20812 13-Mar_ 1570 25 18750 155.8 700 965 92 16800 445 12699 572 13208 16800 226100 1017 17817 0 16800 0 _1555 14254 1593 14801 16800 242900 3148 19948 3150 2800 10500 1332 15586 1235 16036 16450 259350 2567 19017 8400 QO 12250 417 16003 393 16429 20650 280000 810 _21460. on | 1-Mar 13800 1380 0.0 600 832 88 O 2Mar_1510 69.8 2890 69.8 800 836.91 ~~3150 g o oO o o oOo a = o o a SoJO|C|O wo eed a Slo}l|o|o Oo][oO}|Co|Co Oo o o|O oO oa o oJO|o}|oO ojo|o}|o o o 14-Mar_1370 0 20120 155.8 700 938 93 15-Mar_1470 20 21590 175.8 2000 860 93 16-Mar_ 1710 0 23300 175.8 2100 968 94 17-Mar_1780 20 25080 195.8 1800 1138 94 8400 Q 12250 618 16621 602 17031 20650 300650 1220 21870 o]Oo|O 18-Mar_1680 0 26760 195.8 900 1089 93 NO = So o 0 16800 _ 3500 437 17058 434 17465 22400 323050 871 23271 19-Mar_1700 17.6 28460 2134 1900 980 94 0 019600 0 105 17163 538 18003 19600 342650 643 20243 20-Mar_ 1420 35.2 29880 2486 700 1044 93 0 0 8400 5250 1326 18489 1241 19244 13650 356300 2507 16217 21-Mar_ 1480 17.6 31360 266.2 905 500 93 0 0 8400 12250 717 19206 676 19920 20650 376950 1393 22043 23-Mar 1500 0 34330 266.2 600 944 93 0 0 16800 3500 1149 21230 1208 21972 20300 414050 2357 22657 25-Mar_ 1630 15 37500 281.2 1300 1170 90 2100 0 16800 0 2212 24668 2138 25239 18900 449750 4350 23250 27-Mar_1140 1339770 323.2 600 992 63 0 0__11200 0 3839 33510 3724 33827 11200 474950 7563 18763 31-Mar 16700 45250 387.4 1900 1215 92 1050 1050 19600 0 1606 37839 1686 44341 1050 520800 3292 24992 MONTHLY TOTAL KWH PRODUCED 602980 DIESEL KWH PRODUCED PER GAL OF FUEL 11.51 % WIND TURBINE KWH PRODUCED TO TOTAL GRID 13.63% FUEL SAVINGS BY WIND TURBINE OPERATIONS IN GALLONS (AVOIDED GALLONS USED) 7140 ‘03-Mar-98 Production.xls 9/17/99 APRIL 1998 1-Apr_ 1670 13.5 1670 13.5 2000 1019 4200 Q 14000 02094 2094 ~=—2062_~—-2062_—- 18200 ~—-18200 4156 22356 2-Apr___ 1730 0 3400 13.5 900 1030 1050 Q 22400 0 160 2254 164 2226 23450 41650 324 23774 92 96 3-Apr__ 1630 13.6 5030 27.1 2400 879 95 1050 0 8400 _ 7000 726 _2980 737 __2963 16450 _58100 1463 17913 93 83 84 4-Apr___ 1450 0 6480 27. 12400 902 1050 0 017500 491 3471 480 3443 18550 76650 971 19521 5-Apr___ 1330 0 7810 27.1 500 836 1050 0 015750 1403 4874 1358 4801 16800 93450 2761 19561 6-Ap 1110 0 8920 27.1 2100 690 0 0 12250 4202 9076 4114 8915 12250 105700 8316 20566 0 7-Apr__ 1190 7.9 10110 35.0 700 917 85 0 0 11200 5250 4183 13259 4097 13012 16450 122150 8280 24730 8-Apr___ 1700 23.7 11810 58.7 500 1082 96 4200 QO 16800 O 1454 14713 1417 14429 21000 143150 2871 23871 9-Apr__ 1590 _20 13400 78.7 800 972 94 2100 016800 0 937 _ 15650 883 15312 18900 162050 1820 20720 10-Apr_1270 _18 14670 96.7 2000 847 94 0 3150 8400 1750 2451 _=(18101_ 2420 17732 13300 175350 4871 18171 11-Apr__ 1310 0 15980 96.7 _600 88194 1050 0 14000 4986 23087 _4722 22454 15050 190400 9708 24758 12-Apr 920 26.5 16900 123.2 1900 895 85 0 0 8750 4591 27678 4482 26936 _8750 199150 9073 17823 13-Apr__ 1480 0 18380 123.2 1900 972 90 6300 0 14000 1949 29627 1913 28849 20300 219450 3862 24162 i 0 14-Apr__1700 35 _20080 158.2 1000 975 91 8400 10500 ___: 1316 30943 1255 30104 18900 238350 2571 21471 15-Apr__ 1290 0 21370 158.2 2400 875__ 90 4200 8400 5250 _4441 35384 4497 34601 17850 256200 8938 26788, 16-Apr__ 1660 0 23030 158.2 700 1068 95 2100 19600 0 653 _36037 619 35220 21700 277900 1272 22972 17-Apr__1490 61.6 24520 219.8 630 1038 93 0 11200 _5250 315 36352 424 35644 16450 294350 739 17189 5 94 18-Apr__1410 0 25930 219.8 700 907 11200 _ 8750 516 36868 291 35935 19950 314300 807 20757 19-Apr__ 1340 0 27270 219.8 630 820 93 0 15750 596 37464 520 36455 15750 330050 1116 16866 20-Apr__ 1490 0 28760 219.8 600 874 94 14000 _5250 298 37762 269 36724 19250 349300 567 19817 21-Apr__ 1540 0 30300 219.8 2000 915 94 19600 0 153 _ 37915 148 36872 19600 368900 301 19901 0 0 0 0 22-Apr__1640 22.5 31940 2423 700 1040 95 0 19600 0 135 __ 38050 139 37011 19600 388500 274 19874 = 0 0 0 S]LO]O]O]|o}CO ojo|o ojo N Oo '3-Apr__ 1540 0 33480 242.3 620 1060 94 19600 0 475 38525 467 37478 19600 408100 942 20542 24-Apr__ 1220 0 34700 242.3 2100 943 95 5600 5250 _2282 ~+=40807 _2238 39716 10850 418950 4520 15370 95 S-Apr__ 1150 0 35850 242.3 700 1038 0 14000 2682 43489 2646 42362 14000 432950 5328 1932 6-Apr__1510 25 37360 267.3 600 866 94 5251 OQ 12250 362 43851 333 42695 17500 450450 695 1819: '7-Apr__ 1470 0 38830 267.3 2100 986 94 19600 0 553 44404 514 43209 19600 470050 1067 20667 8-Apr__1630 _25 40460 292.3 800 1032 95 2100 19600 0 665 45069 595 43804 21700 491750 1260 22960. 29-Apr__1610 0 42070 292.3 900 1071 94 1050 19600 0 846 45915 767 44571 20650 512400 1613 22263 30-Apr__1640 17.6 43710 309.9 550 1075 94 2100 0 16800 0 246 46161 248 44819 18900 531300 494 19394 43710 46161 44819 531300 N olo|o oo nN o o a NIN Oo ojo|o MONTHLY TOTAL KWH PRODUCED 622280 DIESEL KWH PRODUCED PER GAL OF FUEL 12.16 % WIND TURBINE KWH PRODUCED TO TOTAL GRID 14.62% FUEL SAVINGS BY WIND TURBINE OPERATIONS IN GALLONS (AVOIDED GALLONS USED) 7485 7 7 April 11th - Off line for current assymetry (Phase Imbalance) (04-Apr-98 Production.xis April 17th - Power Outage due to truck hitting power pole 9/17/99 MAY 1998 5250 356 356 382 382 19250 19250 738 1998: 1-Ma 1470 _ 44 1470 44 700 905 94 0 0 8 2-May 1270 0 2740 44 1100 760 94 0 0 0 14000 1084 1440 1098 1480 14000 33250 2182 16182 4-May 1330 39.9 5370 53.1 635 860 94 0 0 11200 5250 1048 3055 1039 3116 16450 67200 2087 18537 5-May 13300 6700 53.1 2100 950 94 0 0 16800 0 1597 4652 1627 4743 16800 64000 3224 20024 6-May 1510 0 8210 53.1 900 1068 __95 0 016800 0 604 5256 606 5349 16800 100800 1210 18010 7-May 1270 14 9480 67.1 600 912 95 0 0 11200 3500 2101 7357 2042 7391 14700 115500 4143 18843, @-May 1010 0 10490 67.1 700 93193 0 0 0 12250 3378 10735 3303 10694 12250 127750 6681 18931 9-May 1100 10 11590 77.1 800 904 93 0 0 0 14000 2698 13433 2606 13300 14000 141750 5304 19304 10-May 1180 0 12770 77.1 2300 895 __95 0 0 0 14000 2505 15938 2435 15735 14000 155750 4940 18940 11-May 1180 20 13950 97.1 2300 852 94 01050 0 12250 2934 18872 2875 18610 13300 169050 5809 19109 12-May 1420 15 15370 112.1 1900 1020 93 0 11200 3500 2266 21138 2288 20898 15750 184800 4554 20304 13-May 1200 4.4 16570 116.5 2200 790 89 0 0 16800 0 2386 23524 2352 23250 16800 201600 4738 21538 15-May 1290 0 19210 1445 700 85491 0 0 11200 5250 1452 26783 1360 26337 16450 234850 2812 19262 16-May 1090 0 20300 1445 700 797__81 0 0 0 14000 2454 29237 2418 28755 14000 248850 4872 18872 17-May_1190__1.5 21490 146.0 700 812 90 0 0 0 14000 _1707 30944 1732 30487 14000 262850 3439 17439 18-May 1400 0 22890 146.0 700 838 91 0 0 14000 3500 809 31753 743 31230 17500 280350 1552 19052 19-May 1430 0 24320 146.0 1500 825 92 0 0 16800 0 1289 33042 1206 32436 16800 297150 2495 19295 20-May 1450 22 25770 168.0 2200 1004 __93 0 016800 0 2003 35045 1914 34350 16800 313950 3917 20717 21-May 1460 0 27230 168.0 2300 1088 94 0 0 _ 19600 0 1727 36772 1665 36015 19600 333550 3392 22992 23-May 1400 __20 29990 206.0 700 895 93 3150 0 0 12250 1361 40260 1286 39336 15400 362950 2647 18047 24-May 1390 20 31380 226.0 2400 938 93 4200 0 0 14000 1461 41721 1405 40741 18200 381150 2866 21066 26-May 1150 20 33740 246.0 1900 930 93 0 0 2800 12250 3620 47738 3512 46554 15050 410200 7132 22162 27-May 1450 0 35190 246.0 2200 85293 1050 0 16800 0 1404 49142 1325 47879 17850 428050 2729 20579 28-May 1550 10 36740 256.0 1400 830 93 1050 0 16800 0 806 49948 732 48611 17850 445900 1538 19388 29-May 1280 0 38020 256.0 600 786 89 0 014000 0 2616 52564 2531 51142 14000 459900 5147 19147 30-May 1430 __31 39450 267.0 2300 849__93 0 0 19600 0 724 +53288 672 51814 19600 479500 1396 20996 31-May 1490 0 40940 287.0 600 87392 0 0 16800 0 333 53621 306 52120 16800 496300 639 17439 MONTHLY TOTAL KWH PRODUCED DIESEL KWH PRODUCED PER GAL OF FUEL 12.12 % WIND TURBINE KWH PRODUCED TO TOTAL GRID 17.56% FUEL SAVINGS BY WIND TURBINE OPERATIONS IN GALLONS (AVOIDED GALLONS USED) 8723 ‘05-May-98 Production.xls 9/17/99 JUNE 1998 i-Jun___ 1430 0 1430 0.0 700 86693 2-Jun__1170 26.4 2600 264 200 649 87 -Jun___ 1390 0 3990 26.4 1400 997__ 90 4-Jun__1460 20 5450 464 700 94491 5-Jun___ 1250 0 6700 46.4 600 1009 __88 -Jun 1320 25 8020 71.4 1100 922 92 0 7-Jun__ 1310 44 9330 75.8 2100 840 92 0 8-Jun___ 1450 0 10780 75.8 1900 97691 0 9-Jun_ 1700 _22 12480 97.8 1100 1014 92 2100 10-Jun_ _ 1600 0 14080 97.8 _700 1217 92 2100 1277__1277 _1193 1193 19600 19600 2470 22071 3891-5168 3730 4923 11200 _ 30800 7621 18821 2659 7827 _ 2537 7460 16800 47600 5196 21991 1973 9800 1889 9349 17850 65450 3862 21712 2646 12446 2541 11890 16800 82250 5187 21987 1781 14227 _1719 13609 14000 96250 3500 17501 1855 16082 1771 15380 16800 113050 3626 20426 1369 17451 1327 16707 16800 129850 2696 19496 303 17754 269 16976 18900 148750 572 19472 1218 18972 1155 18131 18900 167650 2373 21273 11-Jun__1430_13.2 15510 111.0 _600 944 91 0 1610 20582 1548 19679 19600 187250 3158 22758 12-Jun__1270 8.8 16780 119.8 2200 86192 749 21331 731 20410 14000 201250 1480 15481 3-Jun _ 1370 0 18150 119.8 700 834 91 16800 0 1624 22955 1591 22001 16800 218050 3215 2001 14-Jun_ _ 1380 0 19530 119.8 1000 78691 2800 10500 1766 24721 1646 23647 13300 231350 3412 16712 15-Jun__ 1550 0 21080 119.8 700 878 90 Q 12250 923 25644 890 24537 17500 248850 1813 19313 16-Jun__ 1660 0 22740 119.8 _700 97194 0 12250 468 26112 453 24990 19600 268450 921 20521 17-Jun_ 1730 30.8 24470 150.6 _700 1092 95 Q 12250 526 26638 523 25513 20650 289100 1049 21699 18-Jun 1650 17.6 26120 168.2 2100 915 93 0 12250 210 26848 199 25712 20650 309750 409 21059 19-Jun 1530 13.2 27650 181.4 900 90693 Q 12250 784 27632 745 26457 17500 327250 1529 19029 20-Jun 1240 0 28890 181.4 1400 71291 0 14000 1354 28986 1296 27753 14000 341250 2650 16650 21-Jun 1220 0 30110 181.4 1600 830 92 0 15750 _1687 30673 _1650 29403 15750 357000 3337 19087 22-Jun 1340 50 31450 231.4 2200 845 91 16800 1750 _1497 32170 1447 30850 18550 375550 2944 21494 23-Jun__1330__4.4 32780 235.8 800 929 89 14000 OQ 2050 34220 1956 32806 14000 389550 4006 18006 24-Jun__1390 0 34170 235.8 1100 905 90 16800 QO 1652 35872 1605 34411 16800 406350 3257 20057, 25-Jun_ _ 1380 0 35550 235.8 1400 860 90 16800 Q 1599 37471 1519 35930 16800 423150 3118 19918 26-Jun 1330 22.2 36880 258.0 1000 870 90 11200 3500 _1634 39105 1538 37468 14700 437850 3172 17872 27-Jun__ 1360 0 38240 258.0 2100 880 90 0 15750-1184 40289 1099 38567 15750 453600 2283 18033 28-Jun_ 1250 _ 24 39490 282.0 2100 840 91 0 14000 2121 42410 1965 40532 14000 467600 4086 18086 29-Jun__1180 0 40670 282.0 1900 920 90 0 14000 3082 45492 2921 43453 14000 481600 6003 20003 30-Jun 1110 0 41780 282.0 700 940 91 0 14000 _3763 49255 3519 46972 14000 495600 7282 2128; 41780 49255 46972 495600 D wo olo|o SLOLOLO]LO/OLOLO]OIO]oO SLOLOLOLOLOJLOLOLO]OJOIO So a Oo °o °o So = Oo Oo a 8 3 SLO] OL OLOLOLOJOJO[OJOJOI|O Colo oO SLOLOLOLOJLOJO/O]O]O}J|oO Oo np MONTHLY TOTAL KWH PRODUCED 591827 DIESEL KWH PRODUCED PER GAL OF FUEL 11.86 % WIND TURBINE KWH PRODUCED TO TOTAL GRID 16.26% FUEL SAVINGS BY WIND TURBINE OPERATIONS IN GALLONS (AVOIDED GALLONS USED) 8112 ‘06-Jun-98 Production.xls June 12th - D-Line Down 9/17/99 JULY 1998 1-Jul 1430 0 1430 0.0 800 939 90 -Jul 1140 0 2570 0.0 800 743 _ 90 3-Jul 1240 13 3810 13.0 2300 718 90 -Jul 1390 11.6 5200 246 530 846 93 1410 0 6610 246 2400 780 1560 20 8170 446 2100 850 93 -Jul 1520 0 9690 446 700 811 94 1530 23 11220 67.6 1100 914 93 1360 0 12580 67.6 _ 600 849 10-Jul_ 1220 _14 13800 81.6 _600 760 0 11-Jul__1170 0 14970 81.6 900 758 88 12-Jul__1410 0 16380 81.6 _600 804 92 13-Jul__ 1500 0 17880 81.6 2000 93 14-Jul_ 1550 0 19430 81.6 800 875 92 15-Jul_ 1690 _19 21120 100.6 600 999 94 16-Jul_ 1640 30 22760 130.6 1200 968 93 17-Jul__ 1600 0 24360 130.6 900 956 92 18-Jul_ 1510 15 25870 145.6 2100 923 92 Oo 0 12250 2406 2406 = 2265 ~=2265 16450 16450 4671 21121 0 14000 2958 5364 2795 ~—5060 15050 31500 5753 20803 0 14000 1369 6733 _1315 6375 14000 45500 2684 16684 015750 329 7062 302 6677 15750 _61250 631 16381 0 017500 300 _7362 268 6945 17500 78750 568 18068 4200 0 14000 382 7744 347 7292 18200 96950 729 18929 6300 0 10500 849 8593 773 __ 8065 16800 113750 1622 18422 4200 11200 7000 1060 9653 1008 9073 22400 136150 2068 24468 16800 O 1376 11029 1245 10318 16800 152950 2621 19421 5600 7000 __2190 13219 1989 12307 12600 165550 4179 16779 0 14000 2192 15411 2006 14313 14000 179550 4198 18198 0 15750 442 15853 417 14730 15750 195300 859 16609 11200 _ 7000 27 15880 24 14754 18200 213500 51 18251 19600 0 219 16099 212 14966 19600 233100 431 20031 1050 16800 _ 3500 129 16228 128 15094 21350 254450 257 21607 0 19600 0 52 16280 48 15142 19600 274050 100 19700 Q 19600 0 96 16376 110 15252 19600 293650 206 19806 0 19600 0 288 16664 313 15565 19600 313250 601 20201 19-Jul 1530 11.4 27400 157.0 2300 883 92 QO 19600 0 101 16765 113 15678 19600 332850 214 19814 20-Jul__ 1660 0 29060 157.0 2100 906 3150 14000 _ 2100 168 16933 165 15843 19250 352100 333 19583 21-Jul__ 1660 0 30720 157.0 _800 92 8400 0 10500 488 17421 467 16310 18900 371000 955 1985: 22-Jul_ 1750 _8.4 32470 165.4 1100 1095__93 8400 0 12250 212 17633 202 16512 20650 391650 414 23-Jul__ 1681 0 34151 165.4 600 979 90 4200 5250 010500 400 _ 18033 382 16894 19950 411600 782 2073. 24-Jul 1460 35.6 35611 201.0 1000 935 92 4200 0 14000 753 18786 705 17599 18200 429800 1458 19658 25-Jul___ 1330 0 36941 201.0 2300 837 __87 0 0 14000 1243 20029 1136 18735 14000 443800 2379 16379 26-Jul__1190 0 38131 201.0 2100 86188 0 0 14000 2623 22652 2392 21127 14000 457800 5015 19015 27-Jul__1420 18.5 39551 219.5 2100 969 90 0 14000 2100 1583 24235 1460 22587 16100 473900 3043 19143 28-Jul__ 1550 0 41101 219.5 1200 108392 0 19600 0 848 25083 792 23379 19600 493500 1640 21240 29-Jul 1380 25 42481 2445 1300 1100 92 1050 14000 2931 28014 2778 26157 15050 508550 5709 20759 30-Jul___ 1350 0 43831 244.5 1200 1041 93 0 16800 2277 30291-2147 28304 16800 525350 4424 2122. 31-Jul 1420 45 45251 289.5 1100 922 93 0 16800 1190 31481 1081 29385 16800 542150 2271 19071 ~ BQ] Peeler ejeycycyc © © = o oJolololo]ojojojo ojolo g a 0 0 0 SJOJLO]O]OJO]OJoO|oO © ao ao ao wo o a o X So g SCJOJCJ/C}oO|Co o Ny o o & ° oO MONTHLY TOTAL KWH PRODUCED 603016 DIESEL KWH PRODUCED PER GAL OF FUEL 11.98 % WIND TURBINE KWH PRODUCED TO TOTAL GRID 10.09% FUEL SAVINGS BY WIND TURBINE OPERATIONS IN GALLONS (AVOIDED GALLONS USED) 5080 ‘07-Jul-98 Production.xls 9/17/99 AUGUST 1998 1-Aug__ 1540 0 1540 0.0 2100 866 93 3150 Q 15750 293 293 273 273 _+18900 _ 18900 566 19466 0 0 2Aug 15900 3130 0.0 2200 918 93 6300 0 14000 -313_~—«606_~—=«270~—«543~—-20300 ~—-39200 583 20883 3Aug 1440134570 13.0 2300 932 92 0 0 14000 35001068 +1674 ~«976 +1519 17500 56700 2044 19544 SAug 16800 7820 13.0 1100 96892 3150 0 19600 0 800 3631 586 3423 22750 98350 1386 24136 G-Aug 1520 17.6 9340 30.6 1100 93393 0 0 16800 0 1036 4667-962 4385 16800 115150 1998 18798 7-Aug 1620 13.2 10960 43.8 800 971. 0 019600 0 423 5090 391 4776 19600 134750 814 20414 3 i 9 9 13-Aug 160010 21150 103.0 700 986 14-Aug 1540 30 22690 133.0 900 1006 15-Aug 14500 24140 133.0 1200 812 9 16-Aug 1430 15 25570 148.0 2000 827__93 0 0 16800 0 10194 0 9530 16800 306600 0 16800 18-Aug 1520 28.6 28590 196.6 1700 99889 3150 0 14000 144012405 1353 11602 17150 341600 2793 19943 19-Aug 1360 0 29950 196.6 1300 1050 92 2100 0 14000 0 2839 15244 2666 14268 16100 357700 5505 21605 20-Aug 1440 13.2 31390 209.8 2000 90091 1050 011200 5250 1222 +16466_ 1115 15383 17500 375200 2337 19837 2-Aug 1380 0 34230 209.8 600 840 93 0 02800 15750 —-940_~+18271 ~—«844 16991 18550 410200 1784 20334 23-Aug 1580 0 35810 209.8 1600 89092 1050 0 8400 1050058 18329 53 17044 19950 430150 111 20061 24-Aug 1560 19.8 37370 229.6 1900 1015 93 0 0 16800 0 216 18545 216 17260 16800 446950 432 17232 25-Aug _1590 0 38960 229.6 1700 980 __91 0 022400 0 345 _ 18890 330 17590 22400 469350 675 23075 26-Aug 1520 _14 40480 243.6 900 934 3 1050 0 16800 0 1073 19963 1073 18663 17850 487200 2146 19996 0 0 0 0 3150 3, 3 1 4 3150 0 16800 2019 9175 1880 8557 19950 253050 3899 23849 10 3 0 0 0 oO 0 QO 14000 0 889 10064 842 9399 17150 270200 1731 18881 0 19600 0 130 10194 131 9530 19600 289800 261 19861 0 0 0 7-Aug 1570 _16 42050 259.6 1200 958 2100 16800 1252 21215 1152 19815 18900 506100 2404 21304 9 93 8-Aug _1490 0 43540 259.6 _ 800 988 92 0 19600 1338 22553 _ 1181 20996 19600 525700 2519 22119 9-Aug 1700 15.4 45240 275.0 2000 951 92 0 6300 5600 _ 8750 149 22702 121 21117 20650 546350 270 20920 30-Aug 1670 0 46910 275.0 1400 910_ 9. 7350 0 10500 469 23171 429 21546 17850 564200 898 18748 3 0 1-Aug 1790 10 48700 285.0 2000 966 93 0 8400 0 12250 478 23649 431 21977 20650 584850 909 21559 MONTHLY TOTAL KWH PRODUCED 630476 DIESEL KWH PRODUCED PER GAL OF FUEL 12.01 % WIND TURBINE KWH PRODUCED TO TOTAL GRID 7.24% FUEL SAVINGS BY WIND TURBINE OPERATIONS IN GALLONS (AVOIDED GALLONS USED) 3799 , . August 4th - 1600Hr Heavy Loads tripped Power Plant off line (08-Aug-98 Production.xis August 5th - Wind Turbines off line for Maintenance 9/17/99 SEPTEMBER 1998 8400 012250 663 663 592 592 20650 20650 1255 2190: 0 14000 338 1001 314 906 23450 44100 652 2410 5250 11200 _ 3500 301 1302 307 1213 19950 64050 608 2055 0 11200 _7000 122 1424 127 1340 18200 82250 249 1844: 0 17500 105 1529 101__ 1441 17500 _99750 206 1770 0 17500 199 1728 199 1640 17500 117250 398 1789) 0 17500 165 _ 1893 155 1795 17500 134750 320 1782 14000 _ 5250 191 2084 173 1968 19250 154000 364 1961 16800 326 2410 312 2280 19950 173950 638 20588 19600 550 __ 2960 519 2799 19600 193550 1069 2066: 19600 170 _ 3130 156 2955 19600 213150 326 19926 16800 653 _ 3783 576 3531 16800 229950 1229 18029 18528 896 _ 4679 832 4363 18528 248478 1728 20256: 19600 673 _ 5352 613 4976 19600 268078 1286 20886 16800 1190 6542 1113 6089 16800 284878 2303 19103 16-Sep 1620 14 25100 405.0 1000 908 93 2101 16800 782 7324 711 6800 18900 303778 1493 20393 17-Sep _1480 0 26580 405.0 1000 1031 93 19600 920 8244 835 7635 19600 323378 1755 21355 8-Sep 1410 25 27990 430.0 1400 965 94 16800 QO 1490 9734 1380 9015 16800 340178 2870 19671 19-Sep 1540 0 29530 430.0 2100 909 94 0 12250 55__9789 35 9050 17500 357678 90 20-Sep _ 1230 9 30760 439.0 2100 912 93 6300 0 14000 163 9952 172 9222 20300 377978 335 21-Sep _1590 0 32350 439.0 2000 973 93 0 14000 _ 3500 476 10428 451 9673 17500 395478 927 22-Sep 1580 _ 20 33930 459.0 2100 968 94 19600 0 514 10942 477 10150 19600 415078 991 20591 23-Sep 1610 _20 35540 479.0 1200 1029 93 1050 16800 1136 12078 _1112 11262 17850 432928 2248 20098, 24-Sep 1500 _35 37040 514.0 _700 995 92 2100 1850 13928 1726 12988 18900 451828 3576 22476 25-Sep 1350 10 38390 524.0 1200 935 __ 89 0 1758 15686 _1702 14690 16800 468628 3460 20260 26-Sep _1370 0 39760 524.0 2000 888 91 1062 16748 1038 15728 14000 482628 2100 16100 27-Sep 1500 18 41260 542.0 1900 866 92 499 17247 45116179 19600 502228 950 20550 28-Sep _1460 0 42720 542.0 1600 92192 1384 186311410 17589 16800 519028 2794 19594 29-Sep 1400 _43 44120 585.0 2100 1013 __88 3530 22161 _ 3472 21061 16800 535828 7002 23802 30-Sep 1616 0 45736 585.0 _1100 959 92 3151 1433 23594 1362 22423 19950 555778 2795 22745 45736 23594 22423 555778 1-Sep__1770 248 1770 248.0 2200 96693 2-Sep _ 1840 0 3610 248.0 1200 1050 93 1750 0 5360 248.0 1100 988 92 -Sep _ 1550 0 6910 248.0 1100 929 93 1440 0 8350 248.0 900 778 93 6-Sep 1440 17 9790 265.0 1500 840 92 7-Sep _ 1450 0 11240 265.0 2000 817 0 1560 _60 12800 325.0 2000 905 92 1690 0 14490 325.0 1225 1065 __93 315) 10-Sep 1570 0 16060 325.0 1300 939 93 11-Sep 1560 _13 17620 338.0 1000 849 93 12-Sep 1440 20 19060 358.0 1000 800 __93 13-Sep 1440 15 20500 373.0 1000 814 93 14-Sep 1510 18 22010 391.0 2000 856 _93 15-Sep 1470 0 23480 391.0 700 928 92 gle g g @o SLOIDIOIWINIn s olo}ofofolojolo olojolo > © ® g So]o|o|o o STS] oC|oC|o SI[O]SC]SC/C]C]C}]CI|C S[O]SO/OC/OJO/OC]OCICO = o o So 8 3 ojo So]oO/oj|o SJLOJLOITO]C|CI|CO SIOLOJOJOC]C/C|Co 16800 ojo o}|o MONTHLY TOTAL KWH PRODUCED 601795 DIESEL KWH PRODUCED PER GAL OF FUEL 12.15 % WIND TURBINE KWH PRODUCED TO TOTAL GRID 7.65% FUEL SAVINGS BY WIND TURBINE OPERATIONS IN GALLONS (AVOIDED GALLONS USED) 3787 ‘09-Sep-98 Production.xls 9/17/99 OCTOBER 1998 1-Oct 1560 0 1560 0.0 730 1053-93 1050 0 0 1410 1410 1347 1347 15050 15050 2757 17807 2-Oct 1440 22 3000 22.0 1100 902 93 0 14000 3500 875 2285 849 2196 17500 32550 1724 19224 3-Oct 1330 0 4330 22.0 2405 900 92 0 0 5600 10500 1415 3700 1351 3547 16100 48650 2766 18866 4-Oct 1600 0 5930 22.0 2100 939 94 6300 0 0 14000 186 3636 137 3684 20300 68950 273 20573 5-Oct 1670 0 7600 22.0 2100 1008 93 2100 0 16800 1750 TS: 3011 73 3757 20650 89600 148 20798 7-Oct 1710 13.7 10930 57.7 0000 980 92 2100 0 19600 0 582 4720 536 4545 21700 130900 1118 22818 8-Oct 1491 14 12421 71.7 0000 996 89 0 0 16800 0 1529 6249 1398 5943 16800 147700 2927 19727 10-Oct 1530 20 15391 91.7 600 837 92 0 0 19600 0 113° 7517. 73 7069 19600 184100 186 19786 11-Oct 1520 15 16911 106.7 1100 940 92 0 0 16800 0 65. 7562 70 7139 16800 200900 135 16935 13-Oct 1700 0 20171 106.7 2100 961 97 0 7350 0 12250 526 8168 477 7675 19600 237300 1003 20603 16-Oct 1610 37.3 24911 167.1 700 925 94 0 5250 0 14000 241 12269 245 11588 19250 292600 486 19736 17-Oct 1400 1: 26311, 168.1 2100 860 93 0 0 0 15750 504 12773 750 12338 15750 308350 1254 17004 19-Oct 1530 33.1 29261 201.2 2000 905 93 0 0 19600 0 179 13528 161 13088 19600 344750 340 19940 21-Oct 1800 20 32641 240.5 1100 990 940 3150 0 16800 0 260 13927 280 13491 19950 385000 540 20491 22-Oct 2010 0 34651 2405 700 1007 94 3150 0 16800 0 343 14270 305 13796 19950 404950 648 2059: 23-Oct 1540 25 36191 265.5 700 946 93 1050 0 16800 0 601 14871 509 14305 17850 422800 1110 1896 24-Oct 1360 15 37551 280.5 800 850 90 ) 0 14000 0 1395 16266 1304 15609 14000 436800 2699 1669! 27-Oct 1730 0 42181 310.5 1800 961 94 1050 0 16800 0 516 18790 488 18043 17850 491050 1004 28-Oct 1510 50 43691 360.5 0000 900 93 1050 0 16800 0 1566 20356 1468 19511 17850 508900 3034 31-Oct 1490 0 48441 360.5 2000 830 93 0 2100 11200 5250 0 20356 0 19511 18550 563150 0 18550, MONTHLY TOTAL KWH PRODUCED 603017 DIESEL KWH PRODUCED PER GAL OF FUEL 11.63 % WIND TURBINE KWH PRODUCED TO TOTAL GRID 6.61% FUEL SAVINGS BY WIND TURBINE OPERATIONS IN GALLONS (AVOIDED GALLONS USED) 3429 10-Oct-98 Production.xls 9/17/99 NOVEMBER 1998 1-Nov 15300 1530 0.0 1800 827_ HO ( 0 0 0 0 19600 19600 0 19600 2-Nov_ 16500 3180 0.0 1700 1200 95 1050 0 0 O00 20650 40250 0 1610 4-Nov_ 1670 0 6460 46.0 600 103894 ~—-3150 0 030945105 2805 4634 19950 80850 5899 25849 S-Nov 1680 17.6 8140 63.6 600 98294 2100 0 0 1210 6315 1142 5776 18900 99750 2352 21252 @Nov 1440 0 9580 63.6 700 921 94 0 0 02235 8550 2291 8067 16800 116550 4526 21326 7-Nov 155027 11130 90.6 1800 938 94 2100 0 0 154210092 1663 9730 18900 135450 3205 22105 &Nov 1910 13.2 13040 103.8 1100 960 94 1050 0 0 384213934 3694 13424 17850 153300 7536 25386 9Nov 1590 53 14630 156.8 1900 1029 94 2100 0 0525 14459 694 14118 16100 169400 1219 17319 10-Nov 1880 0 16510 156.8 600 115094 5250 0 0347 14806 313 14431 22050 191450 660 22710 12-Nov_ 1790 25 20270 181.8 600 105995 4200 0 0360 15462467 15354 21000 236600 827 21827 13-Nov 1660 __25 21930 206.8 700 1025 95 1050 0 0364 15816 362 15716 20650 257250 716 21366 14-Nov 1580 15 23510 221.8 600 958 94 03150 15750 522 16338 468 16184 18900 276150 990 19890 15-Nov 1630 25 25140 246.8 2300 (1053 95 05250 0 15750741 _~+17079 655 16839 21000 297150 1396 22396 16-Nov 1650 0 26790 246.8 2200 96294 07350 0 12250 1194182731138 17977 19600 316750 2332 21932 17-Nov_1780_ 10 28570 256.8 1800 1196 95 0 8400 0 12250 1475 19748 1424 19401 20650 337400 2899 23549 18-Nov 1950 0 30520 256.8 700 1129 95 0 9450 0 14000217 ~+19965 213 19614 23450 360850 430 23880 19-Nov 1940 35 32460 291.8 700 1121 96 09450 0 14000 19320158 200 19814 23450 384300 393 23843 20-Nov_ 1850 0 34310 291.8 700 1154 96 0 8400 0 14000 207 20365 192 20006 22400 406700 399 22799 21-Nov 1760 0 36070 291.8 2355 1074 ~96 06300 5600 7000~~~—«292~«20657 ~~—«247 ~+20253 18900 425600 539 19439 22-Nov_1740 50 37810 341.8 200 97295 03150 19600 059621253 548 20801 22750 448350 1144 23894 23-Nov 1940 30 39750 371.8 1200 1013 95 05250 16800 0 1439226921314 22115 22050 470400 2753 24803 24-Nov_1780 0 41530 371.8 800 112195 0 7350 14000 0106423756 1044 23159 21350 491750 2108 23458 27-Nov_1740_37.4 46540 462.0 630 106095 3150 0 16800 0270 25213 259 24478 19950 554050 529 20479 28-Nov_1190 26.4 47730 488.4 1700 1010 88 0 0 14000 0 5030 30243 2808 27286 14000 568050 7838 21838 30-Nov 1860 39.6 51180 528.0 2000 1080 94 2100 +1050 16800 0243 32155 76 27362 19950 607600 319 20269 51180 32155 27362 607600 MONTHLY TOTAL KWH PRODUCED DIESEL KWH PRODUCED PER GAL OF FUEL 11.87 % WIND TURBINE KWH PRODUCED TO TOTAL GRID 8.92% FUEL SAVINGS BY WIND TURBINE OPERATIONS IN GALLONS (AVOIDED GALLONS USED) 5013 11-Nov-98 Production.xis | Nov 1-3 WT Production lost due to Navy Utilities construction for 3rd Wind Turbine Installation 9/17/99 DECEMBER 1998 1-Dec 1980 8.8 1980 8.8 1800 1235 95 5250 0 16800 0 774 774 +1512 1512 22050 22050 2286 24336 2-Dec 2040 17.6 4020 26.4 2000 1175 96 7350 0 16800 0 197 971 200 1712 24150 46200 397 24547 4-Dec 1950 32.2 8880 586 700 1009 92 4200 0 16800 0 2498 4799 2417 5602 21000 90300 4915 25915 5-Dec 1460 0 10340 58.6 700 1014 95 7350 0 2800 10500 0 4799 0 5602 20650 110950 0 20650 6-Dec 1750 0 12090 58.6 600 1129 96 7350 0 0 12250 0 4799 0 5602 19600 130550 0 19600 7-Dec 1990 19.4 14080 78.0 1800 1130 97 9450 0 0 15750 4267 9066 5680 11282 25200 155750 9947 35147 10-Dec 2080 0 20090 98.0 700 1200 96 0 6300 19600 0 173 12477 251 14847 25900 225050 424 26324 12-Dec 1720 7 23660 130.0 800 990 95 0 5250 5600 8750 30 12665 286 15318 19600 265650 316 19916 13-Dec 1590 10 25250 140.0 600 900 96 0 5250 5600 8750 1249 13914 1147 16465 19600 285250 2396 21996 15-Dec 1730 37 28580 181.0 1100 1028 96 4200 2100 5600 8750 1260 17740 1675 21203 20650 323750 2935 23585 16-Dec 1780 20 30360 201.0 1200 1056 96 5250 0 16800 0 835 18575 1455 22658 22050 345800 2290 24340 17-Dec 1650 0 32010 201.0 1100 1184 94 1050 0 19600 0 739 19314 673 23331 20650 366450 1412 22062 18-Dec 1490 13.7 33500 214.7 1800 1091 94 6300 0 16800 0 97 19411 90 23421 23100 389550 187 23287 20-Dec 1320 16.3 36350 234.9 2000 1091 86 0 0 16800 0 3892 24830 3804 28708 16800 417550 7696 24496 21-Dec 1860 27 38210 261.9 1800 1154 96 6300 0 14000 3500 951 25781 1062 29770 23800 441350 2013 25813 22-Dec 2010 13 40220 274.9 700 1200 95 5250 0 16800 0 479 26260 498 30268 22050 463400 977 23027 23-Dec 1830 0 42050 274.9 700 1100 97 4200 0 19600 0 449 26709 516 30784 23800 487200 965 24765) 24-Dec 1800 35 43850 309.9 600 1100 96 1050 0 16800 0 194 26903 225 31009 17850 505050 419 18269 25-Dec 1670 15 45520 324.9 600 1016 96 1050 0 22400 0 341 27244 326 31335 23450 528500 667 24117 26-Dec 1610 0 47130 324.9 600 912 96 9450 10500 0 0 117 27361 147 31482 19950 548450 264 20214 28-Dec 1700 30 50410 354.9 2100 1000 95 2100 0 14000 1750 257 27681 218 31785 17850 586250 475 18325 29-Dec 1790 0 52200 354.9 2000 1005 96 3150 0 19600 0 393 28074 350 32135 22750 609000 743 23493 31-Dec 1440 10 55380 373.7 600 874 95 1050 0 8400 7000 1090 29516 1028 33522 16450 645400 2118 18568 OTAL KWH PRODUCED DIESEL KWH PRODUCED PER GAL OF FUEL 11.65 % WIND TURBINE KWH PRODUCED TO TOTAL GRID 8.90% FUEL SAVINGS BY WIND TURBINE OPERATIONS IN GALLONS (AVOIDED GALLONS USED) 5409 12-05-98 Wind Turbines off line ~ 8 hours - Pwr Plant Grid Failure 12-05, 12-06 Computer Display Problem 12-Dec-98 Production.xls 9/17/99 JANUARY 1999 0 14000 606 606 742 742 17150 17150 1348 18498 t-Jan__1510 27.4 1510 27.4 100 803 95 3150 0 2-Jan 1580 0 3090 27.4 2000 88295 5250 0 0 1400044 ~650_~~«90~=«832~*19250 ~+36400 ~—=«d134 19384 3Jan 1540 0 4630 27.4 1900 81794 5250 0 012250267 ~«917~—=«288_~—1120_~+17500 53900 —+555 18055 4-Jan 1700 24.2 6330 51.6 1800 945 95 7350 0 0 12250312 +1229 ~~400 +1520 19600 73500 712 20312 SJan 1770 22 8100 73.6 700 1065 94 7350 0 12250397 1626 ~~—-384_~+1904 19600 93100 781 20381 6-Jan 1840 10 9940 83.6 600 1081 95 7350 0 0 14000292 +1918 + ~—-231 +2135 ~+21350 114450 523 21873 7-Jan 1780 0 11720 83.6 2000 1110 94 8400 0 0 140001493 3411_—«*1402_~—«:3537 22400 ~136850 2895 25295 Jan 1630 0 13350 83.6 1000 1025 94 2100 0 0 1255 4666 1224 4761 18900 155750 2479 21379 oJan 1630 20 14980 103.6 1800 925 95 0 0 0 91 4757 ~~ 180 4941 19600 175350 271 19871 10-Jan 1630 8 16610 111.6 1800 96395 1050 0 0 0 4757_—«34_~«4975 17850 193200 34 17884 12-Jan 2010 10 20420 156.6 800 119095 5250 04325755 598 6103 24850 240800 1030 25880 13-Jan 2090 22 22510 178.6 700 123895 7350 0 35758 100 6203 24150 264950 103 24253 15-Jan 1690 22 26170 222.6 700 10230 2100 0 288 6240 365 7259 18900 306250 653 19553 16-Jan 1430 17.6 27600 240.2 2300 902 95 0 02271 8511 2181 9440 16800 323050 4452 21252 t7-Jan 1410 20 29010 260.2 1900 997 95 2100 0517 9028 473 9913 21700 344750 990 22690 18-Jan 1680 0 30690 260.2 900 928 95 0 0 0 310 9338 307 10220 19600 364350 617 20217 19-Jan 1880 40 32570 300.2 1900 1209 95 5250 0 0687 10025 684 10904 22050 386400 1371 23421 20-Jan 1650 0 34220 300.2 1800 1250 90 4200 0 0 3643 13668 3476 14380 18200 404600 7119 25319 21-Jan 1460 0 35680 300.2 700 1210 86 1050 0 0 4306 17974 4102 18482_17850 422450 8408 26258 22-Jan 1890 35.5 37570 335.7 600 1460 91 4200 0 140001750 326 18300 344 16826 19950 442400 670 20620 23-Jan_ 1880 15 39450 350.7 700 1075 92 8400 0 012250 191 ~+18491 ~~ 180 19006 20650 463050371 21021 24-Jan 1750 19.9 41200 370.6 1900 1110 90 8400 0 0 14000 1608 20099 1590 20596 22400 485450 3198 25598 25-Jan 1870 19.8 43070 3904 1900 1250 91 8400 0 0 12250 1691 21790 1610 22206 20650 506100 3301 23951 26-Jan_ 17800 44850 390.4 1900 1395 ~90_~—~—«9450 0 0 14000 3474252643367 25573 23450 529550 6841 30291 27-Jan_ 2080 0 46930 390.4 800 138195 7350 0 14000 5250 1239 26503 1360 26933 26600 556150 2599 29199 28-Jan 2160 13.2 49090 403.6 600 133095 8400 0 19600 0 333 26836 330 27263 28000 584150 663 28663 29-Jan_ 2060 35 51150 438.6 700 1280 95 7360 0 16800 0 230 27066 237 27500 24150 608300 467 24617 3i-Jan 1650 15 54680 453.6 2300 1031 94 4200 0 _ 14000 0 1863 28982 2577 30333 4200 634550 4440 22640 MONTHLY TOTAL KWH PRODUCED 693865 KWH PRODUCED PER GAL OF FUEL 11.60 % WIND TURBINE KWH PRODUCED TO TOTAL GRID 8.55% FUEL SAVINGS BY WIND TURBINE OPERATIONS IN GALLONS (AVOIDED GALLONS USED) 5111. 01-Jan-99.xls 9/17/99 FEBRUARY 1999 1-Feb 2000 37 _—=-2000 (37.0 2000 1109 95 7350 16800 0 360 360 508 508 24150 24150 868 25018 2-Feb 2090 12.3 4090 49.3 700 A171" 96 4200 16800 _1750 83 443 403 91122750 _46900 486 23236 3-Feb 2150 185 6240 67.8 600 1260 92 9450 16800 0 0 443 140 1051 26250 73150 140 26391 4-Feb 2140 20 8380 87.8 2100 1280 93 7350 0 16800 0 845 1288 1187 2238 24150 97300 2032 2618 5-Feb 1960 37.2 10340 125.0 700 1220 94 4200 2100 _ 16800 O 1127 2415 1083 3321 23100 120400 2210 2531 6-Feb 1790 0 12130 125.0 1800 1105 94 0 5250 8400 7000 1344 3759 1274 4595 20650 141050 2618 2326 7-Feb 1870 17.6 14000 142.6 1900 1115 94 9450 0 14000 781 4540 763 5358 23450 164500 1544 2499. 8-Feb 1950 20 15950 162.6 1800 1242 96 014000 460 5000 433 5791 23450 187950 893 2434 9-Feb___ 1830 0 17780 162.6 1600 1300 96 8400 2800 10500 2092 7092 1962 7753 21700 209650 4054 25754 10-Feb 1840 17.6 19620 180.2 700 1360 93 6300 11200 3500 2602 9694 3548 11301 21000 230650 6150 27150. 11-Feb 1750 14.5 21370 194.7 600 1255 92 6300 _ 14000 QO 2923 12617 2899 14200 20300 250950 5822 26122 12-Feb 1920 33 23290 227.7 600 1324 94 6300 _ 16800 515 13132 546 14746 23100 274050 1061 24161 13-Feb 1820 0 25110 227.7 700 1080 94 4200 _ 16800 852 13984 762 15508 21000 295050 1614 2261 14-Feb 1540 25 26650 252.7 500 1046 _90 4200 _ 14000 3321 17305 _3169 18677 18200 313250 6490 24690 15-Feb 1660 15 28310 267.7 600 1118 94 3150 _ 16800 1903 19208 1855 20532 19950 333200 3758 23708. 16-Feb 1850 0 30160 267.7 700 1153 __ 93 6300 _ 14000 2235 21443 2199 22731 20300 353500 4434 2473. 17-Feb 1880 35 32040 302.7 600 1280 94 6300 _ 14000 2561 24004 2548 25279 20300 373800 5109 25409) 18-Feb 1930 0 33970 302.7 700 1118 94 0 7350 _ 16800 1781 25785 _1669 26948 24150 397950 3450 27600: 19-Feb 1930 181 35900 483.7 700 1201.95 5250 0 16800 0 568 26353 530 27478 22050 420000 1098 23148: 20-Feb 1840 _10 37740 493.7 600 1139 94 7350 5600 8750 _1030 27383 946 28424 21700 441700 1976 23676: 21-Feb 1510 _50 39250 543.7 _ 600 1160 95 5250 0 12250 3505 30888 4069 32493 17500 459200 7574 25074: 22-Feb 1820 0 41070 543.7 600 1172 95 8400 0 12250 1078 31966 1785 34278 20650 479850 2863 23513. 23-Feb 1940 0 43010 543.7 1800 1140 94 9450 0 14000 240 32206 210 34488 23450 503300 450 23900: 24-Feb 1960 _10 44970 553.7 _700 1279 94 6300 14000 _1750 873 33079 978 35466 22050 525350 1851 23901 25-Feb 1850 28 46820 581.7 1900 1300 93 6300 16800 0 2214 35293 2260 37726 23100 548450 4474 27574 26-Feb 1940 0 48760 581.7 _700 114494 6300 16800 0 761 36054 725 38451 23100 571550 1486 24586: 27-Feb 1880 _46 50640 627.7 _700 1060 95 9450 Q 12250 115 36169 114 38565 21700 593250 229 21929) '8-Feb 1860 0 52500 627.7 1800 1049 94 2100 735) 014000 199 36368 720 39285 23450 616700 919 2436 o|o o =) DS) f=) o|C|oO g 3 WIR] oO ojo 0 0 o]oO]oC]|oO ojoj]o|o & o SoJolojo|Co nN ojo|o o oO MONTHLY TOTAL KWH PRODUCED 692353 DIESEL KWH PRODUCED PER GAL OF FUEL 11.75 % WIND TURBINE KWH PRODUCED TO TOTAL GRID 10.93% FUEL SAVINGS BY WIND TURBINE OPERATIONS IN GALLONS (AVOIDED GALLONS USED) 6440 02-Feb-99.xIs 9/17/99 MARCH 1999 1-Mar__1780 0 1780 0.0 2300 1139-93 0 7350 012250 995 9951204 1204 19600 19600 2199 2179! 2-Mar___1830 0 3610 0.0 2000 118192 1050 7350 QO 12250 1892 2887 1770 2974 20650 40250 3662 2431 3-Mar__1710 0 5320 0.0 600 1254 94 0 7350 2800 10500 3385 ~—-6272_~—3355 «6329 ~20650 60900 6740 27391 4-Mar__1760 0 7080 0.0 _700 1305 94 4200 0 16800 0 26338905 2536 8865 21000 81900 5169 2616: 5-Mar__2090 _17 9170 17.0 700 121195 7350 16800 0 2 8907 568921 24150 106050 58 2420 6-Mar__2110 0 11280 17.0 2000 1203 94 9450 2800 12250 211 9118 240 9161 24500 130550 451 2495 a 1860 0 13140 17.0 1900 1265 94 9450 2800 12250 1014 _ 10132 2126 11287 24500 155050 3140 2764 8-Mar__1940 20 15080 37.0 2100 1215 95 8400 11200 3500 1312 11444 1313 12600 23100 178150 2625 2572 9-Mar_ 2020 20 17100 57.0 1900 1352 94 8400 19600 0 1174 12618 1352 13952 28000 206150 2526 3052 10-Mar _1950 0 19050 57.0 1900 138291 8400 16800 2888 15506 2896 16848 25200 231350 5784 30984 11-Mar_ 2030 _55_21080 112.0 _ 800 1395 91 7350 16800 1595 17101 _ 1883 18731 24150 255500 3478 27628: 12-Mar _ 1890 0 22970 112.0 500 1211 93 4200 19600 367 17468 362_ 19093 23800 279300 729 24529) 13-Mar _ 1830 0 24800 112.0 2000 1070 94 3150 16800 375 17843 358 19451 19950 299250 733 20683 14-Mar_ 1920 22 26720 134.0 2000 1069 94 7350 16800 410 18253 421 19872 24150 323400 831 24981 15-Mar_ 1840 _22 28560 156.0 1100 1296 94 7350 14000 1827 20080 1884 21756 21350 344750 3711 25061 6-Mar _ 1970 0 30530 156.0 600 114696 9450 14000 424 20504 622 22378 23450 368200 1046 2449 17-Mar 2290 _ 40 32820 196.0 1900 1228 96 9450 Q 16800 181 20685 346_ 22724 26250 394450 527 26777 18-Mar 2180 _18 35000 214.0 2200 1205 96 7350 3150 _ 16800 0 100 20785 123 22847 27300 421750 223 27523 19-Mar 1980 _13 36980 227.0 700 115196 0 8400 8400 _ 5250 456 21241 656 23503 22050 443800 1112 23162 20-Mar _ 1890 0 38870 227.0 2100 104396 0 8400 014000 338 21579 398 23901 22400 466200 736 23136 21-Mar 1840 35 40710 262.0 700 1027-96 2100 7350 012250 1007-22586 1115 25016 21700 487900 2122 23822 22-Mar _2620 9 43330 271.0 600 1090 96 9450 0 0 10500 2189 ~=24775 _2354 27370 19950 507850 4543 24493 23-Mar 1920 _22 45250 293.0 2000 1128 96 9450 0 5600 10500 1856 26631 1796 29166 25550 533400 3652 29202 24-Mar 2100 0 47350 293.0 600 1244 95 8400 16800 01234 27865 1302 30468 25200 558600 2536 27736 25-Mar_ 2180 _13 49530 306.0 1200 1258 95 9450 16800 0 641 28506 1139 31607 26250 584850 1780 28030, 26-Mar_ 1880 38 51410 344.0 2300 1114 94 6300 16800 O 1566 30072 1840 33447 23100 607950 3406 26506: 27-Mar _ 1960 0 53370 344.0 600 100596 7350 0 14000 0 170 30242 163 33610 21350 629300 333 21683: 28-Mar _ 1950 0 55320 344.0 600 1060 95 1050 8400 2800 __ 8750 752 30994 675 34285 21000 650300 1427 22427 29-Mar _ 1930 0 57250 344.0 2100 1125 94 0 10500 0 14000 115 31109 138 34423 24500 674800 253 24753: 30-Mar 1680 _16 58930 360.0 800 99693 0 8400 0 10500 3607 34716 _3500 37923 18900 693700 7107 26007 31-Mar 1610 20 60540 380.0 900 1325 88 0 7350 0 10500 4989 39705 4942 42865 10500 704200 9931 27781 OLOIN] oO o J ~ = 2 s SLOLO]O]OC/C/C|CO SJOLO]O]O}]O DLAlol|— o|o zy o o a> ° o]o|o MONTHLY TOTAL KWH PRODUCED 786770 DIESEL KWH PRODUCED PER GAL OF FUEL 11.63 % WIND TURBINE KWH PRODUCED TO TOTAL GRID 10.49% FUEL SAVINGS BY WIND TURBINE OPERATIONS IN GALLONS (AVOIDED GALLONS USED) 7099 03-Mar-99.xls 9/17/99 APRIL 1999 1-Apr__ 1890 24 1890 240 700 1340 90 5250 _ 16800 QO 2797 2797 _3006 3006 22050 22050 5803 27853 2-Apr__ 1760 191 3650 215.0 _500 1200 90 6300 8400 7000 2144 4941 3084 6090 21700 43750 5228 26928 3-Apr__ 1590 0 5240 215.0 2400 1225 90 0 14000 3724 8665 3577 9667 19250 63000 7301 26551 4-Apr___ 1590 13.2 6830 228.2 100 1165 90 0 12250 2416 11081 2432 12099 18550 81550 4848 23398 pr__ 1890 0 8720 228.2 2200 1155 90 0 14000 1178 12259 1146 13245 23450 105000 2324 2577. 2070 0 10790 228.2 900 1275 90 0 14000 423 12682 629 13874 23450 128450 1052 2450: 7-Apr___ 1900 41.8 12690 270.0 2300 1222 94 0 14000 2438 15120 2582 16456 23450 151900 5020 28470 8-Apr___ 1910 0 14600 270.0 600 1224 92 0 14000 1880 17000 1802 18258 22400 174300 3682 26082 9-Apr__ 1610 13.2 16210 283.2 800 119392 0 10500 4052 21052 4623 22881 17850 192150 8675 26525, 0-Apr__ 1960 0 18170 283.2 700 1232 93 Q 15750 870 21922 853 23734 25200 217350 1723 2692. 1-Apr__1920 13.2 20090 296.4 1200 1149 94 0 14000 1513 23435 1398 25132 22400 239750 2911 2531 12-Apr__ 2010 13.2 22100 309.6 _600 1205 95 5250 4200 11200 5250 _1243 24678 _1164 26296 25900 265650 2407 28307 13-Apr__2140 39.6 24240 349.2 1000 1230 95 8400 0 16800 0 506 25184 464 26760 25200 290850 970 26170 4-Apr__ 2200 0 26440 349.2 1100 1337 94 9450 0 16800 0 315 25499 284 27044 26250 317100 599 2684: 15-Apr__2100 _ 30 28540 379.2 800 1230 95 9450 0 _ 16800 0 106 25605 105 27149 26250 343350 211 26461 16-Apr__1470 _32_ 30010 411.2 600 1135-95 7350 0 8400 _ 5250 208 25813 370 27519 21000 364350 578 21578 17-Apr__ 1900 0 31910 411.2 600 948 95 0 5250 QO 12250 394 26207 365 27884 17500 381850 759 1825: 18-Apr__1540 35 33450 446.2 2000 870 94 5250 0 12250 645 26852 603 28487 17500 399350 1248 18748 19-Apr__ 1630 0 35080 446.2 1800 1010 94 014000 938 27790 850 29337 20300 419650 1788 22088, 20-Apr 1380 0 36460 446.2 2100 956 94 0 12250 _3254 31044 3049 32386 15400 435050 6303 21703 21-Apr__1300 0 37760 446.2 2100 1125 _ 86 0 10500 _3323 34367 _3248 35634 14700 449750 6571 21271 22-Apr__1600 _9.3 39360 455.5 _ 800 1089 90 0 10500 _2053 36420 1979 37613 14700 464450 4032 18732 23-Apr__1740 29.4 41100 484.9 500 1032 94 8400 2800 14000 379 36799 394 38007 25200 489650 773 25973 24-Apr__1600 0 42700 484.9 2000 905 94 0 16800 0 144 36943 144 38151 16800 506450 288 17088 25-Apr__1470 0 44170 484.9 2000 894 93 0 19600 QO 1071 38014 1026 39177 19600 526050 2097 21697 26-Apr__1670 27.3 45840 512.2 2100 964 94 2100 16800 QO 1408 39422 1383 40560 18900 544950 2791 21691 27-Apr__1600 0 47440 512.2 2000 1030__ 89 0 3150 16800 QO 2313 41735 2270 42830 19950 564900 4583 24533, 28-Apr__1430 _15 48870 527.2 1400 1215 94 1050 0 14000 QO 4605 46340 4620 47450 15050 579950 9225 2427 29-Apr__1500 _15 50370 542.2 700 1204 95 1050 0 16800 OQ 4033 50373 4004 51454 17850 597800 8037 25887 30-Apr__1780 25 52150 567.2 1400 1130 94 3150 0 16800 0 699 51072 668 52122 19950 617750 1367 21317 0 0 52150 567.2 0000 0 0 0 0 QO 51072 0 52122 0 617750 0 MONTHLY TOTAL KWH PRODUCED 720944 DIESEL KWH PRODUCED PER GAL OF FUEL 11.85 % WIND TURBINE KWH PRODUCED TO TOTAL GRID 14.31% FUEL SAVINGS BY WIND TURBINE OPERATIONS IN GALLONS (AVOIDED GALLONS USED) 8712 04-Apr-99.xIs 9/17/99 May 1999 1-Ma 1730___ 301730 30.0 2100 102795 0 010500 712 712 667 __- 66718900 _ 18900 1379 2027: 2-Ma 1520 0 3250 30.0 2000 1011 94 4200 012250 2413-3125 «2366 ~—3033_—« 16450 ~— 35350 4779 2122 }-Ma 1350 10 4600 40.0 2000 1118 94 2100 0 14000 4331 7456 4309 7342 16100 51450 8640 2474 4-Ma 1610 5 6210 45.0 600 1125 95 5250 0 14000 2930 10386 _2834 10176 19250 70700 5764 2501 2040 31.3 8250 76.3 800 1185 95 9450 015750 90 10476 97 10273 25200 _ 95900 187 2538 6-Ma 1900 0 10150 76.3 1100 1051 95 7350 0 14000 453 10929 396 10669 21350 117250 849 2219 7-Ma 1630 14 11780 90.3 800 948 93 4200 8400 7000 1133 12062 1060 11729 19600 136850 2193 2179. 8-Ma 1540 14 13320 104.3 1800 894 94 0 16800 0 679 12741 642 12371 16800 153650 1321 18121 9-Ma 1540 12.7 14860 117.0 1900 942 95 19600 0 511 13252 474 12845 19600 173250 985 2058 10-May 1660 18.9 16520 135.9 2000 952 94 0 19600 0 271 13523 264 13109 19600 192850 535 20135 11-May 1790 _42 18310 177.9 2200 1002.95 7350 8400 _ 8750 87 13610 76 13185 24500 217350 163 24663 1900 12 20210 189.9 1300 1021 95 7350 14000 0 256 13866 243 13428 21350 238700 499 2184 3-May _1900 0 22110 189.9 2300 987 95 8400 1400 0 102 13968 102 13530 9800 248500 204 14-May 1600 3123710 220.9 600 1035 __ 95 4200 11200 01693 15661 1650 15180 15400 263900 3343 18743 15-May 1390 15 25100 235.9 700 865 95 0 3150 0 14000 2263 (17924 2270 17450 17150 281050 4533 21683 16-May 1630 0 26730 235.9 900 890 95 0 3150 8400 _ 8750 294 18218 273 17723 20300 301350 567 20867 17-May 1710 _20 28440 255.9 _700 948 95 1050 0 19600 0 70 18288 66 17789 20650 322000 136 20786 18-May 1710 _18 30150 273.9 500 95395 2100 016800 0 908 19196 849 18638 18900 340900 1757 20657 19-May 1480 20 31630 293.9 1000 980 94 1050 016800 O 2553 21749 2445 21083 17850 358750 4998 22848 20-May 1660 1.3 33290 295.2 600 920 94 0 5250 5600 10500 2123 23872 ~—-2071 + 23154 21350 380100 4194 25544 21-May 1930 10 35220 305.2 _600 1101 95 0 9450 0 14000 64 23936 65 23219 23450 403550 129 23579 22-May _1860 0 37080 305.2 2200 1055 __-95 0 7350 Q 14000 76 24012 71_ 23290 21350 424900 147 2149 23-May 1830 _ 28 38910 333.2 600 1031-95 8400 0 14000 234 24246 212 23502 22400 447300 446 22846 24-May 1680 10 40590 343.2 2200 1005 95 7250 QO 12250 1797 26043 _1709 25211 19500 466800 3506 23006 25-May _ 1800 0 42390 343.2 600 1022.95 8400 O 12250 1163 27206 1085 26296 20650 487450 2248 22898 26-May 1920 18 44310 361.2 1200 1150 94 8400 0 14000 809 28015 766 27062 22400 509850 1575 2397; 7-May 1800 _26 46110 387.2 1900 1070 94 7350 0 14000 1024 29039 937_ 27999 21350 531200 1961 2331 28-May 1610 0 47720 387.2 _700 1000 93 5250 8400 7000 _1490 30529 1427 29426 20650 551850 2917 23567 '29-May _1550 2 49270 389.2 2300 1035 __93 0 16800 0 1590 32119 _1541 30967 16800 568650 3131 19931 30-May 1560 _ 28 50830 417.2 2100 970 __93 019600 0 1268 33387 1197 32164 19600 588250 2465 22065 31-May 1540 0 52370 417.2 2200 905 93 0 16800 0 1032 34419 1016 33180 16800 605050 2048 18848 MONTHLY TOTAL KWH PRODUCED 672649 OlO oe So ojo|o & v2 = 2 | Co oO o wo colo o]o|o a o o D = 2 1 0 10004 0 So]ol|Co Ny ojo|co|o =| ojo DIESEL KWH PRODUCED PER GAL OF FUEL 11.55 % WIND TURBINE KWH PRODUCED TO TOTAL GRID 10.05% FUEL SAVINGS BY WIND TURBINE OPERATIONS IN GALLONS (AVOIDED GALLONS USED) 5851 05-May-99.xls 9/17/99 APPENDIX C: Independent Paper on Wind-Diesel Hybrid Energy System Design and Operation [6] This appendix was included to display some of the earlier design and operational strategy developed by NREL and NREL sub-contractors for the remote wind-diesel system. 45 Report to NREL on the development and operation of a Wind/Diesel Spread Sheet Model The Evaluation of the Wind Resource at San Nicolas Island Alan H. Miller P.S.I. Golden, Colorado Introduction One use of renewable energy in the near term is the addition of wind turbines on isolated, (non-intertied, non-grid- connected), electrical systems such as on small to moderately populated islands or in very remote, sparsely populated areas. Such systems are usually powered by diesel engines. While fairly large wind systems have been installed in diesel powered grids on several of the off-shore islands in Great Britain, and also on some islands off the coast of Norway, none in a class of systems comprising a megawatt or more has ever been installed in the United States, it’s possessions or protectorates. This report provides an assessment of the wind energy potential on one island owned by the U.S. Navy off the southwest coast of California. The wind energy atlases produced by the Pacific Northwest Laboratory and it’s subcontractors over a decade ago, assessed the Channel Islands as having a class two wind resource - a rather low value. A more recent look and a number of qualitative indications gave rise to a program to collect some new wind data in an area appropriate for the installation of wind turbines on the island. This assessment makes use of that data. The assessment has been accomplished in a manner that went beyond the normal resource assessment in that it was done with the current diesel system in mind and in a manner not necessarily intended to maximize the wind energy capture but to integrate as much wind energy as possible without causing any significant disruption to the existing system other than to reduce it’s costs and the effluent of pollutants. The system under study here is the U.S. Navy’s San Nicolas Island, California. The secondary purpose of this report is to document the efficacy of a simple spreadsheet model to perform such resource assessments. The report is comprised of nine sections. Immediately following this Introduction is a section on the Background and history of this assessment and model development task. This is immediately followed by a concise statement of the purpose of the project and the model. The fourth section elucidates the assumptions that had to be made in developing the model. While the terminology ‘assumption’ might elicit the thought that the model is somewhat subjective, the section includes justification for each and every assumption and the potential effect of the assumption. A separate section on the existing power system on the island follows. Though it might well have been placed further forward in the Background section, it was placed where it is so that the technical background material included was in closer proximity to the remainder of the technical development of the system. The model, as will be explained later, was developed based on hourly average data. The description of the data and it’s sources comprises the following section. Thereafter the reader will find the details of the development of the spreadsheet model which lacks a name since it is based on commercially available software. When designing large machinery or complex systems, the ultimate driver for the project is typically, economics. The next section outlines the economics of the system. Obtaining many of the inputs to the economics was tenuous leading to the uneasy feeling that the numbers were something akin to a “guess”. Some of the costs for the turbines is admittedly a guess based on some experience. In all cases, it is believed that the values used are at least conservative. The final two sections show the results of the modeling effort and what conclusions can be drawn from the results. Recommendations have not been included as that seemed inappropriate and self-serving at best. August 8, 1996 } Report to NREL on the development and operation of a Wind/Diesel Spread Sheet Mode! Background Since 1994, the U.S. Navy, with the help of the National Renewable Energy Laboratory, (NREL) has been taking wind and related data on several of the Channel Islands off the coast of Southern California. The purpose of this effort was to evaluate the technical and economic potential of integrating wind energy into the diesel powered grid at a non-grid connected military installation. Further, wind energy needed to be evaluated as a means to both reduce the consumption of diesel fuel and abate the discharge of pollutants into the atmosphere. One of these island facilities among the Channel Island group, San Clemente Island, is moving ahead with the acquisition of wind turbines to integrate into the island power system. The island of interest in this report, however, is San Nicolas Island. While there is no immediate plan to incorporate wind energy on the island, there is a sufficient interest in wind energy to warrant performing this assessment and system operations modeling effort. Several groups have moved in the direction of modeling wind/diesel systems in an effort to demonstrate the utility of adding wind to a diesel system. The number of diesel powered, islanded grids on the face of the earth probably approaches ten thousand. Not all of these have viable wind resources to draw on but many may. The great majority of these are probably systems with a maximum peak demand of less than 100KW - relatively small village communities. Some preliminary analyses of these smaller system leads to the conclusion that the most desirable system and end effect can only be attained if the diesel(s) can be totally shut off. For small systems that can afford some small amount of battery or other storage, this is probably true. In a system comprising a 1000kW or more, the addition of multi-MWh capacity batteries is 1) too expensive and 2) extremely maintenance intensive and, 3) an environmental hazard. Therefore, battery storage was not included in the scenario. One high resolution, wind/diesel model, HYBRID2, is a simulation model with many diverse attributes. It has the capability to simulate minute by minute operations of a wind/hybrid/diesel system that can include storage and any or all of a number of prominent renewable technologies. It was developed by the University of Massachusetts in conjunction with NREL and sponsored by the U. S. Department of Energy. HYBRID2 may well be the epitome of hybrid energy system models with multiple resource modeling capability. However, for cases where only hourly wind and load data are available, a sophisticated model such as this are not justified. Models of this nature may be an expert tool to help define and describe an optimum operating strategy and even to specify optimal component sizes but are not practical tools for looking at low resolution, resource assessment type data. Purpose The purpose of this report, as stated previously, is two-fold. The first and most important is to provide an assessment of the wind resource at San Nicolas and the potential interactions of wind turbines with the diesel system. The second is to document the development and operation of a simple model of a wind/diesel system based on a MicroSoft Excel® Spread Sheet. It should be pointed out that developing a “model” per se, was not in the definition of the task. The resulting model, while being fairly versatile, is not a model in the true sense of the term but could be converted easily enough. The data used in this analysis was also run in the HYBRID2 model and results were obtained by another investigator. At the time this work is being done, the HYBRID2 model had not been fully validated and for the simulation run, the model was apparently allowed to specify the operating strategy of the diesel system including picking the optimum engine(s) to have on line as well as total engine shutdown. While this is an ideal operating situation, it is unrealistic in the example. A simple spread sheet model has the advantage of being as flexible and/or as complex as one is willing to program it. Assumptions in the Spreadsheet Model While the author, a long time advocate of wind energy, would hope that the results of any model would work to the utmost advantage of wind energy, the task of assessing the value of the resource with the spreadsheet model has been carried out very conservatively. The data handling techniques and the modeling development have all been accom- plished with a conservative bent. The idea was to “let the chips fall where they may”. None of the data was modified in any way other than to move a portion of one set as explained below. No exponent was applied to the August 8, 1996 2 Report to NREL on the development and operation of a Wind/Diesel Spread Sheet Model wind speed data nor was their any atmospheric density correction. Since the altitude of the site is only about 700 ft, a density correction would only amount to a factor of about .05 In the modeling and assessment effort, the load and wind data were used as provided. No finer scale data was available to develop power spectral density functions of within-hour load or wind speed variances. As noted above no shear exponent was applied which should have evidenced an increase in wind speeds generally whereas the small density correction would have worked the other direction essentially nullifying the shear induced increase. Since the wind input data is averaged over hourly periods, the subsequent convolution of the wind turbine generator power curve with the hourly average wind data should produce a reasonably conservative estimate of the energy produced. This is due to the fact that while the averaging process is linear, the power available in the wind follows the equation: P= %p av? where P = available power in the wind, p = atmospheric density, and V = the wind speed, and the curve fit to the wind includes a cubic function thereby producing the under-estimates. Wind resource data is usually analyzed and condensed into a convenient characteristic distribution such as the two parameter Weibull or the simpler, single parameter Rayleigh distribution in which the K (shape) parameter is set constant at 2. The concept of fitting a Weibull or Rayleigh distribution function to annual hourly averaged wind data and convoluting the fit with the power curve of a specific wind turbine to ascribe an estimated annual energy capture was successfully challenged over a decade ago. With the large, multi-megawatt wind turbines such as the MOD-1 and MOD-2 that were under development at the time, there was concern that the time derivatives in the operating strategy weren’t being adequately represented. Such methods of estimating annual energy capture did prove to be significantly in error (over-estimates). As mentioned earlier, the HYBRID2 model was apparently allowed to dictate the operating strategy of the diesel power plant. The spread sheet model also applied certain constraints. The only constraint on the diesel system was to set the minimum allowable load remaining on the system for the diesel plant to pick up. For the current presentation, the lower limit was set at 200KW minimum load. (A second iteration with the minimum set at 1OOKW was also run). This was based on the fact that wind turbines incorporating induction generators require some system load to work into or the system voltage and frequency become unstable. Obviously, induction machines also require field magnetization current from the line By setting this constraint it is felt one can maintain system stability (frequency and voltage) and reliability. The number, 200KW, could as well have been 100 or 500. The San Nicolas power system The array of diesel engines that are available to serve load on San Nicolas Island include two Caterpillar diesels with a capacity of 750KW each and three EMD diesels with capacities of SOOKW(2) and 1000KW(1). In a typical large utility operation, pairs of engines would be operated such that either one could pick up the entire expected load should the other engine fail or trip off line for some untoward reason. It is not known if the power plant on San Nicolas is equipped with Woodward Governors and Load Share devices like San Clemente’s powerhouse is, but it is presumed they are. The incorporation of this apparatus makes the operation of the diesel plant fairly simple and manages whatever engines are on line. Frequently, people apply a “rule of thumb” that a diesel generator should never be run below the 40% load level. The author’s experience with large (megawatt and larger) diesels suggested that this was quite possibly hyperbole rather than fact. The Electro-Motive Division (EMD) of Detroit Diesel in LaGrange, Illinois, was contacted to determine the minimum operating conditions for EMD engines. The engineering department indicated that as long as the engines were loaded to a higher level - say 60-80% of rated - for a while before being shut off, there was no problem running them at 10% load, for many hours. They pointed out that railroad locomotives are commonly left running at idle for entire weekends. It has also been pointed out that maintaining engine temperature is mandatory and in these large, sub- and multi-megawatt stationary engines, circulation of coolant and the maintenance of temperature is quite easy and usually accomplished with electrically operated, proportional! controllers rather than mechanical thermo- stats. August 8, 1996 3 Report to NREL on the development and operation of a Wind/Diesel Spread Sheet Model A significant factor in the philosophy of the spread sheet model is related to human engineering. The Navy’s interest is also in maintaining system reliability - that is, a minimum of potential system outages. The typical utility system operating strategy would operate with two engines on line and the load, assuming the presence of LoadShare devices, would be proportionally split. For instance, with the 750KW and a SOOKW diesel running, the 750KW engine would be called upon to provide about 60% of the total load and the SOOKW engine would provide the remaining 40%. The combination of diesels that the operators put on line at San Clemente is a function of the anticipated loads. For example, in the middle of the night with minimum load and no expected operational or abnormal load increases, the load is usually only about 450kW. Under these circumstances the operators would likely have two 500kW generators on line. Each would be running at about 50% load on average with no wind energy installed. In reality, on both islands, the operators are familiar and comfortable with their systems that they frequently operate with only one engine running at any time. If significant wind energy was added to the system there is a finite possibility that more wind energy would be available than needed. With the addition of as many as four 225kW rated wind turbines there is a very good possibility that the island might experience sufficient wind for the wind turbines to be generating their rated power of 225kW. Obviously, with two or more turbines running and the system load at it’s minimum, (~450kW) the system would be unstable. By constraining the wind turbines to only make up the difference between a minimum set point, say 200kW, and the total load, we might be left with only one wind turbine operating (depending on load) but the operating diesel(s) running at a 450kW or a little over 100kW each if the pair were running (about 20% load), well within the safe operating envelope. Obviously, with this sort of operating strategy the system is not minimizing the amount of diesel fuel used but by doing so we have maintained system stability and reliability and the operators are confident that their system will handle the load without interruption. As will be seen later, even with this less than maximum incorporation and utilization of wind, the economics of the entire system appear to be good. The Data The San Nicolas Island wind data is hourly average data. The data were collected on a tower located in reasonable proximity to the area available for wind turbine installation. The data were collected at the 30.5 meter (100’) level, a reasonable approximation to the likely hub height for modern wind turbines. The units of measure are meters/second. They are averaged from 10-minute averages. No other sample statistics such as the hourly standard deviation, skewness or kurtosis were available to the author. The hourly system load data in kilowatts, was transcribed from the operators hand written records to a spread sheet and represent, at least crudely, the operators best guess or “eyeball average” load for the hour. While this is not a terribly satisfying source for such data, looking at daily, weekly and monthly time series plots of the data does not reveal any obvious anomalies in the load and is acceptable for the purposes of this report. The two data sets were not for synonymous periods of time but had 10 months overlap on an annual basis. The wind data collection began on August 1, 1994 and ran for a year while the load data were for the year October 1, 1989 through September, 1990. Since the model is meant to be “representative” of any year, it was decided to shift the load data to mate August and September 1990 loads to the August and September 1994 wind data. It has been pointed out that some form of newer load data is available that indicates an annual increase in load and the suggestion was made that an across-the-board increase be applied to the data but the author has chosen not to do that for a number of reasons. Table 1. The load and wind data overlap. The Aug/Sept Load data were cut and pasted on to the front of the load file. Load 1989 Oct Nov Dec Jan Feb Mar Apr May — Jun Jul Aug Sept Wind 1994 Aug Sept _Oct Nov Dec Jan Feb Mar ____ Apr Mi Jun Jul To represent the wind turbine, the power curve for a current model, 225K W rated wind turbine was used. The manu- facturers literature included a 23 point tabulation of the wind speed vs. power output. This data was truncated at the 18 meter/second inflection point in the curve and a polynomial fit was developed. Second, third, fourth and fifth order fits were then calculated with the results shown in table 2 August 8, 1996 4 Report to NREL on the development and operation of a Wind/Diesel Spread Sheet Mode! Table 2 Results of fitting the Micon power curve data to a polynomial curve of several orders. Order of Fit Residual Variance 2 Coef. Of Determination (R’ 9743 9949 3 4 5 The fourth order was chosen since the residual variance was not reduced significantly by going to the fifth order and coefficient of determination (goodness of fit) was insignificantly better at the fifth order than for the fourth order fit. The power curve when fit with the fourth order polynomial along with the original power curve provided by the manufacturer is shown in Figure | below. Power Curve for the Example Turbine ~ Ss o 3 = o a. Wind Speed (mph) Figure 1. The Manufacturers power curve (the thin, dark, smooth line) and the 4" order polynomial fit to the tabulated data (the thick, gray, jagged line). The Model The model was developed in a Microsoft Excel® format. The spreadsheet was set up in it’s simplest form with the columns assigned as shown in Table 2 below. While it is possible to have a model such as this run totally externally to the data, the task was not to develop such a model but simply to derive the assessment data. The first two columns are self explanatory. They simply provide the identifying date and time stamp. The third column is the hourly averaged wind speed data. This data was originally in units of meters/sec but was converted to miles/hour for the convenience of the reader. Column four is the equivalent (time stamped) system load data and constitutes the data that was edited to place the last two months of data (August and September, ‘90) at the beginning of the file. This was done in case there was any seasonal component to the annual load or wind profile. Column five is the equivalent wind power out of one example turbine. It is calculated by taking the wind speed data in column three and convoluting it with the fourth order polynomial to the WTG power curve. August 8, 1996 2 Report to NREL on the development and operation of a Wind/Diesel Spread Sheet Model The sixth column simply eliminates all WTG power out data below the 5.1kW threshold. Here again, the value of 5.1 could have as easily been 10.1. It was intended to eliminate the questionably low output energy levels. Typically, with the lower wind speeds, the within hour variance is likely to have been high. Therefore, while the average may have been 5.1, there is a high probability that the wind turbine would never have started up because it was not at that speed for long enough uninterrupted periods. The energy contribution, or loss in this case, is so small as to be insignificant. The seventh, ninth, eleventh and thirteenth columns are written to determine if the addition of one, two, three, or four wind turbines at that output level would reduce the load remaining at the power house to below the minimum set point of 200KW. If the test turns out negative, (the addition of the turbine will not reduce the balance of load on the diesels below the minimum), then the flag in the following column is set to a “1” (turbine allowed to operate). If the addition of that turbine would reduce the load on the diesels below the set point limit, then the flag is set to a “0” (the turbine is not allowed to operate) . The leading and intervening columns, numbers six, eight, ten, and twelve are the power available with one, two, three, or four turbines in operation. The 14” and 15" columns simply sum the number of “1s” in “flag” columns seven, nine, eleven, and thirteen and the total wind power that can be utilized under the minimum set point constraint. The value in column 16 is the number representing the wind turbine power available that is not being utilized (the waste wind energy). Column 17 is the balance of load to be picked up by the diesels and column 18 is the equivalent fuel usage by the diesels. This last value was generated assuming a 13kWh gallon" rate. Table 2. Layout of the spread sheet columns. POT MCLs ers Wind Speed Average (mph)* System Load (kW) Power out of one WTG (kW unconstrained) Power out of first WTG (kW constrained** Test #1 for Excess Power Power out of second WTG (kW constrained**) 9 ___| Test #2 for Excess Power 10 — | Power out of third WIG (kW constrained**) ll | Test #3 for Excess Power . ; 12 Power out of fourth WTG (kW constrained**) 13 Test #4 for Excess Power gy allowed on line (kW) 18, Number of turbines on line 16 Waste Wind Energy (kWh not generated due to curtailment of some number of WTGs) _| 17 Diesel load balance (kW load remaining for the diesels to generate) 18 Diesel fuel consumed (gal) at 13 kWh gal” * — The wind speed data was converted from m/s to mph for the convenience of the reader that may have no concept of speeds in meters per second. ** The only WTG constraint was simply to set the output of the turbine to zero if the turbine output was below S5kW since the reliability of such a calculation is somewhat questionable Considerable work has been done to characterize the rate of fuel consumption by diesel engines compared to the percent load on the engine. The results indicate a nearly linear fit with approximately a 25% offset at zero load is appropriate. The operating generator(s) on the islands are never operated below synchronous speed except if they have been taken off line, then they are shut down completely. The use of the Skarstein-Uhlen equation to accumulate fuel use is justified for high temporal resolution models. However, it is the authors opinion that with only hourly sampled load data, applying such an equation is meaningless. Experience with modern, large, low to intermediate speed, stationary diesels shows that the upper end of the August 8, 1996 6 Report to NREL on the development and operation of a Wind/Diesel Spread Sheet Model efficiency spectrum might show as much as 15-17 kWh gal" fuel rate. A record of the fuel utilization on the island was available. The data was in Excel® format and gave the date, the daily energy generated/used in kWh, the engine hours, and the fuel used. When expressed in terms of kWh gal, the annual mean value is 11.6 kWh gal". Further investigation of the fuel use data seemed to be an exercise in futility. The understanding is that 24 engine hours indicates that at no time was there more than one engine on line. Any higher number indicates that for some portion of the day a number of diesels were running and on line. A multivariate analysis of the available data did not help clear up the picture. The relationship of engine hours, to kWh, to gallons of fuel used over the entire year, was uncorrelated. While one could choose to use the 11.6 kWh ga} number as the average, the decision was made to use * the number 13 kWh gal", a more typical value for similar operations familiar to the author. If in fact the real values are lower, such as 11 kWh gal", then the value of wind to the system is even greater. It is important to realize that, though one might expect to see a 200kW diesel load occur frequently in column 17 in this example, the 200kW minimum constraint will not allow the addition of another WTG if that addition would reduce the diesel load below the 200kW set point (limit). Therefore, the number in column 17 will always be between 200kW and 425kw unless there is no wind power available. It must also be kept in mind that this entire model is based on hourly averaged data. In reality, there is little relationship between these apparent ramp rates with the actual ramp rates. One aspect of the control algorithm that was not incorporated into the present model but was “tested” for specific instances is the appropriate addition of load. The appropriate addition of load would occur when the disparity between the minimum set point and balance of diesel load were such that adding some load would allow another wind turbine to come on line thereby reducing the diesel load back to it’s set point. For example; assume that the wind is such that the example wind turbines would be putting out about 200kW; and the system load balance (after reduction by the wind) is 300kW. The addition of another turbine would reduce the load balance to be picked up by the diesel to ~100kW - below the minimum set point assumed to be 200kW. Now, if the control strategy added ~ 100 kW of load to the system, the load balance would be increased to 400kW thus allowing the addition of another wind turbine while reducing the balance of load on the diesel(s) to 200kW (from 300kW), a reduction of 100kW load and ~8 gallons of fuel based on hourly averages. Figure 2 shows one of the model results for a six day period at the beginning of the file. The purpose of the figure is to show how the combination of wind and load governs the operation (on/off) of each of the four wind turbines. While Figure 2 provides a graphical representation of the operation of the wind turbines, the amount of wind energy that is being used is not indicated and the explanation of some of the operations is, thus, obscured. To provide some enlightenment, Figure 3 furnishes a time series chart depicting both the wind energy being used (bottom line bounding the shaded area) in the system and the total system load (top line bounding the shaded area). The wind energy used (from column 14 of the spread sheet) is not necessarily the wind energy available. There are exceptions such as in cases where the bottom line on the chart goes to zero when there is more than 200kW load. To help clarify the content of Figures 2 and 3, the reader is asked to look at the region of each chart on late Julian day 213 and early day 214. There is a period of about ten hours on both charts that appear as a relatively flat floored valley. In Figure 2, the period shows that only one turbine was in operation during the interval. During that same period on Figure 3, it is evident that the single turbine was producing around it’s rated power of 225 kW. If the operating strategy allowed a second turbine to come on line, the balance of load remaining for the diesel(s) to pick up would have been less than 100kW. Therefore, the test for excess power for the second turbine was positive and a second turbine was not allowed to come on line. August 8, 1996 7 Report to NREL on the development and operation of a Wind/Diesel Spread Sheet Mode! Operating Strategy Example, Min.=200kW nN No of peratin r ines Figure 2, Chart showing the number of wind turbines operating during each hour for a six day period in early August. Each diamond represents an hour. The numbers on the abscissa are the Julian Day and are repeated every four hours. This chart does not provide any indication of the wind energy utilization. SNI System Load and Wind Used, Min. = 200kW sed W no DD 3s 8s y 3 = 3 E § & 100 Julian Day Figure 3. A time series plot of the total system load (top line) and the wind energy being used by the system (bottom line) during the same period. The area in the chart that is darkened is the portion of the load that the diesel 8enerators would have to pick up and the clear area from the x-axis to the darkened area constitutes the load picked up by the wind turbines. August 8, 1996 8 Report to NREL on the development and operation of a Wind/Diesel Spread Sheet Model Figures 4 and 5, below, are added to show the effect of changing the minimum set point to 100kW. Figure 4 shows that the number of hours during the period when only one turbine was running under the 200kW set point have been significantly reduced with the lower set point. Note also in Figure 5, that the broad, flat floored valley during late day 213 and early day 214 has become a single hour that the diesels will have to pick up. For clarity, Figure 6 shows what the wind speed was doing during the same period. As can be seen, when there are zero wind turbines operating in both Figures 2 and 4, it is not because of low loads but rather low to no wind. Recall that the power out of the wind turbine below about 10 miles per hour is nil. | Operating Strategy Example, Min.=100kW 2 g u 3 & 8 3 sels RR Jullan Day Figure 4. Similar to Figure 2 but with the minimum set-point for the diesel load at 100 kW. Note that the period of time on late day 213 and early day 214 allows a second turbine to operate for 12 of 13 hours. The Economics of Wind Energy on SNI. The economics of a wind energy project can become quite complex due to financing arrangements. It is assumed for this exercise that the wind energy project is bought and paid for as a capitol development by the Navy. While that simplifies the calculation considerably, there are sufficient unknowns that some values are, at best, estimates. Here again, the estimates are reasonably conservative. The cost, per turbine, installed on the island, is assumed to be $337,500.00 in 1996 dollars. This value simply assumes the cost is $1500.00 per kW installed. While this appears to be an above market price, the costs of installing this small number of turbines on off-shore islands is going to be fairly high since there would be a requirement for the contractor to provide and ship to the island all the necessary heavy equipment including a portable concrete batch plant. This number may be excessively high but, again, it is conservative. The cost of the entire installation of four turbines is about $1,350,000.00 total. Electrical infrastructure and control systems are additional but are a relatively small portion of the total cost. The life expectancy for these turbines according to Navy criteria, is to be 20 years! and thus, the per year amortization amounts to ~$67,500 PA ‘In the wind energy industry, the presumed and design life is typically 30 years however the Navy has made the assumption that the wind turbines will have a life expectancy of only 20 years. August 8, 1996 9 Report to NREL on the development and operation of a Wind/Diesel Spread Sheet Model SNi System Load and Wind Used, Min.=100kW Total System Load aa | j 3 z z se 3 8 = g a Figure 5. Similar to Figure 3 but with the minimum set-point for the diesel load at 100kW. The diesel fuel cost, or in the case of San Nicolas, the cost of JP-5’, their reported fuel of choice, was determined to be ~ $1.14 gal.”' at the site. This value includes a 5¢ charge for administrative handling. Numbers such as this can easily be changed in the model. An additional charge associated with the cost of fuel is the cost of cleaning the barge tanks with each shipment. That cost is $25,000 per barge load. This charge is incurred ~10 times per year for a total cost of $250,000. The diesel power plant also has certain costs. While the original capital costs are ignored here, a “sinking fund” is established to provide for the replacement of the engines. The life of such engines is set at 10 years and the cost of replacement is estimated to be about $700.00/kW or $2,450,000 for all five engines. Amortized over 10 years this is simply taken as $245,000/year. Another charge that needs to be added to the O&M costs for the diesel plant is a $250,000/year contract for overhaul of all five engines after 2000 run-hours. This charge has apparently been nearly annual in the past but is pegged to the 2000 hour run time and is, therefore, less with the addition of wind energy. For the year 1989, the total number of engine hours was 9983 hours, essentially 10,000 hours. With the addition of wind, the number of hours the diesels operate is presumed to be reduced by the ratio of kWh without wind to that with wind. The overhaul contract costs on an annual basis would be reduced to $186,630 if the minimum set point for the diesel plant was 200kW and further reduced to $172,120 if the minimum can be set to 100kW. A very serious consideration concerning the diesel plant and one of the major differences between the HYBRID2 model and this simple spread sheet is fuel use. The HYBRID2 model incorporates a built-in function that calculates the diesel fuel utilization. This function is based on published observations and is a linear function of load with an offset at zero load (idle speed). This is a good example of an attribute of the model that is appropriate with sub- hourly data and a smaller system but is questionable with only hourly averaged data and a large system. In the spread sheet model, and in reality, the engines are never at idle or shut off unless there is another on line. Therefore a simple value for the kWh/gal was assigned based on the best available information and experience with other similar sites and was not calculated via the function . * The diesel engine manufacturers recommend that a “chiller” be added to JP-5 to increase it’s lubricity and reduce wear on the fuel pump, the injection pump and the injectors. August 8, 1996 10 Report to NREL on the development and operation of a Wind/Diesel Spread Sheet Model | SNI Hourly Wind Speed Record Julian Day (In four hour increments) Figure 6. Time series plot of the wind speed during the same period of time as that shown in figures 2,3,4,and 5. The Results Included as an addendum to this report is a copy of the actual spread sheet on 3.5” diskettes. By simply convolving the wind speed to the power curve of the wind turbine for all 8760 hours of the year one gets the number of kWh you might expect assuming 100% availability of the wind turbine. In this exercise, there is no arbitrary assignment of WTG availability. It is the authors contention that it is meaningless to assign an availability since 1) there are sufficient lulls in the wind to perform most maintenance functions and 2) with a constrained, four WTG plant there is nearly always a “spare” turbine to replace one that might be down for some longer duration maintenance. A third and final point would be that today’s wind turbines are exhibiting an availability of >97-98%. Therefore, availability deems to be a “non-issue” in this circumstance. If desired, some sort of Monte Carlo simulation of turbine outages based on experience could easily be added. Also, there were no deductions for losses due to “Array Effects” since there were only to be four wind turbines and the wind has a prevailing direction. Siting of a four turbine array under these circumstances should be straight forward enough to be able to avoid any such effects. There will be some electrical losses in the lines and transformers but these are small enough that the conservative approach to this entire analysis probably has them covered The result of convoluting the wind speed data with the example wind turbine power curve is that a single turbine should produce about 675,000 kilowatt hours of energy per year at the San Nicolas Island site for the year modeled. That is equal to a capacity factor of 34.2% - a number that tends to indicate the resource at the site is extremely good. Again, assuming no artificial availability numbers, with all four turbines allowed to run unconstrained the production swells to 2.7 million kilowatt hours. In fact, however, assuming the constraints imposed by the operating strategy requiring a minimum 200kW load for the diesel(s), the combined, usable energy production of the four wind turbines is only about 1.43 million kWh of energy for the year. The result of re-running the model using a 100kW minimum load constraint showed that the number of usable kWh increases to 1.75 million. The reduction in cost is most easily seen as the fuel cost per kWh (saved) plus the O&M cost/kWh minus the added cost of O&M on the wind turbine. Table 3, below summarizes the best estimates of the apparent cost savings. Figure 7 shows the number of hours that each of four turbines would be running during the year with the 200kW minimum diesel constraint. This is taken directly from the model data and indicates that out of one year (8760 hours) nearly 7400 hours had sufficient wind and system load to have at least one wind turbine running, monotonically decreasing to about 4500 hours with sufficient wind and system load for four turbines to be operating. August 8, 1996 ll Report to NREL on the development and operation of a Wind/Diesel Spread Sheet Model Table 3. Summary of costs and savings assuming the 200kW minimum set point. Item Without Wind Turbines With Wind Turbines Wind Turbine O&M/Yr. @0.01/kWh $0.00 $14,305.45 Diesel O&M/Yr. @0.066/kWh $375,863.76 $278061.61 Diesel Fuel Costs @ $1.14/gal $494,900.31 $369,452.49 Annual Amortization of Wind Turbines $0.00 $67,500.00 Sinking Fund to Replace Diesels $210,000.00 $210,999.00 2000 hour Engine Overhauls $250,000.00 **$186,629.75 aiid Totals $$1,330,764.07 $1,125,949.30 Savings $204,814.76 ** With the set point at 100kW, this number becomes $172,118.74 and the total savings become $ 266,396.14. Hours with # of WTGs Operating Set Point Minimum = 200KW # of Turbines Operating Figure 7. A Histogram of the number of hours that each of the four available WTG would operate with the operating strategy modeled herein. The difference in the estimates is between $205,000 to $266,000 per year saved by adding the four wind turbines. This equates to amortizing one wind turbine every 1.6 or 1.4 years depending on the minimum set point. This seems to provide a good rate of return of investment . For completeness sake Figure 8, shown below, is similar to figure 7 but with the minimum diesel constraint set to 100Kw. August 8, 1996 12 Report to NREL on the development and operation of a Wind/Diesel Spread Sheet Model Hours with # of WTGs Operating Set Poit Minimum = 100KW # of Turbines Operating Figure 8. A Histogram of the number of hours that each of the four available WTG would operate with the operating strategy minimum set to 100 Conclusions Although it is too early to make very bold statements, it appears that this simple wind/diesel model performs quite adequately for the purpose that it is intended. Students getting into the higher mathematics of college frequently do not comprehend the difference between resolution and accuracy. The simple spreadsheet model described herein has all the resolution the data can justify. The accuracy is a question but is certainly within the 10% region. A model such as HYBRID2, afforded the luxury of high temporal resolution data and more quantitative data on such things as fuel utilization, may provide a clearer look into the minute by minute operational details but the accuracy relative to reality is also a question. With the low rate data available for this investigation, the HYBRID2 model is clearly an over-kill and probably can not produce any better (more accurate) results. The Navy has expressed concern over the potential of a wind power plant experiencing large scale changes in output very rapidly thus causing the diesels to work harder to follow load. While this task does not include the documentation of such, it is well documented that the stronger the average winds, the lower the variance and conver- sely, the lower the average winds, the higher the variance, percentage-wise. Further, there has been expressed the concern that the jitter or variance in the wind will add to the variance in the load. While this has never been documented to the best of the authors knowledge, the one thing that has been documented is that when multiple WTGs are on line, their long term mean power is strictly additive while the variances add as the reciprocal of the number of turbines. It is logical that when considering a “noisy source” (the WTGs) interacting with a “noisy sink” (the load), their variances must at least partially cancel. It would be a worthwhile exercise to take a short period (i.e. a week) of high resolution (1sec"') wind speed and direction data and synonymous load data at the island to investigate the apparent effects of adding wind to the San Nicolas Island power system and might be considered a mandatory first step. August 8, 1996 13 APPENDIX D: Wind-Diesel System Operational Guidelines This appendix contains the wind-diesel system operational guidelines that lead to the overall operational scenario. This scenario was developed by NREL and its sub-contractors for the SERDP-funded San Clemente Island wind turbine installation. This appendix conveys information on the amount of wind turbine electrical energy that can be utilized with the SCI diesel system and on the appropriate use of external loads to manage excess wind energy. Fuel savings are still achievable with low demand when the wind-diesel system includes a 225 kW load bank. 46 NREL: SAN CLEMENTE ISLAND WIND/DIESEL GENERATOR LOAD ANALYSIS = z 5 o) v m| m| oO 3 oS = es | LOAD (kW, ry] t9.5[_ 3805] oof 1] svt] saz] 0.0] 1] 110.7] 201.9 2 39.0[ 367.0] 0.0] 2] 1142] 265.8] 0.0] 2] 207.4] 102. [—3[—58.5[ 341.5] 0.0] 3] 171.3] 2087] 0.0] a] 306.1] 118.9] Ts. [478.0] 322.0] 0.0] 4] 226.4] 171.6] 0.0] 4] 4748 225, Ti 79.5] 480.5] 0.0] 1] 57.1] 4a0.9] 0.0] a] 11.7] 301. [2 39.0 461.0] 0.0] 2] 1142] 385.8] 0.0] 2] 237.4] 202. [3 58.5| 441.5] 0.0] 3] 171.3] 328.7] 0.0] 3] 356.1] 143. [478.0 422.0 0.0 4] 226.4] 271.6] 0.0] 4] 474.8] 100.2 7s. |) tt 19.5] 580.5] 0.0f 57.1] 542.9] 0.0) 118. | 2} 39.0f 561.0] off 194.2] 485.8f oof af 237. | sf 58.5] 544.5] off t713f 428.7] 0.0] 3] 356. pat 78.0] 522.0 0.0f a 228.4] 371.6] 0. ay 474. 7oof_4f__ 19.5] 680.5] 0.0f tf 57.4] 642.9] 0.0f tT 118. | 2f 39.0f 664.0] ff 194.2] 585.8f oof af 237. | sf 58.5[ 644.5] Of tt. 3] 528.7 0.0] 3] 356. 4} 78.0] 622.0 0.0 228.4] a7 6.474. oof] 19.5] 780.5] 0.f 57.1 742.9] 0.0f af 178. I 481. 243: 125. 581. 462. 343: IL @ & Claw] olo|wjir[ola]o! = o iz ejee S| = ololo 2 o olo ed 681. N oO x alN So S S 114.2 oe 2 f 562.6] 0.0 | 3] —s58.5| 741.5) 0.0) 3 171.3| 628.7 Oo) 3 356.1] 443.9) 0.0 Pay 78.0] 722.0f0.0f a 228.4] 5716] 0.0] af 474.8] 325.2) jt} 19.5] 880.5] 0.0f tt 57.1 842.9] ft 118.7] 781.3) pat 39.0 861.0] 0.0 1 44.2T 785.8, 0.0, 2] 237.4] 662.6] ps} 58.5 841.5} 0.of 3171.3 728.7] 0.0f 3] 356.1] 543.9] V4 78.0 822.0] 0.0f 228.4] 6716]. 474.8] 425.2 tt eet et ee ee 881.3 0.0 | 2f39.0] 961.0] Of 114.2] 885.8 0.0} 2) 237. 762.6 | 3} 58.5] 944.5] ff 174.3[ 828.7] 0.3 356.1 i eat oa lot ath ee roof tf, 19.5] 1,080.5} 0.0] tt 57.1) ; . . 981.3) | 2t 39.0] 1,064.0] off 114.2 862.6 3 58.5] 1,041.5] 0.0 a 743.9 4 78.0] 1,022.0 0.0 4 228.4 , 625.2] 0.0) 120 1,180.5) 0.0 1 57.1 oo 2 39.0) 0.0 114.2 237.4| 962.6 0.0 | 3] S585] 1,141.5) 0.0 3 171.3 356.1 so Pat 78.0] 1,122.0] 0.0f 228.4 Of af 474.8] 725.2) 0.0) jt} a 0.0 1 : |__ 118.7] 1,181.3] 0.0) 39.0] 1,261.0) 0.0 2 —— } 58.5 — 3 3] 356.1, 943.9] _—02.0J 4 78.0| 1,222.0 0.0 4 ’ | 474.8[ 825.2] 0.0) 1400) 1 19.5] 1,380.5 0.0 1 57.1] 1,342.9] 0.0 tf, 118.7] 1,281.3] 0.0) 2 39.0] 1,361.0 0.0 2 2] 1,285.8 0.0] 2] 237.4[ 1,162.6] 0.0) 3 58.5] 1,341.5 0.0 3 4,228.7 pf [4] 78.0] 1,322.0] 0.0) 4 4] 1,171.6 0.0] 4] 474.8, 925.2 0.0 19.5] 1,480.5 0.0 1 57.1] 1,442.9} 0.0} 1] 118.7] 1,381.3] 0.0) 39.0] 1,461.0 a! 114.2] 1,385.8) 237.4] 1,262.6) 3 58.5] 1,441.5 0.0 3 171.3] 1,328.7 ; 3 356.1] 1,143.9) Lay 78.07 1,422.0] 0.0f 228.4] 1,271.6 . 4] 474.8} 1,025.2 KEY #WG : NUMBER OF WIND GENERATORS WG : WIND GENERATOR LOAD PRODUCTION (kW) : DIESEL GENERATOR LOAD PRODUCTION (kW) : LOAD BANK LOAD DISSIPATION (kW) Page 1 NREL: SAN CLEMENTE ISLAND WIND/DIESEL GENERATOR LOAD ANALYSIS 25 30 35 WG JWG (kW) [DG (kW) [LB (kW) [#WG|WG (kW) _|DG (kW) [LB (kW) J#WG [WG (kW) [DG (kW, 228.6] 0.0 tt 206.0] 194.0] 0.0f tt 227.0] 173.0] 0.0) 342.8[ 132.2 75.0} 2 412.0] 138.0] 150.0] 2] 454.0] 171.0] 225.0) 514.2] 110.8] 225.0) 3 OFF 3 ORE ed Bs So oO} o | OPE na ORE re OP ed if 171.4] 326.6] 0.0], St] 206.0] 294.0] 0.0f tf 227.0] 273.0] 0.0) So So 342.8 _ 163.0/ 75.0] 2] 454.0] 121.0] 75.0 514.2| 135.8] 150.0) 3 618.0| 107.0] 225.0] 3] OFF] || Oe ee ee | 171.4] 428.6] 0.0] 206.0 394.07 0.0] tT 227.0] 373.0] 0.0 | 342.8 257.2] 0.0} 2] 412.0[ 188.0] 0.0] 2] 454.0] 146.0 0.0 |___ 685.6] 130.4] 225.0) OFF a Ore 7 206.0] 494.0] 0.0f, 1] 227.0] 473.07 0.0) 412.0] 288.0] 0.0f, 2 454.0[ 246.0] 0.0) 1OtOR - 225.004 Ore ee] 800 594.0] 0.0] a 227.0] 573.0] 0.0} y 388.0] 0.0], 2] 454.0] 346.0] 0.0) . et 3 681.0] 119.0] _—0.0J i : i : 126.0| 150.0] 4] 908.0] 117.0] 225.0 | 174.4] 728.6] 0.0] 1 206.0] 694.0] 0.0f, tf 227.0] 673.0] 0.0) | 342.8f 557.2] 0.0] 2412.0] 488.0[ 0.of 2] 454.0] 446.0] 0.0) | 514.2] 385.8f 0.0] 618.0] 282.0] 0.0], [681.0] 219.0] 0.0) | 685.6] 214.4f 0p at 824.0] 151.0f 75.0 4] 908.0[ 742.0] 150.0) 6} 0.0} 206.0] 794.0] 0.0 tt 227.0] 773.0[ 0.0) | 2] 342.6] 657.2] oof 412.0] 588.0 oof 2454.0] 546.0] 0.0) | 3st 514.2] 485.8f oof 3] t8.0f 382.0] 0.0] 3] 687.0[ 319.0] 0.0) pa] 685.6] 3144] 0.0 a 824.07 176.0[ 0.0 a] 908.0[ 167.0] 75.0) 1oof_ if t7i.4t 928.6] o.of 206.0] 894.0f 0.0] af 227.0] 873.0] 0.0) S S N ox & 00] Ny i) D 2| 342.8] __757.2 0.0 2 412.0] __688.0 0.0 p24. 646.0] 0.0) 3|___514.2| 585.8 0.0 3 618.0] __482.0 0.0] _3| 681.0] 419.0 0.0} 685.6] 414.4 Oop 4 824.0] 276.0[ 0.0f 4] 908.0] 192.0 0.0] = <= ° S as qq = PID pt 174.4] 1,028.6] 0.0f, i] 206.07 994.0] 0.01] 227.0] 973.0] 0.0) pt 342.8] 857.2] oof 2412.0] 788.0] 0.0 2] 454.0] 746.0] 0.0) 3] 514.2] 685.8f of 3] 618.0] 582.0[ 0.of 3] 681.0 519.0] —0.0) 685.6] 514.4f Of 824.0] 376.0] 0.0 4908.0 292.0] 0.0} 1301 1 : a a ee 2] 342.8] 957.2 0.0) 412.0] 888.0[ 0.02] 454.0] 846.0] 0.0) | 3] 514.2] 785.8f 0.03] 18.0 682.0[ Of S810 619.0 _—0.0) P4685] 614.4f O84 0476.07 0] 408.0] 392.0 _—0.0) 1400 ts et ot 206.0] 1,194.0] 0.0], ~— 1] ~——227.0] 1,173.0] _0..0] 2342.8] 1,057.2 0.0 2 412.0] _988.0/ 0.0] _2| 454.0] 946.0] 0.0) | 3} 514.2] 885.8] 0.0) 3 618.0] 782.0 0.0} 3|___—-681.0_ 719.0] 0.0 Pat 685.6] 714.4] 0.0) 4 824.0] 576.0 O.of 4] 908.07 492.0/ 0.0} 1500] tf 171.4] 1,328.6] 0.0, tt 206.0] 1,294.0] 0.0] 1227.0] 1,273.0] 0.0 Pat 342.8[ 4,157.2] 0.0f 2412.0] 1,088.0[ of 2, 454.0f 7,046.0] 0.0 P38] 514.2] 985.8[ of 3618.0 882.0] 0.0] 3] 81.0] 819.0 _—0.0 Pay 85.6 8i4ay oof 824.0 76.0, ofS TS 08.0f_~—592.0] _—0.0 : NUMBER OF WIND GENERATORS WG : WIND GENERATOR LOAD PRODUCTION (kW) : DIESEL GENERATOR LOAD PRODUCTION (kW) : LOAD BANK LOAD DISSIPATION (kW Page 2 NREL: SAN CLEMENTE ISLAND DEMAND LOAD (kW. |__tt 230.0] 170.0] 0.0) |__2| 460.0] 165.0] 225.0) | 3st Orr] a |i] 230.0] 270.0] ___0.0) EE eee a) a |__ 1} 230.0] 370.0] 0.0) |_ 2] 460.0] 140.0] __—0.0) | OF a | 7oof 1230.0] 470.0] __0.0) |__2]_ 460.0] 240.0] 0.0) |__3}__ 690.0] 160.0] __ 150.0) Lat FF] soot if 230.0] 570.0] __0.0) |__ 2] 460.0] 340.0] 0.0) |__ 3] 690.0] 110.0f __0.0) |_4{__ 920.0] 105.0] 225.0} |_ 1] 230.0] 670.0] ___0.0) |__2{__460.0[__440.0] 0.0} |__3] 690.0] 210.0f___0.0) |_4f__920.0[ 130.0] 150.0} |__ tt 230.0] 770.0] 0.0) | 2| 460.0] 540.0] ___0.0) |__3{__ 690.0] 310.0] 0.0} |_4{__ 920.0] 155.0] 75.0} 1100f__4|__230.0] 870.0] __0.0} |__2|460.0[ 640.0] 0.0} |__3}__690.0[ 410.0] __0.0} |_ 4] ___ 920.0] 180.0] 0.0) 1200] 4|__-230.0] 970.0] 0.0} |__ 2] 460.0] 740.0 ___—0.0) |__3]__690.0[ 510.0] 0.0 4} 920.0f 280.0] 0.0} 1300f__4{__230.0] 1,070.0] __0.0} | 2|_ 460.0] 840.0] 0.0} |__ 3} 690.0] 610.0] 0.0) |__ 4 920.0[__ 380.0] 0.0} 1400] 1] 230.0] 1,170.0] 0.9) | 2] 460.0[ 940.0] 0.0} |__3{__ 690.0 710.0] 0.0} pt 920.0[ _480.0[_ 0.0) 1500f_ if 230.0[ 1,270.0] 0.0} | 2 460.0f 1,040.0] 0.0} |__ 3] 690.0f 810.0] 0.0) Lt 920.0f__ 580.0] ___ 0.0} WIND SPEED (mph) WIND/DIESEL GENERATOR LOAD ANALYSIS : NUMBER OF WIND GENERATORS : WIND GENERATOR LOAD PRODUCTION (kW) : DIESEL GENERATOR LOAD PRODUCTION (kW) : LOAD BANK LOAD DISSIPATION (kW Page 3 NREL: SAN CLEMENTE ISLAND WIND/DIESEL GENERATOR LOAD ANALYSIS INPUT WIND el GEN ae DG LOAD ne DG i LO 474: oe {VCE Te $5 SETA rie Weta a Len A cae 105.0 Page 4 NREL: SAN CLEMENTE ISLAND FUEL CONSUMPTION ESTIMATE DEMAND FUEL DEMAND] FUEL USED] FUEL SAVED 39.0] 761.0 51.6 48.8 2.8 58.5 pS fp 47.7 4 78.0 46.7 WIND SPEED (15 mph DEMAND FUEL DEMAND] FUEL USED] FUEL SAVED bs 4) soo]___1] po] BB 4|___228.4] 571.6 51.6 38.6 WIND SPEED (20 mph DEMAND FUEL DEMAND] FUEL USED] FUEL SAVED! LOAD ew fac ws ary Jos co 800} tf 118.7] 681.3 44.5) 2] 237.4 51.6 38.2 13.4 3] 356.1] 443.9 51.6 31.8 19.8 Pa 474 Bf 3252p Sef 25.4 262] WIND SPEED (25 mph LOAD (kW) [#WG [WG (kW) [DG (kW: gal/hr gal/hr) gal/hr, 800} 1 171.4 628.6 51.6) 41.7 9.9 457.2 51.6 32.5 19.1 285.8 51.6 23.5 28.4 685.6) 1144p Sif 16.6) ; PEED (30 mph DEMAND FUEL DEMAND] FUEL USED] FUEL SAVED LOAD (kW) |#WG |WG (kW) DG (kW) (gal/hr) (gal/hr) 800 i | ___1] 206.0] 594.0 412.0] 388.0 3 618.0 182.0) 4 824.0 126.0) WIND SPEED (35 mph DEMAND FUEL DEMAND] FUEL USED] FUEL SAVED LOAD (kW) [#WG JWG (kW) [DG (kW gal/hr gal/hr, gal/hr, soo} tf 227.0) 573.0) GY 8B 12.9) 2] 454.0] 346.0 51.6 26.6 25.0 681.0 119.0 51.6 16.8) 34.8 3 908.0 117.0 51.6) 16.7} 34.9 WIND SPEED (40 mph DEMAND FUEL DEMAND] FUEL USED] FUEL SAVED! LOAD (kW) [#WG [WG (kW) J[DG (kW) gal/hr)| gal/hr) gal/hr 460.0] 340.0 51.6 26.2 25.4 P 3 690.0f 110.0f Sef teat 35.2] pf 920.0f 105.0f SP 16.2F 35.4] Page 1 NREL: SAN CLEMENTE ISLAND 3 gal/ 75 kW = 0.04 4 gal/ 75 kW = 0.053333 5 gal/ 75 kW = 0.066667 6 gal/ 75 kW = 0.08 7 gal/ 75 kW = 0.093333 (1)WG (2)WG (3)WG_ (4) WG 10 1.3 2.8 3.9 49 15 3.8 6.8 9.9 13.0 20 wal 13.4 19.8 26.2 25 9.9 19.1 28.1 35.0 30 11.8 22.8 32.3 34.5 35 12.9 25.0 34.8 34.9 40 13.1 25.4 35.2 35.4 FUEL CONSUMPTION 40.0 35.0 30.0 o 25.0 ——wel 5 —(2) we = 20.0 (3) WG 8 ——(4) WG 15.0 10.0 5.0 0.0 WIND SPEED (mph) Page 2 NREL: SAN CLEMENTE ISLAND POWER CURVE DATA WIND GENERATOR: MODEL 225kW SPEED} SPEED| POWER | 3.0] 67]. | 4.0] of 8 } 5.0] 11.2] 19.5) | 6.0] 13.4] 31.2) | 8.0] t7.of 87.6) 20 | 9.0f 20.4] 118.7| |__ 10.0] 22.4] 149.6) | 4.0] 24.6] 171.4) | 12.0/ 26.8] 189.3) | 18.0[40.3[ 230.0} | 19.0] 42.5] 225.0) 210.0 210.0 0.0 Vv nN a Generator Data kWh 250.0 200.0 150.0 100.0 50.0 0.0 POWER CURVE oe? ? o ° ooo @ Sd o > a Ad 0.0 10.0 20.0 30.0 40.0 50.0 60.0 Mph @Series1 Mph=(m/s)*(2.237) NREL: SAN CLEMENTE ISLAND PERFORMANCE CURVE DATA DETROIT DIESEL GENERATOR: MODEL 16V-149TI SCCC | bph} KW] gal/hr] |__ 0.0] 0.0] 12.0) | 200.0] 149.2[ 18.0) | 300.0] 223.8] 21.0] | 400.0] 298.4f 24.0) |__ 500.0] 373.0] 28.0) | 700.0] 522.2] 36.0} | 800.0] 596.8] 40.0} 900.0 671.4 1,000. 746. 1,100. 820. 1,200. 895. 1,300. 969. 1,400. 1,500. 1,600. 1,700. 1,800. 1,900. 2,000. ae g © Blolmialo A fe cede cal NV] O0] 6] OR ool. }O a hb es Pa 1,566. 2,200. 1,641. 2,300. |__900.0| | 1,500.0] 4,11 |_ 1,600.0] 1,19 |_ 2,100.0) kW=bhp*0.746 Generator Data PERFORMANCE CURVE 140.0 120.0 ¢ ° ° 100.0 lo ° ° 80.0 4 _ s 2 ° 3 ° 60.0 = ° .° 40.0 = ° ° 20.0 | ° ° ,° 0.0 0.0 200.0 400.0 600.0 800.0 1,000.0 1,200.0 1,400.0 1,600.0 1,800.0 2,000.0 kW @Series1 Graphs Chart 1 Power Quot (KW Manufacturers vs. Polynomial Fit Power Curve 250.0 200.0 3 o 100.0 + 50.0 0.0 — eee oe af 671 895 11.18 13.42 15.65 17.89 20.13 22.36 2460 26.84 29.07 31.31 33.55 35.78 38.02 40.25 4249 44.73 46.96 49.20 51.44 53.67 55.91 Wind Speed (mph) Page 1 Graphs Chart 3 Number of Hours 8,000 7,000 - 5,000 + 4,000 - 3 8 2,000 - 1,000 - Number of Hours with Turbines running Number of turbines running Page 1 Form Approved OMB NO. 0704-0188 Public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this Collection of information, including suggestions for reducing this burden, to Washington ae Services, Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, Arlington, VA 22202-4302, and to the Office of Management and Budget, Paperwork Reduction Project (0704-0188), Washing be 20503. 1, AGENCY USE ONLY (Leave blank) 2. REPORT DATE 3. REPORT TYPE AND DATES COVERED July 1999 Subcontract Report 4. TITLE AND SUBTITLE Performance and Economics of a Wind-Diesel Hybrid Energy System: Naval Air Landing Field, San Clemente Island, California WE903030 5. FUNDING NUMBERS 6. AUTHOR(S) Ed McKenna, NREL Timothy L. Olsen, Tim Olsen Consulting 7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGANIZATION National Renewable Energy Laboratory Tim Olsen Consulting REPORT NUMBER 1617 Cole Blvd. 1428 S. Humboldt Golden, CO 80401-3393 Denver, Colorado 80210 9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSORING/MONITORING National Renewable Energy Laboratory AGENCY REPORT NUMBER 1617 Cole Blvd. Golden, CO 80401-3393 SR-540-24663 11. SUPPLEMENTARY NOTES NREL Technical Monitor: Ed McKenna 12a. DISTRIBUTION/AVAILABILITY STATEMENT 12b. DISTRIBUTION CODE National Technical Information Service U.S. Department of Commerce 5285 Port Royal Road Springfield, VA 22161 13. ABSTRACT (Maximum 200 words) This report provides an overview of the wind resource, economics and operation of the recently installed wind turbines in conjunction with diesel power for the Naval Air Landing Field (NALF), San Clemente Island (SCI), California Project. The primary goal of the SCI wind power system is to operate with the existing diesel power plant and provide equivalent or better power quality and system reliability than the existing diesel system. The wind system is also intended to reduce, as far as possible, the use of diesel fuel and the inherent generation of nitrogen-oxide emissions and other pollutants. The first two NM 225/30 225kW wind turbines were installed and started shake-down operations on February 5, 1998. This report describes the initial operational data gathered from February 1998 through January 1999, as well as the SCI wind resource and initial cost of energy provided by the wind turbines on SCI. In support of this objective, several years of data on the wind resources of San Clemente Island were collected and compared to historical data. The wind resource data were used as input to economic and feasibility studies for a wind-diesel hybrid installation for SCI. 14. SUBJECT TERMS 15. NUMBER OF PAGES wind power, wind turbines, wind power and military applications, wind-diesel applications, Federal Energy Management Program 16. PRICE CODE 19. SECURITY CLASSIFICATION 20. LIMITATION OF ABSTRACT OF ABSTRACT 18. SECURITY CLASSIFICATION OF THIS PAGE 17. SECURITY CLASSIFICATION OF REPORT UL NSN 7540-01-280-5500 Standard Form 298 (Rev. 2-89) Prescribed by ANSI Std. 239-18 298-102