Loading...
HomeMy WebLinkAboutBradley Lake Railbelt Stability Study Phase II Vol. I 1989I RAILBELT STABILITY STUDY PHASE II ALTERNATIVES AND RECOMMENDATION~ I VOLUME I •• TEXT I RAIL BELT STABILITY STUDY PHASE II ALTERNATIVES AND RECOMMENDATIONS VOLUME I·· TEXT Prepared for Alaska Power Authority Contract No. 2800122 PTI Project No. 30.2608 Prepared by: Harrison K. Clark POWER TECHNOLOGIES, INC. Roseville, CA March 30, 1989 PTI Report No. R35·89 Power Technologies, Inc. TABLE OF CONTENTS INTRODUCfiON . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 EXECUTIVE SUMMARY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 115 KV UPGRADE ........................................... . 69 kV Faults ........................................... . 115 kV Faults ...... -..................................... . Other Disturbances . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Kenai-Anchorage 115 kV Reclosing ............................. . Backswing Overvoltages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 MW Export . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Reduced Runback ........................................ . 4 5 7 8 9 11 13 14 Loss Of Anchorage Tie During Export . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 Use Of TSC Rather Than TCR/MSC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 Line!fransformer Drop Compensation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . SVS Size and Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Load Rejection Overvoltages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Brake Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 26 26 28 BRADLEY LAKE LIMITED TO 90 MW . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 SVS ALTERNATIVE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 SVS -Series Capacitor Comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 230 KV LINE ALTERNATIVE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 RELIABILITY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 APPENDIX --Load Models RAILBEL T STABILITY STUDY PHASE ll ALTERNATIVES AND RECOMMENDATIONS INTRODUCTION Phase I of the Railbelt Stability Study explored a number of options for solving stability problems. These included: o Bradley maximum operating power level o Various levels of power run back following disturbances o Various combinations of SVS and series capacitors o Use of a braking resistor at Bradley Lake o Use of stabilizers for dynamic stability Phase I work was based largely on 75 MW export with intermediate to high Kenai load levels, and generation on-line at Bernice and Cooper. The Phase I studies showed that it would be feasible to provide stability without the installation of a 230 kV line from Soldotna to University. Phase II of the Railbelt Stability Study was originally intended to select and optimize one transmission scheme to meet Bradley Lake power transmission objectives. However, as completed, it provides several alternatives addressing different Bradley Lake power limitations and Kenai export levels. Only one (combination of one SVS and three series capacitors) has been tested for a range of operating conditions sufficient to support the subsequent analysis necessary to prepare equipment specifications. Power Technologies, Inc. EXECUTIVE SUMMARY The several alternatives examined in Phase ll are summarized in the following paragraphs. One alternative that was considered in Phase I but not examined further in Phase II is also included to complete the range of alternatives that are available. o Equipment additions limited to that necessary to support 90 MW operation at Bradley Lake. The equipment consists of braking resistors (with associated communications and controls) and stabilizers at Bradley Lake. The braking resistors (two 25 MW units) are complemented by runback to 60 MW upon loss of the Bradley to Soldotna line to provide stability. Other disturbances will not require runback but may require braking. Kenai export limits have not been fully defined for this alternative, but the cases that have been run indicate export is limited only by Kenai load and losses when Bernice and Cooper generators are off (i.e. just over 40 MW). In Phase I this alternative was found to be stable at 76 MW export with Bernice and Cooper providing the additional power for export. Higher exports may be possible under some conditions. This alternative can be extended to handle somewhat higher Bradley Lake power levels, but would require correspondingly greater runback. This alternative does not address heavy Kenai import conditions under which voltage control is a known problem. o Equipment additions to allow 120 MW operation at Bradley. This alternative was studied most thoroughly in Phase ll. The investment is limited by running Bradley back to 90 MW immediately following loss of the Bradley to Soldotna line. The equipment required includes the two braking resistors and stabilizers at Bradley, one SVS at Soldotna and three series capacitors. It has been tested at maximum export when Bernice and Cooper are off and Kenai load is low ( 40 MW). It was also tested and found to be stable at 90 MW export with Bernice and Cooper units on, but such operation is very close to stability limits and the alternative was not designed for such operation on a normal basis (i.e. though the system may be stable at this export under ideal conditions, reliability will be lower when it is operated at this export). 2 Power Technologies, Inc. o Use of two SVSs rather than one SVS and three series capacitors. This alternative is otherwise similar to the one just described. Two SVS will be more costly than one SVS and three series capacitors. There are also technical advantages and disadvantages, but two SVS will offer about the same capability for Bradley Lake power generation and Kenai export. o Two SVS and three series capacitors. This alternative, explored only in Phase I, requires higher investment but does not require Bradley Lake power runback. Though it was not pursued further in Phase II, it bears mention here to complete the array of alternatives. Like all other options, it requires braking resistors and stabilizers at Bradley. This alternative requires upgrading of the 115 kV line from Soldotna to Diamond Ridge to carry the higher power following loss of the Bradley-Soldotna line. o New transmission line from the Anchora~e area to the Kenai area. A new line from the University substation to the Soldotna substation was considered. Though a very expensive alternative, such a line can ·eliminate one series capacitor and the SVS if runback from 120 MW to 90 MW is retained. If Bradley is to be held at 120 MW following loss of the 230 kV line, a small SVS is required at Soldotna. This SVS is also required if the level of dynamic stability performance and margin provided by the 115 kV alternatives is to be achieved. The most attractive feature of this alternative is a reduction in the number of incidents that will separate the Kenai area from the Anchorage area. The compensation package consisting of one SVS and three series capacitors was most thoroughly studied in Phase II. It requires: o Three series capacitors, 50% each in the lines from Bradley to Fritz Creek, Anchor Point to Kasilof, and Soldotna to Quartz. o One SVS +30/-25 MVAR at Soldotna with stabilizer and with control over the existing Soldotna mechanically switched capacitors. o Two 25 MW braking resistors with sensitive triggering and a controller to remove them based on Bradley rotor speed. o Power reduction to 90 MW by the deflector for loss of the Bradley to Soldotna line. 3 Power Technologies, Inc. o Power System Stabilizer on each Bradley Lake machine exciter/voltage regulator. o Bradley automatic voltage regulator line/transformer drop compensation with slow response (optional). The SVS size requirement is based on a number of considerations including first-swing stability, damping, steady state stability, backswing overvoltages, and voltage control during and following islanding. The casework in Phase II makes extensive use of detailed Kenai load models. The model includes large and small motor models and includes the loss of motor load that occurs during faults. The load model is discussed in some detail in Appendix I. The following sections focus on major considerations in selecting the size and location of compensation and control equipment, and on the various · equipment and operating options that can be used to solve the Bradley Lake stability limitations. 115 KV UPGRADE The goal of the Phase ll work was to select the best 115 kV transmission upgrade alternative from among those considered in Phase II and test it for a wide range of disturbances and add or modify the recommendations as necessary to provide an optimum system. The alternative consisting of one SVS and three series capacitors was both the least costly and the best performing of the alternatives examined in Phase II, and was selected for further study. The cases discussed in the remainder of this section address a number of aspects of the combination SVS -Series capacitor alternative: o stability for various 69 and 115 kV faults. o response to other disturbances o control of voltage during and following stability swings and separation from Anchorage. o ability to handle higher export or reduced Bradley runback. 4 Power Technologies, Inc. In addition, a number of other planning and operating options and considerations are discussed. Because construction of a second Kenai to Anchorage line would require much of the same equipment as is required without a new line, much of the discussion in the following subsections regarding equipment characteristics and selection apply also to that alternative. 69 kV Faults A number of 69 kV faults were examined. In all cases the critical clearing time for three-phase faults is between 6 and 9 cycles. Fault clearing times, including relay and breaker times, must be under the critical clearing time at the source (115 kV) end of radial 69 kV lines, and at both ends of looped lines (e.g. the Soldotna-Quartz 69 kV line). The following table summarizes the results of stability cases 182 though 192. FAULT Stable at: Unstable at: Cases Soldotna area 6 7 184, 185 Diamond Ridge 7 9 187, 188 Bernice 7 9 191, 192 Lawing 8 9 189, 190, 190A Dave's Creek 25 kV 20 186 The above clearing times were determined with the braking resistor assumed not to respond to these faults. The braking resistor can make the system stable for 69 kV faults up to several cycles above the critical clearing time without brake use. The voltages at Bradley were recorded for each of these faults, and are below 70% for all cases except the 25 kV fault (which is stable without fast clearing or use of the brake). Hence application of the brake for most close-in 69 kV faults is feasible. Hence where the existing clearing time is not too much above the critical clearing time, the brake may eliminate the need for communications based protection. On radial lines, presumably close-in faults can be cleared within 6 to 8 cycles by zone 1 distance protection. On looped lines, mid-line faults should be cleared promptly at both ends without carrier, and line-end faults will be cleared quickly at the near end and in second zone time at the far end. 5 Power Technologies, Inc. Only two cases have been run to assess use of the brake for 69 kV faults. The first is case 201 which simulates a three-phase fault adjacent to the Soldotna 69 kV bus on the line from Soldotna to Quartz Creek. The Soldotna end is assumed to be cleared at 6 cycles (zone 1 distance protection) the Quartz Creek end is assumed to be cleared in 24 cycles (zone 2 distance protection). The case is unstable. However, this is probably the worst 69 kV fault in that it removes a section of 69 kV line that parallels the 115 kV tie to Anchorage, and the fault depresses transfer to Anchorage significantly both during the first 6 cycles and during the additional 18 cycle line-end fault period in that both faults are directly off the 115 kV path off the line to Anchorage. This case should not discourage a search for locations where communications might be avoided by depending on the brake. The second case demonstrating use of the brake for 69 kV faults is case 212. This case examines reclosing on the 115 kV line from Dave's Creek to Lawing for downstream 69 kV faults. Since the fault clearing time is known only to be between 8 and 10 cycles, a 10 cycle clearing time was used. Because the 10 cycle fault is beyond the critical switching time (see above table and cases 189, 190 and 190A), the braking resistor is used. Reclosing dead times .are 18 cycles, 4 seconds, and 20 seconds. Because faults separated by 4 seconds or more are essentially separate events from a stability standpoint, if stability is achieved for one, all will be stable. However, the 18 cycle dead time for the first reclose places the first unsuccessful reclosing within the "swing period" of the initial fault, making it necessary to consider the two in the same simulation. In case 212 the brake circuit breaker is assumed to close at the instant of fault clearing (this is a conservative assumption, the brake should engage well before the fault is cleared). The brake is removed about .2 seconds later, about .1 second before reclosing again energizes the fault. The brake is not needed for the second fault application because the second fault occurs during the early pan of the Bradley angular backswing, and thus only serves to reduce the extent of the backswing. If the reclosing delay were greater, on the order of .7 seconds or more (i.e. just ahead of or around the peak of the "second swing"), the braking resistor would have to be applied to maintain stability. The second brake application would likely be longer than the first (about .3 seconds) if it occurs near the peak of the second swing. Subsequent brake applications for the 4 second and 20 second reclose attempts would each require about .2 seconds of brake time. The total brake time for the four faults would be about .9 seconds, somewhat greater than the .7 seconds selected based on other disturbances. Since the brake will be triggered by most severe 69 kV multi-phase faults (whether essential for stability or not), it must be capable of handling the reclosing that can occur. Also, in order for the brake to be applied several times in rapid succession, its circuit breaker must be able to provide the required sequence of operations. A magnetic type circuit breaker (e.g. GE Magnab1ast) is capable of rapid close-open operation, but must "recharge" after several close-open operations. This recharge limitation may be less 6 Power Technologies, Inc. significant with a vacuum breaker. Close-open capability of the breakers intended to be used on the braking resistors should be checked to ensure that it will match the fastest 69 kV reclose pattern. The 69 kV clearing times and fault locations that can be accommodated without transfer trip or directional comparison protection (i.e. systems requiring line carrier) by a 40 MW brake must be determined in further exploratory stability cases. Also required for this option is a check of the effects of 69 kV phase-to-phase faults on stability and on the voltage at Bradley Lake. If the brake must be applied for unbalanced 69 kV faults, then the voltage relay logic must be arranged to trigger the brake for such faults (this should not be a problem since the brake must be triggered for unbalanced 115 kV faults also. 115 k V Faults A number of 115 kV faults were examined in addition to a fault on the line from Bradley to Soldotna. Those examined were: o Fault at Bradley, trip Bradley -Fritz Creek --unstable without brake, stable with brake (cases 174 & 175). o Fault at Soldotna, trip Soldotna -Ski Hill --unstable without brake, stable with brake (cases 176 & 177). o Fault at Dave's Creek, trip Dave's Creek -Lawing --stable without brake (case 178). These cases show that 115 kV faults in general are a threat to stability, but can be overcome by use of the brake. These cases emphasize the need for brake triggering that is sensitive to all area faults. These faults cause very low voltage at the Bradley 115 kV bus and are thus easily observed from Bradley. Undervoltage relays thus can be used for brake triggering. It is clear that the Bradley brake should be triggered for multi-phase faults. However, a decision must be made as to whether the brake should be triggered for single-phase faults. Single-phase faults on the Bradley to Soldotna line must trigger the brake, but this is easily done from line relays at Bradley. If the brake is Lo be triggered by ground faults remote from Bradley, the triggering must be based on a one-out-of-three operation of three single-phase voltage relays. There is also the question of Bradley operating level. At 7 Power Technologies, Inc. some low plant output level, the brake will be needed only for faults on the line from Bradley to Soldotna. At low output the brake will not be needed if two units are operating. With one unit operating the brake may be needed when it is near full output. Other Disturbances Three stability cases were run for two miscellaneous disturbances. One is inadvertent braking resistor application, and the other is nuisance trip of a Bradley generating unit under full output. Case 179 is for inadvertent 40 MW brake application at low export. In this case the Bradley angle is low compared to the Anchorage area, so the plant is most susceptible to a disturbance that would slow it and thus take it out of step in a lagging direction (faults accelerate the plant and in an unstable case cause it to pull ahead of the remainder of the system). The brake slows the plant, and thus is a potential threat primarily when the plant angle is initially low. The case shows that the brake is very unlikely to cause instability under any credible operating condition. In fact, the Bradley angle is "turning around" even before the brake is removed at .7 seconds, indicating that the plant is probably stable even if the brake is not removed. If the brake thermal rating is .75 seconds, backup protection would remove the brake within .75 seconds. Case 213 is for inadvertent application of a 50 MW brake under heavy import conditions with one Bradley Lake unit operating in the Kenai. The Kenai load is 75 MW, the Bradley unit is producing 20 MW, and import is 43 MW. The brake is applied for .7 seconds. The case is stable. However, it does show large voltage excursions. Some voltages drop below 50%, and backswing overvoltages reach 120%. The Kenai frequency swings to a high of 60.75 Hz and a low of 59.2 Hz (above the 59 Hz underfrequency relay setpoints). Though the frequency swings are large, and the voltages violate criteria, inadvertent brake application is unlikely to occur because: o the brake should be disabled during import conditions (and low export conditions). o if a local (Bradley) control failure occurs, it should only affect one of the two brakes (i.e. 25 MW instead of 50 MW). o if the braking resistors are energize erroneously by local voltage relays or via carrier from relays on the Kenai-Anchorage lines, it should be removed quickly 8 Power Technologies, Inc. (about .2 seconds) by the same control that will remove it when it is properly applied. Based on the above, inadvertent application of the braking resistors to the extent represented in case 213 is very unlikely to occur, and is thus suggested to not be a problem. Case 180 is for trip of one of two Bradley Lake generating units when the plant is operating at 120 MW. It shows no stability threat or other potential problems. Kenai-Anchorage US k V Reclosing Cases 169 though 173 and 202 and 203 address the question of reclosing on 115 kV lines north of Soldotna. Reclosing on the line from Soldotna to Quartz Creek and the line north of Dave's Creek was examined. The Soldotna to Quartz Creek line has the benefit of the underlying 69 kV line, though this advantage is minimal because of the high impedance of the line and associated transformers (also, the results are slightly conservative because the Quartz Creek transformer was not replaced with the Bernice Unit before cases 169 through 173 were run). Faults further north affect the Kenai area load less, and thus would be expected to be somewhat more stable than those examined below. The cases are summarized in the following table. Case 169 64.3 MW export, Bradley at 120 MW. Bernice & Cooper off, 4 cycle single- phase fault at Soldotna, three-pole reclosing, 15 cycle dead time, no brake, no runback, no load drop. Unstable. Case 170 Same as 169 except brake applied. Stable. Case 172 1 Same as 169 except single-pole reclosing with 25 cycle dead time. Stable. 1 Case 171 is similar except it represents 15 cycle dead time and was, of course, stable. 9 Power Technologies, Inc. Case 173 Same as 169 except 41.8 MW export, Bradley at 90 MW. Stable. Case 202 64.3 MW export, Bradley at 120 MW, Bernice & Cooper off, 4 cycle single- phase fault at Dave's Creek, single-pole reclosing on Dave's Creek -University line, dead time 25 cycles, no brake, no runback, no load drop. Unstable. Case 203 Same as 202 except 15 cycle dead time. Stable. These cases show that three-pole reclosing on the Soldotna to Quartz Creek line is feasible at low export levels (below about 50 MW) without runback or brake operation at Bradley. At higher export levels the Bradley brake can be used to make the system stable. Runback may also provide stability but was not examined. The brake would be more effective and is the method of choice since it represents less risk of load shedding. Single-pole reclosing is stable for dead times up to and including 25 cycles on the Soldotna to Quartz Creek line. Case 172 appears quite stable thus indicating that an even longer dead time could likely be used. Cases 202 and 203 bracket the allowable dead time for single-pole reclosing between Dave's Creek and University. The limit is in the vicinity of 20 cycles. The Bradley brake could be used to extend the allowable dead time for this fault if communications is available to ensure that the brake will be triggered. The single-phase faults and the single-pole-open transfer impedances used in the above cases are based on very rough assumptions of the zero sequence impedances at Soldotna, Quartz, and Dave's Creek, and thus the above results must be considered to be approximate. However, since even the 25 cycle dead time cases are quite stable, the error in zero sequence impedance would have to be quite large to invalidate the above conclusions. Load dropping was not simulated in these cases. Load dropping would make the cases less stable. However, single phase faults north of Soldotna should not cause significant loss of load, and thus would only have a secondary effect on the results. 10 Power Technologies, Inc. Backswing Overvoltages Backswing overvoltages are always a problem when measures such as SVSs, series capacitors, high perfonnance excitation systems, braking resistors, and deflectors are used to provide first-swing stability. The voltages go high as the angle across the system approaches zero degrees during the back swing. At this instant the power flow across the system is very low so that line and transfonner reactive losses approach zero. Reactive power from capacitors and lines must flow to generators, and thus causes a voltage rise during this short time period. The overvoltage is compounded by the high flux in generators that remains from the first swing, or is pushed up by stabilizers attempting to provide damping. Loss of load during the fault also compounds the overvoltage problem. The backswing overvoltages in the Kenai are most severe when Bradley Lake is the only generation operating (case 161) and area load is low (40 MW). A number of 115 kV voltages exceed 110% for about .5s. Anchor Point and Fritz Creek voltages peak just over 118% (Fritz Creek voltage is not plotted), while Soldotna and Quartz Creek voltages peak at about 116%. When Bernice and Cooper generators are on-line, they absorb reactive power during the backswing, measurably reducing the backswing overvoltages (case 163). With one Cooper and one Bernice unit operating (and Kenai load at 60 MW), the backswing overvoltages are somewhat lower, with the highest 115 kV load bus reaching 117%, and others reaching about 113%. Changing the reactive dispatch among the Bernice and Cooper plants and the Soldotna SVS so that the pre-disturbance plant reactive loadings are lower the SVS output is higher pulls the voltages down 2 to 3% (compare cases 163 and 164). The cases discussed in the above paragraph are all based on an SVS at Soldotna with a +20/-15 MVAR dynamic range. Additionally, there is a 30 MVAR capacitor bank operating at Soldotna and a 7.2 MV AR bank operating at Quartz Creek. 2 The -15 MV AR SVS lower limit does not allow the SVS to absorb all reactive power from these capacitor banks. The remaining 22.2 MV AR is a significant contributor to the backswing overvoltages. Case 162 shows that increasing the SVS lower limit by 10 MVAR reduces the overvoltages by 3 to 4 percent. Case 162A shows that reducing the SVS lower limit by yet another 10 MV AR provides an additional 3 to 4% reduction. However, note that 2 The capacitor bank at Quartz Creek was added to improve the voltage proflle between Soldotna and University. It was agreed in the Phase II preliminary report meeting in Anchorage on March 2 to remove this capacitor bank and adjust the SVS size or Soldotna capacitors to compensate for it. Cases above number 205 do not represent this capacitor bank. The drop in voltage is partly offset by raising the 138-115 kV University transformer tap 2.5% and by increased SVS output at Soldotna. 11 Power Technologies, Inc. the drop in overvoltages is most pronounced at and near Soldotna, with voltages closer to Bradley, such as Fritz Creek and Diamond Ridge less improved. Overvoltages will be less severe for less severe faults. The cases discussed above all focus on a three-phase fault near Bradley. Case 165 is similar in all respects to case 161 except that it represents a phase to phase fault near Bradley. The backs wing overvoltages are all 112% or less at all buses except Bradley Lake. Moving a three-phase fault some distance from Bradley, say 5 to 10 miles, would have a similar effect. Only three-phase faults near Bradley will cause the backswing overvoltages exhibited in cases 161-164. The backswing overvoltages are aggravated by the drop in generator power caused by use of the deflector. Delaying cut-in of the deflector will require a larger braking resistor, but will reduce backswing overvoltages. The Bradley stabilizers also aggravate backswing overvoltages. Stabilizers "see" the acceleration of the plant during and immediately after the fault as part of an oscillation, and increase excitation beyond the increase caused by the fault and post-fault power swing, thus driving machine flux up so high that the voltage regulator cannot pull it down during the backswing. Case 205 was run to explore these two effects (it also does not have new shunt capacitors at Quartz Creek). The stabilizer was removed and the deflector is delayed .5 seconds. The highest load bus backswing overvoltage is Fritz Creek, and it is just over 115%. Other voltages are under 115%. Only Fritz Creek voltage is above 110% for more then .5 seconds. The deflector was delayed only one half second in this case, not enough to fully remove its impact on backswing overvoltages. Additional delay could be used. The stabilizer was disabled for this case, but its effect on backswing overvoltages could be limited by cutting stabilizer output to zero when voltage is above some threshold, or reducing stabilizer gain as voltages rise above some threshold (e.g. 108% ). This capability is not normally provided in stabilizer circuitry, but can be added if necessary. The backswing overvoltages can also be reduced by increasing the size of the TCR at Soldotna. Case 206 is similar to 205 except the TCR is increased by 10 MV AR to give the SVS an overall range of +30/-35 MV AR (65 MV AR TCR). In this case only Diamond Ridge and Fritz Creek (and Bradley) voltages go above 110% during the backswing (to a maximum of about 115%). If the SVS "underexcited range" is increased beyond -15 MV AR to control backswing overvoltages, the reactors and thyristors need only be designed to operate briefly in this range. 12 Power Technologies, Inc. 90 MW Export Export levels as high as 90 MW are feasible with one SVS, three series capacitors, two 25 MW braking resistors, and stabilizers at Bradley. However, there is little stability margin at 90 MW export, and backswing overvoltages will be somewhat higher than desired, but if the system is healthy, it will withstand a three-phase fault close to Bradley Lake on the Bradley Lake to Soldotna line. The ability to export 90 MW comes primarily from the fact that the compensation is designed to allow Bradley Lake to operate at 120 MW with Bernice and Cooper generation off. Adding units at Cooper and Bernice is necessary to increase export to 90 MW, and improves stability enough to allow such operation without exhausting the margin built into the reference case (Bradley at 120 MW, low Kenai load, Bernice and Cooper off). At 90 MW export, the system is closest to its first-swing stability limit. Dynamic and steady state stability margins are reduced at 90 MW, but are more substantial than the first swing margin. This difference occurs because the Bernice and Cooper generation contribute somewhat more to dynamic and steady state stability than they do to first swing stability. However, at 40 MW the Bradley brake is adequate for this operating condition. Though case 168 shows that the system may be operated at 90 MW export with one Bernice unit and one Cooper unit on-line, the results cannot necessarily be extrapolated to all conditions under which high export might be achieved. For example, the system stability may be different with Bradley Lake supplemented by just Bernice units or just Cooper units, or with a different dispatch in the Anchorage area. Though these differences should not cause dramatic differences, there may be some under which 90 MW export would be less stable than shown in case 168. If operation at 90 MW export is to occur frequently or must occur under conditions substantially different from those examined in case 168, additional cases for those conditions should be run to ensure that they are stable. Though operation at 90 MW export will be feasible with the compensation equipment outlined above under at least one operating condition, an increased compensation level should be considered if operation at 90 MW is to occur on a regular basis. That is, if it is to occur under emergency or unusual operating conditions that will exist for only a few hours or over a period of a few days each year, then additional compensation is not warranted. However, if such operation is to occur every season for a period of some weeks, then the SVS upper limit should be increased to 25 MV AR. The SVS low limit of -25 MVAR as justified elsewhere in this report should be retained. 13 Power Technologies, Inc. Reduced Runback With Bradley Lake power runback from 120 MW to 90 MW or less following trip of the line from Bradley to Soldotna brings loading on the lines from Diamond Ridge to Soldotna within the summer overload capability of these lines. During the winter, however, the rating is higher, and a lower Bradley Lake power runback could be accommodated if stability can be maintained. Case 166 was run to assess equipment requirements to do this. Power flow case P2CA was frrst run to check post-disturbance conditions with Bradley Lake run back 11 MW to bring loading on the Diamond Ridge to Anchor Point line within its winter emergency rating (assumed to be 110% of the 91.2 MV A continuous winter rating --100.3 MV A) under low Kenai load (40 MW)3 As shown in power flow case P2CA, the Soldotna SVS must provide 25.3 MVAR to hold Soldotna voltage under this condition. To provide margin for dynamic stability and to ensure steady state stability, the SVS was increased to +30/-15 MVAR for the stability test outlined below. Stability case 166 shows that the system will survive fault (three-phase) and trip of the line from Bradley to Soldotna with runback limited to 11 MW when the Soldotna SVS range is + 30/-15 MV AR. However, note that the SVS is close to ceiling in the seconds after the large frrst-swing. In earlier cases with runback to 90 MW (case 161), the SVS was 10 to 15 MVAR from ceiling in the fmal seconds of the simulation. Case 166 thus is closer to stability limits than previous similar cases (e.g. case 161). Note particularly that the oscillations between 2 and 5 seconds in case 166 are lightly damped because the SVS is on ceiling for a total of about 1 second between 2 and 4 seconds (i.e. it has only limited ability to provide damping in this period). An additional 5 to 10 MVAR would be required for case 166 to be as secure as similar cases with Bradley power run back to 90 MW. However, if case 166 represents a condition that will be observed infrequently, the lower margin may be acceptable. Because, as noted above, the steady state SVS output some minutes after the disturbance will be at 25.3 MVAR, and the simulations show poor damping with a 30 MVAR SVS, the SVS upper limit should be increased to 35 MVAR if runback is to be limited to 11 MW. To be fully comparable (i.e. to have the same margin) as the cases with runback to 90 MW, the case with runback to 109 MW would require an SVS with an upper limit of 40 MV AR. 3 Loading on the 115 kV line north of Diamond Ridge is highest when load at Diamond Ridge and Fritz Creek is lowest 14 Power Technologies, Inc. Loss Of Anchorage Tie During Export In order to minimize risk of dropping Kenai load following trip of any of the several line sections between Soldotna and University, the Railbelt utilities have determined that Kenai frequency must remain between 59 and 61.5 Hz following separation from the Anchorage area. Keeping frequency in this range requires: o prompt governor action at Bradley to run the deflectors in at the maximum rate. o similar governor action on any Bernice and Cooper units that are on-line at the time of separation. o coordination of governor characteristics to rrumrruze motoring and the risk of unit trip (by reverse power relays) during power runback (keeping the combustion turbines on-line reduces the risk of collapse of the Kenai island). o droop settings that will allow power sharing among on-line units at the reduced plant loadings. o a change in governor speed reference to bring post-separation frequency into the range 60 to 61 Hz. o frequency regulation capability to accommodate normal load vanat10ns and restoration of any load that is dropped by separation (e.g. the fault that initiates separation). o control of voltage sufficient to not cause generator trip by loss of excitation protection or loss of load due to low voltage. Most of these requirements are associated with the dynamic response of the· power plants, the system loads, and . the SVS. Simulations are essential, for instance, to show that Cfs are not "motored" off the line (tripped by reverse power relays) during the power run back. Initial Overspeed The first task is to limit the initial overspeed. A fault that causes loss of the tie will, itself, cause rapid Kenai generator acceleration by momentarily depressing load and export, 15 Power Technologies, Inc. and by causing motor contactors to open thereby reducing Kenai load. Once the tie is open, the power flow to Anchorage is interrupted, and there is further acceleration. It is desired that the overspeed be limited to 61.5 Hz if practical. The cases described in the following paragraphs examine the basic overspeed problem, and the use of the Bradley brake and Bradley unit tripping options individually and in concert to limit overspeed. All explore a very high export leveL Case 181 is for loss of the tie to Anchorage when export is at 90 MW and Kenai load is low (42 MW). With export high and Kenai load low, the Kenai area overspeed will be the greatest Survival of the Kenai area is most likely if generators at Bernice and Cooper remain on-line through the disturbance because they increase area inertia and will provide better frequency regulation following the initial speed excursion. Case 181 represents a worst-case condition in that the separation is preceded by a three-phase fault at Soldotna (on the line to Quartz). This fault depresses area voltage and accelerates the Kenai generation even before the separation occurs. The overspeed is further aggravated by loss of 20% of the Kenai load (south of Soldoma) during the fault. The load is modelled as 60% motors to ensure accuracy of the simulation ( 1/3 of the motors are dropped during the fault). In case 181 one Bernice generator is on-line and has a typical combustion turbine governor. Because the Bradley governor is not yet defmed, and because it would surely ramp down at the maximum rate once the overspeed is detected, it is represented as a linear ramp from 120 MW down to 30 MW over a 1.2 second period. The actual governor would probably overshoot somewhat, and hold turbine power low until frequency is closer to its droop characteristic (depending on whether the governor is proportional or integral type). If all of the machines have a droop setting of 5%, and are operating near full rated power before the disturbance, the Kenai system frequency will setde at about 62.4 Hz (for the situation of case 181). In case 181 the Bradley Units are ramped back to 30 MW, just 2 or 3 MW below the total area post-disturbance load. As a result, the frequency is dropping only very slowly after the overspeed condition is arrested. The combustion turbine is limited to just a couple of megawatts of reverse power with the model data used in this case, while the actual control is likely to go further negative (because of the large compressor load on the turbine). The SVS is on its lower limit through most of the run, but the output is rising near the end of the run, indicating that the SVS and associated Soldoma capacitors may be able to handle the voltage in such a case without tripping of Soldoma mechanically switched capacitors. However, note that the Bernice Lake excitation is quite low, so tripping some of the capacitors may be a good idea. This can be done under SVS control 16 Power Technologies, Inc. (i.e. if the SVS remains on its lower limit for more than 2 or 3 seconds, it should initiate trip of some of the Soldotna capacitors). The Bradley Lake stabilizers-remain fully active in case 181, and contribute significantly to the high voltages in the first second after the separation. To the stabilizer the overspeed looks like the first of a series of speed oscillations due to poor damping. It is possible to remove stabilizers from service during large speed excursions, but has always been undesirable because it is difficult to distinguish between speed excursions due to damping and those due to other causes. It is also difficult to define logic to bring the stabilizer back on-line. For instance, it is quite possible to have overall system frequency rise rapidly and remain high without loss of interconnections so that the system is in a state in which the damping from stabilizers is essential. Digital stabilizers offer a solution to this problem. The stabilizer output can be clipped when voltage at the plant rises above some threshold such as 110%. This can be done without taking the stabilizer off-line. Since high voltages are momentary (1 to 2 seconds) while the damping problem is a longer-term problem (2 to 10 or 20 seconds), the temporary loss of stabilizer action is not a problem. Also, if the reduction in stabilizer output is made proportional to the voltage excursion above some threshold such as 110%, the stabilizer output is only totally suppressed if the voltage exceeds about 115%. With the duration of total loss of stabilizer effectiveness is very short Case 181A is similar to 181 except that the Bradley brake is used to reduce the overspeed. The brake is applied 6 cycles after the fault is applied (2 cycles after the fault is cleared). A severe fault at Soldotna would trigger the brake somewhat more quickly than this, but an overspeed trigger might be somewhat slower (an overspeed trigger would be needed to apply the brake when separation from Anchorage does not follow a fault). The brake is left on for . 7 seconds because at this time the brake is planned to have a thermal duty cycle of just .75 seconds. A longer duty cycle could be provided if the brake is deemed useful in limiting overspeed following separation from Anchorage. Case 181B is also similar to 181 except that one Bradley Lake unit is tripped about .3s after the fault is applied. The frequency is about 61.4 Hz when the unit trip occurs. The remaining unit is run back to 20 MW and held at that level. The final frequency will be under 62 Hz in this case if the two machines (Bernice and Bradley) have the same droop and start at about the same loading (in percent of rated power). 17 Power Technologies, Inc. To demonstrate the use of the brake, unit tripping, and governor speed reference change, case 211 was run. The following were simulated in this case: o A simple governor with 1.1 second time constant is modeled at Bradley. Its response to overspeed is delayed .4 seconds to conservatively represent the initial delay in the deflector mechanism. o Soldotna switched capacitors (MSC) are taken off 15 MVAR at .4 second and 15 MVAR at .8 seconds (assumed to be under control of the Soldotna SVS). o The Bradley brake is 50 MW (2 X 25 MW) and is applied at .2 seconds and removed at .8 seconds (.1 seconds less than the presently planned thermal capability). o One Bradley unit is tripped at .4 seconds (its brake is left on). o 7.5 MW (about 20%) of Kenai load is dropped during the fault. o The fault is three-phase at Soldotna and causes immediate trip of the 115 and 69 kV lines to Quartz Creek (conservative assumption since the 69 kV line trip would be delayed and would reduce initial acceleration somewhat). o The export is 90 MW when the fault occurs. o Bernice unit 3 is operating at 24 MW. Bradley is at 120 MW. o Governor speed references are dropped 50% at .4 seconds (this would restore the island to 60 Hz if the loading on remaining machines drops to 50% of initial loading). o The Bradley Lake stabilizers are not disabled and do not have gain reduced during the period of overvoltage. The result of these assumptions is overspeed limited to just over 61.4 Hz, voltages below 110% except for two .2s excursions to about 111%, and Bernice combustion turbine power reversal of less than 1 MW, lasting less than .5 seconds. Though many assumptions were made for this case, and all may not be conservative, it is clearly possible to limit Kenai overspeed to less than 61.5 Hz when Kenai export is below 90 MW. 18 Power Technologies, Inc. Case 211 is based on the assumption that three-phase reclosing will not be used on the Kenai-Anchorage tie, or will not be used at the export levels that will require unit tripping to control overspeed. When three-phase reclosing is in use, Bradley unit trip must be delayed and then implemented only if the reclose is not successful. The controls necessary to limit overspeed to 61.5 Hz or less include: o Carrier or microwave signals from relays on the Kenai-Anchorage tie to Bradley and Bernice. Bradley unit trip and/or brake switching and governor speed reference changes should be initiated by relays that will open the tie due to faults or loss of synchronism --when export is above some threshold. The speed reference change should vary with Kenai export if practical. o Local control that will apply the braking resistors when frequency reaches about 60.5 Hz and is rising at a rate of 2 Hz per second or more, and will remove the brake when frequency falls below 60.5 Hz or the brake reaches its thermal capacity. This control ·should not interfere with the· brake controls that apply and remove the brake for faults that initiate stable power-angle swings (such swings do not cause frequency to exceed 60.5 Hz). Planning studies should be sufficient to ensure that the equipment necessary to achieve the desired control will be in place. Operating studies should be conducted to determine how the controls should be set and used under various operating conditions. Unbalanced Dispatch The division of power among the units following the initial overspeed period is importanL If all units operating at the time the tie is lost are to remain on the system, all must be at positive output following the initial overspeed. Any units operating at negative power will be tripped after several seconds by reverse power relays. The operating point of each unit after runback to control overspeed will depend on droop settings, initial operating points, and the amount of runback required. If all units have about the same droop, any unit operating substantially below the other units (in percent of rated maximum power), is likely to be tripped by its motoring protection when a large power runback occurs. The droop can be adjusted to give equal loading or some desired distribution of loading after runback for a given initial set of 19 Power Technologies, Inc. loadings and amount of runback. However, this requires software at a .central location with access to the present loading on each unit, and the ability to reset the droop on each unit to accommodate the present operating condition (i.e. give desired loadings after loss of the tie). The PC based Bradley governor could easily accommodate this novel system. The Bernice Lake crs may not be amenable to such control. If the CTs can accept this control or be modified to accept it, it is suggested consideration be given to such a system. Frequency Regulation If the Kenai area is to survive islanding when only Bradley Lake is in operation, the Bradley governing system must be able to follow load with sufficient response to keep frequency within a reasonable band (59 to 61.5 or better). In addition to normal load variations, there are two phenomena that will impose significant load changes on the system after islanding (loss of the tie to Anchorage). If the Anchorage tie is lost as the result of a fault that is either severe or close to Kenai loads, some motors in the Kenai will be dropped from the system. These motors will be restarted over a five to 10 minute period following loss of the tie. If 10 MW of motors is dropped, and return over a five minute period (pessimistic assumption), the load rise may be as much as 2 MW per minute. If voltages in the Kenai drop as the result of loss of the tie, residential and commercial loads will be reduced. L TCs will restore distribution voltages, thereby causing an increase in load over a period of 2 or 3 minutes after the disturbance. Similar loads not served by LTCs will be partially restored over a period of 10 to 15 minutes by the action of thermostats and similar controls and some manual control actions. In order to keep frequency from falling below 59 Hz as a result of these load increases, the needle valves at Bradley must be able to "follow" the load with only a modest time lag. The needle valves have a full-stroke time of 90 seconds. Hence one Bradley unit can provide .67 MW per second to meet load changes. Two units can provide 1.33 MW per second. This capability should allow the Bradley unit to follow all normal load variations and natural load restoration following faults without excessive frequency variations. However, step changes in load may be troublesome. A sudden large increase in load will cause the needle valves to begin moving quickly, and may cause a momentary drop in pressure at the nozzles. This drop in pressure will momentarily reduce the power produced by the Bradley turbine, and thus will contribute to the frequency decay caused by the step increase in load. This phenomena may limit the size of load step that can be 20 Power Technologies, Inc. accommodated without allowing frequency to drop below 59 Hz. More information on the turbine-penstock system and appropriate modeling (including system effects noted in the next paragraph) is required to determine precisely the limit to the step load that Bradley can handle while avoiding load shedding. The Bradley plant will have the assistance of load-frequency characteristics in preventing large frequency changes when loads increase. The loads themselves will drop with frequency, thereby helping to offset load increases. For instance, if the over-all load- frequency characteristic is 1% drop in load for 1% drop in frequency (it may be more than this), a 1 MW increase in a 50 MW load will cause frequency to drop only 1.2 Hz even without help from generator controls or the natural damping characteristic of the turbine. The benefit of load-frequency characteristics will not be available when the load is low or near zero. When loss of the intertie causes load shedding, restoring loads dropped by load shedding relays could impose sizeable step increases in load (including the cold-load-pickup effect). Restoring any feeders that would hit the system with more than about 2 MW are likely to cause additional or repeated load shedding (detailed simulations mentioned above are needed to provide accurate prediction of the maximum step load that can be accommodated without dropping frequency below 59 Hz.t It may be necessary to restore loads at lower voltages (i.e. in smaller steps) when Bradley is the only generation operating in the Kenai island. Frequency Control Following loss of the tie and runback, the frequency will settle well above the pre- disturbance frequency. If the droop settings are 5% and generator power must be reduced by 60% of rated power to balance load, the speed will increase 3% (to 61.8 Hz). If one Bradley unit is tripped, the final frequency will be about 61 Hz. If one Bradley generator is not tripped when runback of more than about 40% of the Kenai on-line generating capacity is required, the governor speed reference settings on all generators will have to be reduced to keep final frequency under 61.5 Hz. Even when unit tripping will bring the frequency to about 61 Hz without speed reference changes, the reference changes are advisable to get the system as close to 60 Hz (but not below 60 Hz) as practical. 4 Operators can reduce the risk of frequency excursions below 59 Hz following islanding by setting system frequency somewhat above 60 Hz until all loads are restored. 21 Power Technologies, Inc. The speed reference settings can be set quickly from a central location where there is sufficient information to calculate the new settings. The settings must be changed within quickly if the change is to prevent frequency from going over 61.5 Hz. Alternatively, the speed references can be changed based on the initial overspeed event. That is, the initial overspeed could trigger the change in speed references. A risk with this approach is that a system-wide frequency excursion of similar proportions could trigger the change in speed reference, thereby reducing power produced by Kenai generation until the settings can be restored through SCADA or by operator action. However, it is extremely unlikely that the whole system could reach 61 Hz and sustain a rate of rise at that frequency comparable to the Kenai generation following loss of the tie to Anchorage, making this approach worth considering. Bradley Lake Governor The Bradley Lake governor consists of a deflector and needle valves in each jet. At maximum closing rate the deflector can reduce turbine power by 100% in 1.5 seconds. At maximum closing rate the needle valves can reduce turbine power by 100% in 90 seconds. The deflector thus must be used when fast control is required, and the more efficient and precise needle valves can provide steady state regulation. The deflector and needle valves are to be controlled by a programmable controller that is being prepared by Woodward. This control must be designed to take maximum advantage of the deflector response characteristics in order to meet the goal of keeping Kenai frequency within the range 59 to 61.5 Hz following loss of the Anchorage tie. The digital control can be very precise, limited only by its speed sensing circuit and limitations and limitations associated with the deflector linkage and the turbulence of the stream. One control strategy is to supplement needle valve control with the deflector control when frequency rises above the needle valve droop characteristic by more than some amount such as .25 Hz, then return the deflector to the edge of the water stream when frequency returns to within .25 Hz of the droop curve under needle valve control. This requires that both the needle valve control and the deflector control be simultaneously active in some operating conditions, and also requires that the deflector control be idle when it is not needed, and activated when necessary to limit overspeed. To achieve the full potential of the PC based controller, it is essential that the deflector and needle valve operating characteristics be well known and taken into account in its design. Final design should be based on simulations of the complete plant and the associated electrical system. 22 Power Technologies, Inc. Specification and Operations Planning Studies Though the basic mechanisms to limit risk of frequency and voltage problems upon separation between the Kenai and Anchorage areas are well defined above from a planning standpoint, additional power flow and stability cases are essential to specify control equipment to meet all the various system operating conditions that can occur. Beyond this, yet additional simulations are needed to define operating procedures and control equipment settings that will make best use of the available equipment under an even wider range of operating conditions. Such studies must be repeated from time to time as system conditions change. Examples of the details that need attention include: o Calculation of load sharing among the several remaining machines based on droop characteristics. The droop characteristics should each be adjusted specifically to ensure sharing of load when there is a sudden reduction of up to 70% of area generation. These are essentially hand calculations, but demonstration of the load sharing following the initial overspeed can be confinned in the simulations outlined below. o Simulations of the dynamic response of the three plant governing systems to ensure coordinated response to the overspeed and reasonable sharing of the load drop. If loading is not well balanced the Cfs may see sufficient reverse power to develop flame stability problems during the runback. o Adjustment of reverse power relays so they will ride through temporary reverse power following loss of the Anchorage tie (based on simulations listed above). o Simulation of the voltage excursions following loss of the Anchorage tie for several Kenai load levels and generation dispatches to ensure that the SVS controls designed to improve stability will also accommodate islanding. Voltages must be sufficiently well controlled to ensure that generator loss-of- excitation protection will not operate during the initial voltage excursions or later in the post-disturbance steady state condition (SVS control of the Soldotna MSCs will be important). o Simulation to select settings for controls such as the overvoltage and overspeed cutout on the Bradley Lake stabilizers. o Simulations to select settings for controls on the Bradley braking resistors (for both turn-on and turn-off) and Bradley unit tripping. The relay and associated 23 Power Technologies, Inc. carrier or microwave communication details between Soldotna/Quartz Creek/Dave's Creek and Bradley also need to be defined. The maximum overspeed reached in case 181 is about 3.6 Hz, or about 6% above rated speed. The Bernice Lake Cfs have an overspeed trip set at 5% above rated speed. Though the overspeed can be reduced to less than 5% by use of the brake and/or unit tripping even at 90 MW export, it would be desirable to raise the Bernice cr overspeed trip settings to 6% to increase the margin between expected overspeed and the trip settings. Doing so will reduce the risk of nuisance trip should the system not be fully healthy at the time of separation, or some control equipment not function quite as desired following separation. Use Of TSC Rather Than TCRIMSC Early Phase ll studies included a look at the ASEA 7.2 MV AR Minicomp. This is essentially a small SVS in the form of a group of thyristor switched capacitors (TSC). The potential benefits of such a unit would include: o Reasonable cost and high reliability because it is a factory built unit. o Improved performance from distributing the compensation across 4, 5 or 6 substations (several small SVSs are typically more effective than fewer large ones --though the cost per kVAR increases dramatically as the size is reduced). The rninicomp was also considered because in all scenarios examined, the net shunt compensation is postttve. That is, even when a +20/-25 MVAR unit is applied at Soldotna, the net compensation is capacitive because the unit is in parallel with a 30 MV AR mechanically switched capacitor (MSC) bank. Note however, that to avoid need for any thyristor controlled reactor (TCR) capacity, the MSC equipment at Soldotna would have to be replaced with a rninicomp. Power flow cases 2HA and 2HB show, however, that an optimum distribution of the rninicomps is not one per substation along the 115 kV system, but four units (28.8 MVAR) at Kasilof and two at Soldotna. These cases are based on steady state conditions following loss of the Bradley -Soldotna line. Additional TSC capacity at either or both locations would be necessary for stability. Cases 2HA and 2HB show that reactive sources are best placed at Kasilof and Soldotna, though other casework shows Quartz creek to be a 24 Power Technologies, Inc. somewhat more advantageous location. The need to lump the reactive sources as shown in · power flow cases 2HA and 2HB comes about largely from the high resistance lines between Diamond Ridge and Soldotna. Because of the resistance, northerly power flow causes a voltage drop that is offset by southerly reactive flow. Reactive power injected at Quartz Creek and Kasilof flows largely southward. Stability cases showed that the minicomp was no more effective than two large SVSs, and in fact was essentially just that when several 7.2 MV AR units were grouped at Kasilof and Quartz Creek. Additionally, to use the minicomps effectively, some or all of the MSC equipment at Soldotna would have to be replaced with one or more minicomps. Replacing some of the MSC equipment at Soldotna with TSC equipment may be a good idea if doing so is more cost effective than adding a large TCR to offset the capacitors during backswings, oscillations, etc. However, a single large package of appropriate equipment is sure to be less costly than replacing existing MSC with several minicomps. Line/Transformer Drop Compensation Voltage control in the Kenai can be improved by using the Bradley Lake voltage regulator to control voltage well out on the Bradley-Soldotna and Bradley-Fritz Creek lines. This can be done by adjusting the voltage regulator line/transformer drop compensation for an impedance greater than the stepup transformer impedance. This benefit can also be provided manually if the operators are instructed to raise Bradley 13.8 kV voltage when the reactive power out of the plant his high (i.e. indicating low voltages in the Kenai 115 kV system). However, if it is provided automatically by a line/transformer drop compensator circuit, the regulator will not only hold voltage away from the plant in the steady state condition, but will be somewhat more responsive to reactive power swings. This can help first-swing stability. Unfortunately, it pushes up the machine flux during the first-swing, and this flux is sustained during the backswing by the stabilizer. The net effect is backswing overvoltages 2 to 3% higher than they are without the line/transformer drop compensator (compare cases 157 and 159 respectively). Hence line/transformer drop compensation is recommended at Bradley Lake only if it can be provided with its own slow time constant, or is set only to reach into the stepup transformer. It would be highly advantageous to apply the line/transformer drop compensation through a time-lag that would prevent it from responding at first-swing and damping frequencies, but would allow it to adjust the exciter operating point slowly to accommodate steady state conditions. Aside from improved voltage regulation in the Kenai area, the line/transformer drop compensation would improve steady state stability of the plant measurably. 25 Power Technologies, Inc. The series capacitors and the SVS at Soldotna are designed to ensure steady state stability with the Bradley Lake voltage regulator holding Bradley 13.8 kV bus voltage. However, the stability margin will be increased if the Bradley voltage regulators can hold voltage out on the 115 kV lines. Also, extending the voltage control point out on the lines may maintain steady state stability during some operating conditions (unusually high export) or equipment outages (SVS outage) that might otherwise be unstable. An additional note: The system can be expected to be steady state unstable under moderate to high Bradley Lake generation and/or high Kenai export when the Bradley units on "hand control" --i.e. with regulators off and voltage controlled manually. Such operation must be avoided. SVS Size and Design The SVS recommended in this study is to have a dynamic range of 55 MV AR (from +30 to -25). However, the SVS need not be capable of continuous operation over this complete range. Operation below 0 MV AR can and should be made infrequent by proper control over the MSC at Soldotna. Operation below 0 MV AR will be temporary as it will occur during backswing overvoltages and loss of load or system breakup. Hence this portion of the dynamic range can be provided by short-time rated components, or by component overload capability. This opportunity will be covered further in the Phase Ill report. Load Rejection Overvoltages The large amount of capacitors at Soldotna present some risk of "load rejection overvoltages." The potential for damaging overvoltages has only been examined briefly in Phase II, but needs to be considered further before Soldotna substation equipment changes to accommodate the SVS are fmalized. This problem will be considered further in Phase Ill Preliminary Specifications), but some consideration of the problem at this time is useful in that it may affect decisions regarding the SVS size and placement. Load rejection overvoltages occur when line tripping, often involving operation of backup protection, leaves line or cable charging or capacitors connected to a weak source. In the case of the Soldotna capacitors, a problem that opens the lines to Ski Hill, Bernice Lake, and Bradley would leave the SVS and capacitors on the long radial line from University to Soldotna. The line loading will drop, leaving the reactive power from the 26 Power Technologies, Inc. capacitors to flow all the way to the Anchorage area to be absorbed by load or generators. The voltage at Soldotna would be very high, possibly high enough to cause arrestor failure or flashover of the 115 kV line or bus. If the high voltage does not cause a fault the system, it can cause a breaker to fail if that breaker attempts to open under the high voltage and high leading current The overvoltage would be particularly severe if the SVS is operating at maximum output at the instant the system lands in the radial load rejection situation. With the SVS TCR gated off, the capacitors associated with the SVS (20 MVAR) plus the existing capacitors (30 MV AR) would total 50 MV AR. The voltage would be brought down only after 1 or 2 cycles when the SVS overvoltage protection gates the thyristors into full conduction. Flashover or arrestor failure can occur during this brief period of excessive voltage.5 The existing system is probably at higher risk of these potentially damaging load rejection overvoltages than the system will be in the future. The SVS and added line to Bradley will reduce risk of overvoltages. However, the risk may still be excessive, and it is prudent to begin now to determine if the problem wm remain, and if it does, to define solutions. Possible solutions include: o Increase the size of the TCR (though this will not help the first cycle overvoltage when the SVS is on ceiling).6 o Apply metal oxide varistor (MOV) type surge protection to absorb energy and thus hold voltages down until capacitors can be switched off. o Arrange the substation (breakers and protection) so that radial load rejection is very unlikely. 5 The SVS controls can be designed to gate the TCR on full when a fault occurs, and then return to the voltage regulation mode when the fault is removed in order to avoid high voltage upon fault clearing. 6 Where an SVS driven to ceiling by faults presents potential overvoltages immediately following fault removal (or resonance problems during the fault), it may be necessary to gate the TCR thyristors full on during the fault, and then resume voltage control following fault clearing. 27 Power Technologies, Inc. o Initiate trip of the capacitors for any relay or breaker operations that are likely to leave the system in a situation wherein voltages would be excessive. o Locate the capacitors so they are unlikely to become isolated with a weak source. Two power flow cases were run as a preliminary assessment of the potential for damaging overvoltages at Soldotna. With all lines into Soldotna opened, the 30 MV AR of capacitors left on line, and the Quartz 115 to 69 kV transformer removed, the Soldotna voltage is 147% and the University 115 kV bus voltage is 106%. Under the same condition with 50 MVAR on the Soldotna bus, the voltage is 184% at Soldotna and 110% at University 115 kV bus. The Soldotna capacitors could also cause high voltage if isolated with less than 40 to 50 MV A of Bernice Lake generation. The Bradley generators are large enough to avoid a high voltage problem should one of them become isolated with Soldotna capacitors. Brake Control Considerable detail on the application and control of the Bradley braking resistor will be presented in the Phase ill report (Preliminary Specifications). However, several cases were run to determine how critical the brake application time is in terms of stability, backswing overvoltages, and inadvertent brake application. These cases are discussed in this section. Early removal will result in first-swing instability. Late removal will result in large backswing overvoltages, and may cause large power oscillations that the SVS and Bradley stabilizers cannot overcome (i.e. they will be undamped instead of damped). . ~ -·. : . -._ ,• ~ . . Cases 193 through 196 explore the consequences of inappropriate brake application times for the worst-case operating condition (low Kenai load, Bradley at 120 MW, and Bernice and Cooper off) for the worst-case fault (fault and trip of the line from Bradley to Soldotna) with results as follows: 28 Power Technologies, Inc. DURATION .3s As .5s .6s RESULT first-swing unstable near optimum in all respects (.39s is optimum for this system condition) large but tolerable back swing overvoltages excessive backswing overvoltages and large but damped post-first- swing oscillations Though the .3s case (193) is clearly unstable, removing the brake at anything less than the optimum time is risky. Cases 154 and 155, though not otherwise exactly the same as cases 193-196, show the results of removing the brake early but not so early that first- swing instability occurs. In case 154 the brake is too small and is removed too soon, in case 155 the brake size and switching time are optimum. Rather, the subsequent large oscillations result in dynamic instability. The brake duration should be optimum or greater than optimum, but not less than optimum. The .6s case (196) shows large post-frrst-swing oscillations, but they are well damped at the export level examined in this case (damping may be a problem at higher exports if an excessive brake application time initiates large oscillations). It thus does not appear that moderately excessive brake durations carry high risk of dynamic instability as was suspected from some early casework. The improved tuning of stabilizers and other minor differences seem to have reduced this risk. However, leaving the brake on longer than necessary does give rise to larger backswing overvoltages. Some voltages exceed 120% in this case. It should be noted that these cases were run with a detailed load model that includes saturation in motors, transformers, customer loads, and discharge lighting, so the voltages observed are not pessimistic: They are quite representative. Case 204 provides additional information. This case is for trip of the line from Bradley to Soldotna without a fault, and with the brake applied for 0.5s. The deflector is also run in to 90 MW. This case was selected because, the brake, if needed for this case, would have an optimum duration of just .1 or .2 seconds while the duration, if it is to be fixed, must be .4s or more to ensure stability for more severe faults. This case shows that a longer duration is not a problem for this disturbance. 29 Power Technologies, Inc. Case 179 was run for inadvertent application of the brake during low export. Inadvertent application of the brake will cause the largest swing under this condition. This case shows that the brake is not a threat under such conditions. Case 197 shows similar results under heavier export conditions. These cases show that a sensitive brake trigger can be used without risk of causing instability for minor disturbances such as lower voltage faults. The above cases show a relatively large tolerance for the brake duration so long as it is equal to or greater than the optimum time. However, the tolerance is really just .1 seconds or 6 cycles if high backswing overvoltages are to be avoided. The duration will be more critical for higher exports such as 90 MW. That is, extra brake duration will cause increasingly high backswing overvoltages and oscillations as the export is increased. The brake time will also vary somewhat depending on the level of runback. It must remain on longer when runback is reduced to take advantage of the winter Diamond Ridge line overload capability. To ensure that the brake is not removed too quickly, and does not stay on so long that it causes dynamic instability or troublesome backswing overvoltages, it is highly recommended that provision be made to remove it 3 cycles after the peak of the angular swing. This can be done by digital controls at Bradley. The computer would monitor Bradley rotor speed continuously, and each time the brake is applied by undervoltage relays, the computer would send a trip signal to the brake circuit breaker when the rotor speed returns to its pre-disturbance value (or .7 seconds, the maximum brake duration). The logic of the brake controller can be modelled and tested in the same simulation environment used for this study. If digital stabilizers are used at Bradley, the brake control function can be implemented in the same cpu or a tandem cpu in the same equipment. This will be a reliable approach because there will be two stabilizers at Bradley, one on each unit, either of which can control the brake. 30 Power Technologies, Inc. BRADLEY LAKE LIMITED TO 90 MW A stability case was run to extend the conclusions drawn from cases 123A through 123D in the Phase I report. Those cases indicated that with Bradley Lake power at 90 MW and Bernice and Cooper machines on-line, the system would be stable with only the Bradley brake and the Bradley deflector as stability aids. Case 200 was run to determine if stability could also be maintained with just these aids if Bradley Lake is the only generation operating in the Kenai area. The base case includes only Bradley Lake in operation in the Kenai, 40 MW Kenai load, 42.5 MW of export, and all but 10 MV AR of the Soldotna shunt capacitors switched off. Case 200 includes detailed load models and loss of load during the fault. The disturbance is a three-phase fault near Bradley Lake on the line to Soldotna followed by trip of the line at 4 cycles. A 40 MW brake is used, and the deflector is used to run power back to 60 MW over 0.4 seconds starting at the instant of fault clearing. The case is stable. The brake is removed at .35 seconds, indicating some margin for a longer duration fault or a higher bradley Lake power level. The system is shown to be lightly damped, but since the system model includes detailed motor models the simulation should be accurate enough to justify accepting the light damping. Voltages in the Kenai all appear likely to settle between 95 and 100% in the seconds after the fault. The voltages will be even lower after a minute or two when distribution voltages have been restored by LTCs. In order to ensure continued stability in this case, Soldotna shunt capacitors would have to be switched on within about 1 minute after the fault occurs. This would be the case regardless of the cause of trip of the line from Bradley to Soldotna. Alternatively, Bradley could be ramped to a lower level such as 50 MW. SVS ALTERNATIVE Fifty percent series compensation of three of the longer lines in the Kenai area, combined with one SVS at Soldotna has been shown to meet Bradley Lake stability objectives. Hence this solution to the stability problems associated with the existing 115 kV system is recommended for consideration. However, a second alternative utilizing two SVSs and no series capacitors will also meet stability requirements, and may have advantages important to the railbelt utilities. 31 Power Technologies, Inc. Several cases were run to explore an all-SVS option. This option may become imponant if it is found that the lower cost series capacitor option presents design or operating difficulties associated with subsynchronous resonance. The cases focus on SVS size and damping controls, and address only the worst-case disturbance; fault and trip of the line from Bradley to Soldotna. In these cases it is assumed that the deflector will not cut in quickly enough to aid frrst-swing stability (the deflector is assumed to reduce power by 30 MW between .6 and 1.5 seconds). Stabilizers are not used on the SVS in these cases. The MSC at Soldotna are set at 15 MVAR and is unchanged in the simulations. Case 208 shows that two +50/-10 MV AR SVSs, one at Quanz Creek and the other at Soldotna, will provide complete stability. However, the full range of SVS capability is not used at either location. Case 209 shows stability with two +35/-0 MV AR SVSs. Neither SVS reaches ceiling during the first-swing, indicating the possibility of funher size reduction (if allowed by damping and steady state stability). The Quartz Creek SVS peaks at about 33 MVAR while the Anchor Point SVS reaches about 27 MV AR (both hit ceiling during the fault, but reactive supply during this period has negligible impact on stability). Case 209 was also used to assess need for stabilizers on the SVS. SVSs contribute significantly to damping without stabilizers, so the cost of stabilizers can often be avoided. However, it is highly desirable to have some margin, or redundancy, so that the system will remain stable during forced outages and maintenance outages of stabilizing equipment. Case 209 was run without the Bradley stabilizers to determine if the two SVSs alone would provide adequate damping. Oscillations build quickly even though both SVS are well below ceiling, and thus are providing maximum contribution to damping (i.e. the maximum possible without stabilizers). Two additional cases would be useful, one with just one Bradley stabilizer, and one with stabilizers working at Bradley and one SVS out of service. However, even if both of those cases show positive damping, it would be prudent to apply stabilizers to the two SVS to ensure that the system will not be subject to growing oscillations and instability upon loss of a stabilizing device during outage or maintenance of another. Case 210 is similar to 209 except that the Bradley stabilizers are working. The case is well damped. This case shows that +35/0 MVAR is close to an optimum size for the two SVSs. Both peak at close to their 35 MV AR ceiling, :md both settle just under 20 MVAR. The zero lower limit is adequate to keep backswing overvoltages under 110%. 32 Power Technologies, Inc. The two-SVS option does not exhibit large backswing overvoltages that a series compensated system does. The reason is that the post-disturbance system angle is much lower in the series compensated system, and the Bradley machines must swing back down to this angle after the brake is removed. The machines accelerate downward as they move back toward the final angle, thus overshooting it and momentarily pushing transfer down close to zero. With SVSs the final angle is much closer to the angle the Bradley machines are at when the brake is removed. Hence there is much less deceleration following brake removal, and little overshoot. With SVSs the modest drop in angle that does occur. is largely due to the power reduction provided by the deflector. SVS -Series Capacitor Comparison Though all alternatives for raising the stability limits to Bradley Lake power level and Kenai export involve at least one SVS, the decision between an all SVS solution and one making. use of a combination of SVS and series capacitors (SC) should not be based totally on costs and the ability to meet deterministic criteria.7 Some other considerations are presented in this section, and some details will be found in the Phase III report (Preliminary Specifications). These other considerations can affect reliability in ways that are not revealed by tests outlined in deterministic criteria. One of the most important considerations is the ability to handle disturbances beyond the deterministic criteria. Such disturbances are often called "possible but improbable" events or "PBis." Though most planning criteria that addresses PBis indicates that the effects of such disturbances need not be fully mitigated, but should be reduced to the extent practical with modest additional attention to detail or additional equipment. However, when there are two competing alternatives, it is useful to test them against PBis as a measure of inherent robustness (i.e. the ability to hold together beyond the basic criteria tests). If two alternatives are otherwise equal, the one that is inherently more robust should be selected. The benefit of a more robust alternative may also be judged to have monetary value, and thus offset the high It is usually the case that SC based systems or systems using SCs will survive a wider range of more severe disturbances and operator errors than SVS based systems. Flexibility to accommodate future network expansion or upgrade are also attributes that should be considered in the selection of a transmission alternative. In this regard, the SVS 7 Deterministic criteria is a set of specific and well defmed tests that a system plan must pass to be deemed acceptable and fully designed. 33 Power Technologies, Inc. based alternative may be judged to have a small advantage. SVSs, though likely not ideally located for some future system, will usually be useful whatever future development occurs. Series capacitors, on the other hand, may need to be moved or adjusted in size because of line load balancing or SSR problems. The series capacitors require little maintenance, little operator attention, and are very reliable. The absence of controls and the almost total lack of operator attention are probably the most attractive features of series capacitors in the APA system. An SVS, on the other hand, does require adjustment by operators on an hourly or more frequent basis, unless some relatively sophisticated controls are provided to keep the units operating in a mode that will ensure maximum contribution to stability when the system is hit by a fault and line trip. While SVSs have potential harmonic and control problems that must be dealt with in specification studies, series capacitors have their own potential problems. The most significant one is subsynchronous resonance (see the Phase III report for additional detail). It is unlikely, but possible, that subsynchronous resonance problems will rule out use of series capacitors in some or all of the candidate lines in the Kenai. Whether or not this is the case can be determined only after analysis of the potential for subsynchronous resonance, and the possible use of remedial measures to solve the problem. One possible remedial measure is increasing or decreasing the level of series compensation, or remove it from just those lines where it is troublesome. Until such analysis is conducted, it is suggested that the use of two SVSs be considered as a fallback option, and that the economic impact of having to go to this option be recognized as a possibility. If series capacitors cannot be used, then additional study work will· be needed to finalize the sizes of the two SVSs and their locations. The locations --Anchor Point and Quartz Creek --are fairly well defined and are unlikely to change, however, the sizes, indicated to be +35/-0 in case 208, 209 and 210, should be confirmed by further stability cases (only fault and trip of the line from Bradley to Soldotna was examined in cases 208, 209 and 210 --69 kV faults and other 115 kV faults should also be run to confmn the +35/0 MV AR size). If the series capacitor level need only be reduced to avoid SSR problems, then it may be feasible to increase the size of the Soldotna SVS to provide the desired stability performance. It is also possible that increasing series compensation will avoid an SSO problem in which case the Soldotna SVS size can be reduced. 34 Power Technologies, Inc. 230 KV LINE ALTERNATIVE The focus of the Railbelt Stability Study is on alternatives to a new transmission line from the Kenai region to the Anchorage area. However, some analysis of the stability performance of a new 230 kV line has been included as a point of reference from which stability performance of the less costly 115 kV upgrade can be judged. Installing the 230 kV line will allow a reduction in the compensating equipment that would be required for the 115 kV alternative. However, much of the equipment must be retained simply to meet the same stability criteria upon which the 115 kV plan is based. Additionally, if the full potential benefits of a second line are to be realized, some of the compensation equipment that may not be essential to. meet the 115 kV system stability criteria, must be retained. With the 230 kV line added to the system, there are two relatively severe disturbances that dictate compensation requirements. One is fault and trip of the line from Bradley to Soldotna. The second is fault and trip of the 230 kV line. First-swing, dynamic (damping), and steady state stability for the worst-case fault (fault and trip of the Bradley to Soldotna line) require that the braking resistor and two series capacitors between Bradley and Soldotna be retained. Runback is not required for stability for exports up to about 75 MW, but may be required at higher exports.8 The series capacitor between Soldotna and Quartz and the SVS at Soldotna are not essential. Case 167 shows that the system is first-swing stable for fault and trip of the line from Bradley to Soldotna without the SVS and Soldotna-Quartz series capacitor. The braking resistor and runback are essential in making this case stable.9 Cases 198 and 198A show that the system is also first-swing stable for fault and trip of the 230 kV line if the braking resistor is applied. Since the 230 kV fault is remote from Bradley, the brake will have to have sufficiently sensitive triggering based on voltage, or be triggered from 230 kV line protection via carrier. Runback may also be sufficient to provide first-swing stability for the 230 kV fault, but since the brake is required for other faults and avoids runback, it is 8 Runback will be required to avoid thermal overload of the Diamond Ridge to Soldotna line under some Bradley power levels, Kenai load levels, and seasons. 9 As shown in cases on the 115 kV alternative, the braking resistor size can be increased so that runback is not essential for fJISt-swing stability. Runback may be needed for dynamic or steady state stability, but can meet these requirements, but can be delayed and still meet them. 35 Power Technologies, Inc. the better choice. Also, the speed of response of the deflector is in question at this time and may not allow the deflector to contribute significantly to first-swing stability. Steady state stability is not a problem for either of the two worst-case disturbances (loss of the Bradley-Soldotna line and loss of the 230 kV line), but damping is a problem. It was first observed in case 198A which was extended to 5 seconds to check damping. This case shows essentially zero damping, a totally unacceptable situation if it is accurate. Because this case was run with a conventional simplified algebraic load model, and could be the deciding factor in whether or not an SVS is needed with the 230 kV line, it was rerun (case 199) with the complex load model (includes large and small motors, discharge lighting, saturable exciting current model, etc.). This case shows growing oscillations that will lead to loss of synchronism within 20 to 30 seconds after loss of the 230 k V line. The Bradley Lake stabilizers are active in these cases, but are not adequate to handle the damping problem. Adding series capacitors between Soldotna and Quartz will reduce the damping problem, but is not likely to solve it. Running Bradley Lake power back to 90 MW or less will solve the problem as shown in case 207. An SVS at Soldotna, probably limited to + 10/-10 MV AR, would also solve the damping problem. SVSs have been used simply for this purpose in a number of systems, so it is not surprising to have to do so in this case. Though a large SVS would not be required to provide damping, it would have to be operating well within its dynamic range following loss of the 230 kV line. Power flow cases discussed below show that a modest size SVS would go to ceiling or close to ceiling for loss of the 230 kV line. The controller on a small SVS would thus have to be especially designed to provide damping while sacrificing steady state voltage control. If runback is used to solve the problem, it must be triggered within several seconds of loss of the 230 kV line, and should reduce power within several additional seconds. That is, immediate and rapid power reduction is not necessary, but it must be done within about 5 seconds. The amount of runback required is difficult to estimate. Additional stability cases will have to be run to make this assessment. An SVS is the most effective and reliable means to solve the damping problem. While damping will vary with transfer, and reducing transfer can be used avoid damping problems, because damping is so heavily dependent on machine and load characteristics and on the complement of machines that are on-line, studies to define transfer limits (or runback) to achieve damping cannot be highly reliable. An SVS, on the other hand, if 36 Power Technologies, Inc. properly controlled (i.e. not allowed to sit on ceiling) and sized (i.e. with some margin), will easily handle the damping problem for any credible transfer level, dispatch, and assumptions about generating plant and load characteristics. There may be additional justification for providing at least a modest size SVS at Soldotna along with the 230 kV line. Power flow cases P2DA through P2DE cover a range of conditions as follows: P2D Base case with 230 kV line, 25 MV AR 230 kV shunt line reactor at Soldotna, no shunt capacitor bank at Quartz, no series capacitor between Soldotna and Quartz, and the SVS supplying just .4 MVAR (in addition to the 30 MVAR of MSC at Soldotna). P2DA Like P2D except Bradley reduced to 90 MW and the Bradley -Soldotna line open. The SVS is supplying 17.4 MVAR. P2DB Like P2DA except the SVS removed. Soldotna voltage drops from 102.0% to 99.0% P2DC Like P2D except the 230 kV line is out (Bradley remains at 120 MW). SVS is at 9.3 MVAR. P2DD Like P2DC except SVS removed. Soldotna voltage drops from 102.0% to 99.9%. P2DE Like P2D except 230 kV line open at University end. The SVS is absorbing 9.4 MVAR. P2DF Like P2DE except SVS removed. Soldotna voltage rises to 104.2%. Though these voltage variations do not argue strongly for an SVS at Soldotna if the 230 kV line is constructed, they do show that there will be some voltage variations if the line is installed and an SVS is not. If an SVS is not installed, operators can maintain voltages by switching Soldotna capacitors via the SCADA system. In this respect, voltages will be better regulated in the 115 kV alternative. 37 Power Technologies, Inc. If an SVS is to be used to regulate voltage at Soldotna in situations such as those examined in the above power flow cases, it must be sized to provide the voltage regulation and allow for modulation to provide damping. To do this, the SVS would have to have an upward dynamic range of about 20 MV AR, 10 to cover the voltage drop, and an additional 10 for damping. The above power flow cases were done neglecting any cables that will be required for marine sections of the 230 kV line. This should have little impact on the above conclusions so long as the additional charging from the cable sections is fully compensated by additional shunt reactor capacity on the line. Backswing overvoltages are shown in case 167 to be quite close to the Railbelt voltage criteria. 115 kV load bus voltages swing to about 116%, and are above 110% for about .5 seconds (the criteria allows a maximum of 115% and above 110% for no more than .5 seconds). H an SVS is used to provide damping, it should also reduce the backswing overvoltages somewhat (depending on size and initial operating point). The 230 kV reactor size was selected based on just one load flow case, and does not consider charging from submarine cables along the line route. The actual reactor size and location may thus be different from that assumed for this study. However, since the reactor will largely offset charging, and the two have the same voltage and frequency characteristics, adding the 230 kV cable and changing reactor size and location will have only a very insignificant effect on Kenai stability, but may be important from a steady state voltage control standpoint. The charging current not absorbed by the reactor, taking into account the ferranti effect, must be within the capability of Kenai area reactive control equipment (generators and SVS). This will be especially important during 230 kV line energization, following nuisance trip of the University terminal, separation due to instability, etc. RELIABll..ITY The 230 and 115 kV transmission alternatives require about the same compensation equipment to provide stable operation, and both will be affected by outages of that equipment. In both cases, the system as planned does not ensure stability for outages of the stabilizing equipment. That is, outage of the SVS, any one of the series capacitors, or the braking resistor, would make the system unstable for fault (three-phase) and trip of the line from Bradley Lake to Soldotna, and possibly for some other severe faults when Bradley Lake power level is high and/or export is high. The SVS and the brake are the 38 Power Technologies, Inc. two most critical items. Outage of one series capacitor would present a problem only for the most severe disturbance (multi-phase fault close to Bradley Lake on the line from Bradley to Soldotna), while an SVS or brake outage may leave the system unstable for loss of the line from Bradley to Soldotna even without a fault. Of the two, the brake is the most critical. The consequence of not designing for redundancy in the critical stability aids is that Bradley Lake power will have to be reduced during such outages to ensure stability, or to reduce the impact on the system if instability does occur. Alternatively, Bradley Lake power level and Kenai export can remain at or close to the desired level at the risk of instability for major disturbances. There are backup measures that can be applied to support increased transfer during outages of critical components. Two examples are increased Bradley runback, and application of generator tripping. Both of these options can be readily implemented when and if operations planning studies show they are beneficial (that is, the hardware and controls that are outlined in this report can be easily expanded to implement these options). The restrictions that must be applied to meet the normal criteria or a reduced criteria during outage of critical components must be established in operating studies or "operations planning" studies. These studies should consist of power flow and stability studies based on the equipment and conditions that are expected for the future period for which the studies are being conducted. 39 Power Technologies, Inc. APPENDIX I Load Models To ensure accuracy, most of the simulations done in Phase II are based on a detailed Kenai load model consisting of 30% small motors, 30% large motors, 5% discharge lighting and 35% load varying with the 1.5 power of voltage. The reactive load includes a saturable exciting current model set at 5% of the bus real power. Loss of one third of the motors ( 10% small motors and 10% large motors is simulated in most cases in which there is a widespread drop in voltage during a fault.