Loading...
HomeMy WebLinkAboutDelta Junction Area Wind Turbines Plant Impact Study - Dec 2010 - REF Grant 2195370WWW.EPSINC.COM PHONE (425) 883-2833 4020 148th AVE NE, SUITE C, REDMOND, WASHINGTON 98052 FAX (425) 883-0464 PHONE (907) 522-1953 3305 ARCTIC BLVD., SUITE 201, ANCHORAGE, ALASKA 99503 FAX (907) 522-1182 Alaska Environmental Power Golden Valley Electric Association Delta Junction Wind Plant Impact Study Draft Report December 13, 2010 David W. Burlingame, P.E. Dr. James W. Cote, Jr., P.E. Michael J. Bradley, P.E. John D.L. Hieb Alaska Environmental Power & GVEA Delta Junction Wind Plant Impact Study ii Revision Revision Date Revision Description 0 December 13, 2010 Rough Draft to AEP 1 December 14, 2010 Final Draft to AEP/GVEA Table of Contents 1 INTRODUCTION...................................................................................................... 1 2 RAILBELT DATABASE ............................................................................................ 2 2.1 GVEA Generation Additions ................................................................................. 2 2.2 Power Flow Cases ................................................................................................ 2 2.3 BESS Configuration .............................................................................................. 3 2.4 Generation Configuration – Winter Peak .............................................................. 3 2.5 Generation Configuration – Summer Peak ........................................................... 4 2.6 Generation Configuration – Summer Valley ......................................................... 5 2.7 Wind Generation Cases ....................................................................................... 6 3 POWER FLOW CONTINGENCY ANALYSIS ........................................................ 10 4 DISTURBANCES ................................................................................................... 13 5 DYNAMIC STABILITY RESULTS .......................................................................... 14 5.1 Transmission Fault and Trip Scenarios .............................................................. 14 5.2 Generator Trip Scenarios ................................................................................... 16 6 WIND MODEL/ANALYSIS ..................................................................................... 23 6.1 Wind Data ........................................................................................................... 23 6.2 Wind Analysis ..................................................................................................... 25 7 REGULATION ........................................................................................................ 36 7.1 Ramp Rate Analysis ........................................................................................... 37 7.2 Excess Energy – Curtailment ............................................................................. 38 7.3 Voltage Regulation/Coordination ........................................................................ 40 8 WIND RAMP DYNAMIC SIMULATIONS ............................................................... 41 9 INTERCONNECTION ............................................................................................ 44 9.1 Description.......................................................................................................... 44 9.2 Protection Requirements .................................................................................... 44 9.3 Wind Turbine Options ......................................................................................... 45 9.4 Interface ............................................................................................................. 46 9.5 Voltage Regulation ............................................................................................. 46 9.6 Curtailment Procedures ...................................................................................... 46 APPENDIX A_TRANSMISSION LINE DISTURBANCES ............................................. 48 APPENDIX B – GENERATION DISTURBANCES ....................................................... 49 APPENDIX C – WIND RAMP RATES .......................................................................... 50 APPENDIX D – EPS WIND ANALYSIS ....................................................................... 51 APPENDIX E – CLARITY ANALYTICAL WIND ANALYSIS ......................................... 52 APPENDIX F – WIND LOGIC - WIND ANALYSIS ....................................................... 53 APPENDIX G – EPS PROPOSED ONE-LINE FOR AEP WIND FARM ....................... 54 List of Tables Table 1: Winter Peak Load - No Wind Generation .......................................................... 4 Alaska Environmental Power & GVEA Delta Junction Wind Plant Impact Study iii Table 2: Summer Peak Load - No Wind Generation ....................................................... 5 Table 3: Summer Valley Load - No Wind Generation ..................................................... 6 Table 4: Winter Peak Wind Generation Cases ................................................................ 7 Table 5: Summer Peak Wind Generation Cases ............................................................. 8 Table 6: Summer Valley Wind Generation Cases ........................................................... 9 Table 7: Winter Peak Contingency Results – Line Overloads ....................................... 10 Table 8: Summer Peak Contingency Analysis Results – Line Overloads ...................... 11 Table 9: Winter Peak Contingency Analysis - Voltage Violations .................................. 12 Table 10: Contingency List ............................................................................................ 13 Table 11: Dynamic Stability Results Table…………………………………………………. 22 Table 12: Summer Valley Load-Nominal Import…………………………………………… 38 Table 13………………………………………………………………………………………... 38 Table 14: Available Regulation for Wind Ramp Events…………………………………… 41 List of Figures Figure 1: d_sp_brad_d2 Sample Plot ............................................................................ 14 Figure 3: GE LVRT Undesired Tripping Event .............................................................. 16 Figure 4: Winter Peak Maximum Import - Healy Clean Coal Trip .................................. 17 Figure 5: Winter Peak Maximum Import - HCCP Trip + Transfer Trip Healy-Cantwell .. 18 Figure 6: Winter Peak Maximum Import - HCCP trip + BESS Auto-Scheduling ............ 19 Figure 7: Winter Peak Maximum Import - NPCC1 Trip .................................................. 20 Figure 8: Winter Peak Maximum Import - NPCC1 Trip BESS Auto-Scheduled ............. 21 Figure 9: Graph and Chart of Distribution of MW Changes in 2 MW bins……………….. 25 Figure 10: Scatter of Negative Ramp Distribution from Starting Power Level 10 Minutes Ahead………………………………………………………………………………………...… 26 Figure 11: Scatter of Negative Ramp Distribution from Starting Power Level 1 Minutes Ahead…………………………………………………………………………………………... 27 Figure 12: Delta Junction distribution of Hourly 10 minute Min-Max-Avg………………. 28 Figure 13: Scatter of Negative Ramp Distribution from Starting Power Level 10 Minutes Ahead with Negative Ramp Curve………………………………………………………….. 29 Figure 14: Scatter of Negative Ramp Distribution from Starting Power Level 20 Minutes Ahead with Negative Ramp Curve………………………………………………………….. 30 Figure 15: Scatter of Negative Ramp Distribution from Starting Power Level 30 Minutes Ahead with Negative Ramp Curve………………………………………………………….. 31 Figure 16: Delta Junction and Eva Creek Distribution of Hourly 10 minute Min-Max-Avg …………………………………………………………………………………………………... 32 Figure 17: Scatter of Negative Ramp Distribution from Starting Power Level 10 Minutes Ahead with Negative Ramp Curve………………………………………………………….. 33 Figure 18: Scatter of Negative Ramp Distribution from Starting Power Level 20 Minutes Ahead with Negative Ramp Curve………………………………………………………….. 34 Figure 19: Scatter of Negative Ramp Distribution from Starting Power Level 60 Minutes Ahead with Negative Ramp Curve………………………………………………………….. 35 Figure 20: Scatter of Negative Distribution from Starting Power Level 1 Minute Ahead …………………………………………………………………………………………………... 37 Figure 21: Wind Ramp with Insufficient Capacity in Islanded Condition………………... 43 Figure 22: Wind Ramp with Insufficient Capacity Connected System…………………... 44 Alaska Environmental Power & GVEA Delta Junction Wind Plant Impact Study iv Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 1 1 Introduction Alaska Environmental Power (AEP) contracted with Electric Power Systems Inc. (EPS) to perform a grid impact study for the proposed new Delta Junction Wind Plant, interconnected to the Golden Valley Electric Association (GVEA) system near Delta Junction. The Delta Junction wind plant consists of 16 wind turbines rated at 1.6 MW each, step-up transformers, and a collector system connecting to the GVEA system at the Jarvis Creek substation. The study will analyze the operational impacts of the variable wind energy. Power regulation requirements will be recommended based on analysis of 10-minute wind rate ramp analysis. Additionally, the grid impact study will analyze the power flow and transient stability impacts of the Delta Junction wind plant on GVEA’s grid. The protection impacts to the GVEA system as well as the protection requirements for the AEP wind farm are also identified. The proposed 24 MW Eva Creek wind plant will be included in some of the impact studies, evaluated in conjunction with the Delta Junction wind plant for excess energy issues and the impact on power flows and transient stability in a limited number of cases. Both wind projects are assumed to consist of GE wind turbines. The Delta Junction Wind farm is modeled with the “WindInertia” package installed to provide an inertial response from the wind turbines during under-frequency grid events. Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 2 2 Railbelt Database The seasonal power flow cases and dynamic database used in this study were created from the 2020 Railbelt PSS/E database being completed by EPS for the Railbelt System Studies Subcommittee (SSS). The GVEA system loads were updated to reflect the forecast loads for 2015. The generation dispatches for the SSS cases were expanded to create various Anchorage-Fairbanks intertie import scenarios, and wind generation scenarios. 2.1 GVEA Generation Additions The two wind projects (Delta Junction and Eva Creek) result in three wind generation configurations for analysis. The three scenarios are: 1) a base case with no wind, 2) cases with Delta Junction wind only, and 3) cases with Delta Junction wind and Eva Creek wind both online. 2.2 Power Flow Cases The power flow cases were created to represent several GVEA operating situations with respect to the Anchorage-Fairbanks import. The cases are: Case 1 is configured with a nominal import on the Anchorage-Fairbanks Intertie, measured along the Douglas – Stevens section of the 138 kV line. In case 1, the generation is configured so that the intertie flow is equal to GVEA’s portion of the Bradley Lake generation. Case 2 is configured with an assumed maximum import of 75 MW, measured on the Douglas – Stevens 138 kV line section looking north. The import is achieved by taking Bradley Lake generation plus economy sales from the south. Case 3 is configured as an islanded case with the Anchorage-Fairbanks tie line out of service. Each of the three cases were further augmented to include the three different wind combinations of 1) no wind, 2) Delta Junction wind only, and 3) Delta Junction and Eva Creek wind. Additionally, the three load seasons of winter peak, summer peak, and summer valley will be analyzed. There are 27 base case power flows, resulting from 3 load conditions, 3 intertie flow conditions, and 3 wind plant conditions. The power flow case naming convention is as follows: The first section of the case name is the wind configuration o d_ represents the case with the Delta Junction Wind Farm o de_ represents the case with both Delta Junction and Eva Creek Wind Farms o Blank represents the case without wind generation The second section of the case name is the season o wp represents the winter peak case o sp represents the summer peak case Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 3 o sv represents the summer valley case The third section of the case name is the import condition o ‘75’ represents the maximum import case o ‘brad’ represents the nominal import case o ‘open’ represents the open tie case An example case name of d_wp_75.sav would correspond to a maximum import, winter peak case with only the Delta Junction Wind farm generation. 2.3 BESS Configuration The Wilson BESS configuration is important in the configuration of the dynamic cases. GVEA has stated that the BESS will not be used for the regulation of the wind power. However, it is assumed that the primary functions of SVC operation and emergency power injection for extreme under-frequency events are still available from the BESS. 2.4 Generation Configuration – Winter Peak Starting from the 2020 IOC / SSS Railbelt base cases, EPS created the following winter peak load cases. The GVEA generation was changed in order to get the three Anchorage-Fairbanks tie flow conditions discussed previously. Table 1 shows the load, losses, import, unit commitment, and unit dispatch for three winter peak load cases, without any wind generation. Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 4 Table 1: Winter Peak Load - No Wind Generation Generators Capacity MW Spin MW Spin MW Spin Healy Clean Coal 61.9 61.9 0.0 61.9 0.0 61.9 0.0 Healy 1 28.5 28.5 0.0 28.5 0.0 28.5 0.0 EvaCreek Wind 24.0 0.0 0.0 0.0 0.0 0.0 0.0 Chena5 24.0 24.0 0.0 24.0 0.0 24.0 0.0 North Pole 1 64.0 0.0 0.0 0.0 0.0 0.0 0.0 North Pole 2 64.0 54.9 9.1 40.7 23.3 0.0 0.0 North Pole CC1 53.0 53.0 0.0 53.0 0.0 39.7 13.3 North Pole CC2 12.0 12.0 0.0 12.0 0.0 9.9 2.1 DeltaWind 25.6 0.0 0.0 0.0 0.0 0.0 0.0 Univ. AK Gen 22.6 7.6 0.0 7.6 0.0 7.6 0.0 Fort WainwrightGen 20.0 6.0 0.0 6.0 0.0 6.0 0.0 Eielson 25.0 15.0 0.0 15.0 0.0 15.0 0.0 Bradley 15.2 0.0 0.0 14.0 4.6 14.7 4.6 BESS 25.0 0.0 25.0 0.0 25.0 0.0 25.0 Purchase N/A 0.0 0.0 0.1 0.0 59.8 0.0 Total Generation 262.9 34.1 262.6 52.9 267.1 45.0 Required Generation 262.9 262.6 266.9 Total Load 257.4 257.4 257.4 Losses 5.5 5.2 9.5 Additional Military Load (MW) 28.6 28.6 28.6 Auxiliary Loads (MW) 17.1 17.1 17.1 GVEA Net Load (MW) 211.7 211.7 211.7 WinterPeakNo Wind Islanded Nominal Import Maximum import Note that the BESS spin has been unadjusted from the base value of 25 MW. 2.5 Generation Configuration – Summer Peak Starting from the 2020 base cases, EPS has created the proposed summer peak load cases. The GVEA generation was changed in order to get the three Anchorage-Fairbanks tie flow conditions. Table 2 shows the load, losses, import, unit commitment, and unit dispatch for the three cases, without any wind generation. Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 5 Table 2: Summer Peak Load - No Wind Generation Generators Capacity MW Spin MW Spin MW Spin Healy Clean Coal 61.9 61.9 0.0 61.9 0.0 58.0 3.9 Healy 1 26.4 26.0 0.4 26.0 0.4 26.0 0.4 EvaCreek Wind 24.0 0.0 0.0 0.0 0.0 0.0 0.0 Chena5 24.0 24.0 0.0 24.0 0.0 24.0 0.0 North Pole 1 39.5 0.0 0.0 0.0 0.0 0.0 0.0 NorthPole 2 40.0 30.3 9.7 17.5 22.5 0.0 0.0 NorthPole CC1 33.0 33.0 0.0 33.0 0.0 0.0 0.0 NorthPole CC2 7.0 7.0 0.0 7.0 0.0 0.0 0.0 DeltaWind 25.6 0.0 0.0 0.0 0.0 0.0 0.0 Univ. AK Gen 22.6 9.2 0.0 9.2 0.0 9.2 0.0 Fort WainwrightGen 20.0 12.0 0.0 12.0 0.0 12.0 0.0 Eielson 25.0 15.0 0.0 15.0 0.0 15.0 0.0 Pump9Gen(FWP,FGP) 8.0 0.0 0.0 0.0 0.0 0.0 0.0 Bradley 15.2 0.0 0.0 14.0 4.6 14.0 4.6 BESS 0.0 0.0 25.0 0.0 25.0 0.0 25.0 Purchase N/A 0.0 0.0 0.0 0.0 59.8 0.0 Total Generation 218.4 35.1 219.6 52.5 218.0 33.9 Auxiliary Loads 15.94 15.94 15.94 Required Generation 218.44 219.54 220.74 Total Load 215.34 215.34 215.34 Losses 3.1 4.2 5.4 GVEA NetLoad (MW) 163.2 163.2 163.2 Additional Military Load (MW) 36.2 36.2 36.2 Islanded Nominal Import Maximum import SummerPeak No Wind 2.6 Generation Configuration – Summer Valley Starting from the 2020 base cases, EPS has created the proposed summer valley load cases. The GVEA generation was changed in order to get the three Anchorage-Fairbanks tie flow conditions discussed previously. Table 3 shows the load, losses, import, unit commitment, and unit dispatch for the three cases, without any wind generation. Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 6 Table 3: Summer Valley Load - No Wind Generation Generators Capacity MW Spin MW Spin MW Spin Healy Clean Coal 61.9 58.8 3.1 57.9 4.0 38.0 23.9 Healy 1 28.5 26.5 2.0 26.5 2.0 18.0 10.5 EvaCreek Wind 24.0 0.0 0.0 0.0 0.0 0.0 0.0 Chena 5 24.0 24.0 0.0 24.0 0.0 24.0 0.0 North Pole 1 47.0 0.0 0.0 0.0 0.0 0.0 0.0 North Pole 2 48.0 0.0 0.0 0.0 0.0 0.0 0.0 North Pole CC1 43.0 6.5 36.5 0.0 0.0 0.0 0.0 North Pole CC2 10.0 0.0 0.0 0.0 0.0 0.0 0.0 DeltaWind 25.6 0.0 0.0 0.0 0.0 0.0 0.0 Univ. AK Gen 22.6 0.0 0.0 0.0 0.0 0.0 0.0 Fort WainwrightGen 20.0 0.0 0.0 0.0 0.0 0.0 0.0 Eielson 25.0 0.0 0.0 0.0 0.0 0.0 0.0 Bradley 15.2 0.0 0.0 8.0 4.6 8.0 4.6 BESS 0.0 0.0 25.0 0.0 25.0 0.0 25.0 Purchase N/A 0.0 0.0 0.0 31.0 0.0 Totals 115.8 66.6 116.4 35.6 119.0 64.0 Required Generation 115.8 116.4 119 Total Load 113.7 113.7 113.7 Losses 2.1 2.7 5.3 Additional Military Load (MW) 0 0 0 Auxiliary Loads 20.4 20.4 20.4 GVEA Net Load (MW) 93.3 93.3 93.3 SummerValley No Wind Islanded Nominal Import Maximum import 2.7 Wind Generation Cases 2.7.1 Winter Peak Generation Cases The winter peak wind generation cases with wind generation are shown below in table 4. The addition of the wind reduced the amount of generation needed from the North Pole units as compared with the base cases without wind. The maximum import condition with the maximum amount of wind resulted with the North Pole Combined Cycle unit at its minimum while the other units remain near their maximum capabilities. Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 7 Table 4: Winter Peak Wind Generation Cases Generators Capacity MW Spin MW Spin MW Spin MW Spin MW Spin MW Spin Healy Clean Coal 61.9 61.8 0.1 61.9 0.0 57.2 4.7 61.9 0.0 61.9 0.0 59.1 2.8 Healy 1 28.5 28.5 0.0 28.5 0.0 28.5 0.0 28.5 0.0 28.5 0.0 28.5 0.0 EvaCreek Wind 24.0 24.0 0.0 24.0 0.0 24.0 0.0 Chena 5 24.0 24.0 0.0 24.0 0.0 24.0 0.0 24.0 0.0 24.0 0.0 24.0 0.0 North Pole 1 64.0 North Pole 2 64.0 28.8 35.2 15.3 48.7 10.0 54.0 North Pole CC1 53.0 53.0 0.0 53.0 0.0 24.0 29.0 48.5 4.5 45.8 7.2 6.0 47.0 North Pole CC2 12.0 12.0 0.0 12.0 0.0 6.0 6.0 12.0 0.0 11.5 0.6 Delta Wind 25.6 25.6 0.0 25.6 0.0 25.6 0.0 25.6 0.0 25.6 0.0 25.6 0.0 Univ. AK Gen 22.6 7.6 0.0 7.6 0.0 7.6 0.0 7.6 0.0 7.6 0.0 7.6 0.0 Fort Wainwright Gen 20.0 6.0 0.0 6.0 0.0 6.0 0.0 6.0 0.0 6.0 0.0 6.0 0.0 Eielson 25.0 15.0 0.0 15.0 0.0 15.0 0.0 15.0 0.0 15.0 0.0 15.0 0.0 Bradley 15.2 14.0 4.6 14.7 4.6 14.0 4.6 14.7 4.6 BESS 25.0 0.0 25.0 0.0 25.0 0.0 25.0 0.0 25.0 0.0 25.0 0.0 25.0 Purchase N/A 0.0 0.0 59.6 0.0 0.1 0.0 59.5 0.0 Total Generation 262.3 60.3 262.9 78.3 268.2 69.3 263.1 83.5 263.8 37.4 270.0 79.4 Required Generation 262.3 262.8 268.2 257.4 257.4 257.4 5.4 10.8 263 263.7 270 5.6 6.3 12.6 257.4 257.4 Losses 4.9 Total Load 257.4 17.1 17.1 17.1 17.1 17.1Auxiliary Loads (MW) 17.1 28.6 28.6 28.6 28.6 28.6Additional Military Load (MW) 28.6 211.7 211.7 211.7 211.7 211.7GVEA NetLoad (MW) 211.7 WinterPeak No Wind WinterPeak Delta Wind WinterPeak Delta and Eva Wind Islanded Nominal Imp Max Imp Islanded Nominal Imp Max Imp 2.7.2 Summer Peak Wind Generation Cases The summer peak wind generation cases are shown in table 5 below. While creating these cases, the condition with the maximum import, Delta Junction wind, and Eva Creek wind resulted in more generation capability than the load would allow. To resolve this issue, either the wind or import could be reduced. In order to simulate a worst case stability condition EPS kept the import at its limit while curtailing the wind generation. While this may not be a viable operational strategy, it will provide a worst case scenario for a stability simulation. Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 8 Table 5: Summer Peak Wind Generation Cases Generators Capacity MW Spin MW Spin MW Spin MW Spin MW Spin MW Spin Healy Clean Coal 61.9 61.9 0.0 61.9 0.0 41.0 20.9 61.9 0.0 61.9 0.0 38.0 23.9 Healy 1 26.4 26.0 0.4 26.0 0.4 22.0 4.4 26.0 0.4 26.0 0.4 18.0 8.4 EvaCreek Wind 24.0 24.0 0.0 24.0 0.0 16.5 0.0 Chena5 24.0 24.0 0.0 24.0 0.0 24.0 0.0 24.0 0.0 24.0 0.0 24.0 0.0 NorthPole 1 39.5 NorthPole 2 40.0 15.3 24.7 NorthPole CC1 33.0 24.0 9.0 25.7 7.3 22.2 10.8 8.9 24.1 NorthPole CC2 7.0 6.0 1.0 6.2 0.8 DeltaWind 25.6 25.6 0.0 25.6 0.0 25.6 0.0 25.6 0.0 25.6 0.0 16.2 0.0 Univ. AK Gen 22.6 9.2 0.0 9.2 0.0 9.2 0.0 9.2 0.0 9.2 0.0 9.2 0.0 FortWainwrightGen 20.0 12.0 0.0 12.0 0.0 12.0 0.0 12.0 0.0 12.0 0.0 12.0 0.0 Eielson 25.0 15.0 0.0 15.0 0.0 15.0 0.0 15.0 0.0 15.0 0.0 15.0 0.0 Bradley 15.2 14.0 4.6 14.0 4.6 14.0 4.6 14.0 4.6 BESS 0.0 0.0 25.0 0.0 25.0 0.0 25.0 0.0 25.0 0.0 25.0 0.0 25.0 Purchase N/A 60.5 0.0 60.4 0.0 Total Generation 219.0 60.1 219.6 38.1 223.3 54.9 219.9 36.2 220.6 54.1 223.3 61.9 SummerPeakDeltaWind SummerPeakDeltaandEvaWind Max Imp Islanded Nominal Imp Max Imp GVEA NetLoad(MW) 163.2 163.2 Islanded Nominal Imp 163.2 163.2 163.2 163.2 Additional Military Load (MW) 36.2 36.2 36.2 36.2 36.2 36.2 Auxiliary Loads 15.94 15.94 15.94 15.94 15.94 15.94 Total Load 215.34 215.34 Losses 3.7 219.54 215.34 215.34 215.34 215.34 4.2 Required Generation 219.04 223.24 219.84 220.54 223.24 7.9 4.5 5.2 7.9 2.7.3 Summer Valley Wind Generation Cases The summer valley cases are largely similar to each other due to the low levels of load and excess generation capability. It is likely that these cases produce more operational issues than stability issues. It can be seen that the summer valley nominal and maximum import cases are very similar due to the low load levels. When both wind projects are online, all of the cases need curtailment of the wind. In these cases, the minimum generation levels of the coal units may be the limiting condition for accepting wind energy. Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 9 Table 6: Summer Valley Wind Generation Cases Generators Capacity MW Spin MW Spin MW Spin MW Spin MW Spin MW Spin Healy Clean Coal 61.9 39.4 22.5 40.0 21.9 38.0 23.9 38.0 23.9 38.0 23.9 38.0 23.9 Healy 1 28.5 20.0 8.5 18.0 10.5 18.0 10.5 18.0 10.5 18.0 10.5 18.0 10.5 EvaCreek Wind 24.0 17.5 6.5 13.5 10.5 14.0 10.0 Chena5 24.0 24.0 0.0 24.0 0.0 24.0 0.0 24.0 0.0 24.0 0.0 24.0 0.0 North Pole 1 47.0 North Pole 2 48.0 North Pole CC1 43.0 6.5 36.5 North Pole CC2 10.0 Delta Wind 25.6 25.6 0.0 25.6 0.0 25.6 0.0 18.0 7.6 14.1 11.5 14.6 11.0 Univ. AK Gen 22.6 FortWainwright Ge 20.0 Eielson 25.0 Bradley 15.2 0.0 0.0 8.0 4.6 8.0 4.6 0.0 0.0 8.0 4.6 8.0 4.6 BESS 0.0 0.0 25.0 0.0 25.0 0.0 25.0 0.0 25.0 0.0 25.0 0.0 25.0 Purchase N/A 2.0 0.0 0.1 0.0 1.1 0.0 Totals 115.5 92.5 115.6 62.0 115.6 64.0 115.5 73.5 115.5 86.0 115.5 85.0 SummerValley DeltaWind SummerValley Deltaand EvaWind Islanded Nominal Imp Max Imp Islanded Nominal Imp Max Imp GVEA NetLoad (MW) 93.3 93.3 93.3 93.3 93.3 93.3 Additional Military Load (MW)0 0 0 0 0 0 Auxiliary Loads 20.4 20.4 20.4 20.4 20.4 20.4 Total Load 113.7 113.7 113.7 Losses 1.8 115.5 115.5 113.7 113.7 113.7 1.8 1.8 Required Generation 115.5 115.4 115.4 115.4 1.7 1.7 1.7 Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 10 3 Power Flow Contingency Analysis EPS performed a contingency analysis for the 27 power flow cases presented above. EPS outaged each 138 and 69kV transmission line in the GVEA system. Line flow violations seen in the winter peak cases are shown below in table 7. Table 7: Winter Peak Contingency Results – Line Overloads Savecase From Bus To Bus kV From Bus To Bus kV Rating(MVA) % Overload wp_open North Pole Carney 138 Highway Park Dawson 69 29 159.9 wp_open North Pole Carney 138 Dawson Eielson 69 29 139.3 wp_open North Pole Carney 138 Eielson Johnson Tap 69 29 132.4 wp_brad NorthPole Carney 138 Highway Park Dawson 69 29 160.3 wp_brad NorthPole Carney 138 Dawson Eielson 69 29 139.8 wp_brad NorthPole Carney 138 Eielson Johnson Tap 69 29 132.9 wp_75 NorthPole Carney 138 Highway Park Dawson 69 29 166.3 wp_75 NorthPole Carney 138 Dawson Eielson 69 29 140.7 wp_75 NorthPole Carney 138 Eielson Johnson Tap 69 29 134.4 wp_75 Wilson FTWW 138 Gold Hill Aurora 69 43 157.6 wp_75 Wilson FTWW 138 Aurora Zehnder 69 43 138.7 wp_75 Wilson FTWW 138 Gold Hill ChenaTap 69 43 137.8 wp_75 Wilson FTWW 138 University Tap International 69 43 121.7 wp_75 Wilson FTWW 138 ChenaTap University Tap 69 43 121.7 d_wp_open Wilson FTWW 138 Gold Hill Aurora 69 43 153 d_wp_open Wilson FTWW 138 Aurora Zehnder 69 43 134 d_wp_open Wilson FTWW 138 Gold Hill ChenaTap 69 43 133.7 d_wp_open Wilson FTWW 138 University Tap International 69 43 117.6 d_wp_open Wilson FTWW 138 ChenaTap University Tap 69 43 117.6 de_wp_open Wilson FTWW 138 Gold Hill Aurora 69 43 103.6 de_wp_brad Wilson FTWW 138 Gold Hill Aurora 69 43 120.4 de_wp_brad Wilson FTWW 138 Gold Hill ChenaTap 69 43 106.3 de_wp_brad Wilson FTWW 138 Aurora Zehnder 69 43 100.9 de_wp_75 Wilson FTWW 138 Gold Hill Aurora 69 43 183.4 de_wp_75 Wilson FTWW 138 Aurora Zehnder 69 43 165.1 de_wp_75 Wilson FTWW 138 Gold Hill ChenaTap 69 43 159.8 de_wp_75 Wilson FTWW 138 ChenaTap University Tap 69 43 144 de_wp_75 Wilson FTWW 138 UniversityTap International 69 43 144 de_wp_75 Wilson FTWW 138 International PegerRoad 69 43 121.4 de_wp_75 Wilson FTWW 138 PegerRoad SSS Tap 69 43 104.4 de_wp_75 Wilson FTWW 138 SSS Tap S. Fairbanks 69 43 104.4 Contingency Outage Overload Two contingencies result in line overloads. The contingencies are the 138kV lines from Wilson to Fort Wainwright and North Pole to Carney. The North Pole to Carney contingency results in violations on the 69kV line from Highway Park to Eielson substations. When the Delta Junction Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 11 wind farm is providing energy, the loading along the 69kV circuit is reduced under this contingency preventing flow violations. In order to mitigate these violations, the line sections between Highway Park and Johnson Tap could be re-conductored to increase ampacity from the 4/0 cable that is shown in the GVEA impedance diagram. The second contingency of the Wilson-Fort Wainwright 138kV line results in several 69kV line overloads for the lines leaving Gold Hill substation. The line overloads seen from this contingency are quite large. The cases with the Eva Creek wind farm online are more severe than the other cases. When the Eva Creek generation was added to the case, the North Pole generation was reduced. As such more generation is coming out of the Healy area resulting in greater flow on the lines going north into Fairbanks. Additionally, the worst cases are the maximum import cases that also stress the lines going north into Fairbanks. These violations could be mitigated by reconductoring the 69kV lines or unit dispatches that reduce the line flow north into Fairbanks. Due to the apparent severity of the Wilson-Fort Wainwright 138kV contingency, this event was added to the dynamic stability analysis. The summer peak line flow violations are included in table 8 below. Table 8: Summer Peak Contingency Analysis Results – Line Overloads From Bus To Bus kV FromBus To Bus kV Rating(MVA) % Overload sp_open North Pole Carney 138 Highway Park Dawson 69 29 111.9 sp_brad North Pole Carney 138 Highway Park Dawson 69 29 112.0 sp_75 North Pole Carney 138 Highway Park Dawson 69 29 114.8 sp_75 North Pole Carney 138 Dawson Eielson 69 29 102.8 sp_75 Wilson FTWW 138 Gold Hill Aurora 69 43 160.0 sp_75 Wilson FTWW 138 Aurora Zehnder 69 43 145.4 sp_75 Wilson FTWW 138 Gold Hill ChenaTap 69 43 138.1 sp_75 Wilson FTWW 138 University Tap International 69 43 129.1 sp_75 Wilson FTWW 138 ChenaTap University Tap 69 43 129.1 sp_75 Wilson FTWW 138 International PegerRoad 69 43 112.8 d_sp_75 Wilson FTWW 138 Gold Hill Aurora 69 43 135.3 d_sp_75 Wilson FTWW 138 Aurora Zehnder 69 43 120.6 d_sp_75 Wilson FTWW 138 Gold Hill Chena Tap 69 43 116.6 d_sp_75 Wilson FTWW 138 University Tap International 69 43 107.5 d_sp_75 Wilson FTWW 138 ChenaTap University Tap 69 43 107.5 de_sp_open Wilson FTWW 138 Gold Hill Aurora 69 43 107.2 de_sp_brad Wilson FTWW 138 Gold Hill Aurora 69 43 123.7 de_sp_brad Wilson FTWW 138 Aurora Zehnder 69 43 109.1 de_sp_brad Wilson FTWW 138 Gold Hill ChenaTap 69 43 106.2 de_sp_75 Wilson FTWW 138 Gold Hill Aurora 69 43 146.8 de_sp_75 Wilson FTWW 138 Aurora Zehnder 69 43 132.2 de_sp_75 Wilson FTWW 138 Gold Hill ChenaTap 69 43 126.6 de_sp_75 Wilson FTWW 138 ChenaTap University Tap 69 43 117.5 de_sp_75 Wilson FTWW 138 University Tap International 69 43 117.5 de_sp_75 Wilson FTWW 138 International PegerRoad 69 43 101.2 Savecase Contingency Outage Overload Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 12 The summer peak contingencies result in very similar results. Again, the problem contingencies are the Wilson-Fort Wainwright and North Pole-Carney 138kV transmission lines. Due to the lower load levels, the line overloads are not as severe as seen for the winter peak cases. The summer valley cases have no overloads for any transmission outages in the GVEA area. Only the winter peak cases have contingencies that result in voltages below 0.90 pu. The results of the voltage violations are shown below in table 9. Table 9: Winter Peak Contingency Analysis - Voltage Violations Case FromBus To Bus kV Bus kV Voltage (PU) d_wp_75 Jarvis DeltaWind 138 CANTWELL 138 0.88 de_wp_75 Healy Healy Wind 138 CANTWELL 138 0.91 de_wp_75 Healy Healy Wind 138 NENANA 138 0.89 de_wp_75 Jarvis DeltaWind 138 CANTWELL 138 0.89 Outage Violation The loss of generation in the heavy import cases result in a large flow from the Anchorage. These results are misleading as the BESS would auto-schedule for a loss of the Healy-Healy Wind 138kV line. This action mitigates the low voltages seen at Cantwell and Nenana. Also, by losing the Jarvis – Delta Wind 138kV line, the wind generation itself will be lost. The North Pole and Healy governors would respond to the loss in generation reducing the import from Anchorage. This response would mitigate the low voltages seen at Cantwell for this contingency. No remediation actions are required. Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 13 4 Disturbances The list of disturbances run with the power flow cases discussed in section 2 is shown below. The transient stability results were analyzed for possible system stability problems. In these planning cases, the largest unit contingency has been reduced from an AMLP Plant 2 Unit 7 or a CEA Beluga 7 outage to a CEA Southcentral Power Plant (SPP) outage. This largest unit outage is smaller in total MW’s lost than exists in today’s system, and therefore decreases the spin requirements of the entire grid. The SPP outage for the Winter Peak season is 67.5 MW. Table 5 lists the disturbances used for this study. Additionally, for the islanded cases, Healy Clean Coal is the largest unit in the GVEA area at 61.9 MW of capacity. The line fault and trip scenarios were selected as the anticipated worst case fault scenarios as seen from the proposed Delta Junction wind farm. The Douglas – Healy fault will be simulated with the fault being located at both ends of the line as seen in disturbances D1 and D2. The distribution faults will be evaluated to determine the voltage impact at the point of interconnection for the Delta Junction wind farm. Table 10: Contingency List Contingency Voltage (kV) Fault Location Clearing Time (Cycles) Name Line/Generator Local Remote D1 Douglas Healy 138.0 Douglas 5 30 D2 Douglas –Healy 138.0 Healy 5 30 D3 Jarvis –Fort Greely 138.0 Jarvis 5 5 D4 Carney Highway Park 69.0 Carney 5 5 D5 Carney North Pole 138.0 Carney 5 5 D6 Missile Defense Feeder 12.5 12.5 kV Feeder 5 N/A D7 Pump 9 Feeder 13.8 13.8 kV Feeder 5 N/A D8 Wilson –Fort Wainwright 138 Wilson 5 5 G1 Healy Clean Coal unit trip 13.8 N/A G2 SPP G2 unit trip 13.8 N/A Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 14 5 Dynamic Stability Results 5.1 Transmission Fault and Trip Scenarios All of the line fault scenarios exhibited acceptable system performance. All cases were stable. All frequency and voltage swings are within acceptable limits and all cases exhibit a well- damped response and return to nominal voltage and frequency. The worst case line fault and trip scenarios were the d1 and d2 contingencies. In cases with heavy import, the loss of the line causes GVEA to shed load, however, the GVEA system remains stable after the load shed. A sample plot showing the d2 event is shown below in figure 1. The top left plots show the several bus frequencies in the GVEA area and frequency trace in Anchorage. The top right plot shows selected bus voltages in the GVEA area. The bottom left plot shows the relevant wind farm information including the MW, MVAR, and voltage at the wind farm collector buses. The bottom right shows line MW flows as well as BESS output. Based on the line fault and trip transient stability results, the addition of Delta Junction wind farm or the combination of Delta Junction and Eva Creek wind farms have no negative impact on the system stability. Figure 1: d_sp_brad_d2 Sample Plot Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 15 The fault scenario results can be found in Appendix A. 5.1.1 Delta Junction Wind Farm LVRT/ Protection Coordination The Delta Junction Wind farm connects to the Jarvis Creek substation. Distribution feeders that can impact the 138 kV voltage at the wind farm’s POI are the Jarvis Creek substation 24.9 kV Jarvis Creek feeders, the Missile Defense substation feeder, Doyon’s 24.9 kV Fort Greely feeders and the 13.8 Pump 9 substation feeders. An example of a fault and clear of the Pump 9 distribution feeder is shown in figure 2. It can be seen that the Jarvis Creek 138kV voltage is reduced to ~0.45pu. The GE wind turbines have a low voltage ride-through (LVRT) characteristic that will trigger for voltages below 0.75 pu. It is clear that the distribution faults in the Jarvis Creek substation area can cause the voltage at Jarvis Creek to dip into the GE LVRT characteristic. Typically the distribution relays will clear faults on the system much faster than the GE Low Voltage event characteristic, however the initial fault clearing is not necessarily the area of concern with regards to LVRT. Based on EPS experience, the LVRT logic is not properly modeled in the PSSE models provided by GE. A quick discussion is warranted. Figure 2: Summer Peak Maximum Import with Delta Junction Wind - Pump 9 Feeder Fault The GE LVRT scheme uses a seal-in timer that begins timing at the first low voltage event. This timer will continue to count even after voltage returns to a healthy operation point (>0.75 pu) until such a time that the low voltage time is released. Since distribution feeders typically utilize Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 16 auto-reclosing in their protection, the GE LVRT characteristic can be triggered for normally- cleared, nearby distribution faults with reclosing on the second or third reclose event. The following event would result in a trip if the seal-in timer is set at 2 seconds. An example of this undesired wind trip is shown in figure 3. A distribution fault occurs at time = 0”. The voltage at the point of interconnection is reduced below the low voltage pickup of the GE LVRT to 0.4 pu. The relay picks up and clears the fault in 0.1 seconds and the POI voltage recovers to its nominal voltage. The distribution relay then automatically recloses at 2 seconds. The fault still exists and the relay clears the fault again at 2.1 seconds. Even though the fault is active for a total time of only 0.2 seconds, the GE LVRT will activate and trip the wind farm for this scenario. Figure 3: GE LVRT Undesired Tripping Event EPS recommends that GVEA and Doyon review their reclosing schemes in nearby distribution feeders to evaluate the impact of the reclosing on the low voltage ride through of the Delta Junction Wind Farm. The GE LVRT characteristics are programmable and offer a great deal of flexibility with regards to the tripping characteristics. We believe that between a combination of reclosing intervals and programming changes to the LVRT characteristics, inadvertent tripping can be avoided for faults on local distribution circuits. 5.2 Generator Trip Scenarios EPS ran a series of seven unit trip scenarios for each of the power flow cases. In running the simulations for the trip of Healy clean coal unit there were a few scenarios that resulted in out of step conditions along the tie to Anchorage when GVEA is at its maximum import condition. As such, EPS needed to perform additional investigation into the unit trip events. EPS added the following events: G3 – Trip of Healy Clean Coal, and a transfer trip of the Healy-Cantwell line G4 – Trip of Chena 5 unit Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 17 G5 – Trip of Healy 1 unit G6 – Trip of North Pole Combined Cycle unit G7 – Trip of Healy Clean Coal, with auto-scheduling enabled in BESS G8 – Trip of North Pole Combined Cycle, with auto-scheduling enabled in BESS The G1 contingency resulted in out of step conditions for the following cases: Winter peak with maximum import Summer peak with maximum import Winter peak with maximum import and Delta Junction Wind farm. Winter peak with maximum import and both wind farms online. The plot of the winter peak with maximum import case under the G1 contingency is shown below in figure 4. Figure 4: Winter Peak Maximum Import - Healy Clean Coal Trip Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 18 Figure 4 shows the GVEA system going out of step. The red trace in the top left shows the Healy 1 frequency separating from the other GVEA system frequencies. Additionally, the voltage collapse seen in the top right plot suggests the system has gone out of step. The G3 contingency was simulated as a mitigation step for the loss of the HCCP unit. This scenario assumed a transfer trip of the Healy line to Cantwell six cycles after the unit tripped. This simulation was to see if a transfer trip prevents the out of step condition seen in the G1 simulations highlighted in figure 2. The G3 contingency does resolve the out of step conditions seen in the G1 contingencies. The G3 disturbance applied to the winter peak maximum import case is shown below in figure 5. Figure 5: Winter Peak Maximum Import - HCCP Trip + Transfer Trip Healy-Cantwell Figure 5 illustrates that the transfer trip of the Healy-Cantwell tie in response to the trip of Healy Clean Coal can prevent the out of step conditions. All G3 contingency cases resulted in stable simulations. In addition to the transfer trip of the Healy-Cantwell line, EPS simulated the effects of the auto- scheduling function of the BESS. The BESS auto-scheduling is a communication aided BESS mode trip. GVEA provided the following list of seven events which will trigger the auto- scheduling component of the BESS operation. Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 19 1. Healy Unit 1 138kV unit breaker trip 2. North Pole Units 1&2 138kV breaker trip 3. North Pole Units 1&2 13.8kV unit breaker trip 4. North Pole Combined Cycle units 1&2 13.8kV breaker trip 5. Healy to Douglas 138kV trip at healy breaker B17 6. Healy to Wilson 138kV breaker trip at Healy B21 7. Wilson to Healy 138kV trip at Wilson ring-bus breakers B4 & B5 Each of these events communicates with the BESS and pre-loads a unique MW value for each event. When any of the seven events listed below occurs, a high speed communication system calls for the BESS to rapidly ramp to the pre-loaded output level. This BESS enable signal is triggered at the same time the relay decides to trip the breakers, as such, the BESS is assumed to begin ramping at three cycles after the breaker opens to account for some communication delay. This rapid BESS response prevents the system from going out of step in the simulations. The winter peak maximum import condition with the BESS auto-scheduling enabled. Figure 6: Winter Peak Maximum Import - HCCP trip + BESS Auto-Scheduling Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 20 When the BESS provides its maximum output three cycles after the initial HCCP trip, the system remains in synchronism. With the auto-scheduling feature, all generator trip simulations resulted in stable simulations such as this one. The additional unit trips of Chena 5, Healy 1, and North Pole Combined Cycle were expected to be less severe than the trip of HCCP. These were chosen to define a unit trip limit for out of step conditions without the BESS auto-scheduling. The results for the G4 and G5 disturbances result in stable, well-damped systems suggesting that the limit is larger than the 28.5MW seen on the Healy 1 unit. However, the loss of North Pole Combined Cycle 1 for the summer valley case with maximum import results in an out of step condition. In order to mitigate this condition, EPS modeled the auto-scheduling for the NPCC1 unit trip. The results for the out of step condition as well as the mitigation by auto- scheduling the BESS are shown below in figures 7 and 8. Figure 7: Winter Peak Maximum Import - NPCC1 Trip Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 21 Figure 8: Winter Peak Maximum Import - NPCC1 Trip BESS Auto-Scheduled Figure 8 shows the impact of the BESS’s fast response capability to a large loss of generation. By providing MW within 1/10th of a second, the out of step condition is avoided. Assuming GVEA has the capability to start generation before the BESS energy is expended, the loss of the NPCC1 unit is survivable under all contingency scenarios studied. Table 11 shows the simulation results observed in the Delta Junction Wind Farm dynamic simulations. Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 22 Outage G1 G2 G3 G4 G5 G6 G7 G8 PowerFlow Case HCCP SCPP 3 HCCP + TTHly Cantwell Chena5 Healy 1 NPCC1 HCCP +BESS NPCC1+BESS wp_open Stable Stable Stable Stable Stable Stable Stable Stable wp_brad Stable Stable Stable Stable Stable Stable Stable Stable wp_75 Out of Step Stable Stable Stable Stable Outof Step Stable Stable sp_open Stable Stable Stable Stable Stable Stable Stable Stable sp_brad Stable Stable Stable Stable Stable Stable Stable Stable sp_75 Outof Step Stable Stable Stable Stable Stable Stable Stable sv_open Stable Stable Stable Stable Stable Stable Stable Stable sv_brad Stable Stable Stable Stable Stable N/A Stable N/A sv_75 Stable Stable Stable Stable Stable N/A Stable N/A d_wp_open Stable Stable Stable Stable Stable Stable Stable Stable d_wp_brad Stable Stable Stable Stable Stable Stable Stable Stable d_wp_75 Out of Step Stable Stable Stable Stable Stable Stable Stable d_sp_open Stable Stable Stable Stable Stable Stable Stable Stable d_sp_brad Stable Stable Stable Stable Stable Stable Stable Stable d_sp_75 Stable Stable Stable Stable Stable N/A Stable N/A d_sv_open Stable Stable Stable Stable Stable Stable Stable Stable d_sv_brad Stable Stable Stable Stable Stable N/A Stable N/A d_sv_75 Stable Stable Stable Stable Stable N/A Stable N/A de_wp_open Stable Stable Stable Stable Stable Stable Stable Stable de_wp_brad Stable Stable Stable Stable Stable Stable Stable Stable de_wp_75 Outof Step Stable Stable Stable Stable Stable Stable Stable de_sp_open Stable Stable Stable Stable Stable Stable Stable Stable de_sp_brad Stable Stable Stable Stable Stable Stable Stable Stable de_sp_75 Stable Stable Stable Stable Stable N/A Stable N/A de_sv_open Stable Stable Stable Stable Stable N/A Stable N/A de_sv_brad Stable Stable Stable Stable Stable N/A Stable N/A de_sv_75 Stable Stable Stable Stable Stable N/A Stable N/A Table 11: Dynamic Stability Results Table This table highlights two surprising results. First is that the G6 contingency results in an unstable system without wind during a winter peak maximum import case, whereas the cases with wind result in a stable system. The second surprising result is that the G1 contingencies do not result in an out of step condition for the summer peak case with wind, but can run into out of step with no wind. The reason the G6 contingency is causing out of step for the winter peak without wind case is that it is tripping a full 41MW. The cases with wind displace the North Pole Combined Cycle generation. The dispatch level of NPCC is the primary reason for the improvement, not the addition of wind resources. The G1 differences are similar in that by increasing the wind output, the Healy generation will be reduced resulting in a smaller outage. The loss of HCCP is actually the largest contingency seen by GVEA in these cases. The loss of approximately 61MW of generation is larger than the BESS can provide. In islanded cases, the size difference can result in load shedding when spinning reserve is insufficient to cover the 15- 20MW of generation difference. GVEA may need to use its SILOS (Shed in lieu of spin) scheme to balance generation and load since the BESS’s 15 minute design is for 25MW not 46MW. Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 23 The results of the dynamic simulations study are that the cases with wind do not negatively impact the dynamic stability of the GVEA system. 6 Wind Model/Analysis The fundamental step in determining the required amount of regulation reserve lies with the analysis of the wind farm’s power fluctuation. All the power changes from the wind farm must be accounted for and balanced by the Control Area. The response of the Control Area to a change in power is complex and must be examined in different time periods. There are several time frames that must be considered when doing the review and are listed as follows in ascending order: Primary Control – The Primary Control responds to the change in frequency and occurs in the first few seconds. The frequency response from the utility system is provided by the governor controls on the generators. Certain types of loads also respond to changes in system frequency. However, while the governor control attempts to mitigate changes to frequency, it does not attempt to restore system frequency. Secondary Control - The Secondary Control maintains frequency in the minute to minute basis. In the GVEA system, this is achieved through the AGC system which controls GVEA generation and tie-line schedules. Tertiary Control – These are normally manual controls taken by operators that are in the time frame from 10 minutes to an hour. This includes starting units that are off-line, changing hydro schedules or changing interchanges with neighboring utilities. Once tertiary control is undertaken, it generally comes under the control of the utility’s AGC system. The Primary Control is the initial response to the change in frequency, but it does not return frequency back to normal. Secondary Control (GVEA AGC) will correct the frequency error to the extent possible with the resources under AGC control. So long as the unloaded capacity under control of AGC meets or exceeds the amount of energy produced by the wind farm, GVEA will be able to return the system frequency to 60 Hz. The variability of the AEP wind farm will increase the amount of short term frequency error on the system from historical levels due to the increased variability of the wind resource. However, this variability is not expected to exceed the capability of the GVEA thermal generation units expected to provide regulation. The main focus of this section is to characterize Delta Junction Wind Farm expected power output, its variability and to estimate the cumulative effect of adding Eva Creek Wind Farm in conjunction with the AEP project. 6.1 Wind Data EPS’s converted AEP’s MET tower data into a net power output time series to be used for power fluctuations. For this analysis, the MET data was used to get estimated 10-min average and 1 minute average time series to be used for ramp rate analysis. This analysis is not meant to evaluate the available long term energy from the wind farm. EPS also received wind data from GVEA for their proposed Eva Creek wind farm that had been converted to power data by GVEA’s sub-contractor. Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 24 6.1.1 Delta Junction Wind Farm The Delta Junction Wind Farm is located approximately 4.4 miles east of the city of Delta Junction, Alaska. EPS contracted with Wind Logics to do the work to convert the MET tower data received from AEP into 10 minute time series. A chronological listing of the data received can be found in Appendix A that has Windlogic’s Report, “Delta Junction, Alaska Power Ramp Analysis”. Wind Logic’s scope was to analyze the onsite meteorological data collected and to calculate the net energy production of the wind farm based on the proposed 16 tower locations, hub height, and wind turbine type. A second task was to look at wide scale ramps associated with large scale weather patterns. Wind Logic reviewed the MET data from 9/27/2007 through 5/13/2010 and found there were two extended periods in 2009 where there was no data. The data review also confirmed that no sensor errors which were prevalent in sensors of this time period were observed in the data. AEP’s Directwind 54/900 KW turbine was utilized to verify wind shear estimates from the MET tower. Wing Logic recommended that the time series data collected subsequent to January 2010 should not be used for further analysis due to concerns of turbine waking affecting the MET data. As a result of the data gaps in 2009 and Wind Logics recommendation, EPS selected the most recent contiguous year of data for further analysis. Wind Logic’s assessed wide area ramps associated with large scale weather patterns. They defined a ramp event of at least 30% of the capacity as well as the ramp event to be sustained for a period of no less than three hours. This is just a subset of what needs to be examined but the results are still useful. Based on Windlogic’s definition, 350 ramp events, equally distributed between up and down ramps were identified. Winter months had more ramp events than the summer months. The signature of the ramp events varied by season, with the summer events being shorter in duration than the winter months. The summer months are effected more by the heating and cooling cycle. However, the winter months have extreme storms caused more by pressure systems moving through the state. EPS contracted with Clarity Analytical to convert the net 10 minute average power output data received from Wind Logics to a net 1 minute average power output data that is representative of the wind farm. Clarity Analytical also used historical 1 minute generation data from an operating wind farm to perform this task. A chronicle of the work done can be found in Appendix A that has Clarity Analytical’s Memorandum, “Preparation of simulated time series of 1-minute power output of the proposed Delta Junction Wind Farm”. 6.1.2 Eva Creek Wind Farm GVEA is evaluating a project to construct a 24 MW Eva Creek Wind Farm near Healy. The 10- minute averages, completed by V-Bar, LLC and provided by GVEA had the simulated net capacity factor in percent. The simulation used GE-1.5xle/e (82.5-m rotor) with a hub height of 80m but there was no notation of wind farm size. The data range was from 6/24/2008 to 4/20/2010. GVEA stated by e-mail that the net power output to the grid used for our analysis should be 24 MW. The time intervals of the EVA Creek and AEP wind farms are not synchronized, however EPS assumed the data from the wind farms was reasonably synchronized when performing the “net” impacts analysis. Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 25 6.2 Wind Analysis EPS examined the expected ramp rate of each of the data sets looking forward from the starting point. For example, one set of tables and graphs show the number of occurrences of a particular MW change, both positive and negative, for different time periods from the starting point.The time periods are 10, 20, 30, 40, 50 and 60 minutes ahead. This is a rolling calculation and made for every point that data is allowed. 60 minutes would be the difference between the 1st and the 6th value. Next, the distribution of the samples was taken and then split up into ranges and the tables and graphs were made that showed the entire range. Another set of tables and graphs were made that had a smaller range and showed only a portion of the points. This was done to show more detail in the center part of the distribution range. Figure 9: Graph and Chart of Distribution of MW Changes in 2 MW bins. Another set of graphs were made to look at only the negative ramps. Negative ramps are normally the most disruptive to the system since nothing can be done to supply short fall other than the control area’s own generation or the implementation of a BESS. This series of scatter charts show the negative ramp or negative difference in net power output looking ahead some time period from a starting load. The time periods are 10, 20, 30, 40, 50 and 60 minutes ahead. This is a rolling calculation and made for every point that data is allowed. These graphs are very useful to determine the required regulating reserve based on risk of shortfall in the reserves. GVEA has indicated that the control area will be dispatched without wind in order to establish the preferred unit commitment. Once the unit commitment is established, wind will be added to the system to back of generation production, but units will not be de-committed due to the non-firm nature of an isolated wind farm in an islanded network. This procedure ensures that Secondary control, or AGC control will have the ability to provide regulation for all wind variability in the downward direction from the average production level. Power Range 10 minutes 20 minutes 30 minutes 40 minutes 50 minutes 60 minutes lower mw range [>=] upper mw range [<] Delta between 10 min avg [counts] Delta between 10 min avg [counts] Delta between 10 min avg [counts] Delta between 10 min avg [counts] Delta between 10 min avg [counts] Delta between 10 min avg [counts] -27 -25 0 0 0 0 0 1 -25 -23 0 2 4 9 8 13 -23 -21 0 2 10 16 22 23 -21 -19 1 9 16 20 37 33 -19 -17 6 12 28 44 43 70 -17 -15 6 26 46 76 89 116 -15 -13 21 65 95 105 169 199 -13 -11 41 114 188 245 268 293 -11 -9 103 217 315 417 493 532 -9 -7 245 487 629 729 841 887 -7 -5 697 1071 1234 1399 1502 1589 -5 -3 1799 2379 2667 2754 2810 2966 -3 -1 5658 5940 5867 5899 5862 5765 -1 1 35100 31851 30292 29077 28217 27559 1 3 5886 5880 5874 5881 5862 5749 3 5 1972 2510 2698 2830 2964 3021 5 7 702 1082 1310 1477 1495 1617 7 9 260 537 637 745 799 872 9 11 118 246 353 383 490 533 11 13 47 120 190 241 291 324 13 15 18 67 101 151 164 213 15 17 10 38 59 74 109 141 17 19 1 18 34 49 69 76 19 21 1 13 24 36 40 40 21 23 1 4 14 18 26 32 23 25 1 3 5 13 15 20 25 27 0 0 2 3 5 5 Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 26 Figure 10: Scatter of Negative Ramp Distribution from Starting Power Level 10 Minutes Ahead Similar scatter charts were made for the 1 minute data. The same methodology was used except that the time periods are 1, 2, 3, 4, 5 and 6 minutes ahead. These graphs were only done for Delta Junction Wind Farm as opposed to the net impact of the DJWF plus the Eva Creek facility. The minute based graphs plus the 10 minute ahead chart can be used to judge the secondary control requirements. The secondary controls maintain frequency in the minute to minute basis. Normally this would be via automatic generation control. Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 27 Figure 11: Scatter of Negative Ramp Distribution from Starting Power Level 1 Minutes Ahead Lastly, another graph is made to look at the volatility of the data as well as the distribution. The intent of this graph is to show the distribution of the hourly minimum, maximum and average of the 10 minute averages when all the data samples are sorted by the Average from largest to smallest. The hours really indicate the number of samples. All the charts are included in Appendix B and only selected charts are in the main body of the report. Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 28 Figure 12: Delta Junction distribution of Hourly 10 minute Min-Max-Avg 6.2.1 Delta Junction Wind Farm Examining the 10 min ahead scatter graph for DJWF indicates that the predominant maximum negative ramp is equal to the starting power level. When the starting power level reaches 8,000 KW, the knee, the predominant maximum negative ramp becomes less than the starting power level and ultimately reaches a level of -11,000 KW. This is of course excluding spurious points and is plotted in Figure 13. The knee is identified by the circle in the chart which is typical for all of them. Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 29 Figure 13: Scatter of Negative Ramp Distribution from Starting Power Level 10 Minutes Ahead with Negative Ramp Curve Examining the 20 min ahead scatter graph, the point where the knee starts occurs at a higher starting power level than the 10 minute ahead graph and ends ultimately at a lower power level of -15000 KW. This can be seen in figure 14. Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 30 Figure 14: Scatter of Negative Ramp Distribution from Starting Power Level 20 Minutes Ahead with Negative Ramp Curve Examining the 60 min ahead scatter graph, the point where the knee starts occurs at a very high starting power level far beyond where the 20 minute ahead graph knee is. The 60 minute ahead knee is at 20,000 KW starting power level and ends ultimately at a lower power level of - 21,500 KW. This can be seen in Figure 15. Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 31 Figure 15: Scatter of Negative Ramp Distribution from Starting Power Level 30 Minutes Ahead with Negative Ramp Curve In systems which de-commit units based on available wind energy, the amount of required on- line regulation is generally equivalent to either the 10 or 20-minute regulation amounts, depending upon the starting time of the off-line units. Based on the indications from GVEA concerning unit commitment and dispatch conditions, we believe the appropriate regulation level for the GVEA system is the level defined by the 10-minute curve above. It is however recognized that the GVEA system must be able to operate for all instances up to and including the 60-minute regulation requirements. 6.2.2 Eva Creek Wind Farm The simulation for Eva creek indicates more volatility in 10-minute average energy data than the AEP wind farm although the size and type of wind turbine generator are similar. This can be seen by comparing the min-max-average charts between the two wind farms. Figure 16 shows both DJWF and Eva Creek WF next to each other. Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 32 Figure 16: Delta Junction and Eva Creek Distribution of Hourly 10 minute Min-Max-Avg There are two notable differences between the graphs. When compared to the DJWF graph, the Eva Creek WF chart has a greater volatility for a greater range of the output level, even though the wind turbines for the two farms are very similar. For example, beginning at about 20 MW, the Eva Creek wind farm has instances where the output of the wind farm varies between full output and 0 MW. For the DJWF, the average output of the wind farm is at only 10 MW when the variance reaches 0 MW output. The other interesting thing to note is that Eva Creek has 770 hours where the output never drops below full output, compared to 270 hrs for DJWF. The Eva Creek full output duration seems to be abnormally high given the load factor of the plant. However, this graph was based on the wind analysis provided by GVEA. Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 33 6.2.3 Delta Junction and Eva Creek Wind Farm Combined The 10 min ahead scatter graph for the combined net impact to the GVEA system indicates that the predominant maximum negative ramp is equal to the starting power level. When the starting power level reaches 5,000 KW, the knee, the predominant maximum negative ramp becomes less than the starting power level and ultimately reaches a level of -14,000 KW. This excludes spurious points and is plotted in Figure 17. This is a decrease of 3,000 KW when compared to the volatility for DJWF by itself. Figure 17: Scatter of Negative Ramp Distribution from Starting Power Level 10 Minutes Ahead with Negative Ramp Curve Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 34 Examining the 20 min ahead scatter graph, the point where the knee starts occurs at a higher starting power level than the 10 minute ahead graph and ends ultimately at a lower power level of -19,000 KW. This can be seen in figure 18. This is a decrease of 4,000 KW when compared to DJWF by itself. Figure 18: Scatter of Negative Ramp Distribution from Starting Power Level 20 Minutes Ahead with Negative Ramp Curve Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 35 Examining the 60 min ahead scatter graph, the point where the knee starts occurs at a very high starting power level far beyond where the 20 minute ahead graph knee is. The 60 minute ahead knee is at 24,000 KW starting power level and ends ultimately at a lower power level of - 28,000 KW. This can be seen in figure 20. This is a decrease in the magnitude of the power drops of 9,000 KW when compared to just the DJWF by itself. The combination of DJWF and Eva Creek energy results in a significantly less volatile energy input to the GVEA system than if the volatility of both projects were simply added together. Since the exact time stamp of each project is unknown, the exact amount of diversity cannot be accurately determined, however, logically, diversity is expected and does appear present given the assumptions in this study. Figure 19: Scatter of Negative Ramp Distribution from Starting Power Level 60 Minutes Ahead with Negative Ramp Curve Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 36 7 Regulation As defined in the context of this study, regulation is the amount of generation available within the GVEA control area to counteract the variability of wind energy and provide constant frequency regulation and scheduled interchanges with neighboring utilities. Regulation is generally divided into two categories, upward and downward regulation. Upward regulation is the ability of the control area’s generation to increase its power output to make up for a decrease in wind farm output power. Downward regulation is the ability of a control area’s generation to decrease its power output in order to accept more wind power on its system. The DJWF has the ability to curtail or limit its upward generation volatility to match the capability of the GVEA system, therefore downward regulation in a control area can always be met by limiting the upward ramps of the wind farm and is not discussed in this report. A wind farm has no ability to limit or constrain its reduction in power resulting from a reduction in wind. The utility control area must have the capability to increase its power output on its own generation in order to make up for the reduction in power from the wind farm in order to maintain system frequency and interchange schedules. The amount of regulation required by the control area is a function of the acceptable risk to the control area, normal regulation reserves currently available, unit commitment philosophy and starting time of emergency generation units. In islanded systems with single wind farms or wind farms with limited diversity, generation units are usually committed based on the amount of wind that can be considered “firm” to the utility. Many times this results in a unit commitment without consideration of wind energy due to the difficulty in providing accurate forecasts of wind energy for isolated wind plants. In this operating condition, the utility will provide 100% of the regulation requirements for the full output of the wind energy, since wind energy will reduce the outputs of utility generation, but no units will be removed from the system. While the utility does not capture the full benefit of wind energy since it does not reduce the number of operating units, it does capture the incremental cost of energy production by reducing the energy produced by the operating units. As utilities gather more experience with wind characteristics, or they are able to provide unit commitments on less than 24-hour notice, more benefits of the wind energy can be realized as units are de-committed based on the amount of wind energy available to the system. In islanded systems, the ability of the control area to provide energy changes in its generation output through AGC is predicated on there being sufficient capacity under AGC control for the required energy changes. Therefore, although energy changes may only be analyzed in 1- minute increments, capacity requirements necessary to make the energy changes must be analyzed in the time increment required to make capacity decisions in the control area. In the case of GVEA, should units be de-committed based on expected energy from wind farms, the GVEA system must have sufficient capacity available to enable a change in energy schedule by its AGC system for the period of time it takes for additional capacity to be brought on-line. Assuming GVEA can start and load an off-line unit within 10 minutes results in the requirement for GVEA to utilize the 20-minute scatter graphs to determine the required regulation. Changes which occur of thirty minutes would be handled by a combination of the 20-minute reserve capacity and additional unit(s) which would be started to compensate for the ramp event. In practice, this is difficult to administer since the duration of a ramp event can only be judged in Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 37 retrospect, that is, when a ramp event starts it may be 10-15 minutes before a decision is made that the event justifies a unit start. GVEA has indicated that due to the 24-hour scheduling requirements for its hydro energy and inter-utility scheduling requirements, it will be committing its units without consideration of wind and using the wind energy to reduce its incremental cost of energy production. As such, the amount of regulation capacity required by the wind farm is somewhat immaterial as by virtue of unit commitment procedures, GVEA will always have sufficient regulation available. GVEA will however, not realize the 1:1 relationship between increased wind energy and the average cost of displaced thermal energy by virtue of the same procedures. 7.1 Ramp Rate Analysis The intent of this section is to look at the capacity requirement for secondary controls (AGC controlled generation). The analysis primarily uses the scatter charts showing the negative ramp or drop in power from a starting 10-minute average energy power level. The complete set of the graphs can be found in Appendix C for one minute increments of time looking ahead to determine the ability of GVEA’s thermal generation to maintain frequency control with the maximum expected ramp rates. The predominant characterization of the DJWF is that the expected negative change in power output will be greatest in the middle of the wind turbine generator power curve. This can be seen in the Figure 21 for the 1-minute series which shows the drop in power looking one minute ahead with the predominate curve shown as the solid line. Figure 20: Scatter of Negative Distribution from Starting Power Level 1 Minute Ahead Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 38 In general it shows that -2.1 MW/min can be expected in the middle of the power curve on a regular basis, with decreased ramp rates toward the higher or lower starting power levels. Occasional ramp rates approaching -4.0 MW/minute can be expected. Depending upon the time of these ramp rates with respect to the GVEA system loading and the number of units available to respond to this ramp, the GVEA generation should be able to handle these ramp rates. However, during times of extreme volatility, it can be expected that the wind farm will be curtailed by GVEA in order to provide a smoothing impact of the wind farm’s output. Although it is expected that diversity does exist between DJWF and Eva Creek, there will be times when the diversity is additive and as such, the expected ramp rate can be expected to double from the average of -2.1 to -4.2 MW/min with extreme rates approaching -8.0 MW/minute. It is assumed that during extreme volatility, the wind farms will be curtailed to mitigate excessive ramp rates. 7.2 Excess Energy – Curtailment There may be instances where curtailment of the wind farm may be necessary for reasons other than excessive ramp rates. For example, if the amount of energy provided by wind farms exceeds the downward regulating capability of the balancing area’s generation, including a downward regulating margin, the wind farm energy must be curtailed or thermal units de- committed from the system. This condition is often referred to as an “excess energy” condition. EPS did an analysis to see how much curtailment could be expected due to excess energy based on historical loads and estimated energy projections for the Delta Junction and the Eva Creek Wind Farms. Using the 10-minute wind farm 10-minute power averages discussed in Section 6 Wind Model/Analysis, EPS estimated the expected number of hours and energy curtailments for wind energy. When both DJWF and Eve Creek are on-line, the curtailment is not estimated per wind farm, but by total wind energy production of the two plants. The examined data includes 7,740 hours of date from 6/24/2008 through 5/17/2009. A full year’s worth of data (8,760 hours) could not be used due to missing information between the two respective sets of data. Load data was received from GVEA for the system net load as well as the Fort Knox load to allow a sensitivity analysis to be completed both with and without the Fort Knox load. Fort Knox is a very large load on the system that maintains a fairly constant load of approximately 30 MW. The 15 minute time increment load information was converted to 10 minute intervals using linear interpolation so that the wind energy data and GVEA load data could be used in the analysis. The system load was assumed to be net generation to grid, including all losses and loads. The minimum generation capability was obtained in an e-mail from GVEA that stated the minimum load following capability (LFCMIN) of the units shown in Table 7-1. The minimum downward regulating reserve of 3 MW was also obtained from GVEA to account for loss of load during minimum conditions. Table 12 indicates that the minimum generation level is 88 MW. Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 39 Table 12: Summer Valley Load-Nominal Import The minimum generation level sets the threshold of allowable wind energy on the system during off-peak conditions. That is, the total system load minus wind energy must meet or exceed 88 Mw or a curtailment conditions will exist. For purposes of this analysis, the wind farm generation is treated as a negative load. The wind farm’s output is subtracted directly from the net system load and the minimum generation level is used as the curtailment point for every historical 10 minute interval. The negative difference (adjusted load is below the threshold) is accumulated over the analysis time period and converted to MWH to provide an approximation of energy curtailment. The cases that EPS examined were: Case 1 – GVEA net load with Delta Junction Wind Farm Case 2 – GVEA net load with Delta Junction Wind Farm and Eva Creek Wind Farm Case 3 – GVEA net load with Fort Knox load removed and with Delta Junction Wind Farm Case 4 - GVEA net load with Fort Knox load removed and with Delta Junction Wind Farm and Eva Creek Wind Farm To account for changes in downward regulation requirements, EPS completed evaluated the sensitivity by evaluating the regulating reserve at 3, 6 and 9 MW. The results are shown in Table 13. Table 13 Case 1 had no curtailment required for normal conditions with the stated downward regulation reserve of 3 MW. Case 2 had a minimum expected curtailment of 388 MWH. This represents 59 hours were some curtailment was in effect. When the Fort Knox load is removed from the GVEA system, there is more curtailment with Case 3 (just Delta Junction Wind Farm) and MinimumGeneration Scenario Bradley Scheduled 8 MW RegulatingReserve Down 3 MW LFCMIN Healy Clean Coal 35 MW Healy 1 18 MW Chena5 24 MW Minimumallow generation 88 MW RegulatingReserve Down 3MW 6MW 9MW MWH MWH MWH Case 1 0 0 3 Case 2 388 590 878 Case 3 1110 1814 2859 Case 4 9901 12874 16481 Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 40 significant increased curtailment with Case 4 (Delta Junction Wind Farm plus Eva Creek Wind Farm). Case 3’s curtailment represents 189 hours. Case 4 represents 5, 091 hours of curtailment and the curtailed energy represents approximately 8.1% of the combined energy output of the two wind farms. Curtailment of either the Delta Junction Wind Farm or the Eva Creek wind farm or both due to excess non-firm generation does not appear to involve a significant amount of energy. 7.3 Voltage Regulation/Coordination The proposed GE turbines have excellent voltage control characteristics and are capable of regulating the voltage at the POI or providing constant power factor control depending on the control mode selected for the turbine. The wind farm will be interconnected with a relatively large SVC on a weak system. The turbine controls must ensure that control instability or hunting does not occur between the static voltage regulator of the WTG and the SVC at Jarvis Creek station. During conditions when the wind farm becomes islanded on the GVEA system without synchronous generation, there is a possibility of severe overvoltages or abnormal frequency operation until the turbine controls can detect and shut down the machines. The possibility can only be examined through the use of EMTP (electro-magnetic transient programs) and detailed models of the GVEA and wind turbine systems. There are multiple ground sources on the 138 kV system in the area of the proposed wind farm and additional mitigating measures may not be required, however, an EMTP study should be completed during the design stage of the project to ensure insulation coordination with the GVEA system in the area. Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 41 8 Wind Ramp Dynamic Simulations EPS performed a series of wind ramp simulations to determine GVEA’s capability to maintain system stability for a sustained wind ramp event. Based on 1-minute wind ramp analysis, the worst case 1-minute ramp was approximately 4 MW/min for the Delta Junction Wind farm. In order to show a worst case ramp condition, EPS assumed that this 4 MW/min ramp would continue until the wind plant reaches 0MW generation. The permissible ramp rate values used in the AGC model seemed suspect. The turbines at North Pole have been updated with ramp rates of 5%/min. The Healy coal units are modeled with ramp rates of 3%/min but only respond in emergency situations when the frequency dips below 59.90 Hz. In the end, these ramp rates are of little importance for these studies so long as the regulating generators ramp rates are larger than the wind ramp rates. The AGC model was not designed for use in simulating regulation, but rather it was designed to measure the system response during severe under frequency events. In addition, the AGC model for the southern utilities represents the obsolete AGC system no longer in service at AML&P and Chugach. Conclusions based solely on the results of this section of the report should be drawn with care. The actual sustained ramp rates used in the AGC model have been approximated throughout the interconnected system. Additionally, the unit pulse rate, pulse timing, emergency mode logic, and varying control modes have not been verified. The intent of these simulations was not to confirm the absolute regulation capability of GVEA, but rather to evaluate the stability of the GVEA system for sustained wind ramp events and its ability to maintain interchange schedules with neighboring utilities. Four power flow cases were selected for the wind ramp stability simulations. The cases were selected to satisfy the following: The power flow cases must have sufficient spin to cover the loss of all the wind generation, ie no unit starts or BESS required for regulating capacity. An islanded and an interconnected case from both winter and summer should be selected. The winter peak and summer valley cases in nominal import and islanded conditions were selected. Table 14 below shows the regulation amounts with respect to the amount of wind generation online. Table 11: Available Regulation for Wind Ramp Events Power Flow Case Available Capacity BESS Spin Regulation DWF Eva Total Wind Reg Difference d_wp_open 60.3 25 35.3 25.6 0 25.6 9.7 d_wp_brad 78.3 25 53.3 25.6 0 25.6 27.7 d_sv_open 92.5 25 67.5 25.6 0 25.6 41.9 d_sv_brad 62 25 37 25.6 0 25.6 11.4 Regulation = Capacity BESS Spin Generation With a maximum wind generation level of 25.6 MW, and a wind ramp rate of 4 MW/min, the wind can reach 0 MW in 6.4 minutes (384 seconds). EPS used a custom load model to simulate the loss of the wind in a linear ramp. The wind ramp simulations were run to 400 seconds to capture the loss of all wind generation. The system frequencies, GVEA generation, wind generation, and important GVEA line flows were recorded for plotting. Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 42 Two sets of plots are shown below in figures 21 and 22. Figure 21 shows the d_wp_open case. This case only has one unit (North Pole 2) on regulation with room to move. With the wind ramping down at 4 MW/min, North Pole 2 does not have a sufficient ramp rate (3MW/min) to keep up with the wind. As such, the GVEA frequency declines to 59.8 Hz. The top left plot shows the system frequencies on a scale from 59.1 to 60.1Hz. The top right set of traces shows the GVEA unit generation with the orange trace being the North Pole unit 2 generation. The bottom left set of traces shows the wind ramp from 25.6 MW to 0 MW. The bottom right trace shows the important line flows in the GVEA system with the red trace being the tie import. Figure 21: Wind Ramp in Islanded Condition with One Unit Regulating Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 43 Figure 22: Wind Ramp in Connected System with Coal Providing Regulation Figure 22 shows a simulation in which GVEA has generation capacity, and is connected to the rest of the system. It can be seen that the HCCP unit (red trace in top right) does not ramp at the beginning due to the large AGC deadband for this unit. However, at approximately half way through the simulation, HCCP begins to ramp due to AGC pulses. Due to the response of generation in the rest of the grid, the GVEA tie import (red trace bottom right) increases as GVEA cannot keep up with the wind ramp. The wind ramp dynamic simulation results have been included in Appendix C. Based on the results of these simulations, EPS concludes that there are no stability concerns caused by the ramping of wind generation. However, to avoid leaning on the other utilities in the Railbelt, the regulation should be spread across multiple units such that the cumulative ramp rate capability of the units is larger than the maximum expected wind ramp rate. Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 44 9 Interconnection The Delta Junction Wind Farm is located approximately 4.4 miles east of the city of Delta Junction, Alaska. Present there are two operational turbines at the site that are at the site but are not part of this interconnection. This interconnection is for 16 GE 1.6xle wind turbines for a total gross capacity of 26.6 MW. 9.1 Description The Delta Junction Wind Farm (DJWF) substation is a 138 KV station served radially from the GVEA Jarvis Creek switching station via an overhead transmission line. The DJWF substation is the interconnection point for the proposed wind farm project. The proposed DJWF Substation Single Line Diagram (SLD), DJWF-EL-0010 is shown in Appendix D. The point of demarcation at DJWF Substation is the lockable gang operated 138 KV disconnect switch (SW D1). DJWF Substation consists of a 138 KV circuit breaker to a grounded wye-delta, 138 – 34.5 KV, step- down transformer (T1) to a 34.5 KV three breaker switchgear. Between the 138 KV breaker (52-D) and SW D1 are a set of PTs and CTs to be used for the revenue metering. The revenue metering is to be per GVEA standards. There will be an additional 138 KV PT between the circuit breaker 52-D and transformer T1. This PT will be used for synch-check purposes across circuit breaker 52-D. The 34.5 KV switchgear will have three circuit breakers 52-A, 52-B and 52-C. The 34.5 KV switchgear consists of three 1200 amp breakers and rated 500 MVA. The breakers are 52-A, 52-B and 52-C that connect to the step-up transformer T1, collector circuit 1, and collector circuit 2 respectively. The switchgear will also have 3 PTs as well as a three phase station service transformer. The proposed DJWF Collector Circuits SLD, DJWF-EL-0011 is shown in Appendix D. The collection circuits 1 and 2 will have eight wind turbine generators each. Each collection circuit will have the eight turbines and a grounding transformer in a switchable normally open loop system with a radial tap to the collector circuit breaker. The step up transformers will be grounded wye-delta, 575Y/332-34500 V, 1750 KVA OA transformers as well as the grounding transformer. 9.2 Protection Requirements It is recommended that the relaying used for the 138 KV line protection between DJWF Substation and Jarvis Switching Station be SEL-311L relays for both primary and backup at both terminals. These relays should be configured to provide complete redundancy if one relay needs to be taken out of service for some reason. The communication path should be independent if possible to prevent a common point of failure. The primary protection for step-up transformer T1 will be differential protection 87T1 (SEL-387) that covers from the line side CT of breaker 52-D to bus side CT of 52-A. Also within the zone of protection are lightning arrestors (108KV MCOV), one un-fused potential transformers (80.5 KV, 700/1200:1) that is used for synch-check. The primary protection will trip a lock out relay 86T1 which will trip and block close circuit breakers 52-D and 52-A. The backup protection 351T (SEL-351) will use directional ground and phase overcurrent protection utilizing the line side CTs of breaker 52-D and the 138KV PT source. 351T will also provide 27, 59 and 81 O/U protection that must be coordinated with GVEA. 351T will also provide the synch-check function for circuit breaker CB 52-D. It is recommended that closing of circuit breaker 52-D only be Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 45 allowed for DB-DL and DB and LL. The backup protection will trip a lock out relay 86T2 which will trip and block close circuit breakers 52-D and 52-A. Circuit breaker 52-D will have breaker failure protection (SEL-501) and will trip lock out 86BF. 86BF will trip circuit breaker 52-D as well as initiate a direct transfer trip to Jarvis Switching Station via the 87L protection. It is recommend that and immediate re-trip function be used from the SEL-501 to prevent inadvertent operation of the scheme. The primary protection of the 34.5 KV switchgear will be differential protection 87B (SEL-587Z) that covers from the transformer side CT of 52-A and the feeder circuit side of 52-B and 52-C. The differential protection 87B trips lockout relay 86B1 which trips and blocks close of circuit breakers 52-A, 52-B and 52-C. The backup protection 351-A (SEL-351) will use directional ground and phase overcurrent protection utilizing the transformer side CTs of breaker 52-A and the 34.5 KV bus PT source. The backup protection 351-A also provides backup protection for collector circuits 1 and 2. Relay elements 27 and 59 will be utilized for alarming functions. The backup protection 351-A trips lockout relay 86B2 which trips and blocks close of circuit breakers 52-A, 52-B and 52-C. The primary protection of the collection circuits 1 and 2 will be 151-B (SEL-351) and 151-C (SEL-351). The 151-B and 151-C will use directional ground and phase overcurrent protection utilizing the bus side CTs of their respective breakers and the 34.5 KV bus PT source. There is a grounding transformer on each of the collection circuits to maintain a zero sequence source with the collection circuit is tripped for a ground fault. Each grounding transformer neutral will have a CT input into the SEL-351 relay for ground detection. The bus also uses the PTs for a back up zero sequence source for ground detection. There are some possible modification to this scheme and can change during the detail design. Each of the wind turbine generator step-up transformer will have their own high side fuse protection and low voltage protection. The exact configuration may change during the design portion. The GVEA system between Jarvis Creek and Carney Substation will require additional protection upgrades. Due to the limited short circuit current available from the wind turbines, line current differential relaying will be required on each line section between the Jarvis Creek station and Carney Substation. In addition to the line current differential relaying and its associated communication requirements, a communication system and interface to GVEA’s existing BESS control system will be required at the DJWF and each possible point of islanding between Jarvis Creek and Carney substations. 9.3 Wind Turbine Options The following options are recommended to be purchased with the DJWF turbines. Intertial Control GE’s Windinertia option provides simulated inertia from the wind turbines to the power grid during off-frequency performance. This option is required for Railbelt wind turbines to prevent the degradation of frequency due to the loss of inertia in the Railbelt in the event that wind turbines displace higher inertia gas turbines or hydro units on the system. Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 46 WindControl System GE’s Windcontrol system controls several features of the wind farm as a whole, including the ability to provide power frequency droop control, real and reactive power control and real power ramp rate control. Frequency droop control may or may not be beneficial depending upon the operating parameters of the plant and its commercial terms. However, real power control will be required to implement full or partial curtailments and ramp rate control will be required in order to transition from a curtailment period to a non-curtailment period. Ramp rate control may also be required to minimize regulation costs, however the requirement for ramp rate control during normal operation is embedded in the commercial issues associated with wind farm regulation. WindRideThru WindRideThru provides the GE turbines with both low and high voltage ride-through ability. This requirement will not be the typical GE LVRT requirement as this application is not intended for a FERC regulated transmission system. This option will be required in order for the turbines to continue operation through Railbelt grid or GVEA subtransmission or distribution disturbances. Voltage control The FIWf will be operating in a voltage control mode as opposed to a power factor or constant MVAr mode. The FIWF will be required to maintain a constant voltage at its point of interconnection without respect to WTG output. This requirement may force the installation of WindFree as a WTG option. 9.4 Interface The DJWF will have equipment necessary to initiate, control the level of, and remove curtailments. The implementation of the Curtailment Control Interface will allow GVEA’s System Operator to initiate the curtailment, vary the level of curtailment, and remove the curtailment remotely. The Curtailment Control Interface shall include a demarcation cabinet for analog and digital signals and other devices necessary to connect to the GVEA RTU. The Curtailment control interface should provide provisions for an analog and a status feedback indicating the level of curtailment as well as the curtailment still being in effect. When a curtailment lower control signal is received by DJWF’s control interface, the corresponding action shall be initiated without delay. DJWF shall be capable of and allow Company to curtail output at a rate of up to 2 MW/min. 9.5 Voltage Regulation The DJWF shall be capable of voltage regulation at the interconnection. DJWF’s equipment shall deliver reactive capacity at the point of the interconnection from 95% leading to 90% lagging while delivery full power to GVEA. 9.6 Curtailment Procedures During periods of excess energy or other times when the wind farm requires curtailment, the wind farm must have a single point controller, capable of establishing an operating point selectable by the utility system or its designee. When curtailment is instituted, the facility should provide a ramped response to the point of curtailment. When the curtailment is released, the Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 47 controller should provide a ramped response from the curtailed condition to the full generating capability of the plant. The ramp rate control should be selectable to match system capability at the time of the curtailment. Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 51 Appendix D – EPS Wind Analysis Alaska Environmental Power & GVEA Delta Junction Wind Plant Impact Study September 7, 2010 Page 1 Appendix zz – Wind Charts and Tables Scatter Graphs – 10 Minute Distribution of Negative Ramps Delta Junction Wind Farm ………………………………………………………………..zz-2 Eva Creek Wind Farm ………………………………………………………………….…zz-5 Delta Junction and Eva Creek Wind Farms Combined ……………………………….zz-9 10 Minute Distribution of Negative Ramps MW changes Series of Graphs Delta Junction Wind Farm ……………………………………………………………….zz-12 Eva Creek Wind Farm …………………………………………………………………….zz-13 Delta Junction and Eva Creek Wind Farms Combined ……………………………….zz-14 10 Minute Hourly Min-Max-Avg Distribution Graphs Delta Junction Wind Farm ……………………………………………………………….zz-16 Eva Creek Wind Farm …………………………………………………………………….zz-17 Delta Junction and Eva Creek Wind Farms Combined ……………………………….zz-18 1 Minute Delta Junction Wind Farm Graphs Scatter Charts Distribution of Negative Ramps………………………………………..zz-16 Distribution of MW changes Series of Graphs…………………………………………zz-19 Alaska Environmental Power & GVEA Delta Junction Wind Plant Impact Study September 7, 2010 Page 2 Scatter Graphs – 10 Minute Distribution of Negative Ramps Ramp rates from wind farms can occur in both the upward and downward direction corresponding to increases and decreases, respectively, in wind speed. Increases in wind energy are generally ignored when defining regulation requirements for modern wind installations due to the control features available with new turbine systems. Modern control systems have the capability of limiting increases in wind farm energy by controlling the blades of the wind turbine to limits which are acceptable to the utility system. Therefore, should the utility experience increases in wind energy beyond the capability of its generation; it can control the wand farm’s ramp rate to an amount allowable by the utility system. This type of control option is recommended for the DJWF and as such, upward ramp rates are not considered a problem in the GVEA system. Negative ramps are normally the most disruptive to the system since they cannot be controlled by the wind farm itself, but must depend on either utility generation, battery storage or other methods to make up for the loss of wind energy. In the GVEA system, the supply short fall created by the wind farm’s negative ramp is expected to be compensated for by GVEA’s thermal generation. The amount of generation required to make up for expected negative ramp rates is dependent upon the time frame under consideration. For systems with assured capacity, the negative ramp rates can be estimated using minute-minute changes in wind farm output, generally using less than 10 minutes as the interval to define the required energy changes. However, in isolated power systems such as GVEA’s, regulating capacity is limited to the number of units operating in the system. This means that regulation is limited to the capacity defined by the unit commitment schedule of the utility. If the unit commitment schedule uses capacity requirements from 10-minute scatter charts, it means that the utility is only planning on having on-line capacity required to meet the maximum change in wind farm energy that occurs over only a 10-minute period. We would not recommend capacity planning periods of less than 20 minutes in islanded systems due to delay times for starting units, the uncertainty of ramp conditions and the delay between the initiation of a ramp event and the decision to start additional units. The series of scatter charts included in this appendix show the negative ramp or negative difference in net power output looking ahead some time period from a starting load. The time periods are 10, 20, 30, 40, 50 and 60 minutes ahead. GVEA has indicated that thermal units will not be de-committed based on available wind energy from a single wind farm. Unit commitment decisions will be made based on no wind energy input. Following the unit commitment decision, the wind energy will be allowed to reduce the thermal units to their minimum values, assuring GVEA that they have on-line capacity to account for the total loss of wind energy in the GVEA system. Delta Junction Wind Farm The data used for Delta Junction Wind farm is from Wind Logics (WL). The WL data consists of approximately two and one half years of data from 9/27/2007 through 5/13/2010 that developed from the MET tower data from the site. The most recent contiguous year was used which is from 5/17/2008 through 5/17/2009. There are a total of 6 graphs. Alaska Environmental Power & GVEA Delta Junction Wind Plant Impact Study September 7, 2010 Page 3 Figure 1 Scatter of Negative Ramp Distribution from Starting Power Level 10 Minutes Ahead Figure 2 Scatter of Negative Ramp Distribution from Starting Power Level 20 Minutes Ahead Alaska Environmental Power & GVEA Delta Junction Wind Plant Impact Study September 7, 2010 Page 4 Figure 3 Scatter of Negative Ramp Distribution from Starting Power Level 30 Minutes Ahead Figure 4 Scatter of Negative Ramp Distribution from Starting Power Level 40 Minutes Ahead Alaska Environmental Power & GVEA Delta Junction Wind Plant Impact Study September 7, 2010 Page 5 Figure 5 Scatter of Negative Ramp Distribution from Starting Power Level 50 Minutes Ahead Alaska Environmental Power & GVEA Delta Junction Wind Plant Impact Study September 7, 2010 Page 6 Figure 6 Scatter of Negative Ramp Distribution from Starting Power Level 60 Minutes Ahead Eva Creek Wind Farm The data used in the calculation is from GVEA and the estimation made by V-Bar, LLC. The GVEA data is from 6/24/2008 to 4/20/2010. At contiguous year was used which is from 6/24/2008 through 6/23/2009 which is close to the data range for Delta Junction Wind Farm. Figure 7 Scatter of Negative Ramp Distribution from Starting Power Level 10 Minutes Ahead Alaska Environmental Power & GVEA Delta Junction Wind Plant Impact Study September 7, 2010 Page 7 Figure 8 Scatter of Negative Ramp Distribution from Starting Power Level 20 Minutes Ahead Figure 9 Scatter of Negative Ramp Distribution from Starting Power Level 30 Minutes Ahead Alaska Environmental Power & GVEA Delta Junction Wind Plant Impact Study September 7, 2010 Page 8 Figure 10 Scatter of Negative Ramp Distribution from Starting Power Level 40 Minutes Ahead Figure 11 Scatter of Negative Ramp Distribution from Starting Power Level 50 Minutes Ahead Alaska Environmental Power & GVEA Delta Junction Wind Plant Impact Study September 7, 2010 Page 9 Figure 12 Scatter of Negative Ramp Distribution from Starting Power Level 60 Minutes Ahead Delta Junction and Eva Creek Wind Farms Combined The following graphs are of Delta Junction WF and Eva Creek WF combined. The data used in the calculation is from the same sources listed in the individual graphs. Because the time dates for the contiguous years are slightly different, the data show is slightly less than a year. The data is from 6/24/2008 through 5/17/2009. Alaska Environmental Power & GVEA Delta Junction Wind Plant Impact Study September 7, 2010 Page 10 Figure 13 Scatter of Negative Ramp Distribution from Starting Power Level 10 Minutes Ahead Figure 14 Scatter of Negative Ramp Distribution from Starting Power Level 20 Minutes Ahead Alaska Environmental Power & GVEA Delta Junction Wind Plant Impact Study September 7, 2010 Page 11 Figure 15 Scatter of Negative Ramp Distribution from Starting Power Level 30 Minutes Ahead Figure 16 Scatter of Negative Ramp Distribution from Starting Power Level 40 Minutes Ahead Alaska Environmental Power & GVEA Delta Junction Wind Plant Impact Study September 7, 2010 Page 12 Figure 17 Scatter of Negative Ramp Distribution from Starting Power Level 50 Minutes Ahead Figure 18 Scatter of Negative Ramp Distribution from Starting Power Level 60 Minutes Ahead Alaska Environmental Power & GVEA Delta Junction Wind Plant Impact Study September 7, 2010 Page 13 10 Minute Distribution of Negative Ramps MW changes Series of Graphs The intent of this series of graphs and charts are to show the number of occurrences of a particular MW change, both positive and negative, for different time periods from the starting point.The time periods are 10, 20, 30, 40, 50 and 60 minutes ahead. This is a rolling calculation and made for every point that data is allowed. 60 minutes would be the difference between the 1st and the 6th value. Next, the distribution of the samples was taken and then split up into MW ranges or bins and the table and graph are shown. The first graph and table has a larger MW range or bin that covers the total count. The second graph and table has a smaller MW range or bin that shows more details of the distribution centered on zero. Each wind farm sections are covered separately. Delta Junction Wind Farm Delta Junction Wind Farm full count graph and table use 2 MW bins for the full count and is shown in figure 19. Figure 19 is quite useful for the larger deviations but is congested in the +/- 5 MW range. The detailed partial count graph and table use 0.5 MW bins shown in Figure 20 and covers +/- 5.25 MW. An example of the graph and table shown in Figure 19, using the 10 minute ahead, there was one occurrence that happen in the range of -19.0 MW to -21.0 MW. The number of occurrences between -15.0 MW to -21.0 MW would be the sum of the individual ranges (6 + 6 + 1) or 13. Alaska Environmental Power & GVEA Delta Junction Wind Plant Impact Study September 7, 2010 Page 14 Figure 19 Graph and Chart of Distribution of MW Changes in 2 MW bins. Figure 20 Graph and Chart of Distribution of MW Changes in 0.5 MW bins. Power Range 10 minutes 20 minutes 30 minutes 40 minutes 50 minutes 60 minutes lower mw range [>=] upper mw range [<] Delta between 10 min avg [counts] Delta between 10 min avg [counts] Delta between 10 min avg [counts] Delta between 10 min avg [counts] Delta between 10 min avg [counts] Delta between 10 min avg [counts] -27 -25 0 0 0 0 0 1 -25 -23 0 2 4 9 8 13 -23 -21 0 2 10 16 22 23 -21 -19 1 9 16 20 37 33 -19 -17 6 12 28 44 43 70 -17 -15 6 26 46 76 89 116 -15 -13 21 65 95 105 169 199 -13 -11 41 114 188 245 268 293 -11 -9 103 217 315 417 493 532 -9 -7 245 487 629 729 841 887 -7 -5 697 1071 1234 1399 1502 1589 -5 -3 1799 2379 2667 2754 2810 2966 -3 -1 5658 5940 5867 5899 5862 5765 -1 1 35100 31851 30292 29077 28217 27559 1 3 5886 5880 5874 5881 5862 5749 3 5 1972 2510 2698 2830 2964 3021 5 7 702 1082 1310 1477 1495 1617 7 9 260 537 637 745 799 872 9 11 118 246 353 383 490 533 11 13 47 120 190 241 291 324 13 15 18 67 101 151 164 213 15 17 10 38 59 74 109 141 17 19 1 18 34 49 69 76 19 21 1 13 24 36 40 40 21 23 1 4 14 18 26 32 23 25 1 3 5 13 15 20 25 27 0 0 2 3 5 5 Power Range 10 minutes 20 minutes 30 minutes 40 minutes 50 minutes 60 minutes lower mw range [>=] upper mw range [<] Delta between 10 min avg [counts] Delta between 10 min avg [counts] Delta between 10 min avg [counts] Delta between 10 min avg [counts] Delta between 10 min avg [counts] Delta between 10 min avg [counts] -5.25 -4.75 270 404 457 480 515 520 -4.75 -4.25 370 460 588 548 565 675 -4.25 -3.75 404 605 643 684 680 733 -3.75 -3.25 574 710 826 834 876 869 -3.25 -2.75 771 920 911 1035 988 993 -2.75 -2.25 1001 1220 1193 1198 1164 1131 -2.25 -1.75 1373 1336 1398 1463 1424 1431 -1.75 -1.25 1937 1913 1803 1754 1800 1736 -1.25 -0.75 2771 2785 2780 2712 2660 2680 -0.75 -0.25 4889 4196 3967 3678 3666 3516 -0.25 0.25 22282 20422 19281 18658 18053 17472 0.25 0.75 4681 4035 3825 3593 3442 3428 0.75 1.25 2710 2673 2636 2538 2466 2526 1.25 1.75 1825 1742 1711 1720 1769 1662 1.75 2.25 1334 1330 1352 1352 1396 1360 2.25 2.75 1005 1079 1103 1134 1063 1130 2.75 3.25 800 881 929 969 1011 952 3.25 3.75 599 768 827 829 887 859 3.75 4.25 471 600 574 661 712 745 4.25 4.75 344 473 573 597 600 623 4.75 5.25 264 372 456 471 483 524 Alaska Environmental Power & GVEA Delta Junction Wind Plant Impact Study September 7, 2010 Page 15 Eva Creek Wind Farm The data used in the calculation is from GVEA and the estimation made by V-Bar, LLC. The GVEA data is from 6/24/2008 to 4/20/2010. At contiguous year was used which is from 6/24/2008 through 6/23/2009 which is approximately the same as Delta Junction Wind Farm. Eva Creek Wind Farm full count graph and table use 2 MW bins for the full count and is shown in figure 19. Figure 19 is quite useful for the larger deviations but is congested in the +/- 5 MW range. The detailed partial count graph and table use 0.5 MW bins shown in Figure 20 and covers +/- 5.25 MW. Figure 21 Graph and Chart of Distribution of MW Changes in 2 MW bins. Figure 22 Graph and Chart of Distribution of MW Changes in 0.5 MW bins. Power Range 10 minutes 20 minutes 30 minutes 40 minutes 50 minutes 60 minutes lower mw range [>=] upper mw range [<] Delta between 10 min avg [counts] Delta between 10 min avg [counts] Delta between 10 min avg [counts] Delta between 10 min avg [counts] Delta between 10 min avg [counts] Delta between 10 min avg [counts] -27 -25 0 0 0 0 0 0 -25 -23 14 25 36 48 64 74 -23 -21 1 7 14 38 40 54 -21 -19 11 13 32 42 57 71 -19 -17 5 26 54 68 90 112 -17 -15 14 40 76 113 143 175 -15 -13 17 84 126 159 206 256 -13 -11 64 151 226 279 334 376 -11 -9 107 254 350 457 519 585 -9 -7 239 486 620 735 828 834 -7 -5 574 900 1158 1282 1356 1455 -5 -3 1486 2027 2196 2395 2459 2616 -3 -1 5039 5355 5437 5319 5431 5208 -1 1 37545 33923 31994 30651 29576 28969 1 3 4920 5272 5270 5330 5325 5144 3 5 1422 1978 2242 2390 2436 2519 5 7 585 907 1157 1278 1411 1467 7 9 276 523 618 771 815 940 9 11 110 236 390 433 508 586 11 13 54 138 205 283 349 352 13 15 25 76 147 195 221 286 15 17 11 54 82 110 145 183 17 19 10 31 48 72 101 118 19 21 11 19 24 34 47 67 21 23 4 7 18 19 26 34 23 25 15 26 37 55 68 73 25 27 0 0 0 0 0 0 Power Range 10 minutes 20 minutes 30 minutes 40 minutes 50 minutes 60 minutes lower mw range [>=] upper mw range [<] Delta between 10 min avg [counts] Delta between 10 min avg [counts] Delta between 10 min avg [counts] Delta between 10 min avg [counts] Delta between 10 min avg [counts] Delta between 10 min avg [counts] -5.25 -4.75 200 342 388 410 461 459 -4.75 -4.25 263 390 456 499 535 569 -4.25 -3.75 361 456 534 578 564 660 -3.75 -3.25 432 596 622 698 695 732 -3.25 -2.75 666 765 781 815 865 834 -2.75 -2.25 826 939 1048 1039 1101 1070 -2.25 -1.75 1125 1296 1309 1287 1247 1276 -1.75 -1.25 1634 1683 1657 1618 1659 1544 -1.25 -0.75 2467 2322 2151 2083 2110 1950 -0.75 -0.25 4513 4039 3828 3750 3445 3493 -0.25 0.25 26023 23424 22209 21284 20555 20042 0.25 0.75 4346 4006 3625 3443 3385 3292 0.75 1.25 2346 2141 2167 2000 1938 1962 1.25 1.75 1646 1758 1581 1593 1582 1588 1.75 2.25 1101 1236 1333 1320 1450 1263 2.25 2.75 810 907 995 1073 1002 1017 2.75 3.25 587 742 778 785 773 808 3.25 3.75 432 588 700 704 696 744 3.75 4.25 369 488 547 622 574 602 4.25 4.75 243 402 419 463 557 538 4.75 5.25 225 327 374 425 490 459 Alaska Environmental Power & GVEA Delta Junction Wind Plant Impact Study September 7, 2010 Page 16 Delta Junction and Eva Creek Wind Farms Combined The following graphs and tables are for Delta Junction WF and Eva Creek WF combined. The data used in the calculation is from the same sources listed in the individual table and graphs. Because the time dates for the contiguous years are slightly different, the data show is slightly less than a year. The data is from 6/24/2008 through 5/17/2009. The combined wind farm’s full count graph and table use 4 MW bins and is shown in figure 23. Figure 23 is quite useful for the larger deviations but is congested in the +/- 5 MW range. The detailed partial count graph and table use 0.5 MW bins shown in Figure 24 and covers +/- 5.25 MW. Figure 23 Graph and Chart of Distribution of MW Changes in 4 MW bins. Figure 24 Graph and Chart of Distribution of MW Changes in 0.5 MW bins. Power Range 10 minutes 20 minutes 30 minutes 40 minutes 50 minutes 60 minutes lower mw range [>=] upper mw range [<] Delta between 10 min avg [counts] Delta between 10 min avg [counts] Delta between 10 min avg [counts] Delta between 10 min avg [counts] Delta between 10 min avg [counts] Delta between 10 min avg [counts] -46 -42 0 1 2 1 1 0 -42 -38 0 0 1 1 3 5 -38 -34 0 1 0 4 3 5 -34 -30 0 1 1 6 9 11 -30 -26 1 2 9 12 24 37 -26 -22 13 32 55 87 116 133 -22 -18 17 56 122 159 206 269 -18 -14 72 184 287 394 465 545 -14 -10 238 512 755 915 1063 1184 -10 -6 994 1802 2135 2458 2662 2795 -6 -2 5801 6841 7217 7342 7334 7401 -2 2 32800 28392 26107 24443 23475 22443 2 6 5927 6782 6985 7263 7175 7253 6 10 1047 1800 2275 2482 2727 2889 10 14 215 514 783 960 1084 1218 14 18 53 193 294 418 504 595 18 22 22 65 114 153 204 250 22 26 19 36 63 99 126 130 26 30 3 4 8 15 24 33 30 34 0 3 6 2 7 13 34 38 0 0 1 4 6 8 38 42 0 0 0 1 0 0 Power Range 10 minutes 20 minutes 30 minutes 40 minutes 50 minutes 60 minutes lower mw range [>=] upper mw range [<] Delta between 10 min avg [counts] Delta between 10 min avg [counts] Delta between 10 min avg [counts] Delta between 10 min avg [counts] Delta between 10 min avg [counts] Delta between 10 min avg [counts] -5.25 -4.75 379 560 630 598 666 700 -4.75 -4.25 527 615 747 744 759 812 -4.25 -3.75 638 752 852 923 879 928 -3.75 -3.25 715 949 961 981 1000 975 -3.25 -2.75 1030 1107 1111 1173 1131 1126 -2.75 -2.25 1305 1397 1396 1379 1364 1285 -2.25 -1.75 1674 1623 1591 1566 1490 1485 -1.75 -1.25 2231 2079 2050 1834 1863 1769 -1.25 -0.75 3051 2752 2510 2425 2244 2256 -0.75 -0.25 4681 3816 3441 3126 3073 2893 -0.25 0.25 11538 9674 8726 8224 7760 7326 0.25 0.75 4496 3739 3338 3077 2937 2782 0.75 1.25 2862 2652 2465 2333 2181 2144 1.25 1.75 2166 2025 1901 1796 1798 1751 1.75 2.25 1680 1559 1594 1551 1455 1420 2.25 2.75 1206 1281 1299 1327 1247 1266 2.75 3.25 1031 1090 1122 1154 1101 1011 3.25 3.75 807 919 990 968 953 988 3.75 4.25 676 809 765 872 902 878 4.25 4.75 487 624 679 713 787 800 4.75 5.25 394 583 578 600 673 714 Alaska Environmental Power & GVEA Delta Junction Wind Plant Impact Study September 7, 2010 Page 17 10 Minute Hourly Min-Max-Avg Distribution Graphs The intent of this graph is to show the distribution of the hourly minimum, maximum and average of the 10 minute averages when all the data samples are sorted by the Average from largest to smallest. The hours on the graph really indicate the number of samples. The graphs show the volatility of the wind farm across the spectrum of average power output. Delta Junction Wind Farm The data used for Delta Junction Wind farm is from Wind Logics (WL). The WL data consists of approximately two and one half years of data from 9/27/2007 through 5/13/2010 that developed from the MET tower data from the site. Previously it was mentioned that the most recent contiguous year was used which is from 5/17/2008 through 5/17/2009. Figure 25 Delta Junction distribution of Hourly 10 minute Min-Max-Avg Alaska Environmental Power & GVEA Delta Junction Wind Plant Impact Study September 7, 2010 Page 18 Eva Creek Wind Farm The data used in the calculation is from GVEA and the estimation made by V-Bar, LLC. The GVEA data is from 6/24/2008 to 4/20/2010. At contiguous year was used which is from 6/24/2008 through 6/23/2009 which is close to the data range for Delta Junction Wind Farm. It is interesting to see that there are almost a 771 hours where there was no minimum less than full output. Another point to pull away from the graph is volatility. Figure 26 Eva Creek distribution of Hourly 10 minute Min-Max-Avg Alaska Environmental Power & GVEA Delta Junction Wind Plant Impact Study September 7, 2010 Page 19 Delta Junction and Eva Creek Wind Farms Combined The following graphs are of Delta Junction WF and Eva Creek WF combined. The data used in the calculation is from the same sources listed in the individual graphs. Because the time dates for the contiguous years are slightly different, the data show is slightly less than a year. The data is from 6/24/2008 through 5/17/2009 Figure 26 Combined WF Distribution of Hourly 10 minute Min-Max-Avg 1 Minute Delta Junction Wind Farm Graphs The data used for Delta Junction Wind farm is from Clarity Analytical (CA). The WL data consists of approximately two and one half years of data from 9/27/2007 through 5/13/2010 that developed from the MET tower data from the site. Previously it was mentioned that the most recent contiguous year was used which is from 5/17/2008 through 5/17/2009. There are a total of 6 graphs. Negative ramps are normally the most disruptive to the system since nothing can be done to supply short fall other than the control area’s own generation or the implementation of a BESS. This series of scatter charts show the negative ramp or negative difference in net power output looking ahead some time period from a starting load. The time periods are 10, 20, 30, 40, 50 and 60 minutes ahead. This is a rolling calculation and made for every point that data is allowed. There are a total of 6 graphs. The reason for the numerous time periods ahead shows the amount of reserve required depending on the response time. Alaska Environmental Power & GVEA Delta Junction Wind Plant Impact Study September 7, 2010 Page 20 Scatter Charts Distribution of Negative Ramps Figure 27 Scatter of Negative Ramp Distribution from Starting Power Level 1 Minutes Ahead Figure 28 Scatter of Negative Ramp Distribution from Starting Power Level 1 Minutes Ahead Alaska Environmental Power & GVEA Delta Junction Wind Plant Impact Study September 7, 2010 Page 21 Figure 29 Scatter of Negative Ramp Distribution from Starting Power Level 3 Minutes Ahead Figure 30 Scatter of Negative Ramp Distribution from Starting Power Level 4 Minutes Ahead Alaska Environmental Power & GVEA Delta Junction Wind Plant Impact Study September 7, 2010 Page 22 Figure 31 Scatter of Negative Ramp Distribution from Starting Power Level 5 Minutes Ahead Figure 32 Scatter of Negative Ramp Distribution from Starting Power Level 6 Minutes Ahead Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 52 Appendix E – Clarity Analytical Wind Analysis Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 53 Appendix F – Wind Logic - Wind Analysis Electrical Power Systems Inc. Delta Junction, Alaska Power Ramp Analysis November 24, 2010 Confidential Business Information Client Name Electrical Power Systems Inc. Contact Michael Bradley Document ID 1110-835.001-01 Issue B Version Final Date November 24, 2010 Prepared by: Ryan Butterfield Technical Consultant WindLogics 1021 Bandana Boulevard East, Suite 111 Saint Paul, Minnesota 55108 USA 651.556.4200 support@windlogics.com Delta Junction, Alaska C O N F I D E N T I A L Power Ramp Analysis - Page i Final – 1110-835.001-01 DISCLAIMER The following document was prepared by WindLogics Inc. for the use of Electrical Power Systems Inc., 3305 Arctic Boulevard, Suite 201, Anchorage, Alaska, its successors and/or assigns. The report is expressly and exclusively for the sole use and benefit of Electrical Power Systems Inc. and is not for the use or benefit of, nor may be relied upon by, any other person or entity without the advanced written consent of WindLogics Inc. This document is a trade secret and its confidentiality is strictly maintained. Disclaimer: WindLogics Inc. has prepared this report based on available third party historical weather information and use of our predictive software and analysis methods. We cannot guarantee the accuracy of historical weather data. Historical weather information also does not necessarily allow accurate prediction of future weather patterns. WE ARE THEREFORE PROVIDING THIS REPORT TO YOU WITHOUT ANY EXPRESS OR IMPLIED WARRANTY OF FITNESS FOR A PARTICULAR PURPOSE OR MERCHANTABILITY. Our sole responsibility is the preparation and delivery of this report. By accepting this report, you agree that our liability in any situation is limited to the amount paid for our report. In no event will we be liable for any special or consequential damages arising from use, or misuse, of our report or information in it. REVISION HISTORY Issue Issue Date Revision Summary A 11/05/2010 Original Issue B 11/24/2010 Revisions to Meteorological Discussion Delta Junction, Alaska C O N F I D E N T I A L Power Ramp Analysis - Page ii Final – 1110-835.001-01 TABLE OF CONTENTS EXECUTIVE SUMMARY ......................................................................................................................... iii 1. INTRODUCTION .................................................................................................................................. 1 1.1 Overview of Objective and Scope ................................................................................................... 1 1.2 Site Location and Description ......................................................................................................... 1 1.3 Customer-Supplied Data................................................................................................................. 2 2. THEORETICAL WIND FARM PRODUCTION CALCULATION METHODOLOGY ............................. 3 2.1 Meteorological Tower Data Processing Methods ........................................................................... 3 2.1.1 Met Tower Sensor Configuration ............................................................................................................. 4 2.1.2 Met Tower Data Processing Overview .................................................................................................... 5 2.1.3 Data Quality and Data Recovery Results ................................................................................................ 6 2.2 Production Data Processing Methods ............................................................................................. 7 2.2.1 Power Production Data Inventory ............................................................................................................ 7 2.2.2 Nacelle-Mounted Anemometer Filtering .................................................................................................. 8 2.3 Hub-Height Wind Speed Extrapolation ........................................................................................... 8 2.4 Full-Farm Weighting........................................................................................................................ 9 2.5 Energy Production Calculation Methods ....................................................................................... 10 2.6 Project-Specific Loss Analysis Methods ....................................................................................... 11 2.6.1 Wake Losses ......................................................................................................................................... 11 2.6.2 Electrical Efficiency Losses ................................................................................................................... 12 2.6.3 Wind Sector Management Control Losses ............................................................................................ 12 3. POWER RAMP CHARACTERIZATION METHODOLOGY ............................................................... 12 3.1 Ramp Characterization ................................................................................................................. 12 3.2 Ramp Analysis Methods ............................................................................................................... 13 4. RESULTS ........................................................................................................................................... 14 4.1 Meteorological Overview............................................................................................................... 14 4.2 Ramp Occurrences ....................................................................................................................... 15 4.2.1 Monthly Power Ramp Distribution ......................................................................................................... 15 4.2.2 Diurnal Pattern ....................................................................................................................................... 16 4.3 Summary and Discussion of Data Set Limitations ........................................................................ 17 5. CONCLUSIONS ................................................................................................................................. 18 6. REFERENCES ................................................................................................................................... 20 EXHIBIT A – SEASONAL RAMP OCCURANCES ................................................................................ 21 ATTACHMENTS APPENDICES Met Tower Data QA/QC Processing System Appendix A Global Weather Archive Appendix B Delta Junction, Alaska C O N F I D E N T I A L Power Ramp Analysis - Page iii Final – 1110-835.001-01 EXECUTIVE SUMMARY The purpose of this report is to analyze historical wind data at the Delta Junction, Alaska site in order to provide context around wind power ramps for a wind farm planned for development. WindLogics understands that the output data will be used by Electrical Power Systems Inc. (EPS) for power flow modeling, transient stability analysis and other distributional-power grid modeling methods to assess utility operator control strategies. The objective of this effort is to provide time series data deliverables and analysis of power ramp events the Delta Junction wind project. The scope of work is to analyze onsite meteorological data collection (met) towers and available power production data from an operational wind turbine to calculate a time series of full- farm power production data for the sixteen planned GE 1.6XLE wind turbines. The power production data was then analyzed by WindLogics to provide context regarding the positive and negative power ramp events. A power ramp definition and power ramp rate characterization were provided along with a meteorological pattern analysis to determine the drivers behind the temporal fluctuations of energy production. The general approach of the ramp characterization and data analysis is to focus on the fundamentals to derive a theoretical energy production data set, which are: 1) meteorological data collection (met) tower data quality, 2) hub-height extrapolation (wind shear) calculations, and 4) the calculation of full-farm net energy from wind speed and gross energy values. Of particular concern is how these power production simulations relate to the integration of wind energy on the electrical grid network and utility control room. The results of the analysis show a generally equal trend of positive and negative power ramp events by season and an overall drastic increase in power ramps during the transitional and winter seasons. It was also discussed in detailed how the seasonal weather patterns are intrinsic to the daily ramp occurrences. Delta Junction, Alaska C O N F I D E N T I A L Power Ramp Analysis - Page 1 Final – 1110-835.001-01 1. INTRODUCTION The purpose of this report is to analyze historical wind data at the Delta Junction, Alaska site in order to provide context around wind power ramps from a theoretical wind farm. WindLogics understands that the output data will be used by Electrical Power Systems Inc. (EPS) for power flow modeling, transient stability analysis and other distributional-power grid modeling methods to assess utility operator control strategies for the Alaskan utility operator. 1.1 Overview of Objective and Scope The objective of this effort is to provide time series data deliverables and analysis of power ramp events at the Delta Junction wind project. The scope of work is to analyze onsite meteorological data collection (met) towers and available power production data from an operational wind turbine to calculate a time series of full- farm power production data for the sixteen planned GE 1.6XLE wind turbines. The power production data will be analyzed to provide context regarding the positive and negative power ramp events. A power ramp definition and power ramp rate characterization is provided along with a meteorological pattern analysis to determine the drivers behind large fluctuations of energy production. The general approach of the ramp characterization and data analysis is to focus on the fundamentals to derive a theoretical energy production data set, which are: 1) meteorological data collection (met) tower data quality, 2) hub-height extrapolation (wind shear) calculations, and 4) the calculation of full-farm net energy from wind speed and gross energy values. Of particular concern is how these power production simulations relate to the integration of wind energy on the electrical grid network and utility control room. 1.2 Site Location and Description The Delta Junction Wind Farm is located approximately 7 kilometers (km) east of the city of Delta Junction, Alaska. The current makeup of the wind farm is two operational wind turbine generators, a EWT Directwind 54/900kW and a NorthWind 100B, with another sixteen GE 1.6XLE (82m RD) wind turbines currently being planned for construction, for a total of 26.6 megawatts (MW) of nameplate capacity. Figure 1, below, shows a site map that includes the operational turbine locations (white hatch dots) and the planned turbine locations (hollow white dots). Delta Junction, Alaska C O N F I D E N T I A L Power Ramp Analysis - Page 2 Final – 1110-835.001-01 Figure 1: Map of Met Tower and Turbine Locations 1.3 Customer-Supplied Data EPS supplied WindLogics with the following information and data files to be considered in the analysis: Manufacturer-supplied technical specifications for the GE 1.6XLE and EWT Directwind 54/900 kW wind turbine generators. The documentation contained power curves, thrust coefficients and acoustic parameters. Digital photographs of the EWT Directwind 54/900 kW turbine construction including photographs of surrounding land taken from the nacelle. Turbine-location coordinates for the planned GE 1.6XLE wind turbines. Ten-minute average Supervisery Control and Data Acquisition (SCADA) data for an EWT Directwind 54/900 kW wind turbine generator. The data files were transmitted electronically to WindLogics and contained data for wind speed (nacelle-mounted anemometer), wind direction (nacelle- mounted wind vane), active power, reactive power and turbine-specified event codes. Delta Junction, Alaska C O N F I D E N T I A L Power Ramp Analysis - Page 3 Final – 1110-835.001-01 The ten-minute average, met tower data files were transmitted electronically to WindLogics for one (1) onsite meteorological data collection (met) tower. The above information was used by WindLogics to characterize the power performance and analyze power ramp events of the Delta Junction wind farm. Specific descriptions on data handling methodologies will be discussed in Section 2. 2. THEORETICAL WIND FARM PRODUCTION CALCULATION METHODOLOGY This section describes the methods used by WindLogics to create a theoretical data set of full-farm power production data at the Delta Junction site. The methodology for creating time series data for full-farm production data can be summarized in the bulleted points below. Full description of the data handling and energy production methodologies are listed, below in Sections 2.1 to 2.6. The methodologies can be summarized as: On site met tower data were quality checked and poor-quality data filtered from the data set, On site met tower data were then extrapolated to hub-height on a timestep-by-timestep basis using the power law equation, Extrapolated hub-height wind speed data was verified and compared to nacelle-mounted anemometer data collected at 75 meters above ground level (m AGL) from a nearby operational EWT Directwind 54/900 kW wind turbine, Extrapolated hub-height wind speed data was adjusted, on a timestep-by- timestep basis to represent a turbine-location average using power law dynamics, turbine layout information and turbine-specific elevations, The turbine-location adjusted wind speed data was then converted to power production on a timestep-by-timestep basis using manufacturer- supplied power curve information for the GE 1.6XLE wind turbine, turbulence intensity and site-representative air density, Power production data was then scaled to represent gross energy output for the full-farm (16 turbines), and Instantaneous energy production losses were estimated and applied, on a timestep-by-timestep basis. The output of these methodologies were time series data of theoretical wind farm production for the entire data collection period of the Delta Junction met tower and will be the basis for the analysis of power ramp events. 2.1 Meteorological Tower Data Processing Methods EPS supplied WindLogics meteorological data collection (met) tower data for one onsite tower. Table 1, below lists the coordinate locations and data collection Delta Junction, Alaska C O N F I D E N T I A L Power Ramp Analysis - Page 4 Final – 1110-835.001-01 period for the met tower. This section describes the methods used to quality check the met tower data collected at the Delta Junction site. Latitude Longitude Elevation (WGS 84) (WGS 84) (meters AMSL) Delta Junction Met N 64.01375 W 145.59675 403 9/27/2007 - 5/13/2010 Tower Data Collection Period Table 1: On-Site Met Tower Location All data collected at the Delta Junction met tower location following the installation of EWT Directwind 54/900 kW wind turbine was reviewed for possible data contamination due to turbine waking. The affects of turbine waking were not seen in the tower data signature but data collected subsequent to January 2010 should not be considered for further analysis. 2.1.1 Met Tower Sensor Configuration EPS supplied WindLogics with met tower data files in ASCII text format for the Delta Junction met tower location. The file header and metadata contained in these files were used to determine the configuration of the met tower and were entered into the WindLogics Data Management database as-is. The Delta Junction met tower was equipped with wind speed sensors at 30 and 50 m AGL. Full met tower commissioning information, as interpreted from the file header, is listed in Table 2. Delta Junction, Alaska C O N F I D E N T I A L Power Ramp Analysis - Page 5 Final – 1110-835.001-01 Table 2: On-Site Met Tower Configuration 2.1.2 Met Tower Data Processing Overview The ten-minute average, met tower data files, in ASCII text format, were transmitted electronically to WindLogics. The data were quality checked and stored in the WindLogics Data Management database. See Appendix A for more information regarding WindLogics Met Tower Data QA/QC Processing System. It is important to note that NRG sensor defaults for scale and offsets were used in the wind speed data processing. A 90 degree directional offset was applied to the wind direction data based on met tower documentation and configuration notes provided to WindLogics. Following the data filtering, WindLogics examined tower shadow plots for the redundant sensors at 50m and 30m AGL, and some shadowing is seen, which is Delta Junction, Alaska C O N F I D E N T I A L Power Ramp Analysis - Page 6 Final – 1110-835.001-01 to be expected. The effect of tower shadowing or other mast effects have on the data and extrapolation to hub height was minimized by using the maximum value between the redundant sensors at each measurement height. During the analysis process, WindLogics also screened the data for the “NRG #40 vibratory mode issue” (Clark, May 2009) (commonly referred to as sensor dragging). This issue was not seen in the data screening for the Delta Junction met tower. An additional check can be done using NRG’s web interface by inputing sensor-specific serial numbers to ensure the manufacturing of the anemometer did not occur within a defined period by NRG. Serial numbers were not provided in the metadata, as such, WindLogics did not perform this verification. 2.1.3 Data Quality and Data Recovery Results Table 3 and Figure 2, below show the amount of data recovered from the data files and how many timestamps were flagged during the data quality screening process. Channel Name/Height Variable Type Total Timestamps Missed Timestamps Percent Received Flagged Timestamps Total Missing/Flagged Timestamps Percent Not Missing/Flagged Ch. 01 NRG Anem. 50m Wind Speed 138,070 14,572 89.45% 7,691 22,263 83.88% Ch. 02 NRG Anem. 50m Wind Speed 138,070 14,572 89.45% 7,716 22,288 83.86% Ch. 03 NRG Anem. 30m Wind Speed 138,070 14,572 89.45% 8,108 22,680 83.57% Ch. 04 NRG Anem. 30m Wind Speed 138,070 14,572 89.45% 10,229 24,801 82.04% Ch. 05 NRG Anem. 10m Wind Speed 138,070 14,572 89.45% 12,139 26,711 80.65% Ch. 06 NRG Anem. 10m Wind Speed 138,070 14,572 89.45% 11,879 26,451 80.84% Ch. 07 NRG Vane 50m Wind Direction 138,070 14,572 89.45% 8,462 23,034 83.32% Ch. 08 NRG Vane 50m Wind Direction 138,070 14,572 89.45% 7,153 21,725 84.27% Ch. 09 NRG Vane 30m Wind Direction 138,070 14,572 89.45% 7,300 21,872 84.16% Ch. 10 NRG Vane 30m Wind Direction 138,070 14,572 89.45% 9,499 24,071 82.57% Ch. 11 NRG Temp. 0m Temperature 138,070 14,572 89.45% 108 14,680 89.37% Ch. 12 NRG Humid. 0m Humidity 138,070 14,572 89.45% 106 14,678 89.37% Ch. 13 NRG Press. 0m Pressure 138,070 14,572 89.45% 69,743 84,315 38.93% Ch. 14 NRG Volt. 0m Voltage 138,070 14,572 89.45% 0 14,572 89.45% Total: 1,932,980 204,008 89.45% 160,133 364,141 81.16% Delta Junction Met - 09/2007 to 5/13/2010 Table 3: Met Tower Data Recovery Summary . Delta Junction, Alaska C O N F I D E N T I A L Power Ramp Analysis - Page 7 Final – 1110-835.001-01 Figure 2: Tower Data Coverage and Data Recovery Note that in the figure above, the white areas are time periods not part of the data record, but that the light-yellow areas are part of the data record (missing data). Several gaps in the data record exist, but did not occur for an extended period of time, with exception of long data outages in the summer and fall of 2009. The reason for missing data in those months was not documented in the supplied information and is unknown to WindLogics at this time. The minor gaps that were seen with the sensors were periods of calm wind (NRG #40 minimum reading). 2.2 Production Data Processing Methods This section describes the process used by WindLogics to quality check and analyze the power production data from the EWT Directwind 54/900 kW wind turbine. 2.2.1 Power Production Data Inventory EPS supplied WindLogics with turbine production data that were extracted from the SCADA software. The ten-minute average data files were transmitted electronically to WindLogics and contained data for wind speed (nacelle-mounted anemometer), wind direction (nacelle-mounted wind vane), active power, reactive power and turbine-specified event codes, which were entered into the WindLogics Data Management database as-is. For this analysis WindLogics only considered wind speed and wind direction from the nacelle-mounted instruments. All power production data from the EWT Directwind 54/900 kW turbine were deemed to be out of the scope of this analysis and, as such, were disregarded. Delta Junction, Alaska C O N F I D E N T I A L Power Ramp Analysis - Page 8 Final – 1110-835.001-01 It is also important to note that no met towers were installed following construction to allow for wake-free wind speed/direction readings. Consequently, all data collected at the Delta Junction met tower location following installation of EWT Directwind 54/900 kW was reviewed for possible data contamination due to turbine waking. Time series data collected subsequent to January 2010 should not be considered for further analysis. 2.2.2 Nacelle-Mounted Anemometer Filtering To quality check the nacelle-mounted instrumentation for the EWT turbine, the production data files were processed using the ExcelTM and WindographerTM software programs. Automated and manual data filtering algorithms were not used for this data. 2.3 Hub-Height Wind Speed Extrapolation Estimating hub-height wind speeds from measurements at lower elevations required the use of a wind shear exponent, commonly referred to as (alpha), which is a value showing the relationship between wind speeds at an upper and a lower height. This value is then used in the power law equation to calculate estimated hub-height wind speeds. Met tower wind speed data collected at the Delta Junction met tower location was extrapolated to the hub height of 80 m AGL using the wind shear coefficient calculated from data collected at the 30 and 50 m AGL sensor levels. The calculated alpha value was applied via the power law equation to the uppermost sensor velocity on a timestep-by-timestep basis. WindLogics used the data collected from the nacelle-mounted anemometry on the EWT Directwind 54/900 kW turbine to verify wind shear estimates from the Delta Junction met tower. To do this, a comparison of monthly and seasonal alpha values were reviewed. Table 3 below summarizes the calculated alpha values for the met tower and operational turbine. Upper Anemometer Lower Anemometer Upper Anem. Wind Speed Average (m/s) Lower Anem. Wind Speed Average (m/s) Shear Exponent (alpha) Estimated 80m AGL Wind Speed (m/s) 50m(1) & 50m(2) Maximum 30m(1) & 30m(2) Maximum 5.57 4.68 0.297 6.67 EWT - 75m Nacelle-Mounted EWT - 10m 6.84 6.84 0.009 ~6.84 Table 4: Calculated Wind Shear As anticipated, the nacelle-mounted anemometer wind speed was similar to the extrapolated hub-height wind speed at the met tower location. For the same data collection period (Jan 2010 to May 2010), the met tower extrapolated value was 6.96 m/s while nacelle-mounted anemometer had an average wind speed of 6.84 m/s. The cursory comparison added confidence in the extrapolated hub-height wind speeds at the met tower location. WindLogics did not incorporate any other adjustments to the wind speed profile based on the data comparison for two reasons. These are: 1. Nacelle-mounted anemometer is waked from turbine rotor blades which degrade the wind velocity, and Delta Junction, Alaska C O N F I D E N T I A L Power Ramp Analysis - Page 9 Final – 1110-835.001-01 2. Lower wind speed instrumentation, identified at 10 m AGL within the data file, shows an identical average value, which lessened the confidence in the validity of the data. The close geographic proximity, similar elevations and exposure of the met tower and EWT turbine also suggested that the wind speeds would be similar. 2.4 Full-Farm Weighting WindLogics performed a qualitative analysis to determine the representativeness of the met tower location to that of the anticipated turbine locations. The majority of the planned turbine locations are located to the north of the met tower location in lower elevations and with less exposure to the east-southeast (dominant) wind flow, as shown above in Figure 1. To calculate the effects of elevation differences on overall (full-farm) average wind speed, absent any wind flow modeling or detailed information of the spatial variability at the site, WindLogics estimated the average wind speed at the turbine locations using known properties of the change in wind speed with height calculated using the power law equation and met tower data. To do this, the calculated wind shear exponent (alpha) at the met tower location was used in the power law equation to estimate an effective hub-height wind speed representative of all turbine locations. Calculation of a turbine-specific average wind speed was determined by: met met trb mettrb Z ZWSWS (1) Where: WS = average wind speed in m/s at the turbine locations (trb) or met tower location (met) Z = elevation in meters above sea level at the turbine locations (trb) or met tower location (met) = wind shear exponent The power law method used to estimate turbine-location specific wind speed is an oversimplified approximation of the wind flow over terrain and forested areas. A detailed numerical weather or wind flow model is typically used to assess the spatial variability of a wind energy development site to estimate expected energy yields but the simplistic assumptions used here were deemed appropriate because of the relatively small wind farm size and overall intent of the analysis. A detailed wind flow model accounting for atmospheric physics properties (Navier- Stokes equations), stability, surface roughness and high-resolution elevation/land use would likely result in different magnitudes at each turbine location but the aggregate wind farm production, specifically instances of power ramps, will not be materially different. WindLogics estimated that the overall average wind speed at the turbine locations was approximately 2% lower than the measured wind speeds at the Delta Junction, Alaska C O N F I D E N T I A L Power Ramp Analysis - Page 10 Final – 1110-835.001-01 met tower location. The hub-height wind speed data was adjusted on a timestep- by-timestep basis to account for this. The calculated turbine-by-turbine wind speed and full-farm average wind speed is listed in Table 5. Turbine Number Latitude (WGS 84) Longitude (WGS 84) Elevation (m) PowerLaw Wind Speed (m/s) Met 64.0138 -145.5968 403.00 6.67 1 64.0132 -145.5890 390.45 6.61 2 64.0136 -145.5959 404.47 6.68 3 64.0154 -145.5921 388.32 6.60 4 64.0160 -145.5981 407.52 6.69 5 64.0175 -145.5927 391.67 6.61 6 64.0185 -145.5984 395.02 6.63 7 64.0201 -145.5932 373.68 6.52 8 64.0215 -145.5985 387.40 6.59 9 64.0229 -145.5927 370.64 6.51 10 64.0240 -145.5984 377.65 6.54 11 64.0256 -145.5959 372.77 6.52 12 64.0259 -145.5903 364.54 6.47 13 64.0240 -145.5861 363.32 6.47 14 64.0213 -145.5862 366.37 6.48 15 64.0186 -145.5864 371.55 6.51 16 64.0159 -145.5859 373.68 6.52 381.19 6.56Turbine Average Table 5: Calculated Turbine Average Wind Speed It is important to note that this qualitative assessment is based on elevation influences to the wind velocities and does not account for seasonal vegetation porosity, roughness length and is prior to any degradation of wind speeds from inter-turbine waking. 2.5 Energy Production Calculation Methods Hub-height wind speed and air density values were applied to the manufacturer wind turbine power curve to calculate power values, which were then converted to gross energy production and gross capacity factor. The hub-height wind speed and air density were applied to the power curve, on a timestep-by-timestep basis, to calculate a set of power values in kilowatts (kW). For the GE 1.6XLE, with multiple power curves, one for each air density, the air density was used as a lookup value to select the appropriate manufacturer’s power curve, which was then used along with the raw wind speed, to calculate the corresponding power value. It is important to note that the wind speed to power lookup values were interpolated from the manufacturer’s power curve wind speed bins of 0.5 m/s to a resolution within the thousandths (0.001 m/s). The result of this process was a set of gross energy production and gross capacity factor values that took into account both the hub-height wind speed and air density values at the site, on a timestep-by-timestep basis, for the entire data collection period. This data was then weighted to represent the output of 16 wind turbines by simply scaling by a factor of sixteen. Delta Junction, Alaska C O N F I D E N T I A L Power Ramp Analysis - Page 11 Final – 1110-835.001-01 2.6 Project-Specific Loss Analysis Methods This section describes the process used by WindLogics to estimate the amount of wind plant energy losses that will likely affect the energy production profile at the Delta Junction site and thus the rate of power ramps. Again, the intent of this analysis is to create a theoretical wind farm production time series that represents energy production transmitted to the electrical distribution grid. Thus, accrued energy losses for turbine availability, seasonal icing, blade degradation and power curve performance were not considered, nor were losses to energy applied within the data time series. However, instantaneous energy losses due to turbine waking, electrical line losses and high wind hysteresis1 were considered and described in more detail below. 2.6.1 Wake Losses To account for energy losses due to inter-turbine wake effects, WindLogics used professional judgment to estimate the percentage of wake loss. The turbine rotor diameter (82 m) and turbine spacing were assessed. It was determined that the overall wake effect, or loss of energy production, was approximately 8%. Wake losses were applied on a timestep-by-timestep basis using a dynamic wake profile (see Figure 1) that is a function of ambient wind speeds. The resulting net energy was applied to create a load duration curve with a granularity of one tenth of a m/s. 0.01 0.04 0.09 0.14 0.19 0.24 0.29 0.34 Wind Speed (m/s) WakeProfilebased on 8.0% wakeloss Figure 3: Wake Effect versus Wind Speed 1 High wind hysteresis was calculated during the wind speed to power (or power curve calculation) process and not discounted as an operational energy loss. Delta Junction, Alaska C O N F I D E N T I A L Power Ramp Analysis - Page 12 Final – 1110-835.001-01 2.6.2 Electrical Efficiency Losses WindLogics was not supplied with estimates of electrical line losses. Based on professional judgment and operational experience, electrical line losses were estimated to be 2%. This value was applied on a timestep-by-timestep basis to represent the energy loss from each of the turbines locations to that of the interconnection point. 2.6.3 Wind Sector Management Control Losses Unless proper control system precautions are taken, winds from non-prevailing directions could result in a measurable loss of energy with the current turbine placements. WindLogics did not account for energy losses due to turbine shutdown from a manufacturer-defined wind direction sectors or any other turbine control losses. Based on the provided turbine layout however, these losses may affect energy production of the Delta Junction wind farm. It is recommended that the project developer consults the turbine manufacturer for further details on available wind sector management programs. 3. POWER RAMP CHARACTERIZATION METHODOLOGY Increased utility regulatory demands have placed added restrictions on wind plant performance, specifically power ramps. Recent wind turbine generator control system requirements include low voltage ride-through capability, voltage control, power-output control and ramp rate control (UWIG May 2006). For areas with limited wind energy penetration, like Alaska, the sensitivity to variable power sources can be exaggerated. Moreover, system stability studies are highly focused on the correlation between system load profiles and patterns of wind plant output (daily, monthly, and seasonal) as well as small-scale power ramps with the objective to plan for other dispatchable generation sources to meet system demand. The impact of additional wind penetration typically varies depending on many factors. Nonetheless anticipating large energy variations and steep power ramps over a short period of time can be very challenging, especially for balancing areas with long distances to load centers, as is the case for this region of Alaska. Using theoretical wind plant output to simulate the variable wind resource is essential for planning wind integration. Upon commercial operation status and live power flow to the distribution grid it is often critical to have a wind power forecast in place to anticipate wind plant output and plan for power ramp events to ensure continued reliability and stability of the power system. This section describes the methods used by WindLogics to characterize a ramp event and methods used to analyze the full-farm power production data at the Delta Junction site. The intent is to provide a data time series that can be used for future system-stability studies and detailed power-flow modeling. 3.1 Ramp Characterization A power ramp is often described as a steep increase (or decrease) in power output. The exact definition of what constitutes a significant ramp often varies from plant operator to utility off taker. Each utility will have its own thresholds Delta Junction, Alaska C O N F I D E N T I A L Power Ramp Analysis - Page 13 Final – 1110-835.001-01 which will be based on system load, total wind generation and the dispatchable generation available on their system. From the utility perspective, it is also the aggregate wind generation that is most important (barring significant transmission constraints) for defining wind energy ramps For example, the passage of a frontal system will cause a system-wide wind energy ramp, but that ramp will be spread out over a longer period of time than at any individual wind plant. For this analysis a ramp is characterized by two criteria. These can be summarized as follows: 1. Total ramp size must be at least 30% or more of rated capacity of the wind farm. In the case of Delta Junction the rated capacity is 25.6 MW equating to a ramp magnitude of 7.7 MW. 2. To generalize a ramp characteristic, a ramp event is also defined to be sustained for a period of no less than three hours. This is done to isolate power ramps that were caused by large-scale weather patterns and not isolated spikes in power due to independent (yet weather related) wind gusts, which could be the result of various micro-scale meteorological induces. Stated differently, the scope of this analysis was to review large- scale weather patterns influencing wide-spread power ramps and not necessarily an in depth analysis of distinct instances of short-lived and diverse power ramps. It is WindLogics opinion that the above criteria best characterizes a weather- related increase or decrease to wind plant output. Other control strategies or external wind plant operations (imposed curtailments, system-wide faults, etc.) can result in data signatures that mimic that of a power ramp. These cases should be included in any power flow or system stability simulations but are not described in this analysis as power ramps nor are they represented in the derived data set. Moreover, power ramps and wind plant output may show much more variability (or volatility) to a utility control room operator than captured within this analysis. See Section 4.3 for additional discussion on temporal resolution, data smoothing and other assumptions that may limit the use of the derived data set. It is also anticipated that the power ramp rate definition will be sufficient to identify large frequency deviations in weather-related wind plant output. With that being said, the occurrence of events is directly correlated to the power ramp definition. If EPS has a need for a modified power ramp rate based on expectations of both frequency and temporal rate of change, WindLogics will be accommodating in updating the analysis. 3.2 Ramp Analysis Methods The power production time series data was flagged for each time step that met the above criteria. The identified power ramp events were then analyzed in conjunction with WindLogics weather archives using various data analysis software applications. Delta Junction, Alaska C O N F I D E N T I A L Power Ramp Analysis - Page 14 Final – 1110-835.001-01 WindLogics weather archives were used to review the large-scale weather events that corresponded to the same time period as the power ramp events. The weather archive consists of National Weather Service issued forecast model data, of which WindLogics has been actively archiving for over a decade. The extensive weather data archive is obtained by WindLogics in the standard file format for gridded meteorological data (GRIdded Binary or grib format). This data is then converted to a proprietary data format to be accessed across different functioning groups throughout the company. Proprietary scripts to parse through and analyze the surface and low level wind speed/direction data were used in order to corroborate the results throughout this report. Although the horizontal grid resolution ranging from 20 km to approximately 80 km2 is very coarse, especially for the complex terrain of the region, this data was used to review large-scale frontal patterns, pressure systems and overall wind flow patterns. The weather archives were not used to validate the overall wind velocity magnitudes or ramp rates since the smaller-scale features (site-specific terrain and land characteristics) are typically not resolved. The small-scale features can play a significant role in the magnitude of the wind resource. See Appendix B for a detailed description of the weather archives. WindLogics also used the ExcelTM and WindographerTM software programs to analyze the power ramp events and compile the results within this report. 4. RESULTS This section describes the results of the ramp analysis at the Delta Junction site. 4.1 Meteorological Overview During the winter and transitional seasons, the wind regime of this region is highly influenced by transient and developing synoptic-scale weather systems. These cool/cold season weather systems track closely with the position of the upper level jet stream. The position of the jet stream, which is near the Aleutian Islands and the Gulf of Alaska during these seasons, results in the frequent passage of vigorous low-pressure systems south of the Delta Junction site. These low pressure systems often result in power ramps and sustained energy production for multiple days. As the low weakens or moves out of the area, the winds subside creating a sharp drop in production or negative power ramp. During the summer months, the wind regime is influenced by a large seasonally- persistent high pressure system in the eastern Pacific. As a result, wind speeds over much of the region moderate. The Alaskan interior often experiences afternoon thunderstorms in the early summer months (May and June) that drive short-lived ramps in energy production. These often subsist within a few hours. Mountainous terrain plays a very important role in modifying the larger scale synoptic wind patterns that influence the Delta Junction site. The interaction of steep terrain, prevailing winds, and forested valleys results in a variety of wind conditions that include windy mountain tops and sheltered valleys. Specifically the Delta Junction site is located in the plains between two significant terrain 2 Horizontal spatial resolution varies between the archived data sources. Delta Junction, Alaska C O N F I D E N T I A L Power Ramp Analysis - Page 15 Final – 1110-835.001-01 features to the north and south, respectively. Wind flow parallel to the mountains within the Tanana River Valley drives the east-southeastern direction of winds. 4.2 Ramp Occurrences Using the power ramp definition, WindLogics identified a total of 350 ramp events during the met tower data collection period. Table 6, below identifies the total positive and negative power ramp occurrences described by season and time of day. A more detailed discussion of the seasonal and diurnal occurrences of ramp events is located in Sections 4.2.1 and 4.2.2, respectively. Ramp Occurrences Hour Summer Winter Total Summer Winter Total 1-6 6 28 34 16 41 57 7-12 8 37 45 8 41 49 13-18 13 44 57 4 20 24 19-0 16 29 45 10 29 39 Total Ramps 43 138 181 38 131 169 Positive Negative Table 6: Overall Ramp Occurrences Not surprising is the difference in frequency of power ramps between winter and summer. Please note that winter (and transitional) months spanned from October through April, whereas summer months are May through September. The signature of power ramp events also varies by season. In the summer, a ramp up/down is short-lived, ~12-hour, representing the heating/cooling of the surface. Whereas winter time ramp up/down tend to last one to three days (even longer in extreme winter months of January and December) based on the position and duration of low pressure systems in the Bay of Alaska. 4.2.1 Monthly Power Ramp Distribution The monthly distribution of power ramp occurrences is generally correlated to the monthly wind resource, with the transitional seasons having greater instances of large ramp rates. Figure 4 shows the monthly power ramp occurrences. Delta Junction, Alaska C O N F I D E N T I A L Power Ramp Analysis - Page 16 Final – 1110-835.001-01 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 0 5 10 15 20 25 30 Jan. Feb. Mar Apr May Jun Jul Aug Sep Oct Nov Dec MonthlyWind Speed and PowerRampOccurrences Negative Ramps PositiveRamps Monthly Average Wind Speed Figure 4: Monthly Pattern of Power Ramp Events and Monthly Average Wind Speed It is suspected that the months of January and December show a significant decrease in ramp events compared to surrounding months because of the persistence of low pressure systems positioned in the Bay of Alaska during extreme winter months which elongates periods of increased energy production. 4.2.2 Diurnal Pattern The power ramp occurrences were summarized by hour to analyze the daily pattern. This is depicted in Figure 5, below. Seasonal diurnal graphs are included in Exhibit A. Delta Junction, Alaska C O N F I D E N T I A L Power Ramp Analysis - Page 17 Final – 1110-835.001-01 4.0 5.0 6.0 7.0 8.0 9.0 10.0 0 2 4 6 8 10 12 14 16 18 20 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23HourAKST (9 GMT) Annual Daily RampOccurrences PositiveRamps Negative Ramps WindSpeed Poly.(Positive Ramps) Poly.(Negative Ramps) Figure 5: Diurnal Pattern of Power Ramp Events and Wind Speed The highest occurrences of ‘downramps’ occurs as the wind resource statistically is increasing, and the largest number of ‘upramps’ occurs when the wind resource is statistically decreasing. What this likely suggests is that the diurnal cycle isn’t a large factor in creating wind energy ramps – at least not in the annual average. This is not totally surprising given that the sun is up all day in summer, and near or below the horizon in winter. On a seasonal basis the ramp occurrences tend to track well with the diurnal pattern (Shown in Exhibit A). The depiction of diurnal wind patterns overlain with ramp occurrences is a helpful visual of when to plan for positive or negative energy fluctuations in the wind plant operations strategy. 4.3 Summary and Discussion of Data Set Limitations A data set of theoretical wind farm output was compiled to analyze instances of power ramp events and the impact of those events to the utility balancing area. The data set is intended to support the planning of wind energy integration, but certain data-compilation assumptions need to be conveyed in order for proper utilization and data interpretation. Some points of concern are summarized below: Depending on the user requirements, ramp rates are often described as MW change per minute. The onsite data available to perform this analysis was at ten-minute increments. Delta Junction, Alaska C O N F I D E N T I A L Power Ramp Analysis - Page 18 Final – 1110-835.001-01 Average 10-minute wind speed and direction time series data measured at a nearby met tower was used to derive the data. The use of 10-minute data is widely accepted for this type of analysis but does present an issue of data smoothing. Actual power output from an operational wind facility will show much more variability of the minute-by-minute power production. The identification of power ramp occurrences within this report is based on the simplistic assumption that all power ramps are based on weather- related events. Large fluctuations to power output can (and do) occur in any operational wind facility and may be greatly under-represented within the derived data set. The wind farm output is based on calculations from wind distributions measured at a specific-point location that may not be representative of all turbine locations and assumes that a weather system will affect all turbine outputs simultaneously. 5. CONCLUSIONS The objective of this effort was to provide time series data deliverables and analysis of power ramp events the Delta Junction wind project. Specifically, the scope of work was to analyze onsite meteorological data collection (met) towers and available power production data from an operational wind turbine to calculate a time series of full-farm power production data for the sixteen planned GE 1.6XLE wind turbines. The power production data was then analyzed by WindLogics to provide context regarding the positive and negative power ramp events. A power ramp definition and power ramp rate characterization were provided along with a meteorological pattern analysis to determine the drivers behind the temporal fluctuations of energy production. The general approach of the ramp characterization and data analysis was to focus on the fundamentals to derive a theoretical energy production data set, which are: 1) meteorological data collection (met) tower data quality, 2) hub- height extrapolation (wind shear) calculations, and 4) the calculation of full-farm net energy from wind speed and gross energy values. Of particular concern was how these power production simulations relate to the integration of wind energy on the electrical grid network and utility control room. The results of the analysis show an increase in power ramps during the transitional and winter seasons. It was also discussed in detailed how the seasonal weather patterns are intrinsic to the daily ramp occurrences. It is recommended that EPS continue to work with WindLogics to define the ramp rate and ramp characterization that may be more accommodating to EPS and the Alaskan utilities. Specific follow on work that is considered relevant to assist EPS in utility stability analysis include, but not limited to: a more granular review of ramp events, creation of additional data sources to include 1-minute timeseries Delta Junction, Alaska C O N F I D E N T I A L Power Ramp Analysis - Page 19 Final – 1110-835.001-01 distributions, and simulating the effects of wind integration and wind variability at the Delta Junction site. Implementation of wind plant forecasting for both power market operation and utility control room operations is also a recommended long-term step to accommodate increasing amounts of wind penetration. Delta Junction, Alaska C O N F I D E N T I A L Power Ramp Analysis - Page 20 Final – 1110-835.001-01 6. REFERENCES AWS Truewind, LLC. February 10 2010. Overview of the State-of-the Art and Current Issues in Wind Ramp Forecasting. Presentation at UWIG Forecasting Workshop, Albuquerque, New Mexico. Clark, Steven H. “Investigation of the NRG #40 Anemometer Slowdown,” May 2009, AWEA Windpower 2009 Enernex Corporation. May 2006. Xcel Wind Integration Study for Public Service Company of Colorado - Final Report. Knoxville, TN. http://www.nrel.gov/wind/systemsintegration/pdfs/colorado_public_service_windintegstudy.pdf. National Renewable Energy Laboratory. September 2004. Wind Power Plant Behaviors: Analysis of Long-Term Wind Power Data. Golden, CO. http://www.nrel.gov/docs/fy04osti/36551.pdf. Mitchell Jr., Murray. Strong Surface Winds at Big Delta, Alaska – An Example of Orographic Influence on Local Weather. Monthly Weather Review, January 1956. http://docs.lib.noaa.gov/rescue/mwr/084/mwr-084-01-0015.pdf Utility Wind Integration Group (UWIG). Utility Wind Integration State of the Art. Golden, CO. May 2006, http://www.uwig.org/UWIGWindIntegration052006.pdf. Wang, Jianhui, et al. Impact of Wind Power Forecasting on Unit Commitment and Dispatch. Argonne National Laboratory. http://www.dis.anl.gov/pubs/65610.pdf. WindLogics, Iberdrola Renewables. May 26 2010. An Introduction to Wind Ramps: Atmospheric Causes and Forecasting. Presentation at WindPower 2010, Dallas, Texas. Zheng, Haiyang. Kusiak, Andrew. Prediction of Wind Farm Power Ramp Rates: A Data-Mining Approach. Journal of Solar Energy Engineering, August 2003, Volume 131. http://www.icaen.uiowa.edu/~ankusiak/Journal- papers/Wind_Ramp_Rates_2009.pdf Delta Junction, Alaska C O N F I D E N T I A L Power Ramp Analysis - Page 21 Final – 1110-835.001-01 EXHIBIT A – SEASONAL RAMP OCCURANCES 4.0 5.0 6.0 7.0 8.0 9.0 10.0 0 2 4 6 8 10 12 14 16 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23HourAKST (9 GMT) Winter Daily RampOccurrences PositiveRamps Negative Ramps WindSpeedDiurnal Poly.(Positive Ramps) Poly.(Negative Ramps) 4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5 8.0 0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23HourAKST (9 GMT) Summer Daily RampOccurrences Positive Ramps Negative Ramps WindSpeedDiurnal Poly.(Positive Ramps) Poly.(Negative Ramps) C O N F I D E N T I A L Appendix A WindLogics Tower Data QA/QC Processing System Introduction The intent of this document is to provide more information on the WindLogics Tower Data Quality Assurance / Quality Checking (QA/QC) processing system. The QA/QC processing system is used to screen data flows from a variety of sources, including (but not limited to) the following: Prospecting meteorological data collection (met) towers Permanent met towers Turbine-mounted data collection equipment SODAR/LIDAR units (yet to be fully implemented) The QA/QC processing system is made up of the following three phases: Ingest: The first phase of QA/QC processing is to ingest data in various formats. The purpose of this phase is to convert data from the many and various data formats from different equipment vendors to a common format for internal use so that subsequent processing methods are uniform and consistent. Evaluate: The second phase of QA/QC processing is to evaluate the quality of the data by adding QA/QC “flags” (using both automatic and manual methods). It is important to note that at no point during the data evaluation phase are data actually removed from the original data set – in fact, information is added (in the form of the QA/QC flags) during this process. Filter: The third and final phase of QA/QC processing is to filter the data for further processing, based on the QA/QC flags. Only at this point are data removed from the data flow, and only from the output data set, not the original. It is important to note that the needs of various end users have to be kept in mind at this stage, as different end users will likely have different criteria for filtering, depending on what their intended use for the data is. Note: The default filter behaviors listed in the table below can be changed as needed during the output data file creation process. QA/QC Flags The following table lists the types of QA/QC flags and the default filtering behaviors used to generate QA/QC’d data files. Variable Check Value Default Filter Behavior Wind Speed Max. 10-minute Avg. >= 50 m/s REJECT Wind Speed Min. 10-minute Avg. <= 0.4 m/s REJECT Wind Speed Max. 10-minute Std. Dev. > 4 m/s REJECT Wind Speed Min. 10-minute Std. Dev. < 0.1 m/s REJECT Wind Speed Max. 10-minute Delta > 30 m/s REJECT Wind Speed Min. 1-hour Delta < 0.1 m/s REJECT Wind Speed Previous value missing --- ALLOW Wind Speed Negative Shear --- ALLOW C O N F I D E N T I A L Appendix A Variable Check Value Default Filter Behavior Wind Direction Max. 10-minute Avg. > 360 degrees REJECT Wind Direction Min. 10-minute Avg. < 0 degrees REJECT Wind Direction Max. 10-minute Std. Dev. > 75 degrees REJECT Wind Direction Min. 10-minute Std. Dev. < 0.1 degrees REJECT Wind Direction Min. 1-hour Delta < 0.1 degrees REJECT Wind Direction Previous value missing --- ALLOW Speed and Direction Tower Shadow <= 10 degrees of tower orientation ALLOW Temperature Max. 10-minute Avg. > 50 degrees C REJECT Temperature Min. 10-minute Avg. < -40 degrees C REJECT Temperature Max. 10-minute Std. Dev. > 5 degrees C REJECT Temperature Min. 10-minute Std. Dev. < 0 degrees C REJECT Temperature Max. 1-hour Delta > 30 degrees C REJECT Temperature Min. 3-hour Delta < 0.1 degrees C REJECT Temperature Previous value missing --- ALLOW Pressure Max. 10-minute Avg. > 1100 mb REJECT Pressure Min. 10-minute Avg. < 700 mb REJECT Pressure Max. 10-minute Std. Dev. > 5 mb REJECT Pressure Min. 10-minute Std. Dev. < 0 mb REJECT Pressure Max. 1-hour Delta > 40 mb REJECT Pressure Min. 12-hour Delta < 0.1 mb REJECT Pressure Previous value missing --- ALLOW Where the units are: m/s = meters per second degrees = radial degrees (0-360) degrees C = degrees Celsius mb = millibars C O N F I D E N T I A L Appendix B The WindLogics Global Weather Archive WindLogics has compiled a multi-year database of global weather through archival of analysis data from Global Forecast System (GFS). The GFS is a global numerical weather prediction model run by the National Centers for Environmental Prediction (NCEP). This GFS analysis data represents the National Weather Service’s best available, real-time analysis of the state of the atmosphere every six hours over the globe. Observational data from numerous platforms (see below) are assimilated onto a three-dimensional grid to create these datasets. Sophisticated techniques are used to assimilate the observational data while maintaining dynamic consistency. The analysis data are used to initialization runs of the GFS model. WindLogics archives a version of the analysis data and first 3 hour forecast that has 1 degree (111 km at equator, finer spacing in longitude away from equator)) horizontal grid spacing and 26 vertical levels. Data Input to GFS at NCEP Aircraft weather data through the ACARS, ASDAR and MDCARS systems Rawinsondes and dropwindsondes VAD winds from WSR-88D NEXRAD radars SSM/I wind speeds SSM/I precipitation estimates TRMM precipitation estimates MODIS IR and water vapor channel winds GMS, METEOSAT, GOES cloud drift IR and visible winds GOES water vapor channel cloud top winds Quickscat wind speed and direction NOAA and AQUA polar orbiter radiances GOES-12 radiances NOAA 405 MHz wind profilers Boundary-layer (915 MHz) wind profilers Surface / METAR station data Buoy data Ship reports Final Draft Alaska Environmental Power & GVEA Delta Junction Wind Farm Impact Study December 13, 2010 Page 54 Appendix G – EPS Proposed One-Line For AEP Wind Farm