Frequency Response Assessment and Improvement of Three Major North American Interconnections due to High Penetrations of Photovoltaic Generation

Transcription

1 Frequency Response Assessment and Improvement of Three Major North American Interconnections due to High Penetrations of Photovoltaic Generation Yilu Liu 1,2, Yong Liu 1, Shutang You 1, Yingchen Zhang 3, Jin Tan 3, Maozhong Gong 4 1 University of Tennessee, Knoxville 2 Oak Ridge National Laboratory 3 National Renewable Energy Laboratory 4 GE Global Research

2 Project Overview BP 1 (2016): Develop high PV penetration scenarios BP 2 (2017): Study the impact of high PV at both the interconnection and balancing authority levels BP 3 (2018): Develop the mitigation strategies for low system inertia and reduced frequency response. 2

3 Phase 1: Base Model Validation Using Measurements - Introduction Why is the simulated EI primary frequency response significantly higher than measured values? Sensitivity study results. o o o o o o o Governor deadband (major) Governor ratio/spinning reserve (major) Governor droop Load composite The outer loop control Inertia Frequency dependent network Mismatch in frequency responses! 3/27

4 Phase 1: Base Model Validation Using Measurements Governor Deadband Governor deadband is adopted to avoid excessive turbine control actions within normal frequency variation range. Not typically modeled in EI MMWG model. deadband 30 mhz Measurement 49 mhz 4/27

5 Phase 1: Base Model Validation Using Measurements Governor Ratio Governor ratio is the fraction of generation capacity that is providing governor response. The ratio is currently around 80% in the EI MMWG models. Based on FNET/GridEye monitoring data, this ratio for EI is likely lower than 30% to match measurement. Base case Reduced governor ratio Measurement 5/27

6 Phase 1: Base Model Validation Using Measurements Results Measurement locations at EI edges 6/27

7 Phase 1: Base Model Validation Using Measurements Validation Accuracy at Grid Edges (a) Houghton, MI (b) Leroy, NY (c) Andalusia, AL (d) Plant City, FL (e) Charleston, SC (f) Plainsboro, NJ (g) Mduglendive, ND (h) Goodland KS 7/27

8 Phase 1: Base Model Validation Using Measurements WECC and ERCOT results WECC ERCOT Interconnection models frequency response validation 8

9 Phase 1: High PV Scenario Generation Mix Determination The generation mix of each simulation scenario is: Scenario Generation mix of high PV penetration simulation scenarios Instantaneous PV Penetration Level Instantaneous Wind Penetration Level Interconnection Level 5% 15% 20% Scenario 1 Interconnection Level 25% 15% 40% Scenario 2 Interconnection Level 45% 15% % Scenario 3 Interconnection Level 65% 15% 80% Scenario 4 Regional Scenario 100% 0% 100% Total Instantaneous Renewable Penetration Level 9

10 Phase 1: PV Instantaneous Penetration Rate Distribution in the EI 5% PV 25% PV 45% PV PV geographic distribution in the EI 65% PV 10

11 Phase 1: PV Instantaneous Penetration Rate Distribution in the ERCOT 5% PV 25% PV 45% PV 65% PV PV Instantaneous Penetration Rate Distribution in the ERCOT 11

12 Phase 1: PV Instantaneous Penetration Rate Distribution in the WECC PV Instantaneous Penetration Rate Distribution in the WECC 12

13 Inertia (MVA*s) Inertia (MVA*S) Inertia (MVA*s) Phase 1: System inertia in high penetration of renewables EI WECC ERCOT 3,000, ,000 2,500,000 2,000,000 1,500,000 1,000, , % 20% 40% % 80% Renewable penetration level % 20% 40% % 80% Renewable penetration level 200, , ,000 50, % 20% 40% % 80% Renewable penetration level 13

14 Phase 2: EI Frequency Response under High PV Penetration Test resource contingencies in the EI Contingenc y Description Unit Location Generation loss (MW) 1 The largest resource event in Five units in south Indiana (August 4, 4,500 last 10 years 2007 Disturbance) 2 An N-2 contingency Two Braidwood Nuclear Units, Illinois 2,250 3 An N-1 contingency One Browns Ferry Nuclear Unit, Alabama 1, Base Case 20% Renewable 40% Renewable % Renewable 80% Renewable Base Case 20% Renewable 40% Renewable % Renewable 80% Renewable Base Case 20% Renewable 40% Renewable % Renewable 80% Renewable Largest Resource Event (4,500 MW loss) An N-2 Contingency (2,250 MW loss) An N-1 Contingency (1,128 MW loss) The EI frequency responses in different high PV scenarios 14

15 Latitude Wave propagation speed (mile/s) Latitude Wave propagation speed (mile/s) Latitude Wave propagation speed (mile/s) Latitude Wave propagation speed (mile/s) Wave propagation speed (mile/s) 65%PV+15%WT Hz0kVNYISO Additional Study: Impact of High PV Penetration on Electromechanical Wave Propagation Longitude Longitude %PV+15%WT Hz0kVNYISO Longitude 20% renewable 40% renewable % renewable Longitude 80% renewable Electromechanical wave propagation speed in different EI PV penetration scenario Base 20% 40% % 80% Renewable penetration The EI average electromechanical wave propagation speed Regions with high PV penetration tend to have higher propagation speeds. Average wave propagation speed increases dramatically with the increase of PV penetration due to the decrease of system inertia: from 500 miles/s (base case) to 1,800 miles/s (80% renewable). 15

16 Impact on NERC Grid Code Compliance NERC Standard BAL requires sufficient frequency response to maintain Interconnection frequency within predefined bounds by arresting frequency deviations. BAL defined the interconnection frequency response obligation (IFRO) for each interconnection, which is the minimum frequency response for an interconnection to maintain reliability and avoid tripping UFLS relays. IFRO compliance is assessed by dividing the net generation change by the frequency change (Delta P/(f Point A -f Point B ), in the unit of MW/0.1 Hz). Absolute value of IFRO (MW/0.1Hz) Recommended IFROs for Operating Year 2017 EI WECC ERCOT 1,

17 Impact on NERC Grid Code Compliance - EI The NERC-defined EI frequency response value declines consistently with the increasing PV penetration level but still satisfies the current EI IFRO requirement even in the 80% renewable penetration scenario. Scenario 20% renewable 40% renewable % renewable 80% renewable NERC-defined Frequency Response Values of the EI in high PV scenarios A-B difference (Hz) 4.5 GW 2.25 GW 1.13 GW Frequency A-B difference Frequency A-B difference response (Hz) response (Hz) (MW/0.1Hz) (MW/0.1Hz) Frequency response (MW/0.1Hz) , , , , , , , , , , , ,270 17

18 ERCOT frequency response under high PV penetration Test resource contingencies in ERCOT Contingency Description Unit Location Gen. loss (MW) 1 The largest N-2 Two South Texas Nuclear 2,740 contingency Units 2 An N-2 contingency Two Martin Lake Units 1,370 3 An N-1 contingency One Martin Lake Unit UFLS 59.3 Hz UFLS 58.9 Hz 20% Renewable 40% Renewable % Renewable UFLS 58.5 Hz 80% Renewable Contingency 1: 2,740 MW generation loss Contingency 2: 1,370 MW generation loss Contingency 3: 685 MW generation loss UFLS 59.3 Hz The ERCOT frequency responses in different high PV scenarios 20% Renewable 40% Renewable % Renewable 80% Renewable 18

19 Impact on NERC Grid Code Compliance - ERCOT Scenario 20% renewable 40% renewable % renewable 80% renewable A-B difference (Hz) Frequency Response Values of the ERCOT 2700 MW 1370 MW 685 MW Frequency A-B difference Frequency A-B difference response (Hz) response (Hz) (MW/0.1Hz) (MW/0.1Hz) Frequency response (MW/0.1Hz) For the 80% renewable penetration scenario, ERCOT frequency response will be below the IFRO required by BAL-003-1(471 MW/0.1 Hz). The frequency response of 685 MW generation loss in the % scenario can not fulfill BAL due to nonlinearity induced by deadband. 19

20 Without load response Without load response With load response Without load response With load response 58.5 UFLS With load response 58.5 UFLS 58.5 Hz Using existing resources to improve 58.5 UFLS 58.5 Hz 58.5 UFLS 58.5 Hz frequency response in ERCOT FFR provided by load (a)20% renewable (a)20% renewable (b) 40% renewable (b) UFLS 59.3 Hz 59.5 UFLS UFLS 59.3 Hz UFLS 59.3 UFLS Hz 59.3 Hz UFLS 59.3 Hz UFLS 59.3 Hz 59 UFLS 58.9 Hz 58.5 UFLS 58.5 Hz 59 UFLS UFLS UFLS 58.9 Hz t load response Without load response ad response With load response 58.5 UFLS 58.5 Hz ble 59.5 (a)20% renewable 59 UFLS 58.9 Hz 59 UFLS 58.9 Hz Without load response Without load response With load response Without load response With load response Without load response 58.5 UFLS 58.5 Hz 58.5 UFLS 58.5 Hz With load 0 30 response With load response UFLS 58.5 Hz 58.5 UFLS 58.5 Hz (b) 40% (c) renewable % renewable (b) 40% (c) renewable % renewable (d) 80% renewable (d) 8 Figure 1. ERCOT frequency Figure responses 1. ERCOT frequency with fast load responses with (2.7 fast GW load generation response loss ( UFLS 59.3 Hz 59.5 UFLS 59.3 Hz disabled) disabled) 58.5 UFLS 58.5 Hz 58.5 UFLS 58.5 Hz Frequency nadir and settling frequency increased significantly with FFR FFR provided by 57.5load response is highly efficient in supporting frequency 57.5 response Without load when response 57 the governor Without load response 57 of synchronous Without load response generators is insufficient. UFLS 59.3 Hz 59 UFLS 58.9 Hz t load response ad response With load response 58.5 UFLS 58.5 Hz UFLS Hz UFLS 58.9 Hz 59 UFLS 58.9 Hz 59 UFLS 58.9 Hz ble (c) % renewable (d) 80% renewable (d) 80% renewable responses 1. ERCOT frequency with fast load responses with (2.7 fast GW load generation response loss, (2.7 UFLS GW generation loss, UFLS disabled) disabled) With load response With load response Without load response 57 With load response ERCOT frequency responses with fast load response (2.75 GW generation loss, UFLS disabled) 58 20

21 WECC frequency response under high PV penetration Contingency # Description 1 The largest N-2 contingency 2 An N-2 contingency 3 An N-1 contingency Test Contingencies in the WECC Unit Location Loss of the two largest generating units in the Palo Verde nuclear facility. Loss of the two units in the Colstrip coal power plant Loss of one unit in the Comanche generating station Gen. loss (MW) 2,625 1, ,625 MW generation loss 1,514 MW generation loss 804 MW generation loss The WECC frequency responses in different high PV scenarios 21

22 Impact of high PV penetration on WECC UFLS summary of WECC UFLS settings One primary plan + two subarea plans Load shedding block WECC coordinated underfrequency load shedding plan % Customer Load Dropped Pickup Frequency(Hz ) Tripping Time* Load shed ding block NWPP under-frequency load shedding coordinated plan Percent of NWPP Sub-Area Load shedding Pickup Frequen cy(hz) Tripping Time* 1 5.3% cycles 2 5.9% cycles 3 6.5% cycles 4 6.7% cycles 5 6.7% cycles Additional automatic load shedding to correct underfrequency stalling 2.3% seconds 1.7% seconds 2.0% 59.5 seconds Load automatically restored from 59.1 Hz block to correct frequency overshoot 1.1%.5 30 seconds 1.7%.7 5 seconds 2.3%.9 15 cycles *Tripping time includes relay and circuit breaker time % 59.3 No more than 14 cycles 2 5.6% 59.2 No more than 14 cycles 3 5.6% 59.0 No more than 14 cycles 4 5.6% 58.8 No more than 14 cycles 5 5.6% 58.6 No more than 14 cycles Additional automatic load shedding to correct under-frequency stalling 2.3% seconds 1.7% seconds 2.0% 59.5 seconds Load automatically restored from 59.3 Hz block to correct frequency overshoot 1.1%.5 30 seconds 1.7%.7 5 seconds 2.3% seconds *Tripping time includes relay and circuit breaker time. 22

23 Impact on NERC Grid Code Compliance - WECC For all the contingencies in the WECC 80% renewable penetration scenario, the NERC-defined frequency response values will be below the IFRO required by BAL-003-1(906 MW/0.1 Hz). It means the WECC system may not comply with the BAL standard at the 80% renewable penetration level. Scenario 20% renewable 40% renewable % renewable 80% renewable A-B difference (Hz) NERC-defined Frequency Response Values of the WECC 2.6 GW 1.5 GW 0.8 GW Frequency A-B difference Frequency A-B difference response (Hz) response (Hz) (MW/0.1Hz) (MW/0.1Hz) Frequency response (MW/0.1Hz)

24 Phase III work Design remedial schemes for the negative frequency response impacts of high PV generation under the developed multiple PV penetration scenarios (up to 80% instantaneous renewable penetration at the interconnection level and >=100% locally). Test the effectiveness of adjusting current frequency control strategies under multiple PV penetration scenarios for each North American interconnection. Design the artificial inertia/governor/agc controls for PV inverter. Determine whether demand response can provide artificial inertia/governor/agc support. Response time will be the key parameter and validation criteria. Implement artificial inertia/governor/agc schemes on a GE s utility-scale PV inverter product (Brilliance series) and test their effectiveness in GE s RTDS laboratory. Implement the proposed artificial inertia/governor/agc schemes on GE s utility-level PV inverter Test GE s PV inverter with additional artificial inertia/governor/agc functions in GE s RTDS laboratory Summarize PV inverter testing procedures and propose PV inverter frequency control standards. Implement the GE PV inverter control in the EI, WECC, and ERCOT models to demonstrate its effectiveness in frequency response improvement. Host a workshop on PV grid integration and investigate the implications of PV frequency support to existing grid codes and standards (IEEE Standards 1547 and NERC PRC-024-1). 24

25 Publications Journal: [J1] Y. Liu, S. You, J. Tan, Y. Zhang, Y. Liu, "Frequency Response Assessment and Enhancement of the U.S. Interconnections towards Extra-High Photovoltaic Generation Penetrations an Industry Perspective," Power Systems, IEEE Transactions on, [J2] Y. Liu, S. You, X. Zhang, S. Hadley, and Y. Liu, "Study of Advanced Renewable Generation Control in the U.S. Power Grid EI and TI Case Studies," IEEE Power and Energy Technology Systems Journal, [J3] S. You, G. Kou, Y. Liu, M. J. Till, Y. Cui, and Y. Liu, "The Impact of High Renewable Penetration on the Inter-Area Oscillation of the U.S. Eastern Interconnection (EI)," IEEE Access, 2017 [J4] S. You, Y. Liu, G. Kou, X. Zhang, S. Hadley, and Y. Liu, "Non-Invasive Identification of Inertia Distribution Change in High Renewable Systems Using Distribution Level PMU," Power Systems, IEEE Transactions on Conference: [C1] J. Tan, Y. Zhang, S. S. Veda, T. Elgindy, and Y. Liu, "Developing High PV Penetration Cases for Frequency Response study of U.S. Western Interconnection," in The 9th Annual IEEE Green Technologies Conference, Denver, Colorado, March 2017, pp [C2] S. You, X. Zhang, Y. Liu, Y. Liu, and S. W. Hadley, "Impact of High PV Penetration on U.S. Eastern Interconnection Frequency Response," in IEEE Power and Energy Society General Meeting, [C3] S. You, Y. Liu, and Y. Liu, " U.S. Eastern Interconnection (EI) Electromechanical Wave Propagation and the Impact of High PV Penetration on Its Speed, 2018 IEEE PES T&D Conférence & Exposition, [C4] J. Tan, Y. Zhang, S. You, Y. Liu, Y. Liu. Frequency Response Study of U.S. Western Interconnection under Extra-High Photovoltaic Generation Penetrations. IEEE PES General Meeting

26 Acknowledgements This work is funded in whole by the U.S. Department of Energy Solar Energy Technologies Office, under Award Number This work also made use of the Engineering Research Center Shared Facilities supported by the Engineering Research Center Program of the National Science Foundation and DOE under NSF Award Number EEC and the CURENT Industry Partnership Program. 26