NPCC Natural Gas Disruption Risk Assessment Background. Summer 2017

Transcription

1 Background Reliance on natural gas to produce electricity in Northeast Power Coordinating Council (NPCC) Region has been increasing since The disruption of natural gas pipeline transportation capability due to pipeline maintenance (or other factors 1 ) can result in an extended reduction of natural gas supply to the northeast. The associated risks to the reliable operation of the Bulk Power System is mitigated by gasfired generation with dual fuel capabilities (oil and natural gas). Pursuant to NPCC Corporate Goal IV-2 : Assess risks associated with cross sector dependencies and single points of disruptions, 2 this assessment estimates the loss of load expectation (LOLE) and expected use of Emergency Operating Procedures (EOPs) for NPCC (and neighboring Regions) due to unavailable gas-fired generation following reduced natural gas pipeline transportation capability to the northeast Region for a sustained, consecutive period using a multi-area probabilistic reliability approach. Summer The summer assessment estimates the impact of natural gas disruptions on the NPCC and neighboring PJM Area during the NPCC summer peak period. This study was based on the Base Case assumptions used in the NPCC Reliability Assessment for Summer 3 and utilizes the same GE MARS database for its simulations. GE Energy Consulting was retained by NPCC to conduct the simulations. Natural gas disruptions were modeled by reducing the available capacity of natural gas-only generators. Dual-fuel generators were not modified because it was assumed they could switch fuel during the disruption. Table 1 summarizes the number of generators and total capacity of natural gas units assumed in the GE MARS model for August. 1 Disruptions of the natural gas transmission and distribution systems can be caused by many factors, including outages related to related computer technology failures, communication infrastructure and supervisory control and data acquisition systems problems, and pipeline safety considerations. 2 See: Corporate_Goals.pdf. 3 See: Summer.pdf - Appendix VIII. NPCC June 30, Page 1

2 Table 1. Number of natural gas-only powered generators modeled in the NPCC Summer Assessment Area Number of generators Summer (August) capacity (MW) HQ 0 0 MT NE 35 8,567 NY 40 3,815 ON 57 6,397 PJM ,603 Total ,901 Gas disruptions were simulated by a proportional derating of gas-only generator capacity across all NPCC and PJM Areas in 10% increments, as shown in Table 2. Area Table 2. Gas-only generator capacity simulated in each scenario (MW) Base case 10% derated 20% derated 30% derated 40% derated 50% derated HQ MT NE 8,567 7,710 6,853 5,997 5,140 4,283 NY 3,815 3,434 3,052 2,671 2,289 1,908 ON 6,397 5,757 5,118 4,478 3,838 3,199 PJM 74,603 67,143 59,682 52,222 44,762 37,302 Total 93,901 84,511 75,121 65,731 56,340 46,950 The capacity derating in each scenario was applied for different lengths of time. All the disruptions were centered around August 9, the time of the GE MARS program estimated NPCC summer peak load. The disruptions were simulated in one week increments, as summarized in Table 3 and represented in Figure 1. Table 3. Duration of outages Disruption Duration Outage dates 1 week August 6 August 12 2 weeks August 2 August 15 3 weeks July 30 August 19 4 weeks July 26 August 22 NPCC June 30, Page 2

3 Figure 1. NPCC daily peak for the summer assessment, summer peak (red) and study windows (gray) The NPCC GE MARS databases was simulated for 1,000 replications for each of the scenarios described above. For each one of this 1,000 replications, the GE MARS program simulates a pattern of outages for generations and area interfaces, as indicated in the input provided by the Areas. Within each replication, GE MARS considers each one of the seven load levels in the NPCC database. For each combination of replication and load level, the GE MARS programs utilizes generators, contracts, EOPs, etc. to minimize shortages in each NPCC Area and subarea. Once the modeling is done, the output metrics are calculated. When looking at daily loss-of-load metrics, GE MARS examines for each combination of replication and load level and for each day whether an area has a shortage. This is considered to happen when, within a day, any subarea in an Area has a deficiency that cannot be solved through any of the means listed above. One or more deficient hours in a day in one or more subareas count as a single daily deficiency. Total metrics are then calculated for all 8,760 hours in the year. For a single hour, the GE MARS program sums the number of replications and load levels that present a deficiency. Each event is weighted by the load level probability (using the probabilities shown in Table 4) and divided by the number of replications. Table 4. Load forecast uncertainty probabilities in the NPCC GE MARS model Level 1 Level 2 Level 3 Level 4 Level 5 Level 6 Level NPCC June 30, Page 3

4 For example, if at 7 PM on July 10 th the first replication for an Area has a shortage for load level 1, the daily metric is increased by /1000. If both load levels 1 and 2 had a shortage, that metric would be increased by ( )/1000 and so on. When all the daily metrics have been calculated, they then can be aggregated. For instance, the NPCC summer and winter assessments (and this study) report these metrics in a monthly and seasonal basis, while the NPCC Long Range Adequacy Overview reports annual numbers. The estimated use of emergency operating procedures (EOPs) for the expected load level 4 for the May September summer period are summarized below, by Area. Quebec The results for the Quebec Area did not indicate an estimated need for EOP usage (<0.001 days/period) for all the cases simulated for the May September summer period, expected load level. Maritimes Figure 2 and Table 5 show the increase in estimated EOP use across all the cases simulated for the Maritimes Area. Occurrences greater than 0.5 days/period are highlighted. 5 None of the disruption cases resulted in a > 0.5 days/period estimated need to activate EOP steps. 4 The expected load level was based on the probability-weighted average of the seven load levels simulated. 5 Note: likelihoods of less than 0.5 days/period are not considered significant. NPCC June 30, Page 4

5 Figure 2. Estimated use of EOPs for the Maritimes Area May September summer period - Expected Load Level NPCC June 30, Page 5

6 Table 5. Estimated use of EOPs for the Maritimes Area May September summer period - Expected Load Level Gas-only unit derated Scenarios Disruption EOP 0% 10% 20% 30% 40% 50% Activation of DR/SCR Reduce 30-min Reserve week Initiate Interruptible Loads Reduce 10-min Reserve Activation of DR/SCR Reduce 30-min Reserve weeks Initiate Interruptible Loads Reduce 10-min Reserve Activation of DR/SCR Reduce 30-min Reserve weeks Initiate Interruptible Loads Reduce 10-min Reserve Activation of DR/SCR Reduce 30-min Reserve weeks Initiate Interruptible Loads Reduce 10-min Reserve New England Figure 3 and Table 6 show the increase in estimated EOP use across all the cases simulated for the New England Area. Occurrences greater than 0.5 days/period are highlighted. 5 For a 1-week disruption, the 10% derated scenario increased the estimated EOP usage by 15-20%, with that number growing to 45-50% for the 20% derated scenario. The estimated EOP usage roughly multiplied by a factor a 2, 2.5 and 3.5 for the derated scenarios of 30%, 40% and 50%, respectively. All of the disruption cases with a one week or greater disruption duration for the simulated 50% derated scenario resulted in a > 0.5 days/period estimated need to activate the Disconnect Load EOP. The 4- week disruption duration case for the 40% derated scenario also resulted in a > 0.5 days/period estimated need to activate the Disconnect Load EOP. NPCC June 30, Page 6

7 Figure 3. Estimated use of EOPs for the New England Area May September summer period - Expected Load Level NPCC June 30, Page 7

8 Table 6. Estimated use of EOPs for the New England Area May September summer period - Expected Load Level Gas-only unit derated Scenarios Disruption EOP 0% 10% 20% 30% 40% 50% Activation of DR/SCR Reduce 30-min Reserve week Initiate Interruptible Loads Reduce 10-min Reserve Appeals Disconnect Load Activation of DR/SCR Reduce 30-min Reserve weeks Initiate Interruptible Loads Reduce 10-min Reserve Appeals Disconnect Load Activation of DR/SCR Reduce 30-min Reserve weeks Initiate Interruptible Loads Reduce 10-min Reserve Appeals Disconnect Load Activation of DR/SCR Reduce 30-min Reserve weeks Initiate Interruptible Loads Reduce 10-min Reserve Appeals Disconnect Load New York Figure 4 and Table 7 show the increase in estimated EOP use across all the cases simulated for the New York Area. Occurrences greater than 0.5 days/period are highlighted. 5 For a 1 week disruption, the 10% derated scenario increased EOP usage 20-30%, with that number growing to 50-80% for the 20% derated scenario. EOP usage roughly multiplied by a factor a 2.5 and 5 for the derated scenarios of 30% and 40%, respectively. Note: New York initiates the Appeals EOP prior to the Reducing 10-min Reserve EOP. None of the simulated derated scenarios resulted in a > 0.5 days/period estimated need to activate the Disconnect Load EOP. NPCC June 30, Page 8

9 Figure 4. Estimated use of EOPs for the New York Area May September summer period - Expected Load Level NPCC June 30, Page 9

10 Table 7. Estimated use of EOPs for the New York Area May September summer period - Expected Load Level Gas-only unit derated Scenarios Disruption EOP 0% 10% 20% 30% 40% 50% Activation of DR/SCR Reduce 30-min Reserve week Initiate Interruptible Loads Reduce 10-min Reserve Appeals Disconnect Load Activation of DR/SCR Reduce 30-min Reserve weeks Initiate Interruptible Loads Reduce 10-min Reserve Appeals Disconnect Load Activation of DR/SCR Reduce 30-min Reserve weeks Initiate Interruptible Loads Reduce 10-min Reserve Appeals Disconnect Load Activation of DR/SCR Reduce 30-min Reserve weeks Initiate Interruptible Loads Reduce 10-min Reserve Appeals Disconnect Load Ontario Figure 5 and Table 8 shows the increase in estimated EOP use across all the cases simulated for the Ontario Area. Occurrences greater than 0.5 days/period are highlighted. 5 While the estimated EOP usage increases, it is small for derated scenarios below 40%, except for Demand Response EOP activation. None of the simulated derated scenarios resulted in a > 0.5 days/period estimated need to activate the Disconnect Load EOP. NPCC June 30, Page 10

11 Figure 5. Estimated use of EOPs for the Ontario Area May September summer period - Expected Load Level NPCC June 30, Page 11

12 Table 8. Estimated use of EOPs for the Ontario Area May September summer period - Expected Load Level Gas-only unit derated Scenarios Disruption EOP 0% 10% 20% 30% 40% 50% Activation of DR/SCR Reduce 30-min Reserve week Initiate Interruptible Loads Reduce 10-min Reserve Appeals Disconnect Load Activation of DR/SCR Reduce 30-min Reserve weeks Initiate Interruptible Loads Reduce 10-min Reserve Appeals Disconnect Load Activation of DR/SCR Reduce 30-min Reserve weeks Initiate Interruptible Loads Reduce 10-min Reserve Appeals Disconnect Load Activation of DR/SCR Reduce 30-min Reserve weeks Initiate Interruptible Loads Reduce 10-min Reserve Appeals Disconnect Load PJM Figure 6 and Table 9 shows the increase in estimated EOP use across all the cases simulated for the PJM Area. Occurrences greater than 0.5 days/period are highlighted. 5 Estimated EOP usage increases primarily for the activation of the Demand Response EOP. Other EOP usage is small for the derated scenarios below 50%. Only the simulated 50% derated scenario for a 4-week disruption resulted in a > 0.5 days/period estimated need to activate the Disconnect Load EOP. NPCC June 30, Page 12

13 Figure 6. Estimated use of EOPs for the PJM Area May September summer period - Expected Load Level NPCC June 30, Page 13

14 Table 9. Estimated use of EOPs for the PJM Area May September summer period - Expected Load Level Gas-only unit derated Scenarios Disruption EOP 0% 10% 20% 30% 40% 50% Activation of DR/SCR Reduce 30-min Reserve week Initiate Interruptible Loads Reduce 10-min Reserve Appeals Disconnect Load Activation of DR/SCR Reduce 30-min Reserve weeks Initiate Interruptible Loads Reduce 10-min Reserve Appeals Disconnect Load Activation of DR/SCR Reduce 30-min Reserve weeks Initiate Interruptible Loads Reduce 10-min Reserve Appeals Disconnect Load Activation of DR/SCR Reduce 30-min Reserve weeks Initiate Interruptible Loads Reduce 10-min Reserve Appeals Disconnect Load Observations Figure 7 and Figure 8 depict the estimated EOP usage across New England, New York and Ontario across all scenarios. The New England Area shows a gradual increase in estimated EOP use for more severe gas disruptions cases. The New York, Ontario and PJM Areas primarily experience an increase in the estimated Activation of DR/SCR EOP use below the 30% derated scenario. For all three of these Areas and across all EOPs, the results are similar for disruptions that last 2 and 3 weeks. Only the cases with disruptions lasting four weeks have a significant increase in EOP use. NPCC June 30, Page 14

15 Figure 7. Estimated use of EOPs across all Areas May September summer period - Expected Load Level NPCC June 30, Page 15

16 Figure 8. Estimated use of EOPs across all Areas May September summer period - Expected Load Level NPCC June 30, Page 16

17 Next Steps The summer assessment will be expanded to include cases simulating the May September 2019 summer period (expected load level) and the May September 2021 summer period (expected load level) based on the GE MARS model being developed for the NPCC Long Range Adequacy Overview. The model for these future years will be consistent with the data being reported for the NERC Long-Term Reliability Assessment. The results of the risk assessment will be analyzed to compare how the assumed scenarios of natural gas disruptions to the northeast impact the estimated use of EOPs for the future NPCC and PJM Area load and capacity mix Winter The winter assessment estimates the impact of natural gas disruptions on the NPCC and neighboring PJM Area during the NPCC winter peak period. This study was based on the Base Case assumptions used in the NPCC Reliability Assessment for Winter 2016/17 6 and utilizes the same GE MARS database for its simulations. GE Energy Consulting was retained by NPCC to conduct the simulations. Natural gas disruptions were modeled by reducing the available capacity of natural gas-only generators. Dual-fuel generators were not modified because it was assumed they could switch fuel during the disruption. Table 10 summarizes the number of generators and total capacity of natural gas units assumed in the GE MARS model for January. 6 See: Final_Report_including_CP-8.pdf - Appendix VIII. NPCC June 30, Page 17

18 Table 10. Number of natural gas-only powered generators modeled in the NPCC 2016/17 Winter Assessment Area Number of generators Winter (January) capacity (MW) HQ 0 0 MT NE 29 6,003 NY 37 3,775 ON 57 7,073 PJM ,093 Total ,479 Gas disruptions were simulated by a proportional derating of gas-only generator capacity across all NPCC and PJM Areas in 10% increments, as shown in Table 11. Area Table 11. Gas-only generator capacity simulated in each scenario (MW) Base case 10% derated 20% derated 30% derated 40% derated 50% derated HQ MT NE 6,003 5,403 4,802 4,202 3,602 3,001 NY 3,775 3,398 3,020 2,643 2,265 1,888 ON 7,073 6,366 5,658 4,951 4,244 3,536 PJM 78,093 70,284 62,474 54,665 46,856 39,047 Total 95,479 85,931 76,383 66,835 57,287 47,739 The capacity derating in each scenario was applied for different lengths of time. All the disruptions were centered around January 12, the time of the GE MARS program estimated NPCC winter peak load. The disruptions were simulated in one week increments, as summarized in Table 12 and represented in Figure 9. Table 12. Duration of outages Disruption Duration Outage dates 1 week January 9, January 15, 2 weeks January 5, January 18, 3 weeks January 2, January 22, 4 weeks December 29, 2016 January 25, NPCC June 30, Page 18

19 Figure 9. NPCC daily peak for the winter assessment, winter peak (red) and study windows (gray) The NPCC GE MARS databases was simulated for 1,000 replications for each of the scenarios described above. For each one of these 1,000 replications, the GE MARS programs simulates a pattern of outages for generations and area interfaces, as indicated in the input provided by the Areas. Within each replication, GE MARS considers each one of the seven load levels in the NPCC database. For each combination of replication and load level, the GE MARS programs utilizes generators, contracts, EOPs, etc. to minimize shortages in each NPCC Area and subarea. Once the modeling is done, the output metrics are calculated. When looking at daily loss-of-load metrics, GE MARS examines for each combination of replication and load level and for each day whether an area has a shortage. This is considered to happen when, within a day, any subarea in an Area has a deficiency that cannot be solved through any of the means listed above. One or more deficient hours in a day in one or more subareas count as a single daily deficiency. Total metrics are then calculated for all 8,760 hours in the year. For a single hour, the GE MARS program sums the number of replications and load levels that present a deficiency. Each event is weighted by the load level probability (using the probabilities Table 13) and divided by the number of replications. Table 13. Load forecast uncertainty probabilities in the NPCC GE MARS model Level 1 Level 2 Level 3 Level 4 Level 5 Level 6 Level For example, if at 7 PM on January 10 th the first replication has a shortage for load level 1, the daily metric is increased by /1000. If both load levels 1 and 2 had a shortage, that metric would be increased by ( )/1000 and so on. NPCC June 30, Page 19

20 When all the daily metrics have been calculated, then they can be aggregated. For instance, the summer and winter assessments (and this study) report these metrics in a monthly and seasonal basis, while the NPCC Long Range Adequacy Overview reports annual numbers. The estimated use of emergency operating procedures (EOPs) for the expected load level 7 for the November 2016 March winter period are summarized below, by Area. Quebec The results for the Quebec Area indicated a minimal estimated need for EOP usage for all the cases simulated for the November 2016 March winter period, expected load level, as shown in Table The expected load level was based on the probability-weighted average of the seven load levels simulated. NPCC June 30, Page 20

21 Table 14. Estimated use of EOPs for the Quebec Area November 2016 March winter period - Expected Load Level Gas-only unit derated Scenarios Disruption EOP 0% 10% 20% 30% 40% 50% Activation of DR/SCR Reduce 30-min Reserve week Initiate Interruptible Loads Reduce 10-min Reserve Activation of DR/SCR Reduce 30-min Reserve weeks Initiate Interruptible Loads Reduce 10-min Reserve Activation of DR/SCR Reduce 30-min Reserve weeks Initiate Interruptible Loads Reduce 10-min Reserve Activation of DR/SCR Reduce 30-min Reserve weeks Initiate Interruptible Loads Reduce 10-min Reserve Maritimes Figure 10 and Table 15 show the increase in estimated EOP use across all the cases simulated for the Maritimes Area. Occurrences greater than 0.5 days/period are highlighted. 8 None of the disruption cases resulted in a > 0.5 days/period estimated need to activate EOP steps. 8 Note: likelihoods of less than 0.5 days/period are not considered significant. NPCC June 30, Page 21

22 Figure 10. Estimated use of EOPs for the Maritimes Area November 2016 March winter period - Expected Load Level Table 15. Estimated use of EOPs for the Maritimes Area November 2016 March winter period - Expected Load Level NPCC June 30, Page 22

23 Gas-only unit derated Scenarios Disruption EOP 0% 10% 20% 30% 40% 50% Activation of DR/SCR Reduce 30-min Reserve week Initiate Interruptible Loads Reduce 10-min Reserve Activation of DR/SCR Reduce 30-min Reserve weeks Initiate Interruptible Loads Reduce 10-min Reserve Activation of DR/SCR Reduce 30-min Reserve weeks Initiate Interruptible Loads Reduce 10-min Reserve Activation of DR/SCR Reduce 30-min Reserve weeks Initiate Interruptible Loads Reduce 10-min Reserve New England Figure 11 and Table 16 show the increase in estimated EOP use across all the cases simulated for the New England Area. Occurrences greater than 0.5 days/period are highlighted. 5 Only the 50% derated scenario resulted in the estimated need for the Activation of DR and the Reduce 30-min Reserve EOP usage for all the disruption periods. NPCC June 30, Page 23

24 Figure 11. Estimated use of EOPs for the New England Area November 2016 March winter period - Expected Load Level NPCC June 30, Page 24

25 Table 16. Estimated use of EOPs for the New England Area November 2016 March winter period - Expected Load Level Gas-only unit derated Scenarios Disruption EOP 0% 10% 20% 30% 40% 50% Activation of DR/SCR Reduce 30-min Reserve week Initiate Interruptible Loads Reduce 10-min Reserve Appeals Disconnect Load Activation of DR/SCR Reduce 30-min Reserve weeks Initiate Interruptible Loads Reduce 10-min Reserve Appeals Disconnect Load Activation of DR/SCR Reduce 30-min Reserve weeks Initiate Interruptible Loads Reduce 10-min Reserve Appeals Disconnect Load Activation of DR/SCR Reduce 30-min Reserve weeks Initiate Interruptible Loads Reduce 10-min Reserve Appeals Disconnect Load New York The results for the New York Area did not indicate an estimated need for EOP usage (<0.001 days/period) for all the cases simulated for the November 2016 March winter period, expected load level. Ontario Table 17 shows the estimated EOP use across all the cases simulated for the Ontario Area. There are no occurrences greater than 0.5 days/period. 5 NPCC June 30, Page 25

26 Table17. Estimated use of EOPs for the Ontario Area November 2016 March winter period - Expected Load Level Gas-only unit derated Scenarios Disruption EOP 0% 10% 20% 30% 40% 50% Activation of DR/SCR Reduce 30-min Reserve week Initiate Interruptible Loads Reduce 10-min Reserve Activation of DR/SCR Reduce 30-min Reserve weeks Initiate Interruptible Loads Reduce 10-min Reserve Activation of DR/SCR Reduce 30-min Reserve weeks Initiate Interruptible Loads Reduce 10-min Reserve Activation of DR/SCR Reduce 30-min Reserve weeks Initiate Interruptible Loads Reduce 10-min Reserve PJM Figure 12 and Table 18 shows the estimated EOP use across all the cases simulated for the PJM Area. There were no estimated occurrences greater than 0.5 days/period. 5 NPCC June 30, Page 26

27 Figure 12. Estimated use of EOPs for the PJM Area November 2016 March winter period - Expected Load Level NPCC June 30, Page 27

28 Table 18. Estimated use of EOPs for the PJM Area November 2016 March winter period - Expected Load Level Gas-only unit derated Scenarios Disruption EOP 0% 10% 20% 30% 40% 50% Activation of DR/SCR Reduce 30-min Reserve week Initiate Interruptible Loads Reduce 10-min Reserve Appeals Disconnect Load Activation of DR/SCR Reduce 30-min Reserve weeks Initiate Interruptible Loads Reduce 10-min Reserve Appeals Disconnect Load Activation of DR/SCR Reduce 30-min Reserve weeks Initiate Interruptible Loads Reduce 10-min Reserve Appeals Disconnect Load Activation of DR/SCR Reduce 30-min Reserve weeks Initiate Interruptible Loads Reduce 10-min Reserve Appeals Disconnect Load Observations Figure 13 and Figure 14 depict the estimated EOP usage across New England, New York, Ontario and PJM across all scenarios. In general, compared to the results for the summer period, the results for the winter period are less severe for the summer peaking Areas (New York, New England, Ontario and PJM), due primarily to the lower forecasted winter peak loads. NPCC June 30, Page 28

29 Figure 13. Estimated use of EOPs across all Areas November 2016 March winter period - Expected Load Level NPCC June 30, Page 29

30 Figure 14. Estimated use of EOPs across all Areas November 2016 March winter period - Expected Load Level NPCC June 30, Page 30

31 Next Steps The winter assessment will be expanded to include cases simulating the November 2018 March 2019 winter period (expected load level) and the November 2020 March 2021 winter period (expected load level) based on the GE MARS model being developed for the NPCC Long Range Adequacy Overview. The model for these future years will be consistent with the data being reported for the NERC Long- Term Reliability Assessment. The results of the risk assessment will be analyzed to compare how the assumed scenarios of natural gas disruptions to the northeast impact the estimated use of EOPs in the future. NPCC June 30, Page 31