Learning Lessons from the Past Power System Blackouts and using Advanced System Technologies to prevent future ones

Transcription

1 Learning Lessons from the Past Power System Blackouts and using Advanced System Technologies to prevent future ones Presented by: Bharat Bhargava Consulting Engineer Advanced Power System Technologies, Inc. Copyright 2016 Advanced Power System Technologies, Inc. All Rights Reserved 1

2 PRESENTATION OUTLINE Project Objectives Review of some Past Blackouts November 9/10, 1965 Northeast US August 10, WECC Western US August 14, 2003 Northeast US and Canada November 4, 2006 Europe September 8, 2011 WECC SDGE, CFE and IID July 30/31, 2012 Northern, Eastern and Northeastern India Steps we can take to avoid the next big one Application of New Technologies SPMS 2

3 Power System Blackouts Power System Blackouts though rare but do occur result in massive dislocation of services can be a threat to life result in excessive economic losses should be prevented as much as possible If they occur, the power should be restored as soon as possible can be and should be avoided / reduced by using New Advanced Technologies 3

4 TABLE I Impact and Restoration Times of some Power Grid Blackouts(1) Date Area Load lost Number of Restoration MW People Affected Time (Hours) Remarks 11/9/1965 North America 20, M 13 7/13/1977 US- NY 6,000 9 M 13 12/22/1982` US (California) 12,350 2 M 07/2-3, 1996 US-NW 11, M 13 8/10/1996 US- Western 28, M 9 6/25/1998 US - NW M 19 3/11/1999 Brazil 90 M 8/14/2003 NE America 61, M 48+ 9/13/2003 Italy 57 M 9+ 9/13/2003 Sweden +Denmark 5 M 5 11/4/2006 Europe (2) 15,000 5 M 2 11/10/2009 Brazil, Paraguay 17, M 7 2/4/2011 Brazil 53 M 8 9/11/2011 US -SD 4,300 5 M 12 7/30/2012 India 300+ M 12 Est. 7/31/2012 India 660 M 12 Est. (1) Some of this information has been extracted from EPRI documents (2) Europe islanded into three islands and controlled the frequency decay in the low frequency island by automatic under-frequency load shedding (3) The blackouts shown in yellow have been discussed in this report 4

5 Major Power System Blackouts in Last Fifty Years Some of the major Blackouts Northeast US November 9/10, 1965 Western US (WECC) August 10, 1996 Northeast US / Canada August 14, 2003 Europe November 4, 2006 San Diego / CFE / IID (WECC) September 8, 2011 Northern India - July 30, 2012 Northern, Eastern and Northeastern India July 31, 2012 Northeastern US/Canada (2003) was longest over 48 hours Northern, Eastern and Northeastern blackouts in India (2012) impacted most people over 600 million 5

6 Eastern Interconnection Blackout November 9/10,

7 Northeast US Disturbance November 9/10, 1965 First Major wide area System Blackout Complete report submitted to President on December 6, 1965 Restoration helped by a gas turbine in New York area The entire system was restored within nine hours New York restored in less than two hours From: REPORT TO THE PRESIDENT BY THE FEDERAL POWER COMMISSION ON THE POWER FAILURE IN THE NORTHEASTERN UNITED STATES AND THE PROVINCE OF ONTARIO ON NOVEMBER 9-10,

8 Power System Blackouts November 9/10, 1965 Caused by overloading of lines out of Niagara and some faulty relay settings Some areas restored within fifteen minutes A gas turbine in New York area enabled power restoration very fast Total time taken to restore power about 9 hours From: REPORT TO THE PRESIDENT BY THE FEDERAL POWER COMMISSION ON THE POWER FAILURE IN THE NORTHEASTERN UNITED STATES AND THE PROVINCE OF ONTARIO ON NOVEMBER 9-10,

9 Western Electric Coordinating Council Disturbance August 10,

10 WECC System Disturbance August 10, 1996 Highly stressed system conditions and hot weather Transmission lines overload and sag into trees and trip one after the other in Pacific Northwest As the lines trip, system weakens but Operators do not have wide area situational awareness and system stress information Generators are over stressed and trip sequentially System continues to see increase in stress but operators can not monitor and hence no corrective action is taken Large power swings occur between Northwest and Southwest leading to system separation System splits into multiple islands 10

11 WECC August 10, 1996 Event As a result of this disturbance, the WECC system split in to four islands with major loads being dropped in Arizona and California TOTAL WECC System IMPACTS Load lost: 30,489 MW Generation lost: 27,269 MW Customers affected: 7.49 million Outage time: Up to 9 hours 11

12 Growing power oscillations seen at California - Oregon border at Malin substation on August 10, Observed COI Power(Dittmer Control Center) Time in Seconds August 10, COI Power Oscillations at Malin [Source: BPA] 12

13 Growing power oscillations seen at California - Oregon border at Malin substation on August 10, 1996 Oscillations growth increases when a capacitor bank is switched in Malin substation to provide voltage support August 10, 1996 Oscillation Malin 500kV Voltage [Source: BPA] 13

14 Growing oscillations seen on 500 kv busses at various substations August 10, 1996 Oscillation Malin 500kV Voltage [Simulations] A very closely matched simulation conducted by WECC Modeling and Validation Group Switching capacitor bank at Malin causes oscillations to grow faster both in reality and in simulations 14

15 Growing oscillations seen on 500 kv busses at various substations Simulations Maximum voltage amplitude oscillations occur at Malin substation August 10, 1996 Oscillation in 500 kv bus voltages at Malin and other susbtations [Simulations] 15

16 Estimating damping from the above recorded chart 1. It is easy to calculate damping, if the oscillations are of single mode 2. Plot the x and y points, must ensure to include the max and min for each cycle. Minimum two points per cycle 3. Calculate the amplitude of each cycle and the time when the peaks occur 4. Determine the number of cycles (n) and the start (T1) and the end time (T2). 5. Calculate the frequency of oscillations f = Cycles/Time = n/(t2-t1) 6. Calculate A2/A1 ratio for each cycle. If A2 is larger than A1, the oscillations are growing and if A2 is less than A1, the oscillations are damping 7. Take natural log of each ratio A2/A1 8. Calculate the average of natural logs, that is (ln1+ln2+ln3+ln4+ln5)/5 for five cycles. This is the average damping constant (z). If the oscillations are damped this average will be negative, but if the oscillations are growing it will be positive. This constant is known as Damping Ratio that is damping per cycle. 9. The Damping Constant is the damping per second and can be calculated by multiplying Damping Ratio (z) by frequency that is a = z * f 16

17 Estimating growth rate of oscillations before and after capacitor switching August 10, 1996 Oscillation Malin 500kV Voltage [Source: BPA] 17

18 Estimating growth rate of oscillations before and after capacitor switching k i l o V o l t s Notice the increased growth of oscillations after a capacitor bank is switched in Malin area August 10, 1996 Oscillation Estimated Malin 500kV Voltage Time in seconds 18

19 Analysis of Growing power oscillations seen at California - Oregon border at Malin substation on August 10, 1996 Time Voltage Cycle Frequency Amplitude A2/A1 Ratio z a Average Remarks Numbe Seconds kv r f A ln(a2/a1) f*z Damping Before Capacitor switching After Capacitor switching Notice the increase in growth of oscillations after a capacitor bank is switched in Malin area 2. System is already unstable, but switching capacitor made it worse 19

20 kv Oscillation Growth before and after Capacitor Switching in Malin area (From Excel sheet using analysis results) X = A.e -at Sin(wt-q), where A1 = 10 Ac = 525 a= & = 1.57 f= 0.25 Hz q= 0 t = Time in seconds Seconds 20

21 Eastern Interconnection US/Canada Blackout August 14,

22 Western System (WECC) US Disturbance - August 10, 1996 What can we learn Disturbance caused by increasing stress and lack of voltage support at critical (Malin & Captain Jack) substations Increased stress resulted in reduced system damping and growing oscillations Cross tripping islanding scheme was put back into service Models were inaccurate in predicting system behavior Considerable efforts spent by WECC Modeling and Validation group helped in improving simulations and matching actual performance and the modelled performance Wide area monitoring using SPMS could have alerted the operators of the increasing stress 22

23 Northeast US/Canada blackout of August 14,

24 Northeast US-Canada System Disturbance - August 14, 2003 Hot weather and highly stressed system conditions Transmission lines overload and sag into trees and trip one after the other in Michigan and Ohio System stress continues to increase and the system weakens but operators do not have wide area situational awareness Generators are over stressed and trip sequentially System continues to see increase in stress but no corrective action is taken Large power swings occur between Midwest and Eastern US and Canada System separates resulting in blacking out part of Northeast and New York 24

25 Northeast US-Canada System Disturbance - August 14, 2003 Date/ Time of Occurrence August 14, 2003; Impacted Area North Eastern US and Canada Number of People Impacted 50 Million Load Lost 61.8 GW Hours for Restoration 48 + Estimated cost $ 50 Billion 25

26 Generation, Demand, and Interregional Power Flows on August 14 at 15:05 EDT (From NERC Report) 26

27 At 4:13 PM Cascading Sequence Essentially Complete (From NERC Report) 27

28 Area Affected by the Blackout (From NERC Report) 28

29 Growing angle separation between Cleveland and West Michigan Point of no return From NASPI RAPIR Report 29

30 Area hit by August 14, 2003 blackout 30

31 Some lessons we can learn from the blackouts/restoration efforts New York (2003) restoration challenges Restoration took about 48 hours Hydro units at Pumped Storage Gilboa Power Plant were available within 20 minutes, but power could not be restored for three hours because of voltage mismatch at a substation PJM/NY system could not be reclosed because of voltage mismatch and delayed restoration effort Step by step procedure for system restoration could have helped. Wide Area Monitoring & Control Technology such as Synchronized Phasor Measurement Systems could have helped in avoiding cascading and in restoration 31

32 Gilboa Hydro Power Plant in New York State (From NERC Report) 32

33 European System Blackout November 4,

34 European System Disturbance November 4, 2006 Highly stressed system conditions because of heavy wind generation in Northeast and heavy winter load in Southwest Europe Two major 400 kv lines opened for a planned outage Opening the lines resulted in overloading and overstressing the transmission system (Increased Wide Area System stress) Corrective action resulted in stressing the system more separating the System into three islands System disturbance controlled by dropping approximately MW load thru Under Frequency relays Large amount of wind generation is dropped to control frequency System normalized in about two hours 34

35 November 4, 2006 European disturbance Frequency Plot for three islanded areas Source : UCTE 35

36 Map showing three islands of the European system (From UTCE Reports) 36

37 Multiple Trials for Synchronizing three island 37

38 San Diego Gas & Electric, IID and CFE System Blackout on September 8,

39 San Diego - WECC System Disturbance September 8, 2011 San Diego system imports power on two major import paths Hassyampa N. Gila- Imperial Valley Miguel 500 kv South of SONGS Five 230 kv lines (Path 44) Power also flows thru the underlying 220/115/92 kv system from Devers bus to IID and Western Administration Lower Colorado SDG & E has established individual path ratings Hassyampa N. Gila 2200 MW Path 44 south of SONGS 1800 MW SDG & E monitors these thru the EMS / SCADA system SDGE may have been operating beyond safe operational limit 39

40 San Diego - WECC Disturbance September 8, 2011 Sequence of Events Heavily loaded and stressed system conditions and hot weather (115 degrees in IID) Safe operation and (N-1) criteria requires that loss of a path should not result in exceeding the normal rating of other paths. RTCA are generally employed to ensure that the system is operating with in safe operating region The Hassyampa N. Gila line tripped at 15:26 hrs while carrying 1394 MW load due to an operational error Loss of this line resulted in increase of power flow on Path 44 from 1302 MW to 2386 MW which exceeded the path 44 rating of 2200 MW. This indicates that SDGE was operating beyond the safe operational limit 40

41 San Diego - WECC Disturbance September 8, 2011 Sequence of Events - 2 IID flows also increased from 90 MW to 240 MW and resulted in overloading the IID transformers which tripped and increased power flow on path 44 to 2600 MW No action was taken by SDG&E, Cal ISO or WECC to reduce loading on path 44 after H-NG line trip Increased Loading on path 44 also resulted in low voltages in CFE area and tripping of generating units in CFE system which increased loading on path 44 above 3200 MW (15:32:385) Path 44 has relay settings at SCE end that isolate the SDGE, IID & CFE system if the current exceeds 8000 amps or 3186 MVA Power flows continued to stay above 3200 MW and resulted in separation at SONGS. 41

42 Power flow on path 44 after Hassyampa N. Gila line (from FERC/NERC Report) Source: NERC Phase Angle Report May,

43 Power flows to SDG&E, CFE and IID before the N. Gila-Hassyampa line trip SCE 1800 MW (30 deg.) 1302 MW Path 44 (2200 MW) Relay operation set at 3186 MW SONGS SDG SDGE & E / CFE CFE IID Devers busses 90 MW (239 MW) N.Gila 1397 MW (20 deg. ) (1800 MW) Arizona APS/SRP 43

44 Power flows to SDG&E, CFE and IID after the N.Gila -Hassyampa line trip SCE 2000 MW (30 deg.) 2386 MW On Path 44 (2200 MW) Relay operation set at 3186 MW SONGS SDG SDGE & E / CFE CFE IID Devers busses 184 MW (239 MW) N.Gila MW (20 deg. ) (1800 MW) Arizona APS/SRP 44

45 Power flows /Current on Path 44 E v e n t S e q u e n c e Before line trip After line trip Normal Rating Trip level setting Path 44 Current Power flow in MW Power / Current on Path 44 45

46 Recommendations and Lessons learnt from San Diego Blackout that occurred on September 8, 2011 A Wide Area Monitoring System could have warned SDGE operators to take appropriate action with large angle separation and heavy power flow from north on Path 44. Better Coordination with neighbouring Utility (SCE) for Relay settings is necessary Advanced analysis of operating conditions (RTCA) could have alerted operators that they are operating in an unsafe operating zone Large angle difference across the breaker will block the re-closure Inadequate information on restoration issues, however, the system was restored within twelve hours 46

47 North Indian Blackouts of July 30 / 31,

48 Indian Blackout July 30, 2012 Disturbance occurred at 02:33:00 on July 30, 2012 High loading in Northern Region load of MW Generation of MW Imports of 5836 MW mostly from the Western Region Several 400 kv lines out of service because of Planned outages (10) Unscheduled outages (5) Voltage control (6) Western Northern grid connected on Bina-Gwalior-Agra 400 kv line line loaded to 1355 MW (2.2 SIL) Three 230 kv lines 48

49 Generation, Imports/Exports in Indian Regional Grids before Blackout Northern Region MW MW A Western Region MW MW C 535 MW B Eastern Region MW MW North Eastern Region 1367 MW - 53 MW Total Load : NW 49

50 Indian Blackout July 30, 2012 Amount of Load lost 38,200 MW (Estimated) Areas impacted Northern region People Impacted million Estimated cost $ 6 Billion Time to restore power hours Ties lost A & B NR separated from WR and ER Separation initiated by tripping of Bina-Agra circuit on Zone 3 50

51 Northern India Grid and Regional Interconnections Effected Area Northern Region July 30, 2012 Source: Indian Blackout Investigation Report dated August 16,

52 Indian Regional Blackout July 30, 2012 Frequency Profile in Northern Region 52

53 Indian Blackout July 30, 2012 Increased loading on Bina-Gwalior-Agra caused tripping of this line on Zone 3, the other line was out of service Tripping of Bina-Gwalior-Agra line resulted in tripping all 230 kv WR- NR lines separating WR and NR Separation of WR-NR resulted in tripping of all Ties between ER and NR NR left with a deficit of 5686 MW ( 18 % generation deficiency in NR Resulted in rapid frequency decline and NR blackout Power restored in hours 53

54 Power Flows in Indian Regional Grids at the time of Blackout 5686 MW Northern Region MW MW A Western Region C B Eastern Region 53 MW MW North Eastern Region 1367 MW MW 535 MW Total Load :79,479 MW Load Lost: 38,000 MW Ties Lost: A and B 54

55 Indian Blackout July 31, 2012 Occurred in after-noon about 1:20 Hours, shortly after the first one, while the system was still being put together Amount of Load lost MW Areas impacted Northern region Eastern Region North Eastern region People Impacted million Estimated cost $ 10 Billion (Estimated) Time to restore power hours (Estimated) Ties lost A & C separating NR, ER and NER regions from WR 55

56 Indian Blackout July 31, 2012 Increased loading on Bina-Gwalior-Agra again caused tripping of this line on Zone 3, the other line was out of service Tripping of Bina-Gwalior-Agra line resulted in tripping all 230 kv WR-NR lines separating WR and NR Separation of WR-NR resulted in tripping of all Ties between ER and NR NR left with a deficit of 5686 MW ( 18 % generation deficiency in NR Resulted in rapid frequency decline and NR blackout Power restored in hours 56

57 Northern India Blackout Effected Areas July 31, 2012 (From Indian Blackout Investigation Report ) 57

58 Generation, Imports/Exports in Indian Regional Grids before July 31, 2012 Blackout Northern Region MW +4016MW A Western Region MW MW C B Eastern Region 212 MW MW MW North Eastern Region 1014 MW +212 Total Load : NW 58

59 Generation, Imports/Exports in Indian Regional Grids after July 31, 2012 Blackout Northern Region MW +4016MW 212 MW North Eastern Region 1014 MW +212 A Western Region MW MW C B Eastern Region MW MW Total Load : NW Load Lost : MW Ties Lost : A & C 59

60 Indian Regional Blackout July 31, 2012 Frequency Profile in Northern Region 60

61 Indian Blackout Investigation Report Very comprehensive Investigation report details all sequence of events, system configurations All facts and figures are provided making it easy to review and comment Suggests steps that may be taken to improve situation and prevent future blackouts Original reports blamed the states of withdrawing too much power, but indicates that the system had several lines out resulting in reduced inter-region transfer capability Simulations may be improved for better analysis 61

62 Steps Suggested to avoid Blackouts Better Wide Area Visualization using SPMS Internal External at least adjoining areas/critical areas of Grid should be observable ES and EMS in operation Establishing Limits on power flows Use of RTCA Use of Synchronized Phasor Measurement technology Angle measurements and other metrics Withstand loss of ties and maintain frequency within the acceptable band Monitoring Impedance relays zone encroachment` 62

63 Common causes between the SDG&E and Indian Blackouts - 1 Both Systems have two major inter-connections Hassyampa-North Gila & SCE SONGS Path 44 WR and ER Safe operation requires that loss of one interconnection should not result in exceeding the rating of the other path Readjustment necessary after loss of one tie Both systems were clearly operating beyond safe limits No SOL established or being monitored based on the system conditions No adjustments of loading for line outages Excessive imports compared to local area generation Load 4400 MW, imports of 2698 MW (No possibility of survival when both ties are lost ) Load of imports of 5686 loss of tie-line would result in a frequency decline of about 16 % No frequency control by UFLS 63

64 Common causes between the SDG&E and Indian Blackouts - 2 No Situational Awareness, EMS or RTCA in operation Relay setting resulting in system separation Zone 3 relay settings in India Path 44 overload setting at SONGS No system stress monitoring (Angle separation) No previous analysis to define safe operating regions Power flows (Path 44 ) Angle differences (Bina-Gwalior) No use of advanced technologies for real time dynamics monitoring 64

65 Need for Wide Area Monitoring System - We just can t afford wide area blackouts - We need to operate the power systems efficiently and economically - We now have tools available that can help us manage the grid better Synchronized Phasor Measurement Technology 65

66 What is Synchronized Phasor Measurement Technology (SPMT) A state of the art high-speed grid monitoring system which measures and compares voltages, currents and phase angles between different electric system points simultaneously* and let s you know what the heck is happening to the power system *All Measurements taken at the same precise time 66

67 Synchronized Phasor Measurement System References (1 & 2): Technology 1. Use of Synchronized Phasor measurement System for Enhancing AC-DC Power System Transmission Reliability and Capability by John Ballance, Bharat Bhargava and G. D. Rodriguez, Southern California Edison Co., United States of America presented at the CIGRE General Meeting Session, 2004, Paris, France 2. Dawn of the Grid Synchronization by Damir Novosel, Vahid Madani, Bharat Bhargava, Khoi Vu and Jim Cole published in IEEE Power & Energy Magazine, January/February, 2008, pages

68 What is SPM Technology? Measures positive sequence time stamp voltage and currents at different locations Information is transmitted and collected at a central location Information can be received and processed within six cycles Operators can view following system information Wide Area visibility Wide and Local Area System Stress Static and Dynamic stresses Voltage support at critical locations System dynamics Frequency excursions Oscillations and their damping Zone encroachment 68

69 SPMS Technology Synchronized Phasor Measurement System (SPMS) Capabilities has been identified as the key Technology for avoiding blackouts in the February 2, 2006 DOE report to US House and Senate can provide synchronized event recording during disturbances at multiple points can monitor system dynamics in real-time can enable instantaneous assessment of system performance and stability (Situational Awareness) can assist in avoiding major system disturbances can enable quicker restoration of systems after major system disturbances 69

70 SCE Situational Awareness & Analysis Center (From ) 70

71 Synchronized Phasor Measurement System (SPMS) Capabilities SPMS can Enable increased power transfers on existing paths Potentially enable determination of available transmission capacity in real time monitor: Static/dynamic phase angle limits (system stress) Comparing phase angle measurements with bench marked cases and keeping adequate dynamic margin Modal oscillation frequencies and damping Voltage support at critical locations when operating at large phase angles separations Event reconstruction and model validation 71

72 Historical Background and Worldwide Implementation Invented by Dr. Arun Phadke, Jim Thorp and Mark Adamiak during Applied in Western US during thru an EPRI project Southern California Edison Bonneville Power Administration PGE and others American Electric Power Eastern Interconnection, PJM, NYPA, Others India Major thrust after 2012 Blackouts Major rollouts in China, Russia, England, Europe, Mexico etc. 72

73 Can we avoid the next Big One? Technology is fully matured for application Has been implemented at several locations in US WECC / Peak Reliability Eastern Interconnection PJM, NY ISO, ISO-NE, MISO, Southern Co., Entergy, Dominion, Duke etc. ERCOT (Texas) Some of the above organizations are mostly looking at their own system and not taking advantage of wide area applications, which is a must for successful application North American Synchro Phasor Initiative (NASPI) organization in US is trying to advance the applications 73

74 Are we Prepared to avoid the For successful Technology Application, we need to see the entire Operational Control Area next Big One? Data interchange is essential to understand Complex System Dynamics May need a organization for a specific Control area such as WECC / Peak Reliability Eastern Interconnection PJM, NY ISO, ISO-NE, MISO ERCOT India has an Organization dealing with entire system POSOC The individual organizations need to be aware of things happening in other area and should have access to information 74

75 Are we Prepared to avoid the next Big One? For successful Technology Application, we need to Have Excellent Data Quality with percent reliability Have minimum latency and have data available in less then six cycles for processing Train operators to be able to accept and use the technology Develop faster and efficient processing programs Understand what we need to monitor and at what locations Understand and analyze weak spots of the power systems Develop simulated Events for Training The list is long and will continue to grow on and on Although, we have taken some Baby Steps, the challenges are many, many more 75

76 Thanks, for questions, please to : Bharat@advancedpstinc.com 76