Major Findings on Restoration and Resilience of ...

RELIABILITY | RESILIENCE | SECURITY

Major Findings on Restoration and Resilience of Transmission System during Extreme Weather Based on the 2015-2020 TADS dataSvetlana Ekisheva, Principal Data Science Advisor, NERCProbabilistic Assessment WG WebEx MeetingMay 11, 2021

RELIABILITY | RESILIENCE | SECURITY2

• NERC team (Svetlana Ekisheva, Rachel Rieder, Jack Norris) and prof. Ian Dobson (Iowa State University)

• TADS outage events, algorithm development and enhancement Cascading outageso 2019-2020, together with the EAo Panel presentation at the 2020 IEEE PES GM

Weather-related transmission outage eventso 2020-2021, a paper accepted for the 2021 IEEE PES GM (with M. Lauby)

• Restoration and Resilience study for transmission weather-related events Based on outage, restoration, and resilience curves Panel presentation approved for the 2021 IEEE PES GM Analysis Pilot for the 2020 top transmission events in the 2021 SOR

Project


2015-2020 Weather-Related Transmission Eventsby Extreme Weather Type


• Input: 2015-2020 TADS automatic outages All TADS elements, all voltages

• Outages are grouped together into transmission outage events based on the following criteria: Must occurred in same interconnection Similar outage start times or overlap in duration

o If no overlap in duration, an outage must have a start time within 5 mins of another outage in the evento If an overlap in duration does exist, an outage must start within 1 hour of another outage in the event

Weather related events are defined as follows:o If an event contains one or more outages with either an initiating or sustained cause code of “Fire”,

“Weather, excluding lightning”, “Environmental” or “Lightning”, it is considered a weather related event

Outage grouping algorithm

AC Circuit ACDCBTB DC Circuit TF All Elements0-99 kV 683 7 690100-199 kV 32915 26 14 281 33236200-299 kV 12557 107 135 1377 14176300-399 kV 8945 118 1016 10079400-499 kV 83 83400-599 kV 2247 602 2849500-599 kV 220 220600-799 kV 321 129 450All Voltages 57668 251 452 3412 61783

Voltage Class2015-2020 TADS Automatic Outage Counts


• ~62k automatic outages reported in TADS 2015-2020• Overall, weather-related events comprise 36% of the 35,392

transmission outage events for the 6 years Medium events (10-19 outages): 272 weather events and 21 non-weather

events Large events (20-378 outages): 86 weather events and 1 non-weather

event

• The extreme weather that caused a large weather event was determined from the combination of the following data sources: NERC System Awareness Daily reports NERC Event Analysis reports Public sources: National weather service, news, press releases etc.

• A summary for 22 largest events on the next slide

2015-2020 Weather-Related Events


2015-2020 Top Large Weather-Related Events (368-44 outages)

Year Interconnection Extreme/Severe Weather EventEvent Size

Event Duration (Days)

Miles Affected

MVA Affected

TADS elem affected

2017 Eastern Hurricane Irma 368 19.5 6645 125917 3032016 Eastern Hurricane Matthew 279 59.6 6860 100648 2472018 Eastern Hurricane Michael 200 28.5 4659 62051 1842020 Eastern Hurricane Zeta 153 40.7 3731 56740 1272015 Western Strong wind storms 143 5.9 4844 45578 1172020 Eastern Easter Tornado 116 16.1 2630 42085 1092020 Eastern Hurricane Isaias 108 9.4 1352 43404 872017 Eastern Heavy rain and thunderstorms 103 246.0 3303 39253 862020 Eastern Windstorms 74 22.1 1217 26488 732015 Eastern Wide-spread rains and snowstorm 63 1.5 2141 24118 292018 Eastern Blizzard, Severe thunderstorms an 63 1.7 1336 21076 472019 Western Strong winter storms with high wi 55 10.4 2177 23895 332016 Western Heavy snow and freezing rains 52 0.7 1925 21613 372019 Eastern Storm system with high winds, sno 52 81.0 1835 34435 372020 Eastern Hurricane Laura 49 14.6 791 17604 462018 Eastern Nor'easter 48 7.2 840 16573 412018 Eastern Nor'easter 47 2.8 625.7 19966 222020 Western Wildfires 46 87.2 1617.8 19797 432015 Eastern Strong thunderstorms and tornado 45 4.2 1348.8 19963 372015 Western Strong storms with high winds 44 6.2 671.7 7841 422019 Western Lightning storm 44 8.6 2173.6 27475 402020 Eastern Icestorm 44 20.3 923.2 20175 42


2020 Large Weather Events

Event Start

Event Size

Interconnectio

n

Extreme/Severe Weather Event

MVA Affected

Miles Affected

Event Duration

(Days)

TADS Elements Affected

Max Elements

Out

Element-Days Lost

Max MVA Out

MVA-Days lost

1/12 22 Eastern Thunderstorms 5666 564 6.9 18 13 24.8 3102 7670

3/19 35 Western Snowstorms 12984 905 1.0 21 9 3.0 3031 847

4/12 116 Eastern Easter Tornado 42085 2630 16.1 109 67 142.8 18854 40948

5/30 24 Western Thunderstorms 15825 1230 93.1 24 13 151.2 8689 316302

8/4 108 Eastern Hurricane Isaias 43404 1352 9.4 87 50 78.8 19269 27033

8/10 74 Eastern Windstorms 26488 1217 22.1 73 62 233.7 18864 65693

8/16 27 Western Thunderstorms/High Wind 11715 1170 4.2 25 12 16.4 5426 6222

8/17 21 Western Tropical Storm 7190 694 0.7 19 13 2.7 4761 702

8/27 49 Eastern Hurricane Laura 17604 791 14.6 46 41 176.0 15548 55865

8/31 24 Eastern Thunderstorms/High Wind 5464 375 0.1 24 23 0.3 4899 73

9/7 46 Western Wildfires 19797 1618 87.2 43 33 224.7 16258 47126

9/16 21 Eastern Hurricane Sally 6214 361 3.7 20 16 23.8 4093 6251

10/13 29 Western Windstorms 7444 699 2.6 29 23 16.8 5300 3162

10/28 42 ERCOT Icestorm 59730 1770 22.3 40 39 75.1 55606 78769

10/28 44 Eastern Icestorm 20175 923 20.3 42 27 46.5 12252 23515

10/28 153 Eastern Hurricane Zeta 56740 3731 40.7 127 102 197.0 32406 105788

11/15 33 Eastern High Winds 10140 917 3.1 32 18 16.3 4481 3937

11/17 30 Western Wildfires 5448 888 1.5 22 15 12.0 2190 1795


Large Weather Events by Interconnection and Extreme Weather Type

Extreme Weather Type2015-2020 Large

Transmission Events

Thunderstorm, wind 39

Winter s torm, snow 24

Hurricane 10

Tornado 8

Fire 4

Extreme cold 1

Grand Total 86

• 4 out of the 10 hurricane events occur in 2020• The Quebec Interconnection (QI) had no large weather events • The extreme cold weather type is assigned to only one event

from ERCOT (January 2018) with extreme low temperatures identified as the major cause of the event


Large Weather Events Statistics by Extreme Weather Type

• Overall, the average large event size is 46 outages and the duration is 16.0 days

• The large event size varies from 20 to 368 outages• There is a huge variability in the event duration (from 3 hours to 246 days)


Impact of Large Weather Events by Extreme Weather Type

Total Element-Days Lost Total MVA-Days lost

• The element-day loss and the MVA-day loss are NOT equal #Outages*Event Duration or MVA Affected*Event Duration

• These products are an over-estimation because not all elements/MVA are out for the whole duration of the event

• A methodology explained below allows to calculate these and several other statistics precisely.


• The previous information does not provide details about what was happening at every moment in time during an event.

• To uncover and study this history, we can use the TADS data to construct an event “resilience” function, similar to a conceptual graph of a resilience event below (DOE-IEEE Technical Report: Resilience Framework, Methods, and Metrics for the Electricity Sector, October 2020)

How to study transmission outage events


Analysis of Transmission Events with Outage, Restoration, and Resilience (Restoration

Performance) Functions

Example: Hurricane Irma (2017)


• From the TADS data we calculate the outage function O(t) and the restoration function R(t) that count the cumulative number of outages and restores, respectively, occurred in the event by time t.

• From these functions, the resilience function is calculated as R(t)-O(t). It is the negative of the number of TADS elements out at time t.

• Analogs of the element-based functions can be calculated as MVA loss, MVA restored, and MVA out functions.

• The element-based and MVA-based functions and graphs allow us to track a history of the transmission event and calculate several important statistics for outage, restoration and resilience processes.

Outage, Restoration, and Resilience Processes


• What story do they tell?

Hurricane Irma: Element-based and MVA-based Curves


Hurricane Irma: Outage Process

• The event started at 10:16 AM on September 10, 2017• The outage process lasted 42 hours at the average outage rate 9.0 outages and ~43,000 MVA

per hour• 25 utilities affected• There were 378 automatic outages: 372 ac circuit outages (306 sustained and 66 momentary) and 6 transformer outages (all sustained) 344 outages initiated by Weather, excluding lightning; 15 by equipment failures; 4 by vegetation and

power system condition (each) and 11 by other causes 312 elements affected

• ~130,000 MVA affected• 6742 circuit-miles affected• Last outage started at 4:21 AM on September 12, 2017 – the element and MVA loss stopped to

increase.


Hurricane Irma: Restoration Process

• The restoration process started immediately due to a momentary outage at 10:16 AM • The first sustained restore occurred at 11:34 AM • The restoration process lasted 467 hours (the full duration of the event)• The outage and restoration process overlapped by 42 hours (the full duration of the outage

process)• The last remaining outage of a 100-199 kV ac circuits (213 MVA) lasted 11.8 days – out of the

19.5 days of the total event duration• 95% of outages were restored for 89.0 Hours (3.7 days)• 95% of MVA were restored for 86.3 Hours (3.6 days)• Unlike the outage function, the restore function is poorly approximated by a simple piece-linear

function


Hurricane Irma: Resilience (Restoration Performance) Curves

• The element-based resilience function is a negative number of transmission elements out at time t

• The MVA-based resilience function is a negative MVA out at time t• The resilience function decreases when outages occur faster than restores and increases when

restores occur faster than outages (particularly, when outage process stops). • The minimum of the element-resilience function, -182, and the minimum of the MVA-resilience

function, -60346 MVA, are attained at 19:22 PM on September 11, 2017, about 33 hours (1.4 days) after the event start. The system stays at this level for 5 minutes.

• The maximum #elements out ~48% of the total outages; the maximum MVA out ~46% of the total MVA affected

• 388.6 element-days lost (the area between the x (time) axis and the elements-out curve) • 125,581 MVA-days lost (the area between the x (time) axis and the MVA-out curve)


Assessment of the 2020 Large Weather Events


Top Weather Events in 2020: Hurricane Zeta and Texas Ice storm


Top Weather Events in 2020: Windstorm (the EI) and Thunderstorm (the WI)

• The windstorm had the largest element-day loss in 2020 (234 element-days lost)

• The thunderstorm had the largest MVA-day loss in 2020 (~316k MVA-days lost) The longest 2020 event A 500 kV ac circuit outage with a 30-day duration contributed more than 23% to

the total MVA-day loss


Cumulative Impact of Large Events that started on October 28, 2020

• Two of the three events were the 2020 largest (with respect to #outages and MVA affected)

• The cumulate impact was huge but not necessarily equal “the sum of impacts” due to: Differences in the start time of the events Restoration processes started almost immediately (<40 min for each event)

• For each event, the restorations of 95% of elements and 95% of MVA took less than 27% of the event duration


Selected Statistics for Outage, Restoration, and Resilience (Restoration Performance) Processes

for 2015-2020 Large Weather-Related Events


Outages and Restorations

• Typically, the outage process is much shorter than the restoration process

• Unlike a conceptual resilience curve, for a real life transmission event the outage and restoration processes overlap

• The average outage process durations for different extreme weather types are similar except hurricanes, which is ~2-4 times longer.

• The average time to first restore is 47 minutes• The maximum time to first restore is 5 hours 9 min

(hurricane Matthew, 2016)• The outage rates are similar among all types.• The MVA loss rate is the highest for Fire (ac circuits

of higher voltages from WECC are affected)


Maximum Simultaneous Number of Elements Out and MVA Out

• The nadir of an element-based resilience (restoration performance) curve is attained during the outage process – for large events typically soon after the event started.

• MaxElementOut and MaxMVAOut usually happen at the same time (but not necessarily).

• The nadirs are highly correlated with and can be estimated (predicted) from the event size (the next slide).

• Hurricanes cause events with highest maximum number of elements out and MVA out.


• Expected Max#ElementsOut(N)=0.5*N+2.4, where N=the large event size

Maximum Number of Elements Out


Element-Days Lost and MVA-days Lost

• The area between the time axis and a element-resilience curve (the MVA-resilience curve) equals the event element-day loss (the MVA-day loss)

• On average, the losses are higher for hurricanes• The top two events overs the 6 years are thunderstorms: in the

East in 2017 (element-days loss) and in the West in 2020 (MVA-days loss)

• For majority of events a contribution of a final stage of an event to the loss is relatively small since only few (sometimes one) outages remain unrestored before the event ends.

• Can we estimate a time to “almost” restore the system?


• The average times to restore 95% of outages and 95% of MVA are, on average, much shorter than the event duration for all extreme weather types

Events “Almost” Restored


• On average, for the large events the time to restore 95% of outages in an event takes 54% of event duration (from the event start)

• The metric is negatively correlated with the event duration and the event size

95% of Outages Restored


• On average, for the large events the time to restore 95% of MVA affected by the event takes 56% of event duration (from the event start)

• The metric is negatively correlated with the event duration and the event size

95% of MVA Restored


Major Findings on Restoration and Resilience of ...

Documents