developing an adequacy metric resource adequacy forum technical subcommittee meeting october 16,...

Developing an Adequacy Metric

Resource Adequacy ForumTechnical Subcommittee Meeting

October 16, 2009

October 16, 2009 Resource Adequacy Technical Committee

2

Outline

• Current Adequacy Metric: LOLP

• The Problem with LOLP

• LOLP Subcommittee Suggestions

• Next Steps

Current NW Resource Adequacy Metric: LOLP


4

January 1930

-5000

0

5000

10000

15000

20000

25000

30000

35000

40000

45000

1 27 53 79 105

131

157

183

209

235

261

287

313

339

365

391

417

443

469

495

521

547

573

599

625

651

677

703

729

Hour in Month

Meg

awat

ts

Net Demand NW Thermal NW Hydro Unserved Net Imports

Cold

Hydro Limited

GENESYS Simulation Illustrative Example Only


5

What do we Count?• Ideally, we count “significant” events (those

that we want to avoid)• Energy threshold (or contingency resource)

is 1,200 MW for one day or 28,800 MW-hours from Dec-Mar

• Capacity threshold (or contingency resource) is 3,000 MW in any hour from Dec-Mar and from Jun-Sep


6

Curtailment Events(Peaking problems and energy shortages)

0

1000

2000

3000

4000

Hourly Curtailments Dec-Apr (Not all hours shown)

Curt

aile

d M

egaw

atts

Peak Event > 3,000 MW

Energy Event > 28,800 MW-hrs

Each event has a peak and Each event has a peak and duration.duration.


7

Curtailment Events(non-events not shown)

Reliability Events by Game

0

1000

2000

3000

4000

3 12 12 12 12 12 12 15 15 18 18 22 22 25 25 25 25 25 25 33 33 33 33 34 36 36 36 36 36 36 36 37 39 39 39 39 39 39 39 39 41 44 44 46 46 46

Game

Cur

tailm

ent (

MW

)

Seattle


8

Loss of Load ProbabilitySimulated 300 winters (December through March)

Out of 300 winters, 15 had an average curtailment greater than 10 MW-seasons, which means that the Winter Loss of Load Probability (LOLP) = 15/300 = 5 percent


9

Energy LOLP(Sum of Curtailed Energy Dec-Mar)

0102030405060708090

100

0 1 2 3 4 5 6 7 8 9 10Probability (%)

Mag

nitu

de

(MW

-S

easo

ns)

We plot the average seasonal curtailment for every

simulation in descending order. We then observe where that

curve crosses the 10 MW- Season level on the probability

axis - - that identifies the LOLP for this scenario.


10

LOLP vs. Contingency Resource

05

10152025303540

0 10000 20000 30000 40000

Contingency Resource Generation (MW-Hours)

LO

LP

(%

)

28,800 MW-HrsLOLP = 4%

10,000 MW-HrsLOLP = 12%


11

Study StatisticsNumber of Simulations (Games) 50

Hours in the Dec-Mar period 2,904

Total number of hours simulated 145,200

Number of games with at least one hour of curtailment

18

Number of curtailment hours (over all games) 275

Number of curtailment events 83

Average magnitude per event 1,953 MW-hrs

Average duration per event 3.4 hours

Average number of events per game with curtailments

4.6


12

LOLP Type Statistics

Types of events

No. of Games

LOLP

(%)

At least 1 hour of curtailment 18 36

Total curtailment > 28,800 mw-hrs 2 4

Total curtailment > 10,000 mw-hrs 6 12

At least one event > 4 hours 13 26

At least one event > 4,000 mw-hrs 9 18

At least one event > 4-hour & > 4,000 mw-hrs 5 10

At least five hours with curtailment > 1,000 mw 5 10

At least ten hours with curtailment > 1,000 mw 2 4

At least one hour with curtailment > 1,000 mw 11 22

The Problem with LOLP


14

Potential Problem with LOLPSame LOLP – Bigger Magnitude

0

0.2

0.4

0.6

0.8

1

1.2

1 2 3 4 5 6

Probability

Mag

nitu

de


15

Potential Problem with LOLPLower LOLP – Bigger Magnitude

00.20.40.60.8

11.21.41.6

1 2 3 4 5 6

Probability

Mag

nitu

de


16

Are LOLP curves well behaved?(i.e. do they cross?)

0

10

20

30

0 10 20 30 40

Probability of Excedence

Mag

nitu

de (

MW

-Sea

s)

LOLP

Magnitude


17

LOLP vs. Magnitude(If this relationship were dependable, the magnitude of the problem could

be estimated by knowing the LOLP)

y = 64.333x + 54.308

R2 = 0.6418

0

500

1000

1500

2000

2500

0 5 10 15 20 25 30

LOLP (%)

Mag

nitu

de

LOLP Subcommittee Reportand Recommendations


19

LOLP Subcommittee Report• Clearly define all reserve requirements

– Operating reserves– Planning reserves– Wind integration reserves

• Determine which reserve curtailments count toward LOLP• Add temperature-correlated wind as a random variable• Decouple temperature and water condition • Define a “contingency” resource for each month of the

year instead of defining threshold events • Record curtailment events across all months of the year• Consider using other adequacy metrics • Continue to assess climate change impacts


20

LOLP Review Status• Reserves

– Work being done by PNUCC committee• Temperature-correlated wind

– BPA working on a test data set• Decouple temp and water

– Done• Contingency resource

– Work needs to be assigned• Annual metric

– Not yet started• Other metrics

– BPA draft methodology– PSRI review

• Climate change – Ongoing

Next Steps


22

Possible Modifications to the Current Method

• Replace LOLP with an alternative metric

• Use LOLP in conjunction with an alternative adequacy metric

• Use LOLP in conjunction with the magnitude of the most severe event (or an average of the worst 10% of events)


23

Examples of Other Adequacy Metrics

• LOLE: loss of load expectation (%)– Number of hours with curtailment divided by the

total number of hours simulated– Can be more intuitive, i.e. 99.5% reliable– Does not address magnitude

• EUE: expected unserved energy (MW-hr)– Average amount of unserved energy per year– Lacks specific information about severe events


24

Work Plan• PSRI review complete by early 2010

• Benchmark GENESYS by early 2010

• Tech Committee propose new metric and threshold by April of 2010

• Use new metric to assess 3 and 5 year adequacy by June 2010

developing an adequacy metric resource adequacy forum technical subcommittee meeting october 16,...

Documents