Challenges for Balancing Area Coordination Considering High Wind Penetration Robin Broder Hytowitz & Dr. Ben Hobbs, Johns Hopkins University Dr. Ozge Ozdemir, Energy Research Centre of the Netherlands (ECN) TAIC/FERC Workshop, November 7-8 2014

Outline

• Motivation • Background • Initial Results • Future Work

Motivation •  Renewables add variability to system operations •  Balancing area (BA) consolidation has been proposed •  Wind, load diversity reduce impact of variability •  Better resource use

•  Reserve •  Unit scheduling

•  Decrease peak generation & ramping needs

M. Milligan, B. Kirby, and S. Beuning, “Combining Balancing Areas’ Variability  : Impacts on Wind Integration in the Western Interconnection,” in WINDPOWER Conference and Exhibition, 2010.

Motivation •  Tradeoffs •  System size (more nodes & variables) makes solving to

optimality more difficult •  Increasing complexity in systems operations (ex: stochastic UC)

•  Most dramatic results are seen in studies with large consolidation

•  Does not address uncertainty •  What’s more valuable?

•  Challenge of merging different policies, rules, and regulations

B. A. Corcoran, N. Jenkins, and M. Z. Jacobson, “Effects of aggregating electric load in the United States,” Energy Policy, vol. 46, pp. 399–416, Jul. 2012.

Questions

• Assuming full integration is not an option: • How can we incent efficient coordination &

trade? • What coordination comes closest to

integration?

Outline

• Motivation • Background • Initial Results • Future Work

General Market Timeline

DA HA FMM RT

Load & renewable forecasts, generation bids, fixed schedules, updates

Resource schedules (commitment, energy, ancillary services), prices, reliability checks

DA = Day-Ahead, HA = Hour-Ahead, FMM = Fifteen Minute Market, RT = Real-Time

Hurdle Rates •  Definition •  Within models, a transaction cost that represents barrier to

trade between BAs •  “Economic friction”[1]

•  Parameterized to simulate actual amount of trade •  Not actual price

•  Easy to implement in model objective:

HR(TradeA→B +TradeB→A )

General US Markets

DA

RT

DA

RT

BA 1 BA 2

EU Markets (non-market splitting), West Coast

DA

RT

DA

RT

BA 1 BA 2

Bridge Out

EU Market-Splitting/Coupling

DA

RT

DA

RT

BA 1 BA 2

Bridge Out

CAISO – PacifiCorp – Nevada Energy Imbalance Market (EIM)

DA

RT

DA

RT

BA 1 BA 2

CAISO Energy Imbalance Market Details

DA

RT

DA

RT

BA 1 BA 2

HA HA

FMM FMM

Market BA (ex: CAISO)

Allocation of BA Interties

Self-schedule

Generation bids

Dynamic scheduling

Bilateral contracts

Pseudo-tie line

(Loop flow)

etc.

General US Markets

DA

RT

DA

RT

BA 1 BA 2

Actual Markets

DA

RT

DA

RT

BA 1 BA 2

HA HA

FMM FMM

Self-schedule

Pseudo-tie line Generation bids

Dynamic-schedule

Bilateral contracts

Outline

• Motivation • Background • Initial Results • Future Work

Scenarios

Real-Time

No Coordination Hurdle Rate Full Integration

Day-Ahead

No Coordination

Hurdle Rate Old WECC,

EU non-market splitting

Regional authorities CAISO

Full Integration EU market

splitting/coupling (Nordpool, APX)

Consolidation Single RTO

Scenarios

No Coordination

Hurdle Rate

Integration

No Coor- dination

Hurdle Rate

Integration

Day

-Ahe

ad

Hour-Ahead

Model

•  Types of models •  Day-ahead scheduling: unit commitment •  Commits generation units for the next day •  MILP, binary decisions represent commitment

•  Real-time model •  Optimizes (least cost) power flow

•  Both models: •  Subject to transmission & generation constraints (KCL,

KVL, capacity) •  Options to curtail wind and shed load

Model: Integrated Market

min cgPg,t + cgSUvg,t + cg

NLug,t∀g∑

∀t∑

Subject to: Line limits, transmission constraints (BΘ), capacity limits, ramp rates, minimum up & down times, spin & non-spin reserve requirements

Day-Ahead

Real-Time

min cgPg,t∀g∑

∀t∑

Subject to: Line limits, transmission constraints (BΘ), capacity limits, ramp rates, spin reserve requirements

Model: Hurdle Rate

min cgPg,t + cgSUvg,t + cg

NLug,t( )∀g∑

∀t∑ +HR(StAB + StBA )

StAB − St

BA = Pk,tline

∀k∈IT∑ ∀t

Subject to

Line limits, transmission constraints (BΘ), capacity limits, ramp rates, minimum up & down times, spin & non-spin reserve requirements

Day-Ahead

Real-Time

min cgPg,t( )∀g∑

∀t∑ +HR(StAB + StBA )

Model: No Coordination

min cgPg,t + cgSUvg,t + cg

NLug,t∀g∑

∀t∑

Subject to

Line limits, transmission constraints (BΘ), capacity limits, ramp rates, minimum up & down times, separate spin & non-spin reserve requirements

Day-Ahead

Real-Time

min cgPg,t∀g∑

∀t∑

Pk,tline

∀k∈IT∑ = 0 ∀t

System Topography •  Reliability Test System 1996 •  3 Zone (99 generators,

73 buses) •  24 hours

•  Each BA is similar in size •  # generators •  Type of generation •  Wind capacity

[3]

Results – Real Time Costs

Total Cost (Millions $)!

RT!

No Coordination! Hurdle Rate! Full Integration!

DA!

No Coordination!

Hurdle Rate!

Full Integration!

2

2.4

2.8

3.2

RT 2

2.4

2.8

3.2

RT 2

2.4

2.8

3.2

RT

2

2.4

2.8

3.2

RT

200

2

2.4

2.8

3.2

RT 2

2.4

2.8

3.2

RT

2

2.4

2.8

3.2

RT 2

2.4

2.8

3.2

RT 2

2.4

2.8

3.2

RT

300

Results – Prices

Average LMP ($/MWh)!

RT!

No Coordination! Hurdle Rate! Integrated Market!

DA!

No Coordination!

Hurdle Rate!

Integrated Market!

0

20

40

60

DA RT 0

20

40

60

DA RT 0

20

40

60

DA RT

0

20

40

60

DA RT 0

20

40

60

DA RT 0

20

40

60

DA RT

0

20

40

60

DA RT 0

20

40

60

DA RT 0

20

40

60

DA RT

5000

6000

Results - deterministic and stochastic wind scenarios

$2.00

$2.20

$2.40

$2.60

$2.80

$3.00

$3.20

$3.40

0%

1%

2%

3%

4%

5%

6%

7%

No Coordination

Integrated Hurdle Rate- $3

Hurdle Rate- $6

Hurdle Rate- $10

Tota

l Cos

t (M

illio

n $)

Trad

e as

a p

erce

nt o

f dem

and

Detrm - Trade Stoch - Trade Detrm - Total Cost Stoch - Total Cost

Price results – single wind forecast

11% 12% 16%

11% 11%

0% 2% 4% 6% 8% 10% 12% 14% 16% 18%

13.5 14

14.5 15

15.5 16

16.5 17

17.5 18

18.5 19

Perc

ent c

hang

e be

twee

n DA

and

RT

LMP

($/M

Wh)

DA Mean RT Mean Percent Change

Results – All scenarios

0!

0.2!

0.4!

0.6!

0.8!

1!

1.2!

1.4!

1.6!

$2,000,000! $2,500,000! $3,000,000! $3,500,000!

Net!Trade

!as!a!%!of!D

emand!

Total!Cost!($)!

Integration Day-Ahead

Hurdle Rate Day-Ahead

No Coordination Day-Ahead

■ DA ▲ RT Integration ♦ RT HR ● RT NC

Conclusions •  Integrated markets yield most gains from trade •  Barriers on interties impede efficient markets •  Going from no coordination to hurdle rates makes large

difference •  Further lowering hurdle rates most beneficial in DA rather

than RT •  Further work needed to determine the simplifying

effect of hurdle rates relative to actual barriers •  Which barriers are most important to address? •  Who should be responsible for removing inefficiencies?

•  When no RT coordination, there might not be enough generation on-line to meet demand •  Average prices more consistent DA vs RT when there is

no coordination day-ahead

Outline

• Motivation • Background • Initial Results • Future Work

Future Work •  Different size BAs, different generation mix •  Modeling specific barriers on the intertie line •  Self-scheduling •  Dynamic-scheduling

•  COMPETES model – ECN •  Look at European grid

•  Additional balancing areas •  Greatest benefits for renewables

Thank you! Questions?

Email: hytowitz@jhu.edu

References •  [1] Joseph H. Eto, Douglas R. Hale and Bernard C. Lesieutre,

“Toward More Comprehensive Assessments of FERC Electricity Restructuring Policies: A Review of Recent Benefit-Cost Studies of RTOs” The Electricity Journal

•  [2] Frank Wolack, “The Economics of Self Scheduling,” Presentation available: http://www.caiso.com/Documents/Presentation-Economics-Self-Scheduling-MSCPresentation.pdf

•  [3] Jesús María López-Lezama, Mauricio Granada-Echeverri, and Luis A. Gallego-Pareja, “A combined pool/bilateral dispatch model for electricity markets with security constraints” Ingeniería y Ciencia, June 2011.