On the Inherent Inefficiencies of TLR for Trading … · On the Inherent Inefficiencies of TLR for Trading Electricity ... Inefficiencies in Initial TLR Rules No consideration of

On the Inherent Inefficiencies of On the Inherent Inefficiencies of TLR for Trading ElectricityTLR for Trading Electricity‡‡

Fernando L. Alvarado*

IEEE/PES Winter MeetingColumbus, Ohio31 January 2001

(*) Professor, the University of Wisconsin and Senior Consultant, LRCA(‡) Prepared with the help of Rajesh Rajaraman, LRCA

PSERC

Slides available from http://www.pserc.wisc.edu© 2001 F. Alvarado

January 31, 2001 Inefficiencies of TLR

Acronyms, more acronymsAcronyms, more acronyms

TTC, ATCOASISTLRMRDMultilateralLocational pricing:

Zonal, Flowgate, Nodal

Progress

© 2001 F. Alvarado


TTC and ATCTTC and ATC

Total Transmission CapabilityHow much can you send from A to B?

Available Transmission CapacityHow much can you send from A to B and

given other trades and have “room for reserves”?

OASIS: A rational approach to establish and post ATC tofacilitate trade and take into consideration engineering limits

© 2001 F. Alvarado


What is wrong with ATC?What is wrong with ATC?

Economically inefficientCosts not considered

Inherent imprecision and uncertainty leads to:

Under-utilized capacity on some linesOversold capacity on other lines

© 2001 F. Alvarado


TLRTLR

Administrative rule invoked to curtail trades when congestion occurs

Not based on economicsArbitraryArbitrary formula allocates curtailments

∑=

iii

kkknTransactio Initial * PTDF

curtail amount to Total * nTransactio Initial * PTDFamount Curtailed 2

© 2001 F. Alvarado


TLR in theoryTLR in theory

It sounds simpleIt sounds fairIt seems to workIt is necessary

Something must be doneIt is fastIt is not centralized

(really?)

© 2001 F. Alvarado


Inefficiencies in Initial TLR Inefficiencies in Initial TLR RulesRules

No consideration of economicsCongested transmission capacity not

auctioned off to highest bidderMultilateral trades deemed separable

bilateral tradesEffect of counter-flow trades not credited

against other trades over congested lines

© 2001 F. Alvarado


Why Administrative Why Administrative Curtailments are Inefficient (1)Curtailments are Inefficient (1)

Economics 101: In a competitive market, prices are set on the margin

The price of transmission between two locations is equal to the difference in the locational prices

© 2001 F. Alvarado



A B100 MW limitPA = $25/MWh PB = $45/MWh

Price of transmitting power from A to B = $20/MWhTransmission provider collects* = 20*100 = $2000in congestion rents

Line is Congested

* Caution: In a network, the congestion rent calculation is slightly different

© 2001 F. Alvarado



Suppose transmission provider charges $5/MWh to ship power from A to BMarket participants see free money

Marginal profit is 20–5 = $15/MWhCongestion rent “left on the table” is $1500

(15*100)Market participants try to get it by

gaming TLR

© 2001 F. Alvarado



If I can anticipate TLR, I can create positions that will be curtailed “less” by forcing others to be curtailed more

Phantom trades can be scheduledTrades that slip under the TLR radar can be

put together

It is not optimal to setup counterflow trades to minimize flow over congested lines because this leaves “money on the table.”

© 2001 F. Alvarado


Why Taking a Strictly Bilateral Why Taking a Strictly Bilateral Viewpoint is BadViewpoint is Bad

Silly example:A trader has an A to B trade of 100 MW, and

a B to A trade of 100 MW. This combined trade has no effect on flow in the

congested line

TLR rules curtail A to B trade while allowing B to A trade

In networks, there are multilateral trades that have no effect on some congested lines

NERC TLR rules treat them as undesirable

© 2001 F. Alvarado


NERC TLR issues (May 2000)NERC TLR issues (May 2000)

How is TLR reallocation and reloading to be accomplished?

Loading guidelines on implementation of TLR Levels under development

Method under development to find parallel flow effects of native load and network service so that assignment of relief will not double count point-to-point generation

Issues

galore

remain

© 2001 F. Alvarado


Improvements to TLR: Improvements to TLR: Market Redispatch (MRD)Market Redispatch (MRD)

“MRD is a procedure for prearranging market redispatch transactions to provide counterflows on a constrained path to protect transactions that would otherwise be curtailed under transmission loading relief procedures” (NERC, July 2000)

© 2001 F. Alvarado


Advantages of MRDAdvantages of MRD

“Solves” most serious TLR problemsIt is market orientedIt is reasonableIt recognizes the importance of nodal

(rather than regional) redispatch

© 2001 F. Alvarado


Problems with MRDProblems with MRD

No incentive to set up counter-flows on inefficiently priced congested lines

Why forsake “free” congestion rents?For multiple congestion conditions,

optimal trades must be multilateral

No congestion ⇒⇒⇒⇒ One marginal unitOne congested line ⇒⇒⇒⇒ Two marginal unitsTwo congested lines ⇒⇒⇒⇒ Three marginal units

© 2001 F. Alvarado


What is the status of MRD?What is the status of MRD?

FERC approved MRD for summer 1999 FERC accepted NERC's revised MRD

(extended it to December 31, 2000)“The revised MRD represents a best effort to

meet its mandate”NERC will file a report summarizing MRD

results and recommending whether it should be continued, modified, or discontinued.

© 2001 F. Alvarado


Improvements to TLR: Improvements to TLR: Multilateral TradesMultilateral Trades

Trader has the option of creating a packaged trade [Wu, Varaiya 1996]

Counter-flows are credited against flowsCurtailment of packaged trade based

on a normed optimization problem [Bialek et al 2000]

© 2001 F. Alvarado


Advantage of Advantage of multilateralsmultilaterals

Packaged multilateral trades more efficient than bilateral trades

A trader can maximize power transfers and minimize curtailments

More than one flowgate constraint can be accommodated

© 2001 F. Alvarado


Disadvantage of Disadvantage of multilateralsmultilaterals

No incentive to use multilateral trade prior to TLR curtailment

Why forsake “free” congestion rent? Curtailment formula is stillstill inefficient

If packaged trade is curtailed, counterflowsare also curtailed

This is sillyIf only counter-flows are not curtailed, other

traders benefit from counterflow tradeGreater administrative difficulties

managing complex trades© 2001 F. Alvarado


Multilateral trades: commentsMultilateral trades: comments

AfterAfter one round of TLR curtailments, there is no capacity left to sell

“Lucky” participants get 100 MW after first TLRThey sell this 100 MW at prevailing market price

and the market could reach a competitive equilibrium [Wu and Varaiya 1996]

Multilateral trades better than plain TLR Gaming incentives prior to first TLR remain

Multilaterals “great” AFTER first TLR has occurredWhy bother before? No incentive!

© 2001 F. Alvarado


Best Solution to TLR dilemma: Best Solution to TLR dilemma: Price flowgate capacityPrice flowgate capacity

Directly (FGR): Auction capacity on each flowgate separately

Chao-Peck pricingAPX

Indirectly (FTR): Nodal pricing

Both approaches are theoretically equivalent

© 2001 F. Alvarado


What is an FTR?What is an FTR?

Right to collect money based on nodal price differences between nodes

In principle, any node pair or node combination

Purpose: hedge transactionsFTRs derive value exclusively from

the value of underlying flowgatesIf congestion does not affect relative nodal

prices, FTR has no value.

© 2001 F. Alvarado


What is an FGR?What is an FGR?

Right to schedule a transaction across a congested interface

Purpose: hedge transactionsFGRs derive value from their ability to

schedule powerIf congestion does not create price

differences, the FGR has no value

© 2001 F. Alvarado


FGR FGR –– alternative definitionalternative definition

The right to collect money based on the value of the Lagrange multiplier associated with the flowgate of interest

Physical rights: a red herring (unless there arerestrictions on who can participate in physical trades)

© 2001 F. Alvarado


Advantages of direct (FGR) Advantages of direct (FGR) pricingpricing

You can sell the “full capacity”You do not need simultaneous

clearing to be able to trade on the capacity of a single line

However, you must predefine thecommercially significant flowgatesand you are subject to PTDF risk

© 2001 F. Alvarado


Advantages of Indirect (FTR) Advantages of Indirect (FTR) PricingPricing

If you are going from Madison to Sevilla, it is usually better to buy all the flight segments at once

You are not exposed to “PTDF risk”If PTDFs change after the rights are purchased,

a legal or administrative problem may occur

Direct (flowgate): You buy segments for a tripIndirect (nodal): You buy a trip from A to BIn both cases you can go from A to B! However, in the In both cases you can go from A to B! However, in the direct case, if a segment is canceled, you are out of luck!direct case, if a segment is canceled, you are out of luck!

© 2001 F. Alvarado


Arguments against Arguments against locationallocationalpricingpricing

“Any centralized solution is likely to be met with deep suspicion”

“Why should traders give private cost-benefit information to an ISO?”

“Redispatch costs are lower than congestion rents; why shouldn’t marketers pay only redispatch costs?”

“It will increase market power”These arguments are flawed

© 2001 F. Alvarado


“Locational Pricing Methods “Locational Pricing Methods are too Centralized”are too Centralized”

Pop Quiz: Which is fairer? Which is less arbitrary?

An arbitrary curtailment scheme by a centralized authority who does not care what you are willing to pay (TLR)?

A centralized authority (ISO) who curtails you only if what you are willing to pay is lower than the market clearing price?

TLR is no more centralized than Locational Pricing Models; it is just more arbitrary

© 2001 F. Alvarado


“Why should traders give private “Why should traders give private costcost--benefit information to an ISO?”benefit information to an ISO?”

An ISO is a trader of transmission capacity and would like to obtain the best price for this capacity

OK, it is a “regulated” traderUsing bids/offers the ISO solves the

least-cost problem for all nodesSince the ISO is regulated, the ISO cannot

extract monopoly profits from the sale of capacity on congested flowgates

© 2001 F. Alvarado


“Why should traders pay congestion “Why should traders pay congestion rent rather than redispatch cost?”rent rather than redispatch cost?”

Because transmission is a commodity and must be priced on the margin

Comparing congestion rent to redispatch cost is analogous to comparing total generator cost to total payments to generators in an unconstrained system

All infra-marginal generators get the same market clearing price

© 2001 F. Alvarado


What is the right price?What is the right price?

$10/MWh

85 96

Dispatch costs = 11*100+10*85 = $1950Payments = 85*100+11*100 = $9600

Demand

Even without congestion, total dispatchcosts are lower than payments

$100/MWh$100/MWh

© 2001 F. Alvarado


“Proposed method allows marketers “Proposed method allows marketers to exercise market power”to exercise market power”

There is no evidence that such method exacerbates existing market power

Examples in the literature often ascribe existing market power to nodal pricing!

On the other hand, there is a theory that suggests that other methods can make market power worse (Hogan 2000)

© 2001 F. Alvarado


IllustrationIllustration

A B CCongestion

If both the A to C and B to C transactions are treated as transactions from Zone 1 to Zone 2, intra-zonal congestion in Zone 1 limits transactions from B to C, andincreases market power in Zone 2increases market power in Zone 2

Zone 1 Zone 2

© 2001 F. Alvarado


NERC MRD vs. PJM “experiments”NERC MRD vs. PJM “experiments”

NERC: none succeeded (tagging problems)

PJM: many successes

© 2001 F. Alvarado


ConclusionsConclusions

ATC and OASIS have problemsTLR leads to gaming and inefficienciesMRD is better, but still has problemsMultilateral redispatch is “almost” rightThe right methods:

Flowgate-based direct flowgate pricingIndirect flowgate-based (nodal) pricing

© 2001 F. Alvarado


What is an FTR?What is an FTR?

Right to collect money based on nodal price differences between nodes

In principle, any node pair or node combination

Purpose: hedge transactionsFTRs derive value exclusively from

the value of underlying flowgatesIf congestion does not affect relative nodal

prices, FTR has no value.

© 2001 F. Alvarado


What is an FGR?What is an FGR?

Right to schedule a transaction across a congested interface

Purpose: hedge transactionsFGRs derive value from their ability to

schedule powerIf congestion does not create price

differences, the FGR has no value

© 2001 F. Alvarado


FGR FGR –– alternative definitionalternative definition

The right to collect money based on the value of the Lagrange multiplier associated with the flowgate of interest

© 2001 F. Alvarado

Pmax=250$20/MWh

Pmax=250$20/MWh

Pmax=200$80/MWh

G1

G2

G3

G4

G5

System A System B

L1

L2

Pmax=200$20/MWh

Pmax=200$80/MWh

L1: 250 MW (134.4 H) 350 MW (33.6 H)L2: 250 MW (134.4 H) 350 MW (33.6 H)(peak hours are coincident)

Pmax=100

© 2001 F. Alvarado

50 MW$20/MWh

250 MW$20/MWh

0 MW$80/MWh

G1

G2

G3

G4

G5

System A: $20 System B: $20

L1=250

L2=250

200 MW$20/MWh

0 MW$80/MWh

Off-peak conditions(Assume price at G1 slightly higher than $20)

P=50

© 2001 F. Alvarado

200 MW$20/MWh

250 MW$20/MWh

0 MW$80/MWh

G2

G3

G4

G5


L1=350

L2=350

200 MW$20/MWh

50 MW$80/MWh

Peak conditions

P=100G1

© 2001 F. Alvarado


AssumptionsAssumptions

G1 holds a 100 MW FTR from G1 to L2or

G1 holds a 100 MW FGR on flowgateThere is no congestion

All prices the same and equal to $20G1 is supplies part of L2

G2 could do it equally well

© 2001 F. Alvarado

50*20

250*20

0*20

G1

G2

G3

G4

G5


L1=250

L2=250

200*20

0*20

P=50

G1: 50*20+0*100=$1000G2: 250*20=$5000

Observation: All prices are the same, G2 has more income

Off-peak conditions: FTR or FGR viewpoint

50 MW

250 MW

No congestion rentFTR: sales at AFGR: sales to B

© 2001 F. Alvarado

100*20+100*80

250*20

0

G2

G3

G4

G5


L1=350

L2=350

200*80

50*80

Peak conditions: FGR viewpoint

P=100G1

G1: 100*20+100*80=$10000G2: 250*20=$5000G4: 200*80=$16000G5: 50*80=$4000

Because G1 owns the flowgate, it gets to schedule the power

Income fromsales to B

200 MW

250 MW

Payment from B

© 2001 F. Alvarado

200*20+100*60

250*20

0 MW$80/MWh

G2

G3

G4

G5


L1=350

L2=350

200*80

50*80

Peak conditions: FTR viewpoint

P=100G1

G1: 200*20+100*60=$10000G2: 250*20=$5000G4: 200*80=$16000G5: 50*80=$4000

Because G1 owns the TCC, it gets paid the price difference

Incomefrom FTR

200 MW

250 MW

Payment from FTR

© 2001 F. Alvarado


NonNon--unity PTDF caseunity PTDF case

Assume G1 and G2 have different impact (PTDF) on the flowgate for sales to L2

Assume the flowgate limit is 50 MW

12583.3

PmaxPTDF

0.4G2 to L20.6G1 to L2

© 2001 F. Alvarado

150 MW

200 MW

G1

G2

G4

G5


L1=350

L2=350

200 MW

150 MW

0 MW

Possible base case conditions

G1 can sell up to 83.3 MW to L2 G2 can sell up to 125 MW to L2(but its limit is only an additional 50)Either one displaces G5 generation(assume the G5 to L2 PTDF is zero)

(Pmax=50)

© 2001 F. Alvarado

83.3+150 MW

200 MW

G1

G2

G4

G5

System A System B: s=$80

L1=350

L2=350

200 MW

66.7 MW

P=50

An inefficient solution: G1 owns 50 MW FGR

According to Chao-Peck prices at G1and G2 mustmust be different:sG1= sL2−−−−PTDFG1L2* λλλλsG2= sL2−−−−PTDFG2L2* λλλλ

This “solution” is inefficient:If sG1=20, then sG2=40 and PG2=250(If sG2=20, then sG1=-10 and PG1=0)

To make efficient, redispatch both G1 and G2

s=$40

s=$20

λλλλ =100

s: spot priceλλλλ: Lagrange mulitplier

© 2001 F. Alvarado

100+50

200+50

G1

G2

G4

G5

System A System B: $80

L1=350

L2=350

200 MW

50 MW

P=30+20

An efficient solution: G1 owns 30 and G2 owns 20

3020

s=$40

For an FGR:G1 receives 100*20+50*80=6000G2 receives 200*40+50*80=12000

For an FTR:G1 receives 150*20+(80-20)*50=6000G2 receives 250*40+(80-40)*50=12000

s=$2020

30

© 2001 F. Alvarado

150

250

G1

G2

G4

G5

System A System B: $80

L1=350

L2=350

200 MW

50 MW

P=30+20

s=$40

s=$20

G1 owns 30 and G2 owns 20: payment based on Lagrange multiplier

For an FGR based on λλλλ:G1 receives 150*20+30*100=6000G2 receives 250*40+20*100=12000

λλλλ =100


ObservationsObservations

The following are functionally identical:FGRs based on physical schedulesFGRs based on Lagrange multiplier paymentsFTRs

© 2001 F. Alvarado


Two possible differencesTwo possible differences

You can “sell more” FGRs than FTRsYou can sell capacity on all linesOnly certain ones matter

FGRs can be “one sided” (option-like)So can FTRs, but it is not obvious

© 2001 F. Alvarado


“Selling more capacity”“Selling more capacity”

A network with n nodes has n-1 “injection” degrees of freedom

If there are m lines (m>n), only m-1 degrees of freedom exist for these flows

Thus, when clearing an FTR market, at most n-1 flowgates will have value

When selling FGRs, they can be sold on all lines

© 2001 F. Alvarado


The CPOW example variantThe CPOW example variantThree lines cannot congest together

At most n-1 lines can congest together either in the auction or in real time

G1 wants to sell to L3 and be hedged

FGR results (physical)FGR results (financial)FTR results

1

3

2

≤≤≤≤300

≤≤≤≤220

≤≤ ≤≤100

$20

$20

Limits shownX=1 for all lines

$80

© 2001 F. Alvarado

Possible and impossible casesPossible and impossible cases1

3

2

300

200

100

1

3

2

300

220

801

3

2

300

400100

1

3

2

320

22010

0

400

100

380

140

0

0

520

500

120

420

0

540

200

500

0

700


FGR (2 versions)FGR (2 versions) vsvs FTRFTR

1

3

2

300

200

100

400

100

0

500

FTR viewpoint:G1 buys 300 MW FTR 1 to 3Prices are as shownG1 receives 400*20+300*60

$20

$80

$80

FGR (physical) viewpoint:G1 buys 100 on 1 to 2, 100on 2 to 3 and 200 on 1 to 3G1 receives 100*20+300*80

FGR (financial) viewpoint:G1 buys 100 on 1 to 2, 100on 2 to 3 and 200 on 1 to 3From the data, we get thatλλλλ12=60, λλλλ13=60 and λλλλ23=0G1 receives 400*20+100*60+200*60+200*0

© 2001 F. Alvarado


Calculation detailsCalculation details

Prices and Lagrange Multipliers relating according to:

s2=s1+PTDF21*λλλλ1+PTDF22*λλλλ2s3=s1+PTDF31*λλλλ1+PTDF32*λλλλ2

© 2001 F. Alvarado


FTR vs. FGR: Pros and ConsFTR vs. FGR: Pros and Cons

FTR pricing not subject to PTDF riskNot knowing what you need to “own” to hedge

FTR pricing not subject to flowgate riskKnowing which flowgates will actually congest

© 2001 F. Alvarado

Documents

On the Inherent Inefficiencies of TLR for Trading … · On the Inherent Inefficiencies of TLR for Trading Electricity ... Inefficiencies in Initial TLR Rules No consideration of