the ea’s 2-hub ftr market

The EA’s 2-Hub FTR MarketRoger Miller and Grant Read

Presented to EPOC 2011 by Roger Miller

Disclaimer The Authority has just started an RFP process for the FTR

manager, which runs till 31st October RFP is on www.gets.govt.nz Running a fair, open, competitive tender process Information must be given to all potential bidders My presentation will be placed on the Authority’s website Any relevant Q & A’s will be posted on GETS Specific questions can be emailed to [email protected] Any discrepancy with tender documents or Code

amendment - those documents take precedence

NB: This presentation relates to the approved FTR Code Amendment, which comes into force on 1 Oct 2011. It is currently intended to be implemented from Oct 2012.

Overview1. Hubs versus nodes

Why use hubs and how do they work?

2. “All inclusive” FTRs Why use full price difference vs loss exclusive definition

3. Option vs Obligation FTRs Why we need options in a “tidal flow” situation

4. Simplified Grid Model for 2 hub FTR5. Partitioning the Rental Pool for 2 hub FTRAppendix - Loss and Reserve Issues (if time?)


1. Hubs versus nodes Why use hubs and how do they work?


Hubs Hubs are a grouping of one or more nodes Nodes within a hub may have specified weights Settled on the weighted average nodal price FTR hub injections split into component nodal

injections for feasibility testing and rental calculation Trading between a few major hubs reduces market

power concerns Initial FTR “hubs” aligned with ASX futures reference

nodes – BEN2201 and OTA2201 May need to “expand” hubs slightly to handle flows


2. “All inclusive” FTRs Why use full price difference vs loss exclusive definition?


BEN-OTA Monthly Price Components


-100

-80

-60

-40

-20

0

20

40

60Ja

n 20

08

Feb

2008

Mar

200

8

Apr

200

8

May

200

8

Jun

2008

Jul 2

008

Aug

200

8

Sep

200

8

Oct

200

8

Nov

200

8

Dec

200

8

Jan

2009

Feb

2009

Mar

200

9

Apr

200

9

May

200

9

Jun

2009

Jul 2

009

Aug

200

9

Sep

200

9

Oct

200

9

Nov

200

9

Dec

200

9

Jan

2010

Feb

2010

Mar

201

0

Apr

201

0

$/M

Wh

Congestion

HVDC Risk/Reserve

Loss

Wet/North Flow

Dry/South Flow

-30%

-20%

-10%

0%

10%

20%

30%Ja

n 20

08

Feb

2008

Mar

200

8

Apr

200

8

May

200

8

Jun

2008

Jul 2

008

Aug

200

8

Sep

200

8

Oct

200

8

Nov

200

8

Dec

200

8

Jan

2009

Feb

2009

Mar

200

9

Apr

200

9

May

200

9

Jun

2009

Jul 2

009

Aug

200

9

Sep

200

9

Oct

200

9

Nov

200

9

Dec

200

9

Jan

2010

Feb

2010

Mar

201

0

Apr

201

0

BEN-OTA Monthly Loss Factors


Dry/South Flow

Wet/North Flow

“All inclusive” FTRsWhy use full price difference vs loss exclusive definition?

Losses account for a large part of the price differential (especially during HVDC South Flow)

Size and direction of loss price differential depends on generation pattern (unpredictable)

Integrates with futures platform (ASX) better

Simpler for participants to understand and value


3. Option vs Obligation FTRs Why we need options in a “tidal flow” situation


Obligations - What are they?

Price Difference

Payout Pure swap Positive or negative value Symmetrical Linear Simple optimisation

problem

Holder receives payment

Holder must pay


Price Difference

Payout Pure swap Positive or negative value Symmetrical Linear Simple optimisation

problem


Holder must pay


Options - What are they?

Price Difference

Payout

Asymmetrical Non-negative value Non-Linear More complex optimisation

problem


No requirement for holder to pay


Why we need option FTRs Direction of transmission flows depends on generation

pattern (unpredictable) Direction of price differentials depends on direction of

transmission flows Inter-island HVDC flow direction particularly important in

NZ Depends on (SI) hydro inflows (seasonal pattern but

unpredictable from year to year) (“Tidal Flow”) Less important for regions with little generation (e.g.

Upper SI, Auckland? – flow always northward)


Examples


Price

S-N Option

N-S Option

Short/Importing Long/

Exporting

Exercise patterns

Number of exercise patterns = number of permutations of hub orderings by price = n! (assuming options between all hubs, which may not be necessary)

For 2 hubs there are just 2 patterns: S-N, N-S


Ob/Op Constraints

S-N Op + S-N Ob – N-S Ob ≤ S-N Capacity N-S Op + N-S Ob – S-N Ob ≤ N-S Capacity


PS-N Ob = PS-N Op – PN-S Op PN-S Ob = PN-S Op – PS-N Op PS-N Ob = – PN-S Ob

Obligation/Option relationship

Price Difference

Payout

S-N Option

-ve N-S Option

Applies to: Auction clearing prices Payouts Ex-post scaling may distort

these relationships

S-N Obligation



4. Simplified Grid Model for 2 hub FTR


Simplified Grid ModelFor 2 hubs

Traditional approach is to ensure FTRs are simultaneously feasible on forecast FTR grid

For 2 hubs FTRs add or subtract algebraically Establish a maximum flow in each direction FTR grid reduces to a 2 node + 1 line model

Ben-Ota Max

Ota-Ben Max

OtahuhuBenmore


5. Partitioning the rental pool Flow based decomposition for 2 hubs

Motivation Ring fence intra-island rents for a possible

future intra-island locational price risk solution Not yet decided what this should be

(if anything) Avoid cross subsiding the inter-island FTR Maintain consistent inter-island FTR rental

stream if future FTRs hubs or some other intra-island solution are implemented


Different approach used for:

HVDC rents AC rents

HVDC - just collect all rent between Benmore and Haywards HVDC terminals = PricereceivedFlowreceived – PricesentFlowsent


Approach for AC system rents Rents are generated by branch capacities, loss tranche limits,

branch (“equation”) constraints, mixed constraints (if any)

Rent generated by constraint c = σcRHSc

where σc is the constraint shadow price

For each constraint we need to collect rent of σcAssignedCapacityc

where AssignedCapacityc is the maximum LHS constraint loading implied by any feasible FTR flow pattern

Note that if the FTR flow pattern is feasible then AssignedCapacityc ≤ RHSc for all c

For 2 hubs only need to consider 2 extreme flow patterns: BEN-OTA Max; and OTA-BEN Max


Allowing for differences between the FTR Grid and the “On the Day” grid

Extreme injection patterns based on an over-estimated grid (no outages, no contingencies)

Use shift factors to deduce branch/constraint loadings on the “On the Day” grid for each extreme injection pattern

Shift factors avoid having to solve a full load flow for each injection pattern for each trading period

Still collect rent on constraints in series with “On the Day” bottlenecks

Simplified by using lossless shift factors and lossless injection patterns


FTR Injection Patterns Feasible flow pattern is unbalanced (lossy) Approximate by a balanced (lossless) pattern Lossless flow ≥ Lossy flow on each branch

(over estimate)


FTR sinkFTR source FTR flow

Unbalanced flows reduce progressively from source to sink due to losses leaking out at the end of each branch

Balanced flow approximation

Positive injection

at source

Negative injection at sink


branch flowAssigned Branch

Capacity

Actual Flow

Pr LF1

Pr LF2

Pr LFmarg

Cap1 Cap2

AssCap2

AssCap1

SP1

SP2

Loss Rentals

•

SPD uses a piece-wise linear loss approximation

LFj is loss factor of the jth tranche

LFmarg is loss factor of the marginal tranche

Capj is capacity of the jth tranche

AssCapj is assigned capacity of the jth tranche

Pr is price at scheduled flow receiving end node

SPj is shadow price of the jth tranche = Pr (LFmarg - LFj)

Loss rent collected

Pr LFmarg

Marginal Loss Factor

xspot price

The End

the ea’s 2-hub ftr market

Documents

approved ftr code amendment

ftr manager

hub ftr marketroger

hub ftrappendix loss

nodeswhy use hubs

major hubs

hub ftrpartitioning

tender documents