LMPs, FTRs and Congestion:

1
LMPs, FTRs and Congestion A Primer Larry E.
Ruff Texas Nodal Group Austin, TX September 24,
2003
2
Agenda
Intro Who Am I and Why Are We Here? Some
Electricity Basics Background and History of
LMP/FTRs LMP/FTR Definitions and Basics
3
Introduction Who Am I? Why Are We Here?
4
Who Am I? Brief Resume

• BS Physics Caltech 63 PhD Econ Stanford 68
• 3 Years each as Academic (UCSD), bureaucrat
  (USEPA), foundation officer (Ford), energy
  analyst (WRGrace), corporate planner (USSFC),
• Entered consulting (PHB) in 1985, as PURPA was
  beginning the move to competition
• Lived in London 1989-92, invented market based
  on pool pricing and contracts for differences
  (CfDs)
• Was with Hogan in 1992 when he invented LMPs
• Have designed/advised on markets in many
  countries, testifed at FERC, state PUCs, etc.
• Ejected from CA (w. Hogan) for opposing initial
  market
• Now independent consultant and CRA Senior Advisor

5
Why Are We Here?

• ERCOT plans to convert to LMP-based market
• Concepts/techniques are not well understood
• I have been asked to present a primer on LMP

• The objective of this presentation is
• To provide background and introduce concepts
• NOT to argue for or against anything

• When we are done, you should know
• What LMPs and FTRs are and why they are used
• That LMPs/FTRs are very complex at first ( 2nd,
  3rd )
• That there is a lot more to learn
• That I do not know it all

6
Some Electricity Basics
7
The Complexities of AC Networks

• Power flows and voltages on an AC network
• Follow complex physical (Kirchoffs) laws
• Must be maintained within strict and complex
  limits

• The limits on AC network operations include
• Thermal limits on lines transformers (related
  directly to power flows and hence easy to
  understand and apply)
• Voltage limits at locations (related to reactive
  power more than power flows per se and hence
  very complex)
• Stability limits/contingency constraints (system
  must stay stable after loss of 1 or more major
  grid or generation assets very complex and
  judgmental)

Trading/pricing must reflect these factors somehow
8
The DC Approximation

• Most discussions of electricity markets
  (including todays) and even many actual markets
  use a DC approximation of the AC network that
• Largely ignores voltage and reactive power
• Approximates all constraints by limits on power
  flows
• Sometimes ignores thermal losses (as we do today)

• All this simplifies exposition and computation,
  but
• In conceptual discussions, can suggest that some
  market mechanisms are more feasible than they
  actually are
• In practice, requires judgmental approximations
  and can reduce efficiency of dispatch and pricing

9
Basics of Power Flows on a (DC) Network

• In the DC approximation of AC networks
• Each line (meaning any grid element on which
  power flows, e.g., transformers) has a constant
  impedance determined by the electrical
  characteristics of that line
• The power flow on each path between any two
  points is inversely proportional to the total
  impedance of that path
• The net power flow into/out of any point, and
  hence into/out of the grid as a whole, is zero
• Power flows are additive

These rules imply (among other things) that power
will flow over all lines of the grid according to
fixed power distribution factors (PDFs, also
called shift factors)
10
A 3-Node Network
11
A 3-Node Network, Different Impedances
12
A 4-Node Network
It is straightforward (but tedious) to compute
flows and PDFs for complex networks our examples
use 3-node networks with equal impedances on all
lines
13
Background and History of LMPs/FTRs
14
History Contract Paths

• Electricity trading began with monopoly utilities
• Not cost sensitive, cooperated more than competed
• Rough equity in benefit/cost sharing was good
  enough
• Incidental third-party effects were small or
  ignored

• Transmission rights and pricing were needed
• When transactions became larger and longer
• To assist scheduling and dispatch
• To compensate (strongly) affected third parties

• The concept of contract paths
• Was developed and worked for awhile
• Created more problems as wholesale trading grew
• Was totally inadequate to support real competition

15
History Contract Paths
100 A ? B transaction A dispatches 100 more
and B dispatches 100 less than own-load, 100
flows from A to B
C
A
B
100
Loop-flow effects on third parties (C D) are
ignored, uncompensated
D
A-B can pay C for transmitting 100, but (even if
Cs grid can easily handle 100) much of the power
flows through D and E
Dispatch is hard, costs are unfair
16
History Physical Transmission Rights

• As markets developed, some sort of transmission
  rights were clearly needed to
• Keep flows within limits when many wanted to
  trade
• Assure that highest-value transactions were
  scheduled

• Physical transmission rights (PTRs) developed
• Max flows on interfaces are defined
• PTRs equal to max flows are allocated or sold
• PTRs are needed to schedule a transaction
• Trading is supposed to get PTRs to highest-value
  uses

Unfortunately, PTRs work well only in very simple
cases, i.e., with radial lines and no loops
17
Defining Interface Capacity Between Zones

• Gens in A/B are competing to sell into C
• Trades are limited by transmission capacity
• How can A/B ? C capacity be defined so that PTRs
  can be allocated/sold and then traded?

18
What Is the Interface Capacity?
If Energy is Cheaper at B
BUT, if Energy is Cheaper at A
500
If 2,400 PTRA/B-C are put into the market when B
is cheaper, they can then be sold to A when A is
cheaper, and then transmission constraints will
be violated
19
Does Creating More Zones Help?
PTRB-C can be 2,400 if PTRA-C 0 or PTRA-C
can be 1,500 if 0 if PTRB-C 0 Many other
combinations are possible (But NOT PTRA-C 1,000
and PTRB-C 1,600)
Feasible combinations of PTRs cannot be known
until the market solution is known
20
The Birth of LMPs/FTRs

• In about 1982, Professor William Hogan of Harvard
• Realized that contract paths and PTRs were
  inconsistent with the basic physics of
  electricity
• Went back to the earlier work on nodal pricing by
  Schweppe at MIT and the electrical engineering
  literature

• Hogan made LMPs workable by
• Combining UK bid-based pool ideas with nodal
  pricing
• Developing mathematical and programming methods
  for computing prices on real electricity systems

• Hogan then discovered FTRs almost by accident
• While running numerous cases with very different
  dispatches, found that congestion rents were
  constant
• Then did the mathematics to prove this critical
  feature

21
LMP/FTR Definitions and Basics
22
Some Preliminary Definitions

• Locational Marginal Pricing (LMP) An integrated
  dispatch/spot market process that uses voluntary
  bids to determine simultaneously
• A security (or contingency) constrained economic
  dispatch
• A price for energy (PA) at each grid node (A)
• Point-to-point congestion prices (CPAB from A to
  B) defined by differences in LMPs (CPAB PB PA)

Grid Node A commercially and/or electrically
interesting point on the grid, usually an
injection, off-take or intersection point
LMP is also (and more precisely) called nodal
pricing
23
More Preliminary Definitions
Transmission Constraints Limits on power flows
or voltages that must be met by the (pre- or
post-contingency) dispatch(es)
Security Constrained Economic Dispatch (SCED) A
mathematical process for finding a dispatch (and
LMPs) that minimizes the total bid-cost of
meeting demand subject to all transmission
constraints
Transmission Congestion Exists when one or more
transmission constraint(s) is (are) binding and
as a result the SCED uses higher-cost generation
out of merit to displace lower-cost generation
24
Still More Preliminary Definitions

• Financial Transmission Rights (FTRs) Financial
  instruments that
• Convey the right to collect point-to-point
  congestion prices from the ISO a MW of FTRAB
  pays CPAB PB PA
• Are used to hedge congestion prices
• Are also called TCCs, CRRs, and other things

A Power Distribution (or Shift) Factor The
power flow across a specified grid element when 1
MW is injected at a specific node and withdrawn
at a specific sink node
25
Bid-Based Dispatch Pricing (Unconstrained)
S
D1
All suppliers are paid and all loads pay the same
energy price in each hour no matter where they
are
26
Case 0 Single (Radial) Line, No Congestion
GB 0
Constraint FAB ? 1,000
PA 25
PB 25
There is a constraint (FAB 1,000), but it is not
binding, so there is no congestion or congestion
costs, and (ignoring losses) all LMPs are the same
27
Case 1 Radial Line with Congestion
GenCo Offers 1,000 _at_ 40
GenCo Offers 2,000 _at_ 25
B
A
FAB? 1,000
PA 25
PB 40
Demand 800 _at_ 100
Demand 1,500 _at_ 100
Binding constraint ? Congestion, differing LMPs
28
More Definitions

• Congestion Price/Charge (CP) The incremental
  cost of a point-to-point power transfer
• Defined by LMP differences (?s) as CPAB PB
  PA
• Paid to ISO by both spot and bilateral
  transactions
• Rebated by ISO to holders of FTRAB

Congestion Costs The increase in actual
dispatch costs due to congestion
Congestion Rents The net revenue to (in the
first instance) the ISO because congestion has
created LMP differences
In general, Congestion Costs ? Congestion Rents
29
Case 1 Radial Line with Congestion
Congestion Price PB - PA 15
Congestion Cost 500 ? (40 - 25) 7,500
Congestion Rent 1,500?40 - 1,000 ? 25 - 500 ?
40 7,500
In more complex cases, Congestion-Rents gt
Congestion-Costs (often gtgt)
30
Case 2 Two Radial Lines with Congestion
PAU 35
GenCo Offers G5 2,000 _at_ 45
A
GenCo Offers G1 1,200 _at_ 20 G2 1,000 _at_ 35
G1U 1,200
1,500
G2U 300
? 1,000
G5U 0
PBU 35
C
3,000
? 1,600
GenCo Offers G3 1,500 _at_ 30 G4 1,000 _at_ 40
G3U 1,500
PCU 35
1,500
G4U 0
B
Unconstrained Trading/Dispatch Reaches an
Infeasible Solution
31
Case 2 Optimal Dispatch LMPs
PA 20
GenCo Offers G5 2,000 _at_ 45
A
GenCo Offers G1 1,200 _at_ 20 G2 1,000 _at_ 35
G1 1,000
1,000
G2 0
? 1,000
G5 400
C
3,000
? 1,600
GenCo Offers G3 1,500 _at_ 30 G4 1,000 _at_ 40
PC 45
G3 1,500
1,600
G4 100
B
PB 40
With No Loops, 2 (N) Constraints ? 3 (N1) Prices
32
Case 2 Definition and Value of FTRs
GenCo Offers G5 2,000 _at_ 45
A
GenCo Offers G1 1,200 _at_ 20 G2 1,000 _at_ 35
? 1,000
FTRA-C 1,000
C
3,000
FTRB-C 1,600
?1,600
GenCo Offers G3 1,500 _at_ 30 G4 1,000 _at_ 40
Value of FTRA-C 25
B
Value of FTRB-C 5
ISO rebates all congestion rents to FTR holders
a grid user with the right FTRs is well hedged
33
Case 3 A Simple Loop with Congestion
PAU PBU PCU 35
GenCo Offers G5 2,000 _at_ 45
A
GenCo Offers G1 1,200 _at_ 20 G2 1,000 _at_ 35
G1U 1,200
1,500
G2U 300
G5U 0
? 1,000
0
? 750
C
3,000
? 1,600
GenCo Offers G3 1,500 _at_ 30 G4 1,000 _at_ 40
G3U 1,500
1,500
All Lines Have the Same Impedance
G4U 0
B
Unconstrained Trading/Dispatch Reaches Infeasible
Solution
34
Case 3 Optimal Dispatch LMPs
PA 20
GenCo Offers G5 2,000 _at_ 45
A
GenCo Offers G1 1,200 _at_ 20 G2 1,000 _at_ 35
G1 750
1,000
G5 750
G2 0
? 1,000
250
? 750
C
3,000
? 1,600
GenCo Offers G3 1,500 _at_ 30 G4 1,000 _at_ 40
1,250
PC 45
G3 1,500
G4 0
B
PB 40
1 Constraint ? Different price at every node
(even if there are gtgt 3 nodes)
35
Case 3 What Actually Happens w. LMPs/FTRs?
GenCo Offers G5 2,000 _at_ 45
A
GenCo Offers G1 1,200 _at_ 20 G2 1,000 _at_ 35
?1,000
? 750
C
3,000
?1,600
GenCo Offers G3 1,500 _at_ 30 G4 1,000 _at_ 40
B
Traders submit schedules and bids to RTO w/o
FTRs RTO computes/discovers market-clearing Ps
and Qs
RTO settles energy imbalances implicit FTR
trades
Efficient Solution Is Obtained Automatically
36
Case 4 Generation Becomes Cheaper at A
GenCo Offers G5 2,000 _at_ 45
A
GenCo Offers G1 1,200 _at_ 20 G2 1,000 _at_ 24
? 1,000
(Was 35)
? 750
C
3,000
(Was 30)
? 1,600
GenCo Offers G3 1,500 _at_ 36 G4 1,000 _at_ 40
B
Optimal Dispatch, Flows and Prices Change How
Do Flowgate/FGR and LMP/FTR Models Cope?
37
Case 4 New Optimal Dispatch LMPs
GenCo Offers G5 2,000 _at_ 45
A
GenCo Offers G1 1,200 _at_ 20 G2 1,000 _at_ 24
? 1,000
?750
C
3,000
? 1,600
GenCo Offers G3 1,500 _at_ 36 G4 1,000 _at_ 40
B
Spot market finds new, efficient Ps Qs w/o FTRs
FTR holders displaced in dispatch collect FTR
value FTRs can be traded/reconfigured over time
FTRs can provide good hedges even if they are not
traded to match new dispatch/market outcome
38
Case 5 Generation Becomes Cheaper at B
(Was 20)
GenCo Offers G5 2,000 _at_ 45
A
GenCo Offers G1 1,200 _at_ 31 G2 1,000 _at_ 35
? 1,000
? 750
C
3,000
(Was 40)
? 1,600
GenCo Offers G3 1,500 _at_ 30 G4 1,000 _at_ 28
B
Dispatch, flows and prices will change again.
39
Case 5 New Optimal Dispatch LMPs
GenCo Offers G5 2,000 _at_ 45
A
GenCo Offers G1 1,200 _at_ 31 G2 1,000 _at_ 35
? 1,000
?750
C
3,000
? 1,600
GenCo Offers G3 1,500 _at_ 30 G4 1,000 _at_ 28
B
Spot market finds new, efficient Ps Qs w/o FTRs
FTR holders displaced in dispatch collect FTR
value FTRs can be traded/reconfigured over time
Efficient dispatch/pricing/trading is
automatic, even if FTR do not closely match new
dispatch