No Slide Title

Download Report

Transcript No Slide Title

A Comparative Analysis of Actual LMPs
and Estimated SRMCs in the PJM Market
prepared for
Assessing Restructured Electricity Markets: An
APPA Symposium
Washington, DC
February 5, 2007
London Economics International LLC
Dr. Serkan Bahçeci
Contact: [email protected]
Plan of presentation
Introduction
Basics – prices, timeframe, market
SRMC Modeling Methodology
Results of the Study
2
LEI has an extensive regulatory advisory background
 Focus on applied economics with quantitatively rigorous analysis to
policy and regulatory issues in the sector
 Evolved as a result of the infrastructure restructuring in the UK – now
headquartered in US, a global economic, financial, and strategic
advisory professional services firm specializing in energy and
infrastructure
 LEI has a number of complementary and cross-disciplinary practice
areas:




Regulatory economics and market design
Asset valuation, price forecasting and market analysis
Expert testimony and litigation consulting
Strategy advisory
 LEI’s advisory experiences are complemented by a unique set of tools
include network-based simulation models, cost-benefit frameworks,
models of strategic bidding behavior, and real options analysis
 Clients in energy industry include financial institutions, law firms,
market institutions, utilities and IPPs
3
Our assignment is to assess the relationship
between energy prices and marginal costs of
production




Aim to provide insight to the efficiency of market mechanism
Selected timeframe is from January 2003 to July 2006
PJM Classic only (PJM as of January 2003)
Collate actual historical LMPs

Aggregates of the load zone LMPs – Day Ahead Energy
Market
 Simulate the market-clearing price that would have resulted if
generators bid their short-run marginal costs (SRMC)
 Compare actual LMPs to simulated SRMC-based prices

Re-created a price-cost markup index for a sample of
discrete, hourly trading intervals over the timeframe based
on simulated marginal-cost based bidding
4
Plan of presentation
Introduction
Basics – prices, timeframe, market
SRMC Modeling Methodology
Results of the study
5
We have focused on the Day-Ahead LMPs in the
PJM market, as that is one of longest-lived
nodal markets in the US
 Day-ahead LMPs are financially binding and typically cover 60%
of the system load in PJM (2005 average)
 Day-ahead and real time LMPs are actually converging
 In PJM, LMP is composed of generation marginal cost +
transmission congestion costs (other markets have also
included cost of marginal losses, which PJM is also introducing)


LMPs are equal, when transmission system is unconstrained
(ignoring loss component)
LMPs vary by location, when transmission system is constrained
 LMP is based on how energy flows – not contract paths


generators get paid at generation bus LMP
load zone LMPs used for load settlement
 In order to encompass many actual conditions, our modeling
focused on ‘LMP areas’ rather than individual nodes
6
Selected timeframe is long enough to analyze trends
and the selected sample is intended to be representative
of each year
50,000
Daily Average Load (MW)
45,000
40,000
2005 daily load –
365 days
35,000

200 days for each full year
30,000

120 days for 2006
25,000
20,000
50,000
45,000
Daily Average Load (MW)
 We sampled days (rather than
hours) to capture fluctuations
within days
 The sample size is 55%
40,000
35,000
2005 selected
sample –
200 days
 Sampling is done using Latin
hypercube sampling method,
that allows multiple categories
and makes sure each category
is equally represented in the
sample

months
25,000

days of the week
20,000

load level

daily load variation
30,000
7
PJM Classic has 10 transmission zones, based
on the service territories of the electric
distribution companies
8
While coal is concentrated in the west, Eastern
PJM is dominated by gas-fired resources
Primary Fuel Type of
Major Generating Assets
Map also shows
state and county
borders and
location of the 10
load zones
Source: Energy Velocity
9
… and population density (proxy for load) is
higher in the east compared to the west
10
Network topology depends on transmission
interfaces, which are not perfectly aligned with
the boundaries of the load zones
Map is showing:
Eastern
Interface
Western
Interface
Central
Interface

Transmission
lines greater
than 200 kV

Generation units
greater than 100
MW

Transmission
interfaces
recreated from
PJM document
titled “GIS
Applied to PJM
Transmission
System”,
October 18,
2005.
Source: Energy Velocity
11
Our five region topology is consistent with the
differences in zonal LMPs
Central Interface
Eastern Interface
Western Interface
PENELEC
METED (69%)
PPL
BG&E
PEPCO
METED (31%)
AECO
JCPL
PSEG
PECO
DPL
12
n03
M
ar
-0
M 3
ay
-0
3
Ju
l-0
Se 3
p0
N 3
ov
-0
3
Ja
n04
M
ar
-0
M 4
ay
-0
4
Ju
l-0
Se 4
p0
N 4
ov
-0
4
Ja
n05
M
ar
-0
M 5
ay
-0
5
Ju
l-0
Se 5
p0
N 5
ov
-0
5
Ja
n06
M
ar
-0
M 6
ay
-0
6
Ju
l-0
6
Ja
$/MWh
LMPs are consistently lower in the western
zones during the study timeframe
$100
$90
$80
$70
$60
$50
$40
$30
$20
$10
P
M
B
E
D
12 per. Mov. Avg. (M)
$0
Source: PJM
13
Plan of presentation
Introduction
Basics – prices, timeframe, market
SRMC Modeling Methodology
Results of the Study
14
Short run marginal cost includes fuel costs,
environmental costs and variable O&M costs
 Environmental costs cover SO2 and NOx allowances
 Variable O&M costs include

operations supervision


coolants and water
expenditures;
miscellaneous steam power
expenses

maintenance supervision
and engineering

pumped storage expenses

equipment expenses

structures maintenance

steam expenses from other
sources

boiler maintenance

maintenance of reservoirs
expenses of transferred
steam

maintenance of electric plant

miscellaneous maintenance


electric expenses
15
Fuel is the biggest component in SRMC
calculations
$/MWh
$90
$80
Environmental costs
$70
$60
Fuel Costs
$50
Variable O&M Costs
$40
$30
$20
$10
$0
Nuclear
Coal
CCGT
Gas fired peaker
16
n0
M 3
ar
M 03
ay
-0
Ju 3
l-0
Se 3
p0
N 3
ov
-0
Ja 3
n0
M 4
ar
M 04
ay
-0
Ju 4
l-0
Se 4
p0
N 4
ov
-0
Ja 4
n0
M 5
ar
M 05
ay
-0
Ju 5
l-0
Se 5
p0
N 5
ov
-0
Ja 5
n0
M 6
ar
M 06
ay
-0
Ju 6
l-0
6
Ja
$/MMBtu
Average monthly fuel costs have rising trends
through the timeframe
$18
$16
$14
$12
Light Oil
Heavy Oil
Coal
NG
$10
$8
$6
$4
$2
$0
17
Price-cost markup index is unit free, allowing
comparisons across regions and time
 Price-cost markup (in dollar terms) is informative but cannot be
used for binary comparisons
 Price-cost markup index (markup over price) is used by PJM as
well as other ISOs, and measures the percentage of the pricecost markup of the actual LMP
Actual LMP  SRMC ( Modeled LMP)
Actual LMP
18
We simulated the market using POOLMod, our
proprietary market simulation software
 For each hour we used actual historical

load

transmission flows

imports and exports
 Daily fuel prices are used for thermal plants

For gas and oil fired units the closest NG or oil hub (spot) price is
used

Coal prices take long-term contracts and transportation costs into
account
 The availability of each thermal plant is determined using EPA’s
CEMS database and other sources

CEMS cover more than 90% of the nameplate capacity
 For hydro plants we used monthly production data as submitted
in EIA Form 906
19
POOLMod simulates the actual market conditions to
calculate SRMC-based market clearing prices
Stage 1- Commitment (daily)
No
Not committed
for dispatch
Stage
Stage
2 - Dispatch
2
(hourly)
Is plant
available?
Competitive
SRMC
bidding
assumed
Yes
Schedule hydro
based on optimal
duration of operation
LMPs set equal to the
bid of the most
expensive dispatched
resource
(MW)
(MW)
Review technical
capabilities of
unitsunits
and offers
Resources dispatched
based on offer price subject
to transmission limits, as
well as plant-specific
technical constraints
Hourly
demand
Demand duration
curve plus
reserves for a
typical day
20
Plan of presentation
Introduction
Basics – prices, timeframe, market
SRMC Modeling Methodology
Results of the Study
21
Monthly averages of price-cost markup indices
show variation across regions
25
20
15
10
Peak markup
index, %
20% + %
20
15% - - 15%
10%
0%
5% - 1 5%
0% -
5
0
P
M
B
E
D
22
… and through time
14%
12%
Peak
Off-Peak
Markup index, %
10%
8%
2 per. Mov. Avg.
(Peak)
2 per. Mov. Avg.
(Off-Peak)
6%
4%
2%
0%
Ja
n0
M 3
ar
-0
M 3
ay
-0
3
Ju
l-0
Se 3
p0
N 3
ov
-0
Ja 3
n0
M 4
ar
-0
M 4
ay
-0
4
Ju
l-0
Se 4
p0
N 4
ov
-0
Ja 4
n0
M 5
ar
-0
M 5
ay
-0
5
Ju
l-0
Se 5
p0
N 5
ov
-0
Ja 5
n0
M 6
ar
-0
M 6
ay
-0
6
Ju
l-0
6
-2%
Load-weighted monthly peak and off-peak price-cost markup
indices
23
Monthly price-cost markup indices and monthly
averages of regional load are not correlated
30%
P - Peak
B - Off-peak
P - Off-peak
E - Peak
M - Peak
E - Off-peak
M - Off-peak
D - Peak
B - Peak
D - Off-peak
Price-cost markup index (%)
25%
20%
15%
10%
5%
0%
-5%
0
5,000
10,000
15,000
20,000
Regional load (MW)
25,000
30,000
24
-5%
Peak Markup index
-10%
30%
-10%
Ja
n03
M
ar
-0
3
M
ay
-0
3
Ju
l-0
3
Se
p03
N
ov
-0
3
Ja
n04
M
ar
-0
4
M
ay
-0
4
Ju
l-0
4
Se
p04
N
ov
-0
4
Ja
n05
M
ar
-0
5
M
ay
-0
5
Ju
l-0
5
Se
p05
N
ov
-0
5
Ja
n06
M
ar
-0
6
M
ay
-0
6
Ju
l-0
6
Overall Markup index
n03
M
ar
-0
3
M
ay
-0
3
Ju
l-0
3
Se
p03
N
ov
-0
3
Ja
n04
M
ar
-0
4
M
ay
-0
4
Ju
l-0
4
Se
p04
N
ov
-0
4
Ja
n05
M
ar
-0
5
M
ay
-0
5
Ju
l-0
5
Se
p05
N
ov
-0
5
Ja
n06
M
ar
-0
6
M
ay
-0
6
Ju
l-0
6
Ja
n03
M
ar
-0
M 3
ay
-0
3
Ju
l-0
3
Se
p03
N
ov
-0
3
Ja
n04
M
ar
-0
4
M
ay
-0
4
Ju
l-0
4
Se
p04
N
ov
-0
4
Ja
n05
M
ar
-0
5
M
ay
-0
5
Ju
l-0
5
Se
p05
N
ov
-0
5
Ja
n06
M
ar
-0
M 6
ay
-0
6
Ju
l-0
6
0%
Ja
Ja
n03
M
ar
-0
3
M
ay
-0
3
Ju
l-0
3
Se
p03
N
ov
-0
3
Ja
n04
M
ar
-0
4
M
ay
-0
4
Ju
l-0
4
Se
p04
N
ov
-0
4
Ja
n05
M
ar
-0
5
M
ay
-0
5
Ju
l-0
5
Se
p05
N
ov
-0
5
Ja
n06
M
ar
-0
6
M
ay
-0
6
Ju
l-0
6
Region P (Penelec) has relatively higher markup
indices compared to the other regions
30%
30%
25%
25%
20%
20%
15%
15%
10%
10%
5%
5%
0%
-5%
Off-Peak Markup index
-10%
30%
All periods
25%
25%
20%
20%
15%
15%
10%
10%
5%
5%
0%
0%
-5%
-5%
Off-peak periods
-10%
Peak periods
25
15%
Off-peak periods
n03
M
ar
-0
3
M
ay
-0
3
Ju
l-0
3
Se
p03
N
ov
-0
3
Ja
n04
M
ar
-0
4
M
ay
-0
4
Ju
l-0
4
Se
p04
N
ov
-0
4
Ja
n05
M
ar
-0
5
M
ay
-0
5
Ju
l-0
5
Se
p05
N
ov
-0
5
Ja
n06
M
ar
-0
6
M
ay
-0
6
Ju
l-0
6
20%
Overall Markup index
Ja
25%
n03
M
ar
-0
3
M
ay
-0
3
Ju
l-0
3
Se
p03
N
ov
-0
3
Ja
n04
M
ar
-0
4
M
ay
-0
4
Ju
l-0
4
Se
p04
N
ov
-0
4
Ja
n05
M
ar
-0
5
M
ay
-0
5
Ju
l-0
5
Se
p05
N
ov
-0
5
Ja
n06
M
ar
-0
6
M
ay
-0
6
Ju
l-0
6
n03
M
ar
-0
M 3
ay
-0
3
Ju
l-0
3
Se
p03
N
ov
-0
3
Ja
n04
M
ar
-0
4
M
ay
-0
4
Ju
l-0
4
Se
p04
N
ov
-0
4
Ja
n05
M
ar
-0
5
M
ay
-0
5
Ju
l-0
5
Se
p05
N
ov
-0
5
Ja
n06
M
ar
-0
M 6
ay
-0
6
Ju
l-0
6
Ja
30%
Ja
n03
M
ar
-0
3
M
ay
-0
3
Ju
l-0
3
Se
p03
N
ov
-0
3
Ja
n04
M
ar
-0
4
M
ay
-0
4
Ju
l-0
4
Se
p04
N
ov
-0
4
Ja
n05
M
ar
-0
5
M
ay
-0
5
Ju
l-0
5
Se
p05
N
ov
-0
5
Ja
n06
M
ar
-0
6
M
ay
-0
6
Ju
l-0
6
Ja
Region E (AECO, JCPL, PECO, PSEG) had high
markups through 2004 to 2005
30%
25%
Peak Markup index
Off-Peak Markup index
20%
15%
All periods
10%
10%
5%
5%
0%
0%
-5%
-5%
-10%
30%
-10%
30%
Peak periods
25%
25%
20%
20%
15%
15%
10%
10%
5%
5%
0%
0%
-5%
-5%
-10%
-10%
26
25%
20%
Overall Markup index
0%
-5%
-10%
30%
Off-peak periods
25%
20%
15%
10%
5%
0%
-5%
Ja
n03
M
ar
-0
3
M
ay
-0
3
Ju
l-0
3
Se
p03
N
ov
-0
3
Ja
n04
M
ar
-0
4
M
ay
-0
4
Ju
l-0
4
Se
p04
N
ov
-0
4
Ja
n05
M
ar
-0
5
M
ay
-0
5
Ju
l-0
5
Se
p05
N
ov
-0
5
Ja
n06
M
ar
-0
6
M
ay
-0
6
Ju
l-0
6
Ja
n03
M
ar
-0
M 3
ay
-0
3
Ju
l-0
3
Se
p03
N
ov
-0
3
Ja
n04
M
ar
-0
4
M
ay
-0
4
Ju
l-0
4
Se
p04
N
ov
-0
4
Ja
n05
M
ar
-0
5
M
ay
-0
5
Ju
l-0
5
Se
p05
N
ov
-0
5
Ja
n06
M
ar
-0
M 6
ay
-0
6
Ju
l-0
6
30%
an
-0
3
ar
-0
3
ay
-0
3
Ju
l-0
3
ep
-0
3
ov
-0
3
an
-0
4
ar
-0
4
ay
-0
4
Ju
l-0
4
ep
-0
4
ov
-0
4
an
-0
5
ar
-0
5
ay
-0
5
Ju
l-0
5
ep
-0
5
ov
-0
5
an
-0
6
ar
-0
6
ay
-0
6
Ju
l-0
6
Ja
n03
M
ar
-0
3
M
ay
-0
3
Ju
l-0
3
Se
p03
N
ov
-0
3
Ja
n04
M
ar
-0
4
M
ay
-0
4
Ju
l-0
4
Se
p04
N
ov
-0
4
Ja
n05
M
ar
-0
5
M
ay
-0
5
Ju
l-0
5
Se
p05
N
ov
-0
5
Ja
n06
M
ar
-0
6
M
ay
-0
6
Ju
l-0
6
Region M (PPL and portion of METED) has many
statistically insignificant off-peak indices
30%
25%
Peak Markup index
Off-Peak Markup index
20%
15%
15%
10%
10%
5%
5%
0%
-5%
All periods
-10%
30%
25%
20%
15%
10%
5%
0%
-5%
-10%
-10%
Peak periods
27
Peak periods
15%
15%
10%
10%
5%
5%
0%
0%
-5%
-5%
-10%
30%
-10%
30%
Off-peak periods
n03
M
ar
-0
3
M
ay
-0
3
Ju
l-0
3
Se
p03
N
ov
-0
3
Ja
n04
M
ar
-0
4
M
ay
-0
4
Ju
l-0
4
Se
p04
N
ov
-0
4
Ja
n05
M
ar
-0
5
M
ay
-0
5
Ju
l-0
5
Se
p05
N
ov
-0
5
Ja
n06
M
ar
-0
6
M
ay
-0
6
Ju
l-0
6
20%
Overall Markup index
Ja
25%
n03
M
ar
-0
3
M
ay
-0
3
Ju
l-0
3
Se
p03
N
ov
-0
3
Ja
n04
M
ar
-0
4
M
ay
-0
4
Ju
l-0
4
Se
p04
N
ov
-0
4
Ja
n05
M
ar
-0
5
M
ay
-0
5
Ju
l-0
5
Se
p05
N
ov
-0
5
Ja
n06
M
ar
-0
6
M
ay
-0
6
Ju
l-0
6
n03
M
ar
-0
M 3
ay
-0
3
Ju
l-0
3
Se
p03
N
ov
-0
3
Ja
n04
M
ar
-0
M 4
ay
-0
4
Ju
l-0
4
Se
p04
N
ov
-0
4
Ja
n05
M
ar
-0
M 5
ay
-0
5
Ju
l-0
5
Se
p05
N
ov
-0
5
Ja
n06
M
ar
-0
M 6
ay
-0
6
Ju
l-0
6
Ja
30%
Ja
Ja
n03
M
ar
-0
3
M
ay
-0
3
Ju
l-0
3
Se
p03
N
ov
-0
3
Ja
n04
M
ar
-0
4
M
ay
-0
4
Ju
l-0
4
Se
p04
N
ov
-0
4
Ja
n05
M
ar
-0
5
M
ay
-0
5
Ju
l-0
5
Se
p05
N
ov
-0
5
Ja
n06
M
ar
-0
6
M
ay
-0
6
Ju
l-0
6
Region B (BGE, PEPCO and remaining METED)
has a smaller difference between peak and offpeak indices
30%
25%
Peak Markup index
Off-Peak Markup index
20%
All periods
25%
25%
20%
20%
15%
15%
10%
10%
5%
5%
0%
0%
-5%
-5%
-10%
-10%
28
n03
M
ar
-0
3
M
ay
-0
3
Ju
l-0
3
Se
p03
N
ov
-0
3
Ja
n04
M
ar
-0
4
M
ay
-0
4
Ju
l-0
4
Se
p04
N
ov
-0
4
Ja
n05
M
ar
-0
5
M
ay
-0
5
Ju
l-0
5
Se
p05
N
ov
-0
5
Ja
n06
M
ar
-0
6
M
ay
-0
6
Ju
Ja l-0
6
n03
M
ar
-0
3
M
ay
-0
3
Ju
l-0
3
Se
p03
N
ov
-0
3
Ja
n04
M
ar
-0
4
M
ay
-0
4
Ju
l-0
4
Se
p04
N
ov
-0
4
Ja
n05
M
ar
-0
5
M
ay
-0
5
Ju
l-0
5
Se
p05
N
ov
-0
5
Ja
n06
M
ar
-0
6
M
ay
-0
6
Ju
l-0
6
Ja
n03
M
ar
-0
M 3
ay
-0
3
Ju
l-0
3
Se
p03
N
ov
-0
3
Ja
n04
M
ar
-0
4
M
ay
-0
4
Ju
l-0
4
Se
p04
N
ov
-0
4
Ja
n05
M
ar
-0
5
M
ay
-0
5
Ju
l-0
5
Se
p05
N
ov
-0
5
Ja
n06
M
ar
-0
M 6
ay
-0
6
Ju
l-0
Ja 6
n03
M
ar
-0
3
M
ay
-0
3
Ju
l-0
3
Se
p03
N
ov
-0
3
Ja
n04
M
ar
-0
4
M
ay
-0
4
Ju
l-0
4
Se
p04
N
ov
-0
4
Ja
n05
M
ar
-0
5
M
ay
-0
5
Ju
l-0
5
Se
p05
N
ov
-0
5
Ja
n06
M
ar
-0
6
M
ay
-0
6
Ju
l-0
6
Ja
Region D (Delmarva peninsula) has the highest
markup index values in late 2004
30%
15%
30%
25%
Overall Markup index
Peak Markup index
25%
20%
Off-Peak Markup
index
20%
15%
Peak periods
10%
10%
5%
5%
0%
0%
-5%
-5%
30%
-10%
30%
-10%
All periods
25%
Off-peak periods
25%
20%
20%
15%
15%
10%
10%
5%
5%
0%
0%
-5%
-5%
-10%
-10%
29