Energy and Ancillary Service Dispatch for the Interim ISO New England Electricity Market1
Kwok W. Cheung, Sr. Member
Payman Shamsollahi, Member
ALSTOM ESCA Corporation
Bellevue, WA 98004 USA
David Sun, Member
Jim Milligan
Mike Potishnak
ISO New England Inc.
Holyoke, MA 01040 USA
Abstract: Since the Federal Energy Regulatory Commission’s
(FERC) order No. 888 has mandated the establishment of
unbundled electricity markets in the newly deregulated
environment, competitive bidding of ancillary services, along with
bidding of energy, becomes increasingly important. In this paper,
an optimization-based framework for solving a multi-commodity
electricity market dispatch problem is presented. In compliance
with New England Power Pool (NEPOOL) Market Rules and
Procedures, a hybrid dispatch method which combines the
sequential dispatch method with the joint dispatch method is
proposed to solve the energy and ancillary dispatch problem for
ISO New England (ISO-NE). Numerical results on a 6-unit test
system and the 324-unit ISO-NE system are included.
Keywords: Deregulation; Energy and reserve dispatch; System
operator; Automatic generation control; Electricity market
I. INTRODUCTION
ISO New England Inc. (ISO-NE) was established on July 1, 1997
following its approval by the Federal Energy Regulatory
Commission (FERC). In the newly deregulated environment, ISONE will operate both the bulk power generation and transmission
systems and the wholesale electric power market for New England.
In compliance with FERC’s orders, like other competitive power
pools, ISO-NE is designing and implementing new methods of
operation and dispatch that are based on bids and offers submitted
by market participants [1].
In this “brave new world” of deregulation, energy price is no longer
cost-driven but energy itself is treated as a competitive commodity
[2]. Market participants offer energy into the competitive pool for
each trading interval. At the first glance, it may seem that the
energy dispatch process in a price-based competitive pool is very
similar to the conventional, cost-based dispatch except the
incremental costs of energy production are replaced by bid prices.
However, it should be noted that bid prices can vary according to
bidding strategies that a supplier might follow and may be quite
different from actual costs which are much more predictable and
manageable from a dispatcher’s point of view.
The other challenge for operation and dispatch is the issue of
dealing with competitive bidding of other products such as
contingency reserves and regulation capacity, or the so-called
ancillary services. As the electric industry moves toward full
competition and deregulation, the various services previously
provided by utilities are being unbundled. Much of the attention
given to some ISO development has focused on the structures of
markets for energy and transmission, but the design of markets for
ancillary services also requires serious attention and is getting more
critical.
Beside ISO-New England Electricity Market, other electricity
markets including New Zealand Electricity Market [3], Australian
National Electricity Market [4], WEPEX in California [5], also
allow competitive bidding of energy and ancillary services. In these
markets, market participants can bid in ancillary services, similar in
principle to their bids for the energy commodity. The ISO would
then dispatch these other types of bids, along with energy bids,
according to applicable market rules.
In the literature, several alternative methods are available for
energy and ancillary service dispatch: merit-order-based dispatch,
sequential dispatch, and joint dispatch. These methods reflect
tradeoffs among the rigor of feasibility and optimality for efficient
market operation, the practicality of results for physical power
system operation, and the complexity of implementing bid clearing
application software [6]. This paper presents experience gained
from the design and implementation of the ISO-NE Electricity
Market System project. In particular, our focus is on our proposed
dispatch method for energy and ancillary services, and the
framework which supports the development of such a dispatch
method. The rest of the paper is organized as follows. Section II
describes the basic market structure of ISO-NE. An optimizationbased framework for solving multi-commodity electricity market
dispatch problem is presented in Section III. In compliance with
NEPOOL Market Rules and Procedures [7], a hybrid dispatch
method which combines the sequential dispatch method with joint
dispatch method to solve the energy and ancillary dispatch problem
of ISO-NE is presented in Section IV. Numerical results on a 6-unit
test system and the 324-unit ISO-NE system are included in
Section V. Section VI concludes the paper.
II. MARKET STRUCTURE OF ISO-NE
The evolution of ISO-NE operation to market orientation includes
the following seven wholesale electricity market commodities to be
provided by NEPOOL market participants:
1.
1
0-7803-5478-8/99/$10.00 © 1999 1EEE.
Energy
2.
3.
4.
5.
6.
7.
Automatic Generation Control (AGC)
Ten Minute Spinning Reserve (TMSR)
Ten Minute Non-Spinning Reserve (TMNSR)
Thirty Minute Operating Reserve (TMOR)
Operable Capability
Installed Capability
Of these seven commodities, only the first five which affect system
operation decisions in real-time are addressed in this paper. All
commodities except the first one are considered as ancillary
services.
Energy is the primary commodity of all markets. An energy bid is
an offer to supply or consume energy (MWh) at a price ($). The
energy bid must include the entire capability range of the resource,
which may be modeled, for each hour, by up to ten blocks of
energy with monotonically non-decreasing prices. Zero priced
energy block may be used to represent self-scheduled energy.
AGC is the regulating capability, under automatic generation
control, that responds in an effort to continuously balance the ISONE Control Area’s supply resources with minute to minute load
variations in order to meet the NERC and NPCC Control
Performance Standards. An AGC bid will be in the form of $ per
hour to provide AGC service. Regulating capacity is converted to
an AGC commodity called a “Reg” which is a weighted average of
the MWs of response available in ten minutes and in sixty minutes.
The scheduling function optimizes the clearing of bids for energy,
regulating capacity and reserve to meet forecasted hourly demand.
The scheduling horizon should be variable and selectable by the
operator, but not to exceed 48 hours. The length of each trading
period for scheduling is one hour except for the first period which
is the fractional hour remaining in the current hour. Only valid bids
from available resources will be scheduled. The scheduling
function will not perform inter-temporal optimization except for
ramp rate constraints from the previous trading interval.
The dispatch function runs automatically every five minutes. Its
bid-clearing results are used to update generation setpoints as input
to automatic generation control. In addition, five-minute dispatch
computes clearing prices for the ancillary service markets and
calculates Lost Opportunity Costs (LOC) for TMSR and AGC
markets.
For the AGC market, the dispatch function ranks eligible AGC bids
according to rules reflecting system condition and regulation
capacity price, and recommends to the operator which generators
should be selected for AGC duty. Based on the rank from the
dispatch function, the operator can make rational decisions of
assigning generators on or off AGC.
III. OPTIMIZATION FRAMEWORK
TMSR is a resource capacity synchronized to the system, which is
(a) able to immediately begin to supply energy or reduce demand,
(b) fully available within ten minutes, and (c) able to be sustained
for a period of at least thirty minutes to provide first contingency
protection. A TMSR bid submitted by a market participant is an
offer to supply TMSR to the market at a price ($ per MW).
The bid-clearing model of the ISO-NE dispatch and scheduling
functions is formulated as a constrained optimization problem. The
objective is to maximize market benefits which is equivalent to
minimizing the sum of costs to energy and ancillary service offers.
The objective function of the bid-clearing model is described as
follows:
TMNSR is a resource capacity non-synchronized to the system,
which is (a) able to supply energy or reduce demand, (b) fully
available within ten minutes, and (c) able to be sustained for a
period of at least thirty minutes to provide first contingency
protection. A TMNSR bid submitted by a market participant is an
offer to supply TMNSR to the market at a price ($ per MW).
(1)
TMOR is a resource capacity non-synchronized to the system,
which is (a) able to supply energy or reduce demand, (b) fully
available within thirty minutes, and (c) able to be sustained for a
period of at least sixty minutes to provide second contingency
protection. A TMOR bid submitted by a market participant is an
offer to supply TMOR to the market at a price ($ per MW).
For the interim market, market participants submit hourly bids for
the next day for all energy and ancillary services. ISO-NE uses this
information to dispatch generating resources based on bid-clearing
methodology. The interim market methodology calls for
dispatching and scheduling based on optimized use of daily bids
for energy, reserve and regulating capacity. Most information
necessary for the settlement process will be derived from market
data and dispatch solutions. The new dispatch and scheduling
subsystems require interfaces with the existing Energy
Management System (EMS), a bidding and settlement system, and
a resource commitment application.
The objective of scheduling and dispatch is to provide optimal
solutions to the ISO New England market using bids of energy,
regulating capacity and reserve to satisfy the demand for energy
while maintaining the required levels of generating reserve and
regulating capacity and operating within the safe capability of
generation and transmission equipment.
Min
 TBMW (mni, i, tt, ti, t, ns) * TBP (mni, i, tt, ti, t, ns)
mni i
tt
ti
ns
where mni is an index to a market node; A market node defines the
location of a market participant; n denotes an index to a bus
(reference node) of the network; tt denotes an index to a bid type in
the bid type set {ENOF, LDOF, AGC, TMSR, TMNSR, TMOR};
ENOF and LDOF denote energy offer and demand bid,
respectively; ti denotes an index to a trader; t denotes the study
interval within the study period; ns denotes an index to a MW band
of a bid.
TBMW(mni,i,tt,ti,t,ns) denotes the band MW of band ns of bid type
tt at market node mni, reference node i at time t by trader ti. By
convention, Band MW is negative for demand bids and positive for
all the other bids.
A. Band (Bid Block) MW Limits
Band MW is constrained by band MW limit.
For energy and reserve bid blocks (positive):
(2.1)
0  TBMW (mni, i, tt, ti, t, ns)  TBMW max (mni, i, tt, ti, t, ns)
For energy demand bid blocks (negative):
(2.2)
TBMW max (mni, i, tt, ti, t, ns)  TBMW (mni, i, tt, ti, t, ns)  0
Note that TBP(mni,i,tt,ti,t,ns) denotes the band price ($/MW) of
band ns of bid type tt at commercial market node mni, reference
node i at time t by trader ti. Band price, TBP(mni,i,tt,ti,t,ns), is
calculated with reference to the reference node based on the
original band price at market node mni and intra-regional loss
factor as in (3):
(9)
TMNSR(mni, ti, t ) 
 TBMW (mni, i, TMNSR, ti, t, ns),
i
TBP raw (mni, i, tt, ti, t , ns)
TBP (mni, i, tt, ti, t , ns) 
IARLF (mni, i, t )
(3)
(10)
TMOR (mni, ti, t ) 
ns
 TBMW (mni, i, TMOR , ti, t , ns),
i
ns
where TBP raw(mni, i, tt , ti , t , ns) denotes the raw band price ($/MW)
at market node mni submitted by trader ti, and IARLF(mni,i,t)
denotes the intra-area loss factor at market node mni connected to
reference node i at time t.
are the capacity for TMSR, TMNSR and TMOR, respectively. For
generators on AGC, the following additional constraints apply:
B. Ramp Rate Constraints
(10) MNGEN (mni, ti, t )  DnAGC (mni, ti, t )  ALL(mni, ti, (mni, ti), t )
A generating unit has limits on its ability to move from one level of
MW generation to another within a specified time period.
Generation ramp-rate constraints are modeled as follows:
where DnAGC (mni, ti , t ) denotes the capacity assigned to AGC in
the downward direction, AHL(mni,ti,  ,t) and ALL(mni,ti,  ,t) are
the automatic high and low limits of the selected AGC range for the
AGC-capable generator of trader ti at market node mni at time t,
respectively.
Up-ramp rate constraints are:
MNGEN (mni, ti, t )  MNGEN (mni, ti, t  1)   * RR (mni, ti)
(4.1)
Down-ramp rate constraints are:
(4.2) MNGEN (mni, ti, t  1)  MNGEN (mni, ti, t )   * RR (mni, ti)
(9) MNGEN (mni, ti, t )  UpAGC(mni, ti, t )  AHL(mni, ti, (mni, ti), t )
The following inequalities describe the constraints of ten-minute
reserve coupling and thirty-minute reserve coupling for generators,
respectively:
(11)
where
MNGEN (mni, ti, t )
(5)


TBMW (mni, i,1, ti, t , ns ),
if (mni, ti, t ) is dispatchable

i ns
 FixedMW
(mni, ti, t ),
if (mni, ti, t ) is non - dispatchable


denotes the MW energy generation at market node mni by trader ti
at time t;  denotes the dispatch period in minute: 60 for 60-min
pre-dispatch scheduling, and 5 for 5-min dispatch;
FixedMW(mni,ti,t) is the fixed basepoint at market node mni by
trader ti at time t if the generator at time t is considered as fixed.
(6)

ARR (mni, ti, ar ),

RR (mni, ti)  
ar  ( mni,ti )

 MRR (mni, ti),

if ( mni, ti) is on AGC
if (mni, ti) is off AGC
denotes the ramp rate limit (MW/min) for the generator of trader ti
at market node mni; ARR(mni,ti,ar) and MRR(mni,ti) denote the
automatic response rate and manual response rate, respectively; NR
denotes the number of AGC ranges of trader ti at market node mni;
 ( mni, ti ) denotes the AGC range selected by the dispatcher for
each generator on AGC.
C. Capacity Constraints
(7) MNGEN (mni, ti, t )  UpAGC(mni, ti, t )  TMSR(mni, ti, t )
 TMNSR(mni, ti, t )  TMOR (mni, ti, t )  MHL (mni, ti, t )
where MHL(mni, ti , t ) denotes the maximum of capacity available,
UpAGC(mni, ti , t ) denotes the capacity assigned to AGC in the
upward direction,
TMSR (mni, ti, t ) 
 TBMW (mni, i, TMSR, ti, t , ns),
i
ns
10 * MRR (mni, ti)  TMSR (mni, ti, t ), if (mni, ti) is online

if (mni, ti) is offline
Claim10(mni, ti),
TMNSR(mni, ti, t )  TMOR (mni, ti, t )
30 * MRR (mni, ti)  TMSR(mni, ti, t ), if (mni, ti) is online

if (mni, ti) is offline
Claim 30(mni, ti),
Claim10(mni,ti) and Claim30(mni,ti) are the capacity of the trader
ti at market node mni can be claimed in ten minutes and thirty
minutes, respectively, when the trader is offline.
D. System Requirements
The following equality and inequalities characterize the system
requirements of ISO-NE:
(13)
 MNGEN (mni, ti, t )   (t ),
(14)
mni ti
(15)
(17)
 TMSR(mni, ti, t )   (t ),
mni ti
 TMNSR(mni, ti, t )   (t ),
(16)
mni ti
 TMOR (mni, ti, t )   (t ),
mni ti
UpReg (mni, ti, t )   (t ),
mni ti
(19)
Generators have limited capacities to produce energy and at the
same time to provide ancillary services. These are defined as
capacity constraints as follows:
(8)
(12)
TMNSR(mni, ti, t )
(18)
 DnReg (mni, ti, t )   (t ),
mni ti
 Reg (mni, ti, t )   (t ),
mni ti
where  (t ) is the system demand which includes area net
interchange and load,  (t ) ,  (t ) , and  (t ) are the requirements
for TMSR, TMNSR, and TMOR, respectively,  (t ) ,  (t ) , and
 (t ) are the requirements for upward, downward, and total AGC
regulation, respectively.
(20)
UpRe g (mni, ti, t )
 min[ Re g (mni, ti, t ), AHL(mni, ti, , t )  MNGEN (mni, ti, t )]
(21)
DnRe g (mni, ti, t )
 min[ Re g (mni, ti, t ), MNGEN (mni, ti, t )  ALL(mni, ti, t )]
(22)
Re g (mni, ti, t )
 k * min[ 10 * ARR , AHL  ALL]
 (1  k ) * min[ 60 * ARR , AHL  ALL]
where k is a tunable constant used to weight the mix of ten minute
and sixty minute response for the AGC commodity.
Note that the clearing of the AGC market (20)-(22) is a set of
procedures based on ranked price and energy dispatch targets. As a
result, the optimization framework does not directly enforce AGC
requirements for the system. However, as part of the hybrid
dispatch method described in the next section, the effects of AGC
clearing are fully recognized by the energy and other reserve
markets.
The AGC Bid Evaluation process calculates potential Regs
(up/down & total) and computes AGC ranking price for each
eligible AGC range using an estimated energy-only solution of
LP1. The AGC ranking price includes estimates of payments for
AGC capacity reservation, AGC service, lost opportunity, and
uplift. It also has some look-ahead features to improve stability of
the AGC market.
The AGC Suggestor process suggests generators and their ranges
for AGC duty, based on AGC ranking price, until the upward,
downward, and total Reg requirements are met. The AGC
suggestion process is only a recommendation to the operator for the
dispatch function. For the scheduling function, AGC suggestion
could actually be used for clearing the AGC market.
IV. HYBRID DISPATCH METHOD
[LP1] E
Sequential dispatch and joint (simultaneous) dispatch are two
current alternatives for dispatching energy and ancillary services.
Based on a priority sequence of market commodities, the sequential
approach progressively reduces available capacity of each resource
to meet system requirements for each commodity. This approach is
intuitive. However, its inability to determine the best tradeoffs in
sharing limited resource capacity for energy and ancillary services
may result in higher prices or even insufficient supply for the lower
priority commodities. On the other hand, the simultaneous
approach is based on formulating the dispatch problem in the
context of constrained optimization which provides improved
coordination of energy and ancillary service dispatch to achieve the
most secure and economic solution.
AGC Bid Evaluation
AGC Suggestor
[LP2] E(A)
For the interim market of ISO-NE, it is necessary to explicitly
evaluate TMSR lost opportunity, AGC lost opportunity, and the
clearing of AGC market which require the intermediate solutions of
a sequential dispatch. In order to meet the functional requirements
for the interim market, a hybrid approach of combining sequential
dispatch with joint dispatch is developed to clear all markets. The
hybrid dispatch method consists of a series of Linear Programming
(LP) solutions as depicted in Figure 1 which not only provide
intermediate solutions for the calculation of lost opportunity costs
as well as the ranking and the clearing of AGC market, but also try
to minimize the negative impact on performance and effectiveness
that sequential clearing may result. Note that the prime character [’]
associated with each market in a particular LP solution denotes that
the dispatch solution of that market is frozen when the LP solution
is performed. For example, E’ in LP4 means the energy dispatch
solution is frozen in the fourth LP solution.
AGC Capacity
Reservation
[LP3] E+A'+T
[LP4] E'+A'+T'+N+O
LOC Calculation
Figure 1: Flow Chart of ISO-NE Hybrid Dispatch Method
There are four LP solutions in the present configuration for each
dispatch run:
[LP1] E - pure energy dispatch (E) honoring manual operating limit
and manual response rate constraints
[LP2] E(A) - pure energy dispatch honoring AGC range limit and
auto response rate constraints for units selected to be on
AGC
[LP3] E+A’+T - simultaneous dispatch of energy (E) and TMSR
(T) honoring AGC constraints and AGC capacity reservation
for units selected to be on AGC
[LP4] E’+A’+T’+N+O - simultaneous dispatch TMNSR (N) and
TMOR (O) honoring previous clearing of the other markets.
The AGC Capacity Reservation process will reserve upward AGC
capability for units selected for AGC so that five minutes of
upward AGC response is excluded from the TMSR market.
All market clearing prices, except for the AGC market, are set by
the shadow prices of their corresponding market requirement
constraints, which may be over-written by floor prices under
specific, special conditions. The AGC Clearing Price (AGCCP),
on the other hand, is calculated by the following procedures:
I.
For each generator selected for AGC, calculate the Reg for the
range and response rate being used
II.
Divide its AGC bid price in $/hr by its Reg calculated in (22)
III. Set AGCCP to the highest values of $/Reg-hr
For settlement purpose, intermediate energy dispatch solutions for
LP1 and LP2 are buffered so that they can be compared against that
of LP3 for the purpose of calculating lost opportunity cost for AGC
and TMSR markets, respectively. For example, the TMSR Lost
Opportunity MW (TMSRLOMW) is determined by comparing the
generation assignment (Gen) between LP3 and LP2:
TMSRLOMW (mni, ti , t )  Gen( LP2)  Gen( LP3)
where
This section presents the results of case studies. Two case studies
are performed, one using a 6-unit system and another using ISONE system data. The first study is used to illustrate the steps
involved in the hybrid dispatch method as described in the previous
section. The second study is used to demonstrate results and the
impact of ancillary services on clearing prices on a real system. The
scheduling function is employed to obtain results for both cases.
A. 6-Unit Test System
Gen( LPn)  MNGEN (mni, ti , t ) for LPn
In this system, system demand is 1650 MW. TMSR, TMNSR and
TMOR requirements are 130 MW, 120 MW and 120 MW,
respectively. Total, up and down regulating reserve (AGC)
requirements are 90, 45 and 18 Regs, respectively. It is assumed
that the operator has chosen units C and G to be on AGC. Bid
price data for the system is shown in Table 1.
TMSR Lost Opportunity Cost ($/MW) is computed as:
TMSRLOC(mni, ti , t )
Gen ( LP 2 )
(23)
 ECP (t ) 
 EBP (mni, ti, t, MW ) dMW
Table 2 shows the MW dispatch of each unit in LP1. It also shows
the Manual High Limit (MHL), Manual Low Limit (MLL) and the
energy bid price of each unit. Note that, in this LP1 solution, units
F and G are constrained at their lower limits because of their high
energy bid price while units A, B and C are at their upper limits
because of their relatively cheap energy.
Gen ( LP 3)
TMSRLOMW (mni, ti , t )
EBP
Unit
A
B
C
D
F
G
ECP
Energy
($/MW)
10
20
30
40
50
60
AGC
($/hr)
6
5
3
2
5
1
TMSR
($/MW)
0
1
2
3
14
14.5
TMNSR
($/MW)
16.7
12
12
5.6
6.7
14
TMOR
($/MW)
16.7
12
12
5.6
6.7
14
Table 1: Bidding Prices for Different Markets and Units
0
Gen(LP3)
Gen(LP2)
MHL
MW
TMSRLOMW
Unit
A
B
C
D
F
G
Figure 2: Calculation of TMSR LOC for Settlement
where
n
(24)
EBP (mni, ti , t , MW )   TBP (mni, i, ENOF , ti , t ,  )
Energy
Cleared
(MW)
400
400
400
350
50
50
MHL
(MW)
MLL
(MW)
400
400
400
400
400
400
100
100
100
100
50
50
Energy
Bid Price
($/MW)
10
20
30
40
50
60
Table 2: Results of LP1 (Energy Solution Only)
i 1
such that
 1
TBMW
max
(mni, i, tt , ti , t , ns)  MW
ns1
and

MW 
TBMW
max
(mni, i, tt , ti , t , ns)
ns1
EBP denotes the energy bid price as a function of MW,  is the
index pointing to the bid band of MW, and ECP denotes the Energy
Clearing Price. Note that the shaded area in Figure 2 represents the
payment of TMSR Lost Opportunity for (mni,ti,t). The AGC lost
opportunities are also computed similarly in the LOC Calculation
module.
V. NUMERICAL RESULTS
The next step after LP1 in the flow chart (as shown in Fig. 1), is
AGC Bid Evaluation based on which AGC Suggestor works. At
this point, AGC Suggestor chooses units D and G to be on AGC. It
is important to note that the outcomes of AGC Suggestor is solely a
recommendation to operators. However, if the suggestion were
implemented, AGCCP would be $0.047.
Since unit C has actually been selected by the operator to be on
AGC, it is backed down from 400 MW to 300 MW in order to
honor its Automatic High Limit (AHL). In this case, the AGCCP
becomes $0.083 which is higher than that of the AGC Suggestor.
Table 3 shows the energy cleared, Automatic High/Low Limits
(AHL/ALL) and Up/Dn AGC capacity reserved.
Unit
A
B
C
Energy
Cleared
(MW)
400
400
300
AHL
(MW)
ALL
(MW)
300
300
300
200
250
250
Up AGC
Capacity
(MW)
Dn AGC
Capacity
(MW)
15
D
F
G
400
70
80
300
300
150
250
250
80
down AGC regulating requirements are 600, 210 and 90 Regs
respectively. AGC, TMSR, TMNSR and TMOR clearing prices
are $0.1, $5, $0 and $0, respectively.
30
Table 3: Results of LP2 and AGC Capacity Reservation
After LP2 and AGC Capacity Reservation, LP3 is performed which
clears energy and TMSR markets simultaneously while freezing
capacities reserved for AGC. The mutual coupling of energy, AGC
and TMSR markets are clearly observed. As a result, the most
expensive energy bid is now cleared at $50 instead of $40 in
LP1and LP2. This is because of the fact that generating capacity is
limited. Therefore, part of capacity cleared for energy has to be
backed down to make room for reserve requirements. This will
make the system to use capacity of a more expensive unit for
energy market, resulting in higher ECP. Table 4 demonstrates the
result of this study. As it is seen in the table, unit D has been
backed down 40 MW to provide TMSR. Therefore, a TMSR LOC
is expected for this unit. Unit F has picked up the extra 40 MW
(hence the reason for $50 ECP). Up/Dn AGC Regs provided by
each unit are also shown in the table.
Energy
Cleared
(MW)
400
400
300
360
110
80
Unit
A
B
C
D
F
G
TMSR
(MW)
AGC
UpReg
(Reg)
AGC
DnReg
(Reg)
50
40
40
36
63
Table 4: Results of LP3 and AGC Regs
If regulating and reserve requirements are kept constant, energy
price will vary with demand. This effect has been shown in Figure
3. In this figure, series 1 represents the results with TMSR,
TMNSR and TMOR requirement equal to 800 MW, 450 MW and
400 MW, respectively. For series 2, these requirements are 1200
MW, 550 MW and 500 MW. For both series, the AGC
requirements are kept constant.
For demands higher than 15300 MW in series 2, the simulated
system has deficiency in capacity and therefore no price is shown.
As it is seen in the figure, when reserve requirements increase, the
system goes to deficiency state sooner and the price is higher for
near saturation demands.
VI. CONCLUSIONS
This paper presents the design and implementation of the energy
and ancillary service dispatch for the interim ISO-NE electricity
market. An LP-based bid-clearing framework is presented and a
hybrid dispatch method is proposed to solve the multi-commodity
electricity market dispatch problem for ISO-NE. As of this writing,
April 1, 1999 is the tentative market opening day for the interim
electricity market in New England. After a review of the existing
market rules, potential areas of improvements which include
location-based transmission congestion pricing, demand-side
bidding, and multi-settlement have already been identified for the
final market of ISO-NE. As the ISO-NE Market System continues
to mature, experiences gained from operation will help facilitating
enhancements and implementations of market rules and operation
procedures in the future.
LP4 is the last LP to be solved. In this step, the previous solutions
for energy, AGC, and TMSR are frozen and the non-spinning
reserve markets are dispatched simultaneously. Results of LP4 are
shown in Table 5.
A
B
C
D
F
G
Energy
Cleared
(MW)
400
400
300
360
110
80
AGC
UpReg
(Reg)
AGC
DnReg
(Reg)
TMSR
(MW)
36
50
40
40
TMNSR
(MW)
TMOR
(MW)
25
Energy Price ($/MW)
Unit
30
Series1
Series2
20
15
10
5
120
120
63
Table 5: Results of LP4 (Final market solution)
TMSR and AGC lost opportunity compensation is accounted in the
last step of the flow chart, LOC Calculation. Since unit D has been
backed down to provide TMSR. This creates 40 MW of
TMSRLOMW and 10 $/MW of TMSRLOC.
0
11200
11700
12200
12700
13200
13700
14200
14700
15200
15700
Demand (MW)
Figure 4: Energy Price vs. Demand for ISO-NE System
VII. REFERENCES
B. ISO-NE System
The ISO-NE system consists of 324 generating units in 180 market
nodes. There are about 2500 constraints and 3800 variables, a big
portion of which is due to sophisticated reserve and ramp models.
The simulated data contain a total number of 1300 bid blocks.
System demand and energy clearing price are 15500 MW and $20
respectively. Reserve requirements of TMSR, TMNSR and TMOR
are 800 MW, 450 MW and 400 MW respectively. Total, up and
[1] K. Cheung, P. Shamsollahi, S. Asteriadis, J. Milligan, and M.
Potishnak, “Functional Requirments of Energy and Ancillary
Service Dispatch for the Interim ISO New England Electricity
Market”, Proceedings of IEEE/PES 1999 Winter Meeting,
New York, New York, USA.
[2] F. C. Schweppe, M. C. Caramanis, R. D. Tabors and R. E.
Robn, Spot Pricing of Electricity, Kluwer Academic
Publishers, 1998.
[3] Rules of New Zealand Electricity Market, http://www.emco.co.nz/
[4] Australian National Electricity Code, Version 2.0,
http://electricity.net.au/codrule.htm
[5] California ISO Scheduling Applications Functional
Requirements (Revised 1/21/97),
http://www.energyonline.com/wepex/reports/reports2.html#ISO_bus
[6] X. Ma, D. Sun and K. Cheung, “Energy and Reserve Dispatch
in a Multi-Zone Electricity Market”, to appear in IEEE/PES
Trans. on Power Systems.
[7] ISO New England Market Rules and Procedures,
http://www.iso-ne.com/market_rules_and_procedures/documents/
VIII. BIOGRAPHIES
Kwok W. Cheung received his B.S. from National Cheng Kung
University, Taiwan, in 1986, his M.S. from University of Texas at
Arlington, in 1988, and his Ph.D. from Rensselaer Polytechnic
Institute, Troy, NY in 1991, all in Electrical Engineering. He joined
ALSTOM ESCA Corporation in October 1991. His current
interests include deregulation applications and power system
stability. Dr. Cheung is a registered Professional Engineer of the
State of Washington.
Payman Shamsollahi received his B.S. (1989) and M.S. (1990)
degrees from Tehran University, Iran, and Ph.D. (1997) degree
from University of Calgary, Canada, all in electrical engineering.
He joined ALSTOM ESCA Corporation in 1997. His main
interests are deregulation applications, artificial intelligence and
power system control.
David Sun joined ALSTOM ESCA Corporation in June 1980. He
received his B.S. and M.S. from Rensselaer Polytechnic Institute,
and his Ph.D. from University of Texas at Arlington, in 1974,
1976, and 1980, respectively, all in Electrical Engineering. His
current focus is on the planning and development of deregulation
applications.
James Milligan received his B.S. in Electrical Engineering from
Western New England College in 1976. He has been with
NEPOOL and ISO-New England since 1976 and is presently
Manager, Engineering for ISO-New England System Operations
and Reliability. In this capacity his primary responsibility is the
overall coordination of the Energy Management System support for
ISO-NE Operations functions.
Mike Potishnak received his B.S. in Electrical Engineering from
Newark College of Engineering (NJIT) in 1972. He was an
Engineer with Con Edison from 1972-1976. He received his M.S.
in Mechanical Engineering from Colorado State Univesity in 1978.
From 1978 to 1989, he was a Senior Engineer of Public Service
Company of Colorado. Since 1989, he has been with NEPOOL and
is currently a Principal Engineer of ISO New England.