Você está na página 1de 8

IEEE Transactions on Power Systems, Vol. 10, No. 3.

August 1995 1211

Reliability Differentiated Pricing of Spinning Reserve


by
Shams N. Siddiqi, Member Martin L. Baughman, Senior Member
Department of Electrical and Computer Engineering
The University of Texas at Austin
Austin, Texas 78712

ABSTRAm This paper extends the RDP model to derive an optimal price for
spinning reserve and an optimal level of spinning reserve from a societal
This paper derives the spinning reserve pricing policy implied by welfare point of view. Rather than incorporating a minimum spinning
the reliability differentiated pricing model and illustrates the importance reserve constraint, the reliability differentiated pricing model includes
of such a pricing policy. The purchase of additional spinning reserve customers' outage costs into the overall objective function to reflect the
capacity at the prices derived leads to benefits for the customers, through idea that outages created by insufficient generation or transmission
reduced real and reactive power prices and improved reliability of the actually cause a loss of welfare.
system, and for the system as a whole, through reduced societal costs
and lesser investment in reinforcements for the system. The magnitudes A substantial body of theoretical and empirical work in
and behavior of the optimal prices for real and reactive spinning reserve government, industry, and academia has been devoted to developing
are illustrated in two case studies. One case study determines reliability estimates of customer's outage costs. For example, the U. S.
differentiated prices of real and reactive power for a sample four-areas Department of Energy sponsored the National Electric Reliability Study
system with and without spinning reserve purchase and the optimal [4] that surveyed the methodologies and estimates of short-term costs of
purchase prices for both real and reactive power spinning reserve electricity supply interruptions, costs of prolonged electricity outages,
capacities. The results of this case study show that when reliability non-utility mitigation mechanisms, and compared service interruption
differentiated pricing is extended to both real and reactive power reserve cost studies. The Electric Power Research Institute [5] updated this
it leads to benefits for the customers and the system as a whole. A literature by compiling recent findings on the demand for reliability
second case study examines the locational dependence of the optimal among major classes of customers and discussing methods of
prices for real and reactive reserves on the operating status of equipment quantifying the evaluations. More recently, The Energy Journal 16)
in the 24-bus IEEE Reliability Test System. It is concluded that even devoted an entire issue to the topic in which both the domestic and
small amounts of reliable reserve capacity purchase from neighboring international practices for estimating customers' outage costs were
utilities can improve the reliability of the system and lower expected real presented. Incorporating outage costs into utility planning studies is
and reactive power prices leading to gains in social welfare. becoming routine. For example, EPRI recently hosted a Value-Based
Transmission Resource Analysis Training Workshop [7] presenting
KEY WORDS transmission planning tools that account explicitly for outage costs in
their evaluation of reliability value. In essence, the topic of the present
Spinning Reserve, Reliability, Outage Cost, Real-Time Pricing, paper is nothing more than value-based pricing of real and reactive
Reliability Differentiated Pricing, Social Welfare. reserve.
I. INTRODUCTION This paper extends the reliability differentiated pricing model to
include spinning reserve constraints in Section lI and 'derives optimal
Spinning reserve is generating capacity that is spinning and prices for real and reactive power in Section III.The spinning reserve
synchronized with the system and available to serve load on a moment's pricing policy implied by the RDP model is derived in Section IV.
notice. Utilities must maintain or purchase some generation capacity as Section V presents two case studies illustrating the behavior of Lhc
spinning reserve to serve loads in the event that operating generating reliability differentiated prices of real and reactive spinning reserve. It is
units, transmission lines, or other equipment that is serving load concluded that the spinning reserve purchase pricing policy implied by
suddcnly or unexpectedly fails. Historically, the amount of spinning the RDP model results in benefits to both the customers and the system
reserve that it is prudent for the utility to maintain has been determined as a whole.
by rules of thumb, such as some multiple (i.e., one or one and one-half
'times) of the rating of the largest operating unit, or as some percentage U. THE RELIABILITY DIFFERENTIATED PRICING MODEL
(perhaps 15 percent) of the system load being served. Caramanis, Bohn,
and Schweppe [ l ] follow this idea closely by incorporating a minimum Customers are modeled as price-taking expected profit-
"spinning reserve" constraint into their extended formulation of the basic maximizing f m s . In the model, the customer's value-added function
real-time pricing model. The result is that a price for spinning reserve is includes both real and reactive power demands (for notational simplicity,
created whenever the spinning reserve constraint is binding. the time t subscript of all variables is implied in the description of
constraints below):
Reliability differentiated pricing (RDP) [2,3] integrates various
quality attributes of electricity into the price of electricity. Under this
pricing system, the price of electricity reflects the true marginal costs of where Pid = real power demands of customers at bus i,
providing electric power service, including any costs of unrcliability that
are imposed on the customers as a result of forced outages of generating d
Q. = reactive power demands of customers at bus i,
units or transmission lines. The prices are spatially differentiated to
reflect any constraints that the bulk power system Configuration may pi = prices of real power for customers at bus i,
have on supply costs and they change with system operating conc!itions. qi = prices of reactive power for customers at bus i,
The resulting prices provide for maximum economic efficiency from a w = exogenous economic and weather related random factors
societal welfare point of view (i.e., maximum producer plus consumer
surplus). (i) Load Flow Fauations
The load flow equations, based on Kirchoffs laws, that govcrn
the flow of real and reac&e power for bus i are:
94 SM 527-2 PWRS A paper recommended and approved
by t h e IEEE Power System Engineering Committee o f t h e
IEEE Power Engineering S o c i e t y for p r e s e n t a t i o n a t
t h e IEEEfPES 1994 Summer Meeting, San F r a n c i s c o , C A ,
J u l y 24 - 28, 1994. Manuscript submitted August 2 , j
1993; made a v a i l a b l e f o r p r i n t i n g May 3, 1994. where the vertical bars denote the magnitude of complex quantity
bracketed and where
Pp = real power gcneration at bus i.

0885-8950/95/$04.00 0 1994 IEEE


1212

= reactive power generation at bus i, advance. for the short-run operating problem it is assumed that Sp and
Pu = unserved real power demand of customers at bus i, Sf are fixed and that they are perfectly reliable. The constraint on the use
QU = unserved reactive power demand of customers at bus i, of reserve real and reactive power Pf and at bus i is then given by:
Pf = use of real power spinning reserve purchased at bus i, 0 s Pf 5 sp
Qs = use of reactive power spinning reserve purchased at bus i, 0 1 Qf I sf
Yij = the ij* term of the bus admittance matrix,
Obiective:
Vi = the voltage at bus i, The objective of the short-run operating and optimal pricing
8 . .- phase angle of the admittance Yij, problem for a welfare maximizing utility using the criterion of
6i = the voltage angle at bus i. maximizing expected consumers' plus producer's surplus can be stated
as follows:
li0 Generation Limits
The generating unit capacity constraintsmay be represented as:
0 5 Pf S Kfai (4)
-Kfai I e 5 K4ai (5)
where KP= maximum real power output capability of generator at bus i.
K: = maximum reactive power output capability of generator at
bus i,
a, = a stochastic binary variable which is equal to zero when the
where f ,Qf, Py, Q:, P:,
where
e, P:, and e
are the decision variables, and
generator at bus i is out of service. E[*]designates the expected value of the argument,
If the generator at bus i has only reactive power generating OC,t(PY.Q:) = outage cost function (includes loss in customer
capability, for example, capacitor banks or synchronous condensers, benefit) of customer i,
then Kp would be zero for that bus while Kf would be non-zero. PCiJPft,@J = cost of producing real and reactive power at bus i,
EC.,(Pyt,Qyt) = cost of use of reserve real and reactive power at
The transmission capacity constraints between buses i and j are bus i.
modeled here as thermal limits on the volt-ampere flows, namely:
rrijl 1 P '"J b i j (6) AU decision variables, except those involvingreactive power,
are non-negative. All constraints of the model have continuous
where p" = the maximum volt-ampere flow admissible over the line second partial derivatives. The outage cost function is usually
IA
connecting the buses i and j. expressed as a linear function of the real power interruption. The
lTi,l = the magnitude of the volt-ampere flow over the line production cost function at a bus is the s u m of the production costs
connecting the buses i and j which is given by of the generating units at the bus and these are typically expressed as
polynomial functions of real power generation by the units. The use
n-4'J =
,- of real and reactive power reserve cost function is nothing but the
Pi, = the real power flow over the line connecting bus i to bus j "energy" cost of purchasing power and is a linear function of the
and which, assuming the shunt admittance is negligible, amount of power purchased.
is given by
pi, = ivii lvjl iYijlcos(eij+Gj-si)- Ill. OPTIMAL RELIABILITY DIFFERENTIATED PRICES
codij,
Qi, = the reactive power flow over the line connecting bus i to Consider the short-run operating problem of the welfare
bus j,
b.. - a stochastic binary variable which is 0 when there is an maximizing utility. Given the customer demands that the utility tries to
outage on line connecting bus i to bus j. supply, the short-run problem of a welfare maximizing utility is to
operate the system so as to minimize societal costs. "his problem may be
stated as follows.
liv) Voltaee Limits
Voltage limits refer to the requirement for the system bus
voltages to remain within a narrow range of levels. Devices that consume
electricity are often designed so that their performance is degraded if the
voltage departs significanrlyfrom the nominal for prolonged periods of
time. A s in [8], the voltage limits can be expressed by the following
constramt:

v y 5 lVil I VY" (7) subject to constraints (2)-(9).


This problem can be solved to obtain optimal values of all
where V?'" and vYaxare the minimum and maximum voltage levels, variables as functions ofthe of this problem, namely pit
d and
respectively, that are acceptable at bus i.
C$, assuming that there is an optimal solution that is a regular point and
fv) Suinnine Reserve Constraint that satisfies the second-ordersufficiency conditions for a minimum (see
Let the Cost of acquiring real and reactive reserve of the Sensitivity Theorem [9] or Envelope Theorem [lo] for proof). The
amounts Sf and Sf,respectively. at bus i be SCi(Sp, S'?). This spinning objective function Can be rewritten as:
reserve represents capacity in addition to the excess capacity of the
generating units that are on-line and must be acquired by running
cOpt
=
t
( ~OC;(<,Q;,
I
+ pc:(p;,~t) +E C ~ ~ , Q ; )) (12)
additional units at minimum load or by purchasing capacity from
neighboring utilities. Since spinning reserve must be purchased in

I
1213

corresponding "unreliability" component in the optimal price of reactive


where<= ( P y t ; v i c I ) power. Previous theories for optimal electricity pricing have ignored this
outage cost component. The other components of price, i.e. the terms
Q f = (Qft;ViE I )
apc; aEc;
and where symbols in bold type represent expected values. - +- ,represent the energy charge for power prod~cedand/or
Since the customer sets his net marginal benefit from apt spit
consumption of electricity equal to the price of electricity, the optimal purghased to 'skrve the customer's demand.
prices for real and reactive power, respectively, that induce the
customers to behave in a social welfare m a x i m i i g manner are given by: The real-time implementation of reliability differentiated pricing is
described in [2.3], and the importance of pricing reactive power is
discussed in [8].
IV. IMPLICATIONS FOR PRICING SPINNING RESERVE
Utilities must maintain some generation capacity as spinning
reserve to serve loads in the event that operating capacity that is serving
load suddenly or unexpectedly fails. The purchase of spinning reserve
is a short-termcontract made to protect against loss of customer load due
These optimal prices of real and reactive power essentially consists of to unplanned outage of generating units or transmissionlines. Spinning
two parts: (i) the expected marginal cost of supplying the electrical reserve can be purchased from neighboring utilities. independent power
producers (IPP).and/or cogenerators in order to provide customers with
apc; aEc; apc;
energy to the customer, i.e. 7+ - for real power and - higher levels of service reliability in the short-run. The price paid for
apjt a$ aQ; + such capacity should exactly equal the value, or marginal benefit, that
customers derive from that additional capacity. To detennine an optimal
aEc; price of spinning reserve purchase, consider the short-run operating
- for reactive power; and (ii) the expected marginal change in problem of the welfare maximizing utility given in Equations (2) - (9).
a: J.
and (11).
outage costs of other customers due to incremental change in the As before, this problem can be solved to obtain optimal values of
aoc; aocyt for all variables as functions of the parameters of the problem, in this case
customer's demand. i . e . C - for real power and - taken to be sp and s;. The objective function be hnen as:
i-cj a$ i+j a$
reactive power. Of course. both of these components of real and reactive
power prices are affected by the operating conditions and constraints
Copt = ( coCf$Sp,Sf) + pcs(Sp
t i ' Sf) + ECS(SP
t i'
Ss) ] (15)
i
included in the model [Equations (2) - (9)]. The net effect of where the bold type design* expected values.
consideration of the constraints imposed on system operations in the
RDP model is to create premia, or amounts that-needtobe added, to the Using the Sensitivity Theorem [9]. the optimal spinning reserve
optimal prices whenever the constraints are binding. Whether any of the purchase prices at bus j for real and reactive power capacity.
constraint+ are binding or not, however, there are still only two prices, respectively, are given by:
pj and qj, for real and reactive power, respectively, that need to be
communicated to an electricity user to induce optimal consumption
decisions. The only thing is that the prices are higher than they would be
otherwise whenever there are constraints that bind system operations.
Thus, the optimal prices are i n f l u e n d by:
Marginal cost of generation;
Marginal cost of emergencypower purchase;
Marginal outage cost;
Generator real and reactive power output constraints;
Transmission capacity constraints; and As above, the optimal spinning reserve purchase prices for real and
Voltage regulation requirements. reactive power capacity consists of two parts: (i) The expected change in
the marginal cost of supplying electrical energy due to incremental
Let us now examine the two components of the optimal price axs aEc:
more carefully. The marginal generation cost corresponds to the added
fuel and variable operations and maintenanceexpense that the utility must
change in spinning reserve purchase, i.e. - 2- for real -
incur to serve an increment in demand; it is commonly included in real asp asp
power prices and it varies little from customer to customer at the same apt: a E c ;
bus. The expected marginal outage cost corresponds to the expected power capacity or - -- -for reactive power capacity; and (ii)
increase in outage costs imposed on the rest of the customers of the as: as:
system by an increment in demand it is not normally included in real The expected marginal reduction'in outage costs of all customers due to
power prices and it depends greatly on the customer's own outage cost.
aoqt
This is because an increase in demand of a customer with relatively high
outage cost results in increased expected outages for customers with
lower outage costs than that of the customer. Also an increase in demand
incremental change in the spinning reserve purchase, i.e. -E -aspi
of a customer with relatively low outage cost does not affect the expected
outages for customers with higher outage costs than that of the customer.
Hence this component of price depends on the outage cost of the for real power capacity or -E aoqt
i
-for reactive power capacity. Of
asp
customer, being higher for customers with higher outage costs even at
the same bus. This component of price may be thought of as the capacity course, both of the components of s $ i g reserve purchase prices for
charge which depends on the level of reliability of that purchased either real or reactive power capacity are affected by the operating
capacity. What it reflects is that an incrementof demand anywhere in the conditions and constraints included in the model [Equations (2)-(9)].
system leads to some increase in unreliability of operating the system, Thus, the optimal prices are, as before, influenced by the marginal cost
and added outage costs in the system. Since outage costs are often of generation, the marginal cost of emergency power purchase, the
considered to be quite high, with values as high as $5.00 or more per marginal outage cost, generator real and reactive output constraints,
kilowatt-hour sometimes quoted in the literature [11,12], the marginal transmission capacity constraints, and voltage regulationrequirements.
outage cost term will be a significant component of the optimal real
power price particularly at times of high load. There is also a
1214

In the next section it will be seen that the purchase of spinning purchase, i.e. use of any reserve purchased, at Buses 3 and 4 are
reserve has little impact on the total generation costs of the system. The $50/MWh and $60/MWh, respectively, for real power, and $2O/MVARh
dominant component of spinning reserve purchase prices for real and at both buses for reactive power.
reactive power capacity is the component which gives the expected
margin?' reduction in outage costs of all customers due to incremental Two alternative cases of spinning reserve purchase are studied.
change m the spinning reserve purchase. This is also the result one One places limits on the use of purchased spinning reserve equal to the
would expect, since the additional capacity purchased would be utilized amount of spinning reserve purchased while, for comparison the other
to offset the expected customer loss of load in the event of unplanned places an upper bound of zero on the use of purchased reserve.
forced outages of the generating units.
The model was formulated in GAMS (General Algebraic
Since a reduction in the expected marginal outage costs will Modeling System) and solved using the MINOS optimizer on the IBM
almost always result from the supply of more spinning reserve, the 3081. The results obtained from the analysis are presented below.
optimal price for spinning reserve will always be non-zero. The
expected outage cost function is a decreasing function of the amount of Behavior of Real Power Prices
spinning reserve. Since the cost function for supply of spinning reserve The variations of the optimal expected reliability differentiated
at bus j, SC.(SP,Sq), is a non-decreasing function of SP and S'!, thus price of real power at Bus 3 for a 24 hour period with and without
J J J J spinning reserve purchase are shown in Figure 2.
some equilibrium supply of spinning reserve exists where marginal cost
equals marginal benefit. If this desired supply of spinning reserve is The expected real power price eajectories are closely related to
obtained, then the optimal price of spinning reserve is exactly the demand, except that the percentage change in prices is greater than the
marginal cost (or marginal benefit) at this equilibrium. corresponding percentage change in demand. The real power prices at
BUS3 for the system with external spinning reserve purchase range born
Admittedly there has been little market activity in spinning $24.55/MWh at the time of lowest demand to $56.93/MWh at peak
reserves in the past since each utility sought to provide its "share" of the demand, whereas for the system without spinning reserve purchase. the
necessary real power reserves to the pool in which it operated and real power prices range from $24.98/MWh to $93.88/MWh. The reason
reactive power markets have been virtually non-existent. In the future, for this large difference in peak prices is that customer outage costs are
however, with electricity services being unbundled and supply of usually much greater than the costs of generation and in the system with
services being provided by nonutility sources and via contract with spinning reserve purchase the expected customer outage cost is
customer service companies, future wholesale markets for these negligible compared to the system with no spinning reserve purchase.
commodities will be needed and will likely develop.
V. CASE STUDY RESULTS: BEHAVIOR OF OPTIMAL PRICES
140
120
- System with spinning reserve purchase
The Four-Areacase Study
A four-area power system, shown in Figure 1 with each area
- - System without spinning reserve purchase
100-
represented by a bus, is used to gain insights into the behavior of
reliability differentiated prices with and without spinning reserve 80-
purchase and the prices of reserve capacity purchase.
60-
40 -
20

0 4 8 12 16 20 24
T14 Hour of Dav
Figure 2 Variations of Expected Real Power Prices at Bus 3 for a 24
Hour Period
In real power systems, since every piece of equipment has a non-
L2 zero forced outage rate, there is a probability that the spinning reserve
contracted for will not be available - but this probability is usually very
Bus 3 low. Therefore, there is always some probability that customer loads can
not be served. The expected unserved real power demands at Buses 3
and 4 for the system with no spinning reserve purchase are shown in
Figure 1 Four-Area Power System for Case Study Figure 3 (the system with spinning reserve purchase has negligible
expected unserved real power at both buses). Reliability differentiated
The regions of the sample system are interconnected by high
-
voltage transmission lines, as shown in Figure 1. The specifications of
the elements of the sample system are described in the Appendix. The
impedances and admittances of the transmission network are given in
Table A.3. At Buses 1 and 2, there are utility-controlled generating
plants whose characteristics are given in Table A.l. A controllable
k3
3,
a
8-( - Bus4 1
reactive power source, described in Table A.2, is available at Bus 3. 6-
Customer loads exist at every bus and the loads are assumed to have the
same load shape over the 24 hour time period. The real and reactive 4-
power load curves of the system are shown in Figures A.l and A.2. The
peak loads and customer outage costs at the different buses are:
2-
Bus 1: 120MW, $600/MWh Bus 2: 65MW, $200/MWh
Bus 3: 84MW, $SOO/MWh Bus 4 67MW. $400/MWh 0
Buses 3 and 4 are connected to neighboring utilities from which ( 1 5 ~ ~ . 0 4 8 12 16 20 24
15MVAR) and (20MW. 20MVAR) of spinning reserve capacity, Hour of Day
respectively, can be purchased. The prices of emergency power Figure 3 Variations of Expected Unserved Real Power for
24 Hours Without Spinning Reserve Purchase

1
1215

pricing takes into account the inherent unreliabilityof power systems in customer outage costs is the dominant component of the s p h g reserve
designing its pricing policy. purchase price. the variations of spinning reserve purchase p i c a of real
power are closely related to the expected unserved real power costs for
Behavior of Reactive Power Prices the system. The real power spinning reserve purchase prices vary from
When there is a shortage of reactive power supply, the utility $O/MW at times of low demand to as high as %Z.M/MW at the time of
must ration both real power and reactive power in order to reduce peak load.
reactive power load. This is because reactive power is usually consumed
in conjunction with real power and reactive power consumption can not Behavior of Reactive Suinninp. Reserve Purchase Prices
be reduced without a corresponding reduction in real power The variation of the optimal spinning reserve purchase prices of
consumption. Hence an outage cost for real power is incurred whenever reactive power at Buses 3 and 4 for a 24 hour period are shown h
reactive power needs to be rationed. Figure 6. Whenever there is a shortage of reactive power supply, the
utility must ration both real power and reactive p o w a in order to reduce
Figure 4 shows the variation of the optimal expected reactive reactive power load. Hence the reactive power spixming reserve purchase
power price at Bus 3 for a 24 hour period with and without spinning price is by no means negligible. The prices vary from $O/MVARh at
reserve purchase when taking into account equipment operating times of low demand to as high as $18.30MVARh at the t h e of peak
uncertainties. The expected price of reactive power varies from load.
$OSl/MVARh at the time of lowest demand to $12.64/MVARh at peak
demand for the system with spinning reserve purchase. Without The Affect of Spinning Reserve Purchase on Societal Cost
spinning reserve purchase, the expected price of reactive power varies Societal cost is defined here to mean the s u m total of customer
from $1.61/MVARh to $24.25/MVARh. An important point to note is outage costs, utility operating costs, costs of emergency purchases and
that the expected price of reactive power at Bus 3 at peak demand the costs of purchasing additional spinning reserve capacity. Societal
($12,64/MVARh)is about half of the expected price of real power at the costs for the system with and without spinning reserve purchase for a 24
time of lowest demand ($24.55/MWh) with spinning reserve purchase. hour period are shown in Figure 7. It is observed that the total expected
Without spinning reserve purchase, the correspondingprices are almost costs for the system with spinning reserve purchase is lower at all hours
equal. Thus, the price of reactive power is quite significant when as compared to that without spinning reserve purchase. This is because
compared to the price of real power and both prices are required in order the benefits of purchasing spinning reserve in the form of reduced
to send efficient price signals to customers. customer outage costs more than offsets the reserve capacity purchase
costs even though there are no reserve capacity costs when spinning
reserve is not purchased. Hence the purchase of spinning reserve
8 40
- System without spinning reserve purchase
capacity at the derived prices results in benefits to the customers through
reduced real and reactive power prices and improved reliability of the
$ -
U
3o
- System with spinning reserve purchase
system, as well as benefits to the system as a whole through redbced
societal costs.

> - Bus3
F
3
Q
25- - Bus4

5 15-
n
16 20 24
$
'p.
v)
10-

2
2 5-
b)
'1

1
Hour of Day

c
- BUSS Figure 6 Variations of Reactive Power Spinning Reserve

2
a
Q 25- - Bus4 Prices at Buses 3 and 4 for a 24 Hour Period

$
?!c
20-

15:
g
c
50.
- System without spinniig r e m e purchase
- System with spinning reserve purchase
M

.?a
.M 10:
C
A

? 5-

';;; 0--, I I I , I

0 4 8 12 16 20
Hour of Dav
Figure 7 Variations of Expected Societal Costs for a 24 Hour Period

1
1216

The 24-Bus Case Study


The 24-bus IEEE Reliability Test System (RTS), shown in
Figure 8 with interconnections to two neighboring utilities, was also
used as a second case study system. All the data specified in Reference
[13] for the RTS was used in the analysis. Thus, the system peak load
is 2,850MW and the total system generating capacity is 3.405MW.
Additional assumptions were made about the load. The load at each load
bus in the RTS was subdivided among a set of classes with an outage
cost assigned to each class load. The outage costs used were $1.77,
$11.28, and $8.10 per KWh for residential, commercial, and industrial
customers, respectively [14]. The compositionof class loads at each bus
is giveninTableB.1.

It was assumed that two other utilities with spinning reserve to


sell are connected to the 24-bus system. Utility 1 is assumed to have tie-
line links with this system at Buses 16, 17 and 18, and Utility 2 is
connected at Buses 2.7, and 8. The problem is to find optimal prices
for real and reactive reserve capacity that the utility operating the RTS is
willing to pay to these neighboring utilities given that the energy charge
for real power purchase from both utilities is $45/MWh and for reactive
power purchase price is $O/MVARh.
The number of possible contingencies (configuration states) that
can arise due to the random forced outage rate of generating units,
transmission lines,and transformers is immensely large even for this 24-
bus system. System configuration states are determined by the Utility 1
availability or unavailability of generating units, transmission lines. and Peak Off-peak Peak I Off-Peak
transformers. Obviously, it is not practically feasible or desirable to
evaluate the model for all possible configuration states. Importance
sampling techniques (151 were used to reduce the required sample size
1 =l( )
Lacti5,!$%VAR) ::2 0.21
0.04
8.16
2.68 I :E
by "distorting" the sampling of states to favor the appearance of states
with high costs. Six important elements were identified and one random

7 7
Utilit!
A

BUS 14
Synch.
cond.
BUS 13
Utility 2

*+-j
Utility 1
Peak I Off -Peak Peak I Off-peak
[Real 6 M W ) 407.8 I 0.21 413.8 I 5.38
BUS 11 IReactive($lhiVAR) I 0.55 I 0.03 I 2.60 I 0.21 ]
Table 2 Optimal Expected Spinning Reserve Prices-Case 2
(Operating Reserve = 155MW)
In a third case, called Case 3, rather than losing service from
nuclear unit specified in Case 2, it is assumed that something forces the
three lOOMW oil-fired units at Bus 7 out of service on the peak day.
This results in operating reserves in this case of about 255MW.
approximately 8.9 percent of the peak load. The effects on the expected
optimal prices for real and reactive reserves are shown on Table 3.
Though the operatingreserves are higher in this case than in Case
2, there are some dramatic increases in the optimal prices for reserves
from Utility 2. the utility assumed to be interconnectedwith the RTS in
the neighborhood of the three oil-fired units out of service. The optimal

At Bus'
t p yI I , I
Utility 2

Table 3 Optimal Expected Spinning Reserve Prices- Case 3


Figure 8 Illustration of the IEEE Reliability Test System
(Operating Reserve = 255MW)

1
1217

prices of real and reactive power reserves from Utility 2 climb to


$1,404/MW and $1,856/MVAR. respectively, at the time of peak. The
interpretation of this is that the loss of the three oil-fiied units has
significant reliability implications for the RTS in the neighborhood of
Utility 2. making its reserves very valuable to the RTS (while creating
little change in the optimal value of reserves from Utility 1). In this case,
the value of reactive reserves (in $/MVAR) even exceeds the price of real
reserves (in $/MW). This is because the reactive power capability of the
three oil-fired units whose service was lost in this case were critical for
voltage support in the portion of the RTS 138KV system interconnected
with Utility 2. Though the optimal expected price of reactive reserve
from Utility 1 is higher in this case than in the other cases, it is still
orders of magnitude less than the corresponding price of reactive reserve
from Utility 2. This is because the load flow constraints make it near
impossible to transmit VARs from Utility 1 to the neighborhood of
Utility 2. o ! , , , I . , * I . , . i
0 4 8 12 16 20 24
V. CONCLUSION Hours
The results of this paper point to the importance of an optimal Figure A.l Test System Real Power Load Curve for a 24 Hour Period
pricing policy for pricing spinning reserve. The purchase of additional
spinning reserve capacity at the optimal prices leads to benefits for the
customers, through reduced real and reactive power prices and improved
reliability of the system, and for the system as a whole, through reduced
societal costs and lesser investment in reinforcements for the system.
Mathematical derivations showed that the reliability differentiated pricing
model provides an optimal policy for pricing spinning reserve capacity
purchase from a societal welfare point of view.
A case study determined reliability differentiated prices of real
and reactive power for a 4-area system with and without spinning
reserve purchase and the purchase price of real and reactive power
spinning reserve of fixed capacities. The results of the study illustrated
that even small amounts of reliable spinning reserve capacity purchase
from neighboring utilities can improve the reliability of the system and
lower expected real and reactive power prices leading to gains in social
welfare.
o] . , , , , , . I '
A case study of the optimal spinning reserve prices in the 24-bus 0 4 8 12 Ib
20 24
IEEE-RTS for three alternative unit commitment configurations Hours
illustrated. the dependence of the prices both on location and on the Figure A.2 Test System Reactive Power Load Curve for a 24 Hour
operating status of equipment within the system. Period
APPENDIX A
APPENDIX B
Table A.l Generating Unit Characteristics
WAR Forced tor at Each Bus)
w Coefficientsof Cost Curve Capacity Outage Bus Base Year Percent Perm Percent
Bussize Type Fuel (Pi)o (P;J1 ( P i p Min.Max. Rate No. Load(MW) Residential Jommercial hdustlial

1 350 Thermal Coal 8160.0 1622.37 169.50 -75 350 0.00


2 200 Thermal Gas 25091.0 2098.90 441.89 -80 300 0.10
1
2
~

3
z:;
179.6
100
100
0
0
0
0
0
0
100
4 74.1 0 0 100
5 71.3 50 25 25
Table A.2 SynchronousCondenser Characteristics 6 136.8 50 50 0
7 125.4 50 50 0
8 171.0 50 50 0
Device BUS MVAR Cauacitv FOR 9 173.9 50 50 0
10 193.8 50 50 0
SynchronousCondenser 3 50 Reactive 200 Capacitive 0.20 11 0.0 0 0 0
12 0.0 0 0 0
13 265.1 0 0 100
Table A.3 TransmissionLine Characteristics 14 193.8 0 0 100
15 316.4 50 50 0
Line, Th& Forced 16 99.8 50 50 0
busto Length R X Charging LLnit outage 17 0.0 0 0 0
bus mi ohm ohm WAR perunit rate 18 333.5 50 50 0
19 182.4 50 50 0
1-2 40 8 32 4.1 3 .O 0.0 20 128.3 100 0 0
1-4 30 6 24 3.1 0.8 21 0.0 0 0 0
0.0
2-3 30 6 24 3.1 3.0 22 0.0 0 0 0
0.0 0
24 80 16 64 8.2 3.0 0.3 23 0.0 0 0
3-4 50 10 40 5.1 3.O 0.0 24 0.0 0 0 0
'Total 2850.0 4314 31.0 25.6
1218

REFERENCES
[ 141 Havemann, Steven D., "Estimates of Customer Outage Costs for
[ l ] Caramanis, M. C.. R. E. Bohn, and F. C. Schweppe, "System the 1990 Resource Planning Project," Corporate Planning
Security Control and Optimal Pricing of Electricity," Electric Power Department, Houston Lighting and Power Company, Houston TX,
and Energy Systems. Vol. 9, No. 4, October 1987, pp. 217-224. May 10,1990

[2] Siddiqi, S . N., and Baughman, M. L., "Reliability Differentiated [15] Danzig, G.B., P.W. Glynn, M. Avriel, J.C. Stone, R. Entriken.
Real-time Pricing of Electricity." IEEE Transactions on Power and M. Nakayama, Decomposition Techniques for Multi-Area
Systems, Vol. 8 , No. 2, May 1993, pp. 548-554. Generation and Transmission Planning Under Uncertainty, Final
Report for EPRI Proiect 2940-1, August. 1989.
[3] Siddiqi, S . N., Reliability Differentiated Pricing and Optimal
Planning for Electrical Power Systems, Ph.D. Dissertation, The
University of Texas at Austin, Austin, Texas, May 1993. Shams N. Siddiqi was born in Chittagong,
Bangladesh on September 1, 1964. He
[4] United States Department of Energy, The National Technical received a B.Sc. (Engineering) degree from
Reliability Study: Technical Study Reports, DOE/EP-0005, Bangladesh University of Engineering and
Washington, D. C., April 1981. Technology in 1988. He received the Master
and Ph.D. degrees in Electrical Engineering
[5] Criterion, Incorporated, The Value of Service Reliability to from the University of Texas at Austin in
Consumers. EPRI EA-4494, Project 1104-6, Electric Power 1989 and 1993, respectively. He was a
Research Institute, Palo Alto, CA, May 1986. Graduate Research Assistant with the Center
for Energy Studies at the University of Texas
[6] The Energy Journal-- Special Electricity Reliability Issue, at Austin from September 1988 to April 1993,
International Association of Energy Economics, Washington, D. C., where he is now working as a Research
Vol. 9, 1988. Associate.
His research interests are in the areas of
[7] Value-Based Transmission Resource Analysis (VBTRA), workshop comprehensive electrical systems planning, power systems economics,
sponsored by the Electric Power Research Institute, October 25-28, and electricity pricing.
1993. Atlanta, CA.
[8] Baughman, M. L., and Siddiqi, S . N., "Real-Time Pricing of Martin L. Baughman (S, 72) was born on February
4l Reactive Power: Theory and Case Study Results," I E E E 18, 1946 in Paulding, Ohio. He received his BS in
Transactions on Power Systems, Vol. 6, No. 1, February 1991, pp. Electrical Engineering from Ohio Northem
23-29. University in 1968 and his MSEE and PbD.
degrees in electrical engineering at MIT in 1970 and
[9] Luenberger, D. G.. Linear and Nonlinear Programming, Addison- 1972, respectively.
Wesley Publishing Company, 1984. Dr.B a u h a n was a Research Associate at
Massachusetts L t i t u t e of Technology from 1972 to
[lo] Silberberg, E., The Structure of Economics: A Mathematical 1975, at which time he joined the University of
Analysis, Second Edition, McCraw-Hill, New York, NY, 1990. Texas at Austin as a Senior Research Associate. In
1976 he pined the faculty of the Department of Electrical and Computer
[ l l ] Munasinghe, M., and A. Sanghvi, "Reliability of Electricity Engineering as an Assistant Professor. In 1979 he co-authored a book
Supply, Outage Costs and Value of Service: An Overview," The with Paul Joskow on electricity supply planning entitled Electricitv in the
Energy Journal, Vol. 9, 1988, pp. 1-18. United States: Models and Policy Analvsis. From 1984 to 1986 he
chaired the National Research Council Committee on Electricity in
[12] Woo, C. K., and K. Train, "The Cost of Electric Power Economic Growth. He has served as a consultant to several agencies,
Interruptions to Commercial Firms," The Energy Journal, Vol. 9, including Edison Electric Institute, the MIT Energy Laboratory, the
1988, pp. 161-172. Economic Council of Canada, and the Ministry of Planning in Saudi
Arabia, and the Electric Power Research Institute.
[13] IEEE Task Force, "IEEE Reliability Test System," IEEE Dr.Baughman is a member of the International Association of
Transactions on Power Arpratus and Svstems, PAS-98, Nov/Dec Economists and registered Professional Engineer in the State of Texas.
1979, pp. 2047-2054.

Você também pode gostar