Ancillary Services4

8/8/2019 Ancillary Services4

1/7

498 IEEE Transactions on Power Systems, Vol. 14, No. 2, May 1999Com petitive Procurement of Ancillary Servicesby an Independent System Operator

Harry Singh Alex PapalexopoulosMember Senior Member

Pacific Gas and Electric CompanySan Francisco, CaliforniaAbstract

This paper discusses the competitive procurement ofAncillary Services by an Independent System Operator(ISO). The paper assumes the existence of an underlyingmarket for energy and explains why this energy marketmust be accompanied by a market for Ancillary Services.These services include operating reserves and AutomaticGeneration Control (AGC) both of which may requiregenerators that are infra-marginal in the energy market tochoose between supplying energy or Ancillary Services.The paper discusses the design of an auction for theseservices that has formed the basis for the Califomia ISOsAncillary Services market. The paper also discusses thereal-time dispatch of Ancillary Services.Keywords: Electric Power Deregulation, Auctions,Operating Reserves.

1. IntroductionCompetitive energy markets are being instituted in manyjurisdictions as electricity supply industries are restructuredin steps towards deregulation. In general, thesearrangements require some form of a central auction forelectricity to be delivered over short time periods. There aretwo distinct fonns of market structure that have alreadybeen implemented around the world; a real-time market,such as the one that has been instituted in the Australianstate of Victoria, and day-ahead (forward) markets such asthose that have been instituted in England and Wales,Norway, and California [1,2,3].Since electricity is delivered instantaneously, it is obviousthat a day-ahead market requires a balancing service with anassociated real-time market. In England and Wales, thebalancing function is provided at the day-ahead price. InCalifornia, the balancing function is priced separately at anex-post price that is determined in the real time marketbased on an auction for imbalance energy and the associatedPE-427-PWRS-0-06-1998 A paper recommended and approved bythe IEEE Power System Analysis, Computing and EconomicsCommittee of the IEE E P ower Engineering Society for publication in theIEE E Transactions on Power System s. Manuscript submitted January5 , 1998 ; made ava ilable for printing June 12, 1998 .

dispatch instructions issued by the Independent SystemOperator (ISO).The operation of the real time market is a fundamentalresponsibility of the ISO. Since the IS 0 does not ow n anygeneration, it must ensure that there is sufficieni. unloadedcapacity among on-line generators to participilte in thebalancing market and secure the required amount ofreserves needed for maintaining the reliability of the system.The IS0 must also ensure that there is a sufficient quantityof Automatic Generation Control (AGC) capability andquick-start generation capacity, and conform to mandatedreliability criteria [7]. One way the IS0 can meet theserequirements is to create a competitive market for AncillaryServices.A premise of this paper is that competitive markets forenergy require competitive markets for Ancillary Services.This is particularly true for those services that bear acomplementarity to energy markets. Two particularexamples are spinning reserves and AGC. Both theseservices require generating units to maintain on-lineunloaded capacity that can respond within 10 minutes.These services are used in real-time by the ISO, either forbalancing purposes or for replacing energy that had beenscheduled to be provided by a unit that malfunctions.The main characteristic of the ancillary services discussed inthis paper is that capacity is reserved and prcicured inadvance and then incremental energy is dispe.tched inresponse to real-time imbalances. This means that there aretwo relevant prices to be paid to generators. One is the pricepaid for reserved capacity and the other is the pricc: paid forenergy that is dispatched under certain defined conditions.Since the generator cannot anticipate or control the amountof energy it will be called on to produce, it is entitled to avariable energy payment. Since balancing markets are likelyto be much smaller than day-ahead markets, it is essentialthat market designers and regulators ensure thai: reservegenerators do not achieve market power in these real-timeenergy markets.Efficient dispatch may require some generating units thatare infra-marginal in the energy market to contribute

0885-8950/99/$10.00 0 1998 IEEE


2/7

49 9towards the reserve requirements in a control area. Theseinfra-marginal units may incur an opportunity cost forsetting aside a part of their capacity to provide reservesinstead of selling energy in the competitive market. Oneapproach to pricing the reserve services is to simultaneouslyprocure reserves and energy in a single combined auctionand compensate the units providing reserves at theiropportunity costs derived from the energy auction. Thisapproach can work in a market structure where the entityresponsible for procuring Ancillary Services is alsoresponsible for operating the energy market.Another approach is to set up a separate auction forprocuring reserves. A separate reserve auction is bettersuited for market structures where the Independent SystemOperator (ISO) is separate from the Power Exchange (PX),the entity that operates the forward energy market. Thispaper discusses the competitive procurement of AncillaryServices by an IS0 under such a market structure. Section2 of the paper describes the operation of the day-aheadenergy market. Section 3 discusses the California model forprocuring Ancillary Services while Section 4 presents therole and the financial responsibilities of the purchasers ofthe ancillary services. Section 5 describes how opportunitycosts are incurred in providing reserves while Section 6presents the structure of the reserves auction. Section 7presents the interactions between the various energy andancillary services markets. Section 8 describes howancillary services are dispatched in real-time and Section 9addresses practical limitations of th e proposed ancillaryservices auction model arising from market power concerns.

2. The Day Ahead Energy MarketGiven the close relationship between the energy and theancillary services markets it is desirable to start by brieflydescribing the structure of the day-ahead energy markets.The market structure assumed in this paper consists of aseparate grid operator or IS0 and market operator or PowerExchange (PX). This model has been adopted in Califomia.It is useful to begin with a conceptual description of the PXenergy auction. The PX auction is actually a series of 24auctions, one for each hour, conducted simultaneously.Generators submit price curves that describe their energyoffers as a function of price; the quantity of energy offeredmust be a monotonically increasing function of price.Loads submit similar curves which must be decreasing. Thebid curves may be different for different hours.The PX aggregates the curves into aggregate supply anddemand curves for each hour. It determines the point atwhich each hour's supply and demand curves intersect.This point sets the PX Market Clearing Price (MCP) for thathour. The bid structure forces generators to bid as if their

total cost curves are convex (that is, the bid curves areconvex and if one assumes generators bid their costs thenthey would be claiming to have convex total cost curves).Furthermore, the PX auction model adopted in Californiaand discussed here recognizes no operational or inter-temporal constraints. There may be iterations in the energyauction to help bidders satisfy their operationalrequirements, but they are not significant in describing theancillary services procurement model [2,3].This means that the PX prices and associated dispatchpatterns are not likely to lead to the conventional cost-minimizing dispatch for each hour. Units need not bid purevariable costs and are encouraged to internalize all physicalconstraints. This helps reduce the opportunities for arbitragebetween the PX and the ISO's balancing market and makesthe pricing process more transparent.The goal of the PX auction is not necessarily to minimizeproduction costs but to facilitate trade. The PX auction isintended to discover the prices at which parties are willingto transact.

3. Procurementof Ancillary ServicesThe Ancillary Services which the Califomia IS0 isresponsible for procuring include spinning reserves, non-spinning reserves, AGC, replacement reserves, voltagesupport and black start'. The Western Systems CoordinatingCouncil (WSCC) definition refers to the first tw o asoperating reserves and the first three are as contingencyreserves [7]. Spinning reserves refer to unloaded generatingcapacity that is on-line and synchronized to the system andall of which can be made available within 10minutes. Non-spinning reserves refer to unloaded generating capacity thatcan be made available within 10 minutes but is notnecessarily on-line or synchronized to the system. Thereserves and AGC services are usually provided bygenerators by setting aside unloaded capacity in the energyauction for possible use in real time by the ISO. The firstfour services can be procured by the IS0 by means of dailycompetitive auctions. These auctions occur after the PXenergy auction is completed. The sequence of the auctionsis AGC, followed by spinning reserves, non-spinningreserves, and replacement reserves. This sequence allowscapacity bids that are not selected in one auction (e.g. AGC)to be considered in a subsequent auction (e.g. spinningreserves). The last two ancillary services, i.e., black startcapability and VAR support, are more suitable forpurchasing based on long-term contracts from among thoseunits physically capable of providing them. Spinning and

' The definitions of th e various Ancillary Services used in thispaper are based on those used by the California I S 0 and areconsistent with standards set by theWSCC.


3/7

500non-spinning reserve requirements for the IS0 are definedby traditional reliability criteria [712. In addition, the IS0needs to make sure it has enough capacity online for a well-functioning balancing market. For this, the IS 0 may arrangefor replacement reserves which are defined as unloadedgeneration capacity (not necessarily on-line) that can bemade available within 60 minutes.One can interpret operating reserves as a call option. Thestrke price associated with the call option could be theenergy price that is agreed to at the time of purchasing thereserves. The exercise value of the option is the price thatwould otherwise be paid for balancing energy. However, animportant characteristic of reserves, as distinct fromfinancial options, is that they are procured not based on anydynamic hedging strategy but based on preset reliabilitystandards defied by NERC and WSCC that relate thereserve requirement to overall generation.

4. Ancillary S ervices SettlementsThe costs incurred by the IS0 for purchasing reserves canbe classified into two categories, (a) those that are mandatedby reliability protocols, and (b) those that can be assigned tothe ISOs customers as a function of their observedperformance and behavior. For our purposes it is sufficientto assume that the ISOs customers are represented byScheduling Coordinators (S CS)~ consistent with theCalifornia energy market model. Since the reliabilityrequirements are proportional to scheduled load, the reserveresponsibility can be allocated pro-rata to SchedulingCoordinators in proportion to their submitted schedules.The one exception is the case of replacement reserves whichare procured (at least in part) on the basis of what the I S 0expects to be the difference between its own load forecastand the sum of the SC submitted schedules. In case thereplacement reserves are actually dispatched, their costs areallocated to the Scheduling Coordinators in proportion topositive imbalances.

The wscc requires an operating reserve requirement of 7percent of scheduled demand in addition to any provisions madefor interruptible imports and firm exports.A call option is a financial instrument that gives the holder aright to buy the underlying asset at a certain time at a certain price.A Scheduling Coordinator submits a set of balanced generation

and demand schedules to the IS0 in the day-ahead market andsettles with the IS0 for any imbalances in the real-time market.The PX is also a Scheduling Coordinator.For a given Scheduling Coordinator, an imbalance is defined asthe difference between the actual aggregate load and actualaggregate generation. An imbalance is positive when actual loadexceeds actual generation.

5. Opportunity CostsBefore proceeding with the design of an auction forreserves, it is important to understand the struclure of thecosts incurred in providing reserves. There are two types ofcosts that must be represented. These include theopportunity costs of providing reserves and the costsincurred in case the generator is actually dispatched. Tounderstand how opportunity costs arise, consider the case ofa single reserve category, spinning reserves. We usespinning reserves as a prototype to motivate the zialysis butthe same logic applies to other ancillary services $,haring hecharacteristic that capacity is reserved in advance and thenused as dispatched generation in real-time.Suppose that for a given hour the system load requirementis 6000 MW and that the spinning reserve requirement is200 MW. Assume that 5600 MW of demand is met by abase load nuclear plant operating at h l l output. For theremaining 400 MW of demand and the 200MW of spinningreserves, there are two units A and B with incrementalenergy bids of 4ckWh and 5cikWh respectivdy and acapacity of 300 MW each. In the absence of any rampingrate limits, the economic solution would be to dispatch unitA at its upper limit of 300 M W and unit B at 100MW. UnitB would provide the spinning reserve from its unloadedcapacity. Since unit B is the marginal unit in the energydispatch, it does not incur any opportunity cost in providingspinning reserves.Next assume a more realistic scenario where both units areconstrained by ramping rate limitations; each i s able toprovide a maximum of 100MW in 10minutes. In this case,the spinning reserve must be provided by both units andboth are dispatched at 200 MW. Unit A is an infra-marginalunit, that incurs an opportunity cost to be able to providespinning reserves. In current operation, it is common to seeinfra-marginal units incurring opportunity costs inproviding reserves during peak load hours. This suggeststhat bidders should be allowed to bid in a coinpetitiveauction for reserves their assessment of opportunity costs interms of a capacity reservation bid,The one case where a capacity bid may reflect actual costsinstead of opportunity costs is AGC. AGC requires agenerator to operate at a level higher that its rninirnumoutput, in order to allow movement of the unit in downwardas well as upward directions. For example a unit Qith costsof 5 ckWh would not normally operate at levels higher thanits minimum output when the price in the energy marketfalls below 5 c/kWh. In order for it to do so, it is entitled tobid for a capacity payment to cover the actual costs the unitmay incur.


4/7

5016. The Reserves Auction

This section presents the basic structure of an auction forcompetitively procuring ancillary services. The prsotgdmarket structure assumes that there are separate au c t i w o reach category of reserves (spinning, non-spinning andreplacement). Each bidders submits a separate capacity bidR, ($/MW), and energy bid E, ($/MWh) for each service.Successful bidders are paid a capacity reservation paymentthat is intended to reflect their opportunity costs. Biddersare also paid for energy if they are called on to provide it inreal time. Depending on the specific case (as determinedlater in this section by the value of a weighting factor x) thisenergy price can be the real-time market price or a strikeprice as if an option were being exercised in the real-timebalancing market.The auctions for reserves differ from the PX auction in onevery important aspect. The goal of the PX auction is tofacilitate trade because there will be many participants onboth sides, i.e., buyers and sellers6. Ancillary services arefully the ISOs responsibility, so the IS0 acts as the singlepurchaser in reserve auctions to meet its reliabilityobligations. If the purchaser of energy from reserves couldbe identified in advance it could be forced to schedulecapacity. Because the energy purchase is a random event,the IS0 as the agent for all market participants buys thereserves according to a preset formula. As a single buyerthe IS0 can seek to minimize its costs; the formula thatgoverns the purchase should prevent strategic withholdingof reserve demand.An important property of opportunity costs is that the moreinfra-marginal a unit (the deeper in the loading order), thegreater its opportunity costs in providing spinning reserves.Thermal units capacity reservation bids should reflect theiropportunity costs, which in turn reflect lost profit (marketenergy price minus marginal cost). The capacityreservation bids R, hould be inversely related to the energybids E;, if energy bids are cost-reflective as they ought to bein a competitive market. Thus, to minimize the expectedcost of reserves, the IS0 cannot consider only the capacityor only the energy bids, but must estimate the probabilitythat reserves will be utilized in real-time. It must include aprobability factor represented by a parameter x in the bidevaluation, i.e., bids are ranked on the basis of Ri+XEi. Inorder to keep the auction transparent, x must be set by theIS0 prior to bid submission.The ISOs reserve auction can then be expressed as thefollowing cost minimization problem:

The PX energy auction is a double auction while the ISOsancillary services auctions are single sided auctions.

where Q, is the quantity of reserves offered by resource i, R,and Ei are the associated capacity reservation and energybids, and QFaXs the maximum capacity that the resourcecan offer based on its ramp-rate. QTq is the quantity ofreserves that needs to be procured.Next altemative auction designs are examinedcorresponding to different values for the parameter x.The (x = 1 ) case:By setting x=l, the IS0 can determine the maximum extentof its payments [ l] . Under assumptions of cost revealingbids, the x=l scenario allows a bidder to be indifferentbetween providing spinning reserve or energy. Successfulbidders are paid a capacity payment at a price Ciwhere

C i =max {R + E i } - E ,i;isnand ZZ is the set of all successful bidders. They are paid theirenergy strike price Ei for any energy they are called uponto generate in real-time.The (0 5 x I 1 ) case:In the design described in [2], the IS0 proposed an auctionloosely based on the notion of purchasing call options onthe real-time balancing market. In order to pursue a costminimizing bid evaluation, the IS0 would use an unbiasedestimate of the probability x (between 0 and 1). In this case,the bids are ranked according to R, + xEi. Successfulbidders are paid at a price

C i = m a x { R i + x E , } - x E ii; isf lA reserve generator is paid its energy strike price Ei if it iscalled upon to generate in real-time, irrespective of thebalancing market price.The (x=O) case:The (0 I x S 1) case may be criticized as being too complexand susceptible to being gamed. It is only an approximationto the true cost minimization, as the probability of callingupon a successful bidder to generate in real-time is likely tovary according to the energy bids. Presumably, intelligentbidders could game this market structure to increase theirrevenues. One scenario involves bidders lowering theirenergy bids in order to increase their capacity payments.However, this also increases the probability of the biddersgetting called upon to generate in real-time. Ideally, abidder would like to lower its energy bid as much aspossible without altering its position in the merit orderstack. Finally, the option-based case does not result in auniform price for reserve capacity payments making it lesstransparent.


5/7

502

An alternative design, is to rank bids solely on the capacityreservation: i.e., the IS 0 takes bids in order of reservationbids, and pays each successful bidder the highest reservationbid among those accepted.Since the energy bids are ignored in the bid evaluation,there is less justification to consider them as fixed strikeprices. It might appear reasonable to assume that thereserve mils are price-takers in the real-time market, i.e.,they cannot set the real-time price which they must accept.Clearly if all reserve units are price takers (and there is noobligation to replace reserves that are used), the IS0 couldend up dispatching all the reserves without impacting thereal-time price.To solve this problem, reserve units are deemed to bid their(preset) energy prices into the balancing market plus anyadditional costs the IS0 may incur. Reliability Councilrequirements mandate that the IS0 should replace spinningreserves that are used in real-time within the hour (i.e., ifenergy is taken from a spinning reserve unit, the IS0 needsto procure additional spinning reserve within a certain time-frame in order to get its reserves back to the WSCC-mandated level). The IS0 must factor the cost of suchreplacements into dispatch decisions. To do so , the energybids are increased by additional costs such as capacityreplacement.Under this design, energy bids are predetermined, but notpayments. For their generation, bidders will be paid theuniform balancing market price. This might seem counter-intuitive from a cost minimization perspective. However, itemerged as the best alternative in the California electricindustry restructuring debate under the assumptions of acompetitive real-time market.The three alternatives for bid evaluation are illustrated byusing a simple example. Suppose there are three generatingunits that submit bids as shown in Table 1 into the spinningreserves auction.

Table I : Bids fo r spinning reserves au ctionUnit ($/MW) ($NWh) (MW/min) capacity (MW)

22 10012 20 10 100

Suppose that the IS0 needs to procure 180 MW of spinningreserves. Depending on the value chosen for x, the outcomeof the auction can be different. The results of the ISO'sauction for different values of x are shown in Table 2 and

Table 3. Table 2 shows the selected quantities and Table 3shows the capacity prices that are paid to the selectedbidders.

Table 2: Spinning reserve auction resultsSelected MW quantities

800.5 80 1001 0 100

~~

Table 3 : Capacity payments to successful b,'dders

120.5 4 10.5

1 0 10The analysis put forth to support this alternative assumedthat reserve bidders would bid their true energ)' costs, andthat their capacity reservation bids woullil be thecorresponding opportunity costs. Those asseitions werebased in turn on the assumption that the reserve marketwould be competitive as well as the market fo:: balancingenergy. That second assumption may not hold ifopportunity costs are large enough that non-reserve units areeliminated from the balancing market. Therefore, theargument against an explicitly cost-minimizing solution isless compelling than in the case of the PX auction.

7. Interactions Between the Various MarketsThe ancillary services markets are embedded in a series ofenergy and capacity markets including the forward energymarkets (day-ahead and hour-ahead) and the real-timebalancing market. It important to manage the interactionsbetween the two markets.For example, the IS0 is responsible for ensuring reliability.Therefore it needs to make sure there is sufficientgeneration available to match with demand spikes. It buysthat energy from the real-time market but because there isno price set on reliability (and no known cost for failure toserve) it cannot assume that the real-time price will sufficeto secure sufficient supply. Therefore, the IS0 may have toschedule replacement reserves, in excess of the WSCCrequirement for reserves, to ensure liquidity in the real-timemarket.Just like generators, loads can choose between fie forwardand real-time markets. In other words, a load without


6/7

503prescheduled supply can just choose to appear in the real-time market. In fact, if forward prices exceed incrementalcosts but real-time prices do not - because they are based onincremental bids from already-operating units, or becausefixed costs of operation are covered by reservationpayments - we should expect massive desertion of loadfrom the forward markets into the real-time markets. Thesolution for mitigating against this possibility that has beenput forward is to make real-time market load deviationsresponsible for replacement reserve capacity payments[ V I .There are also relations between the various reservemarkets, and the real-time market. The IS 0 may not alwaysbe able to choose the cheapest source of energy from thereal-time market. For example, a unit receiving capacityreservation payments as replacement reserve, while it mayhave the lowest energy bid among available generators, itmay not be able to respond to the ISOs call in a timelymanner. The IS0 may have to take a spinning reserve unitinstead. On the other hand, the IS 0 will probably try toavoid taking energy from spinning or 10-minute reserveunits whenever possible because of the high associatedreplacement costs.This means that when the IS 0 looks for real-time generatior.it looks only at a subset of the available balancing units.However, one cannot consistently identify the relevantenergy market as associated with a particular reservecategory. Thus, the size of the reserve energy market issmall, raising market power concerns; these concerns, albeitserious, are hard to quantify.

8. Dispatch of Ancillary ServicesAll schedules received by the IS 0 from the PX and otherSC s are balanced. In real-time, deviations from scheduledgeneration and load levels are inevitable, either as a resultof normal load and generation variation or as a result oflarger unplanned outages, or as a result of strategicdecisions by market participants. Such uninstructeddeviations from schedules result in an energy imbalance thatis manifested in the Area Control Error (ACE). Thegenerators on AGC are the first to respond to theimbalances. This action takes place within a time-frame ofseconds. In order to periodically, restore the AGC units totheir set-points, the IS0 must procure energy by means ofinstructed deviations from resources that have submitted theappropriate energy price/quantity bids to the ISO. Thisaction takes place on a slightly slower time-frame; say everyten minutes. The energy bids (Ei) from the various resourcesincluding the three types of reserves can be aggregated toform a supply curve as shown in Figure 1.

Price---PG e n e r a t i o n > Load Load > G e n e r a t i o n

Figure 1: Supply curvefor real-time dispatchEvery ten minutes, the IS0 conducts a single sided a~ctionfor buying or selling energy depending on whether there isexcess demand or excess generation in the system. In thissense, the reserves are used along with other availableresources in the real-time balancing market. The highestprice resource used in the interval sets the price for that tenminute interval. A weighted average of the ten minuteprices is calculated at the end of the hour to apply to alluninstructed deviations. A detailed description of the real-time market is beyond the scope of this paper.

9. Market Power andRegulationIn order to secure the regulators permission to sell energyat market based prices, utilities typically must present adetailed analysis showing the absence of market power.Similarly, they must show that they lack market power inthe ancillary services markets in order to be able to sellancillary services at market based prices. There have beenfew, if any, attempts to analyze market power in the case ofancillary services. This means that utilities must agree toaccept cost-based rates for ancillary services andparticularly for reserve capacity. However, determining thecost based rates can be equally arbitrary. Moreover, bychoosing a single price to serve as an upper bound or capfor all the hours creates more problems than it solves. Forexample, in peak hours the opportunity cost of supplyingspinning reserves for a generator can be higher than the costbased cap. This would eliminate incentives for infra-marginal generators to make unloaded capacity available forreserves instead of supplying energy. The net consequencewould be a shortfall in ancillary services. On the other hand,in off-peak hours, because of the acknowledged possibilityof market power, it is quite possible that the prices forreserves could approach the cost-based rates, which can behigher than the competitive prices. This observation seemsto be the case in the initial experience in the Californiamarket. The prices and quantities for spinning reserves areillustrated in Figure 2 for a given day. The pricesthroughout the day were at or near the bid caps. During thehours 1-6 and 22-24, the prices were at or near 7.40 $/MW

t any given instant, the I S 0 is either buying or selling energy.The IS0 does not attempt to optimize the dispatch bysimultaneously selling and buying energy.


7/7

504and during the hours 7-21, the prices were at or near 9.50$/MW8. The quantities procured in the hours 7-21 were lessthan in the lower priced hours, suggesting insufficientsupply at the capped prices. In cases of such price-capinduced shortfalls, the IS0 may be forced to rely on theincreased use of generators on long-term Reliability MustRun (RMR) contracts, at costs that can easily outweigh anysavings that result from the price caps.

Spinning Reserves Market ( 42 /98 )- 10

8 8- - 4

I , - 0.-

I-MW -$/MW/ 2 rt-d cucu

Figure 2: Spinning reserves market resultsThe reserve capacity markets are only loosely coupled tothe other parts of the restructured energy market. The real-time balancing market, on the other hand, is much moreclosely related to the forward energy markets. Balancingenergy is freely substitutable for forward energy. On theother hand, balancing energy is supplied largely from unitsreceiving reserve capacity reservation fees. If the regulatorsaccept that utility market power is mitigated in the forwardenergy market, but not in the ancillary service market,where does the real-time market fall? A determination thatbidders in the real-time market can only receive cost-basedrates can have serious repercussions on the prices in theforward energy markets.

10 . ConclusionCompetitive markets for electricity require competitivemarkets for reliability or ancillary services. This paperpresents a market structure for the competitive procurementof such services and discusses the role of the opportunitycosts incurred in providing these. The design of an auctionfor ancillary services such as reserves presents an interestingdilemma between strict cost-minimization and efficientincentives for bidders. The proposed design seeks to chooseefficient incentives over strict cost-minimization under theassumptions of a well functioning and competitive market.

~ ~ ~ ~ p p8 In the actual implementation of the Califomia ISOs ancillaryservices auction only those bidders that are not subject to price-caps are eligible to receive the market clearing price. Bidderssubject to price-caps are paid according to their bids. Thisdeparture from a uniform price auction may also contribute to anincrease in prices.

DisclaimerThis paper does not necessarily reflect the positions ofPG&E. Any errors or omissions are the sole responsibilityof the authors. References[11 Federal Energy Regulatory Commission (FEW). Jointapplication of Pacific Gas and Electric Company, San DiegoGas and Electric Company, and Southem California EdisonCompany for authorization to convey operational control ofdesignated jurisdictional facilities to an Independcnt SystemOperator, (Phase I IS0 filing), FERC Docket No. ER96-1663-000, April 29, 1996.[2] Federal Energy Regulatory Commission (FERC).PhaseI1 filings for the California IS0 and PX, FERC Clocket No.ER96-1663-001, March 31, 1997.[3] Federal Energy Regulatory Commission (FERC).Amended Phase I1 filings for the Califomia IS0 and PX ,FERC Docket No. ER96- 1663-003, August 15, 1997.[4] E. Hirst and B. Kirby, Creating Competitivi: Marketsfor Ancillary Services, Oak Ridge National L Iboratory,October 1997.[5] L.D. Kirsch and H. Singh, Pricing Ancillary ElectricPower Services, The Electricity Journal, October 1995.[6] Wilson, Robert, Priority Pricing of Ancillary Services,Report to the Trust for Power Industry Restructuring, May17, 1997.[7] Minimum Operating Reliability Criteria (MORC),Western Systems Coordinating Council (WSCC),September 1996.Harry Sinah received a Ph.D. in Electrical Engineering from theUniversityof Wisconsin-Madison in 1994.He is current y with thePacific Gas and Electric Company in San Francisco where he hasworked on the implementation of the Califomia PX and ISO.Harry was one of designers of the Califomia ISOs ancillaryservices market. His research interests include power systemsanalysis, energy economics, and mathematical programing. He isa member of IEEE, SIAM, and Sigma Xi .Alex D. aualexououlos received the Electrical and h4echanicalEngineering Diploma from the University of Athens, Greece in1980 and the M.S. and Ph.D. degrees in Electrical Engineeringfrom the Georgia Institute of Technology, Atlanta, Georgia in1982 and 1985 .respectively. Alex is heavily involvcd in theelectric industry restructuring efforts in California where he was akey developer of th e market rules for th e California I S 0 and th ePX and he led the implementation of th e IS0 business systems. Heis a senior member of IEEE and a member of Sigma X i and theTechnical Chamber of Greece. Alex has published iumerouspapers in IEEE and he is the 1992 recipient of PG&Ej Wall ofFame Award and th e 1996 recipient of IEEEs PES Prize PaperAward.

Ancillary Services4

Documents

Transcript of Ancillary Services4