Dispatching Reactive Power Considering All Providers in Competitive Electricity Markets

Transcription

1 IEEE-PES General Meetng, Mnneapols, July Dspatchng Reactve Power Consderng All Provders n Compettve Electrcty Markets Hossen Haghghat, Claudo Cañzares, Fellow IEEE, Kankar Bhattacharya, Senor Member, IEEE Abstract Ths paper proposes a level playng feld for the supply of reactve power ancllary servces, wheren not only synchronous generators, but other provders of reactve power are also pad for ther servces. An Optmal Power Flow (OPF)-based reactve power dspatch model s proposed based on the reactve power payment mechansms exstent n Ontaro. Novel cost models are proposed for Statc VAR Compensators (SVCs) and Statc Synchronous Compensators (STATCOMs) and ncluded n the dspatch model. The proposed methodology s tested on a dspatch model of Ontaro power grd, and the results show that the proposed technque can sgnfcantly reduce the cost of reactve power dspatch whle mantanng system securty. Index Terms- Electrcty markets, reactve power dspatch, SVC, STATCOM. I. NOMENCLATURE A. Parameters HOEP: Hourly energy Ontaro prce n $/MWh. P G : Actve power generaton at bus n Q mn G : Mnmum reactve power lmt of a generator. P D : Actve power demand at bus n Y : Element of admttance matrx n θ : Angle assocated wth Y n radans. Q D : Reactve power demand at bus n Q mn : Mnmum reactve power of generator g, n Q SVC : Rated VAR capacty of SVC n Q STATCOM : Rated VAR capacty of SVC n B1 : Prce of upward balance servces n $/MWh B2 : Prce of downward balance servces n $/MWh P max B1 : Maxmum upward balance servce at bus n P max B2 : Maxmum downward balance servce at bus n V max : Maxmum allowable voltage at bus, n V mn : Mnmum allowable voltage at bus, n Ths work was ontly fnanced by the Ontaro Centres of Excellence and MITACS, Canada. H. Haghghat, C. A. Cañzares, and K. Bhattacharya are wth the Department of Electrcal & Computer Engneerng, Unversty of Waterloo, Waterloo, Ontaro, Canada, N2L 3G1 (emal: h2haghg@engmal.uwaterloo.ca; kankar@ece.uwaterloo.ca; ccanzar@uwaterloo.ca). P max : B SVC_L B SVC_C I STATCOM_L I STATCOM_C b SVC : a SVC : Maxmum power flow from bus to bus, n SVC mnmum nductve susceptance n SVC maxmum capactve susceptance n STATCOM maxmum nductve current n STATCOM mnmum capactve current n Fxed cost component of O SVC, $/ Varable cost component of O SVC $/ B. Varables: O SVC : SVC offer curve. Q G : Generator reactve power at bus n P B1 : Upward balance servce at bus n P B2 : Downward balance servce at bus n V : Bus voltage magntude n δ : Bus voltage angle n radans. P : Power flowng from bus to bus, n δ : Power flowng from bus to bus, n Q SVC : SVC reactve power n Q STATCOM : STATCOM reactve power n B SVC : SVC susceptance n I STATCOM : STATCOM current n Loss SVC : SVC actve losses n Loss STATCOM : STATCOM actve losses n Loss : Actve losses n generatng unt g n R II. INTRODUCTION EACTIVE power dspatch s a crtcal short-term functon carred out by power system operators n order to operate the system n a secure manner. The tradtonal reactve power dspatch paradgm based on mnmzaton of losses has gradually gven way to new crtera such as reactve power payment mnmzaton [1]. In the recent lterature, a two-ter structure for the management of reactve power has been proposed n the context of compettve electrcty markets [2]-[4]. The latter propose that the problem of reactve power management be splt nto a procurement problem and a dspatch problem, wth the procurement problem beng essentally a long-term ssue of contractng approprate set of generators for the servce provson, whereas the dspatch problem deals wth allocaton of reactve power generaton to the unts n the

2 IEEE-PES General Meetng, Mnneapols, July real-tme. In the procurement problem, the Independent System Operator (ISO) seeks to dentfy the generators that are crtcal for provdng reactve power support, consderng the overall system securty. In [2] and [4], the problem s treated as a generator bddng process on a seasonal bass to avod potental problems assocated wth the effects of prce volatlty of energy markets on reactve power prces. Wth the help of frst a maxmum loadablty Optmal Power Flow (OPF) and then a securty-constraned OPF, the ISO determnes, based on reactve power offers, zonal reactve power prce components and the key sets of supplers. Once the reactve power prces are known from the procurement stage, the dspatch problem s carred out close to real-tme to optmally allocate the system reactve power demand to supplers; ths process s based on an OPF that mnmzes the total ISO costs assocated wth reactve power dspatch subect to securty constrants [3], [4]. In ts 1996 Order No.888, the Federal Energy Regulatory Commsson (FERC) had recognzed reactve power supply and voltage control servces from synchronous generators as one of the sx ancllary servces that transmsson provders must nclude n ther open access transmsson tarff. It s also stated that reactve power from capactors and FACTS controllers, that form part of the transmsson system, were not separate ancllary servces [5]. However, there are recent recommendatons from FERC to recognze reactve power provsons from sources other than synchronous generators, as ancllary servces, so that they are elgble for fnancal rembursement [6]. In Ontaro, synchronous generators are pad for the real power losses (MW-losses) ncurred when operatng at nonunty power factors. are made at the Hourly Ontaro Electrcty Prce (HOEP) rates based on calculated generator losses for the hour. There s no payment for reactve capablty wthn the standard power factor range; n other words, there s no payment n Ontaro for the costs of equpment such as excters whch are deemed essental for real power producton by all generators. To avod economc dstorton because of possble large revenues for generators operatng solely wthn standard power factor ranges, capablty payment for reactve power s gnored n Ontaro. Alternatve var supplers such as capactors, reactors, Statc Var Compensators (SVCs) and Statc Synchronous Compensators (STATCOMs) are elgble for payments for ther costs of nstallng and mantanng the equpment [7]. It has been dscussed and demonst n [8] that f synchronous generators are the only reactve power ancllary servce provders n the system, sgnfcant possbltes of market power can arse at certan buses n the system. It was suggested theren that n order to allevate such stuatons, the reactve power market be a level playng feld wheren all reactve power provders are consdered ancllary servce provders and be elgble for payment. Such a suggeston s also n lne wth FERC recommendatons [6]. In [9], a reactve power capacty market s proposed wheren the ISO procures reactve power capacty through annual auctons; the optmal capacty s determned consderng offers from generators and other sources of reactve power such as capactors and SVCs. In vew of the above, the man obectve of ths paper s to propose and present a level playng feld reactve power dspatch model that can address the problems of reactve power market neffcences and brng n more farness and competton n reactve power ancllary servce provsons. The proposed dspatch model seeks to mnmze the ISO s total cost of reactve power dspatch, ncurred through payments to the servce provders, whle mantanng the securty of the system. It should be ponted out that, unlke [2], [3], [4] and [8], ths work does not consder var prce offers from the servce provders, so that t s more n lne wth the practce of the Independent Electrcty System Operator (IESO) of Ontaro, whch s the Control Area operator n Ontaro responsble for admnsterng the wholesale electrcty market. The reactve power supplers are pad for ther actve power losses (MW-losses) ncurred because of reactve power provson, at the HOEP rate as n Ontaro. The approach proposed n ths paper dffers from that of [2]-[4] n some other aspects as well. Frstly, the procurement model s not consdered here, snce the framework does not requre submttng prce offers for reactve power. Secondly, the ISO s payment obectve functon has been modeled n ths paper to resemble the IESO s practce, whereas n [3] and [4] the payment obectve was based on reactve power dspatch and actve power redspatch. Lastly, the work takes nto account the presence n the dspatch model of other sources of reactve power such as capactors, reactors, SVCs and STATCOMs, whch are recognzed elgble for payment for ther reactve power ancllary servce. The rest of the paper s organzed as follows: In Secton III the cost of reactve power provson for statc VAR compensators s dscussed. The reactve power dspatch model s ntroduced n Secton IV, and s based on the optmal allocaton of reactve power demand to supplers whle mnmzng ther MW-losses,.e. mnmzng the reactve power dspatch costs for the ISO. The results from the applcaton of the models to the IESO-controlled grd dspatch model, whch s comprsed of 2,833 buses and 4,205 branches, are presented and dscussed n Secton V. Concludng remarks and a hghlght of the man contrbutons of the paper are provded n Secton VI. III. COST OF REACTIVE POWER FOR SHUNT COMPENSATORS The domnant cost of reactve power producton from shunt compensators such as SVCs and STACOMs can be decomposed nto two cost components: (1) cost of nstalled var capacty, and (2) cost of operaton. The nvestment cost

3 IEEE-PES General Meetng, Mnneapols, July of SVC/STATCOM s typcally n the range of 40 to 50 $/kvar [10]-[11]. Consderng a lfetme of 20 years and a dscount rate of 8%, the amortzed cost s n the range of 4.07 to 5.1 $/kvar-year, or equvalently 0.46 to 0.58 $/Mvarh. Actve power losses consttute a maor part of the SVC operatng cost, and s typcally n the range of 0.5 to 0.75% of the reactve power output [12]-[13]. For STATCOMs, the varable cost s essentally assocated wth the losses n the converter whch, at the output, are hgher than for comparable SVCs [14]; these losses are typcally less than 1% of the reactve power output [10]. In order to arrve at a typcal range for the varable cost of SVCs and STATCOMs, the followng assumptons are made: The SVC/STATCOM s fully utlzed 40% of the tme. The HOEP s n the range of 80 to 100 $/MWh, whch s a typcal hgh prce n the Ontaro market. The total actve power losses n the SVC/STATCOM are approxmately 1% of the reactve power output. Gven the above assumptons, the total gene reactve power for a 1 Mvar SVC/STATCOM n a year would be 40% 1 Mvar 8760 h = 3504 Mvarh, and the correspondng actve power losses are then 1% 3504 Mvarh = MWh. Therefore, the annual cost of losses at the HOEP rate would be MWh $/MWh = $2,803-$3,504, whch yelds an approxmate varable cost of $/Mvarh. These calculated values can be used for both SVC and STATCOM, provded the aforementoned assumptons hold true. In addton to the actve power losses whch vary wth the output level of the SVC/STATCOM, the devce ncurs losses even when t does not supply any reactve power. These no-load losses are typcally estmated at 0.1% of the SVC/STATCOM ratng [10]. In order to ensure that the nvestment cost n the SVC/STATCOM s recovered, the devce should be pad for both ts fxed and varable costs. Hence, the offer curve for reactve power suppled by the SVC/STATCOM would consst of two components as follows: OSVC bsvc QSVC asvc QSVC $/h bsvc a [0.46,0.58] $ / M varh [0.032,0.040] $ / M varh SVC The parameters n ths equaton are based on the aforementoned dscussons. Equaton (1) can also be used for the STATCOM. The actve power losses n SVCs and STATCOMs can be calculated based on ther loss curves, whch yeld the actve power losses n the devce wth respect to the reactve power output level [15]. The actve power losses n an SVC n at an operatng pont Q SVC are gven by: 2 svc ( svc ) svc (1) Loss Q Q (2) And smlarly for the actve power losses n the STATCOM. The compensator s then pad for actve power losses at the HOEP rate, as per the IESO current practces, plus an uplft payment to account for the fxed costs, as descrbed n the next secton. Smlarly to SVCs and STATCOMs, the cost ncurred by a generator for reactve power producton can be decomposed nto two parts. A fxed part assocated wth the dfference between the plant buldng costs wth and wthout a reactve power margn, and wth the equpments needed to mantan that margn [16]; ths fxed component s gnored n the reactve power dspatch. The varable part s prmarly due to actve power losses (MW-losses) assocated wth the reactve power output. For synchronous generators, these losses are classfed nto Joule, eddy, hysteress and stray losses, mechancal losses, and excter losses. Typcal expermental curves for losses n the rotor, stator and step-up transformer as a functon of reactve power necton are provded n [16] for a gven nomnal actve power. Based on these loss curves, the loss functon approxmatng the total real power losses n the man and auxlary parts of a typcal generator can be represented by: Loss Q Q (3) MW where Q s the generator reactve power n Mvar for generator g. Observe that the total losses are mnmal for a certan amount of absorbed reactve power snce Q < 0, whch s not the case for nected reactve power. The loss functons approxmatng total MW-losses for generators and SVCs/STATCOMs (2) and (3), respectvely, are used n the Q-dspatch model to calculate the MWlosses and the correspondng payment to reactve power supplers, as descrbed next. IV. PROPOSED LEVEL PLAYING FIELD Q-DISPATCH MODEL Ideally, reactve power should be dspatched n an economcal manner to mnmze transmsson losses and actve losses of generatng facltes whle keepng the system secure. Consderng the complextes nvolved n supplyng reactve power n deregulated electrcty markets, a method for reactve power dspatch s proposed here that s sutable for real-tme applcatons. Gven that the actve power dspatch levels of the generators are known from energy market clearng, the ISO can determne the reactve power dspatch usng an OPF-based model whch mnmzes the total actve power losses of reactve power supplers subect to system securty constrants. In ths approach, the supplers of reactve power ncludng generators, SVCs and STATCOMs are assumed to be pad for ther real power losses, whch s the payment mechansm presently used n Ontaro. It must be mentoned that synchronous generators n the Ontaro electrcty market are pad f they are requred to operate outsde ther standard power factor range and thereby reduce ther real power output. In such crcumstances, generators are pad for ther lost opportunty

4 IEEE-PES General Meetng, Mnneapols, July cost based on the reducton n generaton level ncludng losses at the HOEP rate [7]. Snce ths stuaton seldom arses n Ontaro, n the dspatch model presented here t s assumed that generators are not allowed to operate n the opportunty regon and are therefore only pad for ther actve power losses. Based on the loss equatons (2) and (3) for generators and SVCs/STATCOMs, the level playng feld reactve power dspatch problem can be formulated as a securtyconstraned OPF to mnmze the total actve power losses of reactve power supplers. The obectve functon of the problem can then be defned as follows: J 1 g m k HOEP Loss ( HOEP Loss b Q ) SVCm SVCm SVCm ( HOEP Loss b Q ) STATCOMk STATCOMk STATCOMk where J 1 represents the total cost of reactve power dspatch n $/h, whch s based on the calculated losses and the HOEP rate n $/MWh. The frst term n (4) refers to the actve power losses n the generator, the second term represents the actve power losses n the SVC, and the last term denotes the actve power losses n the STATCOM. Note that n (4), SVCs and STATCOMs are pad for both the varable and fxed costs of reactve power generaton. If the devce s not dspatched, the payment assocated wth the varable component wll be zero; however, t s assumed that t wll be pad for the fxed cost component f avalable for dspatch. Observe that these fxed costs do not affect the optmzaton process and can be removed from (4); however, they are taken nto account durng the payment process. The reactve power dspatch s then based on the followng OPF model: mn. J 1 (5) s.t. P P V V Y G D QG QD QSVC QSTATCOM VV Y (4) cos (6) sn V mn P V V max V P max (7) (8), (9) QSTATCOM VSTATCOM ISTATCOM (12) B B B m (13) SVC_ Lm SVCm SVC_ Cm I I I k (14) STATCOM _ Ck STATCOMk STATCOM _ Lk Equatons (6) and (7) represent the nodal actve and reactve power flow equatons, respectvely. Constrants (8) and (9) mpose securty lmts on the bus voltages and transmsson flows, and (10) restrcts the generator reactve power to be wthn ts lmts. Equatons (11) and (12) model the reactve power gene by the SVC and STATCOM, respectvely, as the former s bascally an mpedance based controller, whereas the former s fundamentally a controllable voltage source [10], [15]. The relevant lmts on the SVC susceptance and STATCOM current are mposed through constrants (13) and (14), respectvely. The soluton to the OPF problem (5)-(14) yelds the optmum levels of reactve power dspatch and the payment to each suppler. Note that n ths OPF problem, the reactve power suppled by capactors and reactors has not been consdered; nevertheless, f the loss curves of these devces are avalable, the ncluson of the correspondng costs s straghtforward. It s assumed that the MW-dspatches of generators are known and reman unchanged durng the reactve power dspatch process. Based on current IESO procedures, f there s a need for real power reschedulng n the OPF model (5)-(14) due to reactve power dspatch or voltage control needs, the requred amount s allocated to the slack bus n the system. In certan market structures, balancng servces may be used to make up for the real power unbalance between the energy dspatched and the demand, as dscussed n detal n [3], [4]. In ths case, the obectve functon can be changed to: J 2 g m k HOEP Loss ( HOEP Loss b Q ) ( HOEP Loss b Q ) P B1 B1 B2 B2 SVCm SVCm SVCm STATCOMk STATCOMk STATCOMk P (15) where the last term denotes the payment towards balance servces. The requred energy upward or downward balance servces are represented by P B1 and P B2, respectvely, wth the correspondng costs B1 and B2, and lmts: max B1,2 PB 1,2 P (16) Q mn Q Q max g (10) QSVC VSVC BSVC 2 (11)

5 IEEE-PES General Meetng, Mnneapols, July Fg. 1 the IESO-controlled grd [18] In ths case, the nodal actve power flow equatons need to be modfed as follows: PG PB 1 PB 2 PD VV Y cos (17) Therefore, the OPF n ths case conssts on mnmzng J 2 subect to constrants (7)-(14), plus (16) and (17). The reactve power dspatch problem (5)-(14), or ts correspondng balance servce verson, s a Nonlnear Programmng (NLP) problem that was mplemented n the AMPL envronment and solved usng the IPOPT solver [17]. The soluton bascally yelds the optmal reactve power dspatch for each suppler. V. ONTARIO GRID CASE STUDY The IESO-controlled grd was used to test the applcaton of the proposed Q-dspatch model. Ths s the porton of the Ontaro power system that s controlled by the IESO, and ncludes all transmsson lnes at voltage levels 50 kv or greater. The system nterconnects wth two provnces n Canada (Mantoba and Quebec) and three states n the Unted States (Mchgan, Mnnesota and New York). The total length of the transmsson lnes s about 31,000 km. The nstalled generaton capacty n Ontaro s about 31,000 MW wth a peak demand of nearly 27,000 MW [18]. The reduced verson of the IESO-controlled grd employed for dspatch purposes by the IESO was used for the studes presented and dscussed here. Ths grd model

6 IEEE-PES General Meetng, Mnneapols, July Bus TABLE I LOSSES FOR DIFFERENT COMPENSATOR LOCATIONS Output (Mvar) P Reschedule Trans. No SVC & TABLE II PAYMENTS FOR DIFFERENT COMPENSATOR LOCATIONS Bus Gen. Gen. Total No SVC ,449 61, ,210 61, ,318 63, & ,083 62,246 TABLE III DISPATCH RESULTS FOR DIFFERENT COMPENSATOR LOCATIONS: CONTINGENCY CASE (LINE OUTAGE IN GTA) Bus Output (Mvar) Trans. Gen. No SVC & conssts of 2,833 buses and 4,205 branches, and ts man features are depcted n Fg. 1. The system peak load of 27,000 MW was chosen as the base case, and thrteen N-1 and N-2 crtcal contngences were consdered, as per IESO recommendatons. The proper placement for the SVCs/STATCOMs was determned through a senstvty analyss of the maxmum transfer capablty of the system wth respect to reactve power compensaton. Thus, maxmum loadablty margns, whch were used n leu of system transfer capabltes, were calculated usng a maxmum loadablty OPF model [2], [4]. The Lagrange multplers assocated wth the soluton of ths OPF model provde the requred senstvtes; thus, the optmal buses for shunt compensaton placement are those wth the largest Lagrange multpler values. Ths procedure yelded ffty buses out of 2,435 elgble buses under normal operatng condtons, all located n the Greater Toronto Area (GTA), whch accounts for 40% of Ontaro s total demand, wth heavy power transfers from Southern and Western Ontaro durng peakload condtons. Each optmal locaton was then examned aganst the IESO recommended crtcal contngences; f for a gven ratng the compensator was able to mantan the system secure for the consdered contngences, the correspondng bus was ranked hgher. Ths analyss dentfed Buses 760 and 842 n the GTA as the best locatons for SVC/STATCOM placements for both normal and contngency condtons. The output level of the SVCs/STATCOMs was then determned from the soluton of the proposed dspatch OPF model as dscussed next. TABLE IV DISPATCH RESULTS WITH AND WITHOUT BALANCING SERVICES (SVCS AT BOTH BUSES 760 &842) Model Wthout balancng servces Wth balancng servces Method Trans. Q Balance Servce + Resched. Total ,321 2,880 62, ,093 3,350 63,442 TABLE V DISPATCH COST COMPARISONS Q-dspatch Loss Slack-bus Total Proposed 59,449 1,800 61,249 Exstent 67,844 2,760 70,604 The MW-losses from all reactve power provders were assumed to be rembursed at an HOEP of $120/MWh, whch s a typcal hgh energy prce durng peak-load condtons. The upward and downward balance servces were assumed to be prced at $110/MWh and $90/MWh, respectvely, based on the IESO payment procedures for these knds of servces. The same loss equaton was used for all generators, rrespectve of ther type. The reactve dspatch problem was then solved for the followng cases: no SVCs; an SVC placed at each one of the two dentfed crtcal buses; and two SVCs placed at the correspondng crtcal buses. Smlar scenaros were consdered and studed for STATCOMs, but snce the results obtaned were practcally the same as those for the SVCs, these are not presented here. The dspatch results for the varous cases consdered are shown n Tables I and II. Observe that wth no SVCs n the system, transmsson losses are the hghest, whle wth two SVCs these losses are the lowest, as expected. In all cases, SVCs supply reactve power to the system, and there s a need for real power reschedulng, as antcpated. The requred MW-rescheduled amount s suppled by the slack bus and s pad by the IESO at the HOEP. The total payment to the SVC ncludes payments for losses at the HOEP, and for fxed costs at a prce of 0.50 $/Mvarh. Table III shows the dspatch results for a contngency case, correspondng to a crtcal sngle lne outage n the GTA where both SVCs are located. As expected, the SVC output ncreases due to the need for addtonal reactve power supply to mantan the system secure. Observe as well that transmsson losses ncrease wth respect to normal operatng condtons. Table IV shows the OPF results for the models wth and wthout balancng servces, and wth both SVCs placed n the system. Observe that the IESO payments and transmsson losses are hgher for the model wth balancng servces.

7 IEEE-PES General Meetng, Mnneapols, July Lastly, Table V compares the total IESO payments for the proposed method wth respect to the exstent reactve power dspatch approach n Ontaro. For the latter, the IESO uses ordnary power flows to determne the generator termnal voltages that mantan the system secure; the generators are then pad for the MW-losses (measured quanttes) requred to mantan the requested termnal voltages. For these studes, no SVCs were placed n the system and the requred real power reschedulng was allocated to the slack bus. Observe that the total cost of Q- loss payment and the total cost of reactve power are lower for the proposed method compared to the exstent approach. Thus, the proposed method would brng savngs to the IESO of $9,355/h under peak loadng condtons, whch take place about 5% of the tme durng the year, resultng n over $4 mllon yearly savngs. It should be noted that the proposed reactve dspatch models were computatonally effcent. Thus, t requred on average 43 teratons correspondng to about 7.2s of CPU tme n an IBM server wth 4 Intel Xeon 2.8GHz processors and 32 GB RAM runnng 32-bt MS Wndows. VI. CONCLUSIONS A dspatch method for reactve power whch mnmzes the actve losses n reactve power supplers whle consderng system securty was proposed and descrbed n detal. Generators and statc var compensators such as SVCs and STATCOMs were consdered and represented n the proposed dspatch model, assumng that ther servces are pad n terms of ther actve power losses at the market energy prce, as per Ontaro s reactve power payment procedures. The proposed model was tested and compared usng the Ontaro s dspatch grd model, demonstratng that, even though there s no reactve power dspatch mechansm per say n the Ontaro s grd and electrcty market, the proposed method may reduce the cost of reactve power dspatch whle mantanng system securty. It s also shown that the method s computatonally effcent and approprate for real-tme dspatch applcatons. REFERENCES [1] K. Bhattacharya and J. Zhong, Reactve power as an ancllary servce, IEEE Trans. Power Syst., vol. 16, pp , May [2] I. El-Samahy, K. Bhattacharya, C. A. Cañzares, M. Anos, and J. Pan, A procurement market model for reactve power servces consderng system securty, IEEE Trans. Power Syst., vol. 23, pp , Feb [3] C. A. Cañzares, K. Bhattacharya, I. El-Samahy, H. Haghghat, J. Pan, and C. Tan, Re-defnng the reactve power dspatch problem n the context of compettve electrcty markets, to appear n IET Generaton, Transmsson and Dstrbuton, specal ssue on Markets and Economcs n Power Systems, [4] I. El-Samahy, Secure provson of reactve power ancllary servces n compettve electrcty markets, PhD thess, Dept. of Electrcal and Computer Engneerng, Unversty of Waterloo, ON, Canada, [5] FERC Order No.888, Promoton of wholesale competton through open access non-dscrmnatory transmsson servces by publc utltes and recovery of stranded costs by publc utltes and transmttng utltes, Issued Aprl [6] FERC Staff Report, Prncples for effcent and relable reactve power supply and consumpton, Feb [7] FERC reactve power IESO comments 2005, [Onlne]. Avalable:http// AD pdf. [8] J. Zhong, E. Noble, A. Bose and K. Bhattacharya, Localzed reactve power markets usng the concept of voltage control areas, IEEE Trans. Power Syst., vol. 19, pp , Aug [9] P. Frías, T. Gómez, and D. Soler, A reactve power capacty market usng annual auctons, IEEE Trans. Power Syst., vol. 23, pp , Aug [10] Y. H. Song and A.T. Johns, Flexble AC Transmsson Systems (FACTS). IEE Power and Energy seres: [11] A. Sode-Yome, N. Mthulananthan, and K. Y. Lee, A comprehensve comparson of FACTS devces for enhancng statc voltage stablty, IEEE PES General Meetng, [12] IEEE Gude for Statc Var Compensator Feld Tests, IEEE Standard 1303, [13] IEEE Gude for the Functonal Specfcaton of Transmsson Statc Var Compensators, IEEE Standard 1031, [14] M. Noroozan and C. W. Taylor, Benefts of SVC and STATCOM for electrc utlty applcaton, IEEE PES Transmsson and Dstrbuton Conference and Exhbton, Sept [15] N. Hngoran and L. Gyugy, Understandng FACTS: Concepts and Technology of Flexble AC Transmsson Systems. IEEE press: [16] J. Barquín, D. Soler, O. Largo, G. Relaño and I. de la Fuente, On the cost of the reactve power generaton and voltage support servce, n Proc. Bulk Power Syst. Dynamcs Contr. IV: Restructurng, Santorn, Greece, Aug , [17] Modelng Language for Mathematcal Programmng (AMPL) [Onlne]. Avalable: [18] Independent Electrc System Operator, Ontaro transmsson system [Onlne]. Avalable Hossen Haghghat receved hs BSc from Shraz Unversty, Shraz, and MSc and PhD all n Electrcal Engneerng from Tarbat Modares Unversty, Tehran. In he was a post-doctoral fellow at the Department of Electrcal and Computer Engneerng, Unversty of Waterloo, Canada. Hs research nterest s power system optmzaton and deregulaton. Claudo Cañzares (S'86, M'91, SM 00, F 07) receved the Electrcal Engneer degree from the Escuela Poltécnnca Naconal (EPN), Quto- Ecuador, n 1984 where he held dfferent teachng and admnstratve postons from 1983 to Hs MSc (1988) and PhD (1991) degrees n Electrcal Engneerng are from the Unversty of Wsconsn-Madson. He has been wth the E&CE Department, Unversty of Waterloo snce 1993, where he has held varous academc and admnstratve postons and s currently a Full Professor, the Hydro One Endowed Char and an Assocate Drector of the Waterloo Insttute for Sustanable Energy (WISE). Hs man expertse s n the areas of stablty, modelng, smulaton, control, optmzaton and computatonal ssues n power and energy systems wthn the context of compettve energy markets and smart grds. He has been the recpent of varous IEEE-PES Workng Group awards, and also holds and has held several leadershp postons n varous IEEE-PES techncal commttees, workng groups and task forces. Kankar Bhattacharya (M 95, SM 01) receved the Ph.D. degree n electrcal engneerng from Indan Insttute of Technology, New Delh, n He was wth the Faculty of Indra Gandh Insttute of Development Research, Bombay, Inda, durng , and the Department of Electrc Power Engneerng, Chalmers Unversty of Technology, Gothenburg, Sweden, durng Snce January 2003, he has been wth the Department of Electrcal and Computer Engneerng, Unversty of Waterloo, Canada, and currently he s a Professor. Hs research nterests are n power system dynamcs, stablty and control, economc operatons plannng, electrcty prcng and electrc utlty deregulaton. Dr. Bhattacharya receved the 2001 Gunnar Engström Foundaton Prze from ABB Sweden for hs work on power system economcs and deregulaton ssues.