Conference on te promotion of Distribted Renewable Enery Sorces in te Mediterranean reion 11-12 December 29 Nicosia Cyprs (Paper ref No: 134) Tertiary reserve in Greece s electricity market: Te need for peakers 1 Panaiotis Andrianesis Geore Liberopolos and Pandelis Biskas Abstract-- Many recent power system enineerin stdies ave brot into attention te need to incorporate peaker nits (or peakers ) in power systems. Peakers are electricity eneration nits tat can cover te peak load and provide tertiary spinnin and non-spinnin reserve. In Greece s electricity market te increasin interest in Renewable Enery Sorces (RES) mostly wind plants makes te need for peakers even more pressin becase more tertiary reserve is ired to preserve system secrity in cases of sdden canes in RES injection. In tis paper we address in detail te establisment of a tertiary reserve market by introdcin te commodity of tertiary reserve (bot spinnin and non-spinnin) in te Day- Aead Scedlin (DAS) problem. To provide an estimate of te revenes of peakers from participatin in sc a market we simlate te Greek electricity market wit a tertiary reserve market add-on sin istorical data of eneration nits system load and system reserve irements. Or preliary reslts sow tat te estimated profits are sinificant indicatin tat te prospect of a tertiary reserve market can provide te rit incentives for investin in peakers. Finally we discss potential extensions of or analysis and directions for frter researc. Index Terms-- Electricity market tertiary reserve. I. NOMENCLATURE A. Sets - Sbsets U Generation nits U t Termal eneration nits ( Ut U) U Hydro eneration nits ( U U Ut U U ) U peak Peaker eneration nits ( U peak Ut U ) B. Parameters Generation nit U Hor (time period) = 1 H SL System load in or NImp Net imports in or RES RES injections in or MHyd Mandatory ydros injections in nit or Pmp Pmpin in or PR Primary reserve irement SRU Secondary reserve p irement in or SRD Secondary reserve down irement in or P. Andrianesis is wit te Department of Mecanical Enineerin University of Tessaly Volos 38334 Greece (pone: +3 (2421) 746; fax: +3 (2421) 7459; e-mail: andrianesis@t.r). G. Liberopolos is wit te Department of Mecanical Enineerin University of Tessaly Volos 38334 Greece (e-mail: lib@mie.t.r). P. Biskas is wit te Hellenic Transmission System Operator (HTSO) S.A. 22 Asklipio Str. Krioneri (Atens) 14568 Greece (e-mail: pbiskas@desmie.r). TR Tertiary reserve irement in or Q Tecnical imm of nit Q Tecnical imm of nit AGC Tecnical imm of nit nder AGC AGC Tecnical imm of nit nder AGC PR Primary reserve of nit SRR Secondary reserve rane of nit TR Tertiary reserve of nit SDC St-down cost of nit C Cost of enery eneration of nit MU Minimm ptime of nit MD Minimm downtime of nit avail ST Availability stats of nit or pr sr tr P Price of enery (eneration) of nit or P Price of primary reserve of nit or P Price of secondary reserve of nit or P Price of tertiary reserve of nit or ST Initial stats of nit (at or ) X Nmber of ors nit as been ON at or W Nmber of ors nit as been OFF at or C. Decision variables G Total eneration (otpt) of nit or PR Primary reserve of nit or SRU Secondary reserve p of nit or SRD Secondary reserve down of nit or TR Tertiary reserve of nit or ST Stats (condition) of nit or Binary variable: 1 = (LINE) = OFF(LINE) AGC AGC condition of nit in or Dependent binary variable: 1 = In AGC mode = Not in AGC mode V Stdown sinal for nit in or Dependent binary variable: 1 = Stdown = No stdown X Nmber of ors nit as been ON at or since last startp Inteer variable W Nmber of ors nit as been OFF at or since last stdown Inteer variable II. INTRODUCTION HE liberalization of national electricity markets wic T was establised by te Eropean Directive 96/92/EC led to fndamental canes in te oranization and
2 fnctionin of te electricity sector witin te member states of te EU. Sinificant empasis was placed on te wolesale market rles overnin te scedlin of te eneration nits and te enery tey are asked to prodce. In te case of Greece tese rles are defined by te Grid Control and Power Excane Code for Electricity [1]. Te market is spervised by te Relatory Atority for Enery (RAE) and is operated by te Hellenic Transmission System Operator (HTSO). Te operation of te wolesale market is based on te Day-Aead Scedlin (DAS) market problem wic clears te enery s (prodction) aainst te load declarations (consmption) in a mandatory pool. Te orsepower of te Greek electricity market sector is linite. Apart from linite nits te market also incldes combined-cycle and open-cycle as-fired nits oil nits lare ydro plants and Renewable Enery Sorces (RES) sc as wind parks small ydros biomass potovoltaic and small coeneration. Te installed capacity prodction profile in Greece is sown in Fi. 1. Moreover it is commonly areed tat it is more efficient to dispatc flexible nits for some peak ors drin te day tan keepin expensive nits at teir tecnical imm for te wole day. Peakers are nits tat can cover te peak load and provide tertiary non-spinnin reserve at off-peak ors; ence tey are ideal for tis operation. Teir main disadvantae owever is teir i installation and operational costs wic raises te qestion of weter tey can be covered tro a liberalized electricity market. Conseqently market desiners face te callene of providin adeqate incentives to investors wo are interested in installin sc nits. In tis paper we address in detail te establisment of a tertiary reserve market by introdcin te commodity of tertiary reserve (bot spinnin and non-spinnin) in te dayaead market. More specifically we present te matematical formlation of te DAS market problem and we simlate te Greek electricity market wit a tertiary reserve market add-on sin istorical data from year 28 in order to estimate te revenes of peakers from participatin in sc a market. Or preliary reslts sow tat te estimated profits are sinificant indicatin tat te prospect of a tertiary reserve market can provide te rit incentives for investin in peakers. In or concldin remarks we discss possible extensions and directions for frter researc. Fi. 1. Installed capacity prodction profile in te Greek electricity sector Te total installed capacity is approximately 12 MW. Witot RES te capacity drops to approximately 11 MW; te termal nits installed capacity amonts to abot 8 MW wile ydro nits provide approximately 3 MW. In 28 te system orly load varied approximately from 34 MW to 12 MW. Greece is a member of te Union for Coordination of Transmission of Electricity (UCTE) and is interconnected tro overead AC lines wit Albania FYROM and Blaria in te Nortern borders and tro an HVDC cable wit Italy in te Nort- West. In order to ensre a reliable system operation HTSO sets irements for te so-called ancillary services. In tis paper we only consider fency-related ancillary services ereinafter referred to as reserves. Reserves are separate commodities tat can be traded in te day-aead market ts establisin reserve markets. Te primary reserve irement is set at 8 MW wile te secondary reserve irement varies between 15-3 MW for secondary reserve p and between 5-15 MW for secondary reserve down. Tertiary reserve irement is set for te moment at abot 5% of te system load. Crrently tere are separate primary and secondary reserve markets; tertiary reserve is not remnerated bt te relative irement enters te DAS proram as a constraint. Te increasin interest in RES mostly wind plants creates te need for more tertiary reserve to preserve system secrity in cases of sdden canes of RES injection. III. THE DAS MARKET PROBLEM Te DAS problem is solved daily simltaneosly for all 24 ors of te next day. Te objective is to imize te cost of matcin te enery to be absorbed wit te enery to be injected in te system wile meetin te reserve irements and te eneration nits tecnical constraints. Te DAS soltion defines ow eac nit sold operate in eac or so tat te social welfare of te electricity market is imized. It also deteres te clearin prices of te enery and reserves. In wat follows we present a model of te DAS problem were we incorporate an add-on tertiary reserve market. We inclde te most important featres of te real DAS model tat is rn every day by te HTSO wile keepin or formlation as simple as possible so tat it can be explained and nderstood witin te scope and limitations of tis paper. In or model we consider all available eneration nits termal and ydro plants net imports injections from te RES mandatory injections from te ydro plants and te pmpin stations. Reserves inclde primary reserve secondary p and down and spinnin and non-spinnin tertiary reserve. Te system load and te reserve irements are exoenosly detered by te HTSO sin forecastin metods and estimates tat are beyond te scope of tis paper and are considered as parameters of te optimization problem. Te transmission constraints wic in te case of Greece s electricity market amont to a Nort-to-Sot transmission corridor constraint are not inclded in or model for simplicity. Te prodcers sbmit enery s for eac or of te followin day as a stepwise fnction of price-qantity pairs wit sccessive prices bein strictly non-decreasin. For
3 simplicity reasons and witot loss of enerality we assme a sinle price bid for enery. Te prodcers also sbmit reserve bids as price-qantity pairs and stdown costs. Te tecnical caracteristics of te eneration nits tat constitte te constraints of te DAS problem inclde te tecnical imm and imm otpt te AGC imm and imm te imm reserve availability and te imm ptimes and downtimes. Ramp p/down limits are not considered in or analysis alto or model can be easily extended to inclde tem. Te DAS problem can be formlated as a Mixed Inteer Proram (MIP) problem. In te followin and nless oterwise mentioned refers to te eneral set U. pr P G P PR sr f P ( SRU SRD ) (1) G PR SRU SRD TR ST AGC V X W tr P TR V SDC sbject to: G SL NImp RES MHyd Pmp (2) PR PR SRU SRU SRD SRD TR TR G PR SRU TR Q ST AGC AGC AGC ( ) G PR SRU TR MHyd Q ST AGC AGC AGC ( ) (3) (4) (5) (6) U (7) t U (8) G Q ST U (9) G MHyd Q ST U (1) G SRD U t Q ( ST AGC ) AGC AGC (11) G SRD MHyd U Q ( ST AGC ) AGC AGC (12) PR ST PR (13) SRU SRD AGC SRR (14) TR ST TR (15) AGC ST (16) avail ST ST (17) ( X MU )( ST ST avail ) ST (18) 1 1 ( W MD )( ST ST ) (19) 1 1 V ST (1 ST ) (2) 1 X ( X 1) ST (21) 1 W ( W 1)(1 ST ) (22) 1 ST ST X X W W wit (23) (24) (25) G PR SRU SRD TR. t Te objective as defined in (1) is to imize te as-bid cost for enery and reserves as well as te stdown cost. Eqation (2) expresses te enery balance constraint. Te enery injected in te system mst eqal te system load s te net imports s te injections from te RES s te mandatory injections from ydro plants pls te pmpin load. Constraints (3)-(6) ensre adeqacy of te different types of reserves (primary secondary p secondary down and tertiary bot spinnin and non-spinnin). For simplicity te formlation of tese constraints does not take into consideration te ierarcical natre of te different types of reserve and ence does not allow te sbstittability of lower qality reserves by ier qality ones. Ineqalities (7) and (8) refer to te nits tecnical imm constraint. More specifically ineqality (7) refers to te termal nits and states tat te eneration otpt of eac nit pls te contribtion for primary secondary p and tertiary reserve cannot exceed te nit s tecnical imm. In case te nit s secondary p reserve i.e. te nit is in AGC mode te tecnical imm is lowered to te AGC imm. Ineqality (8) refers to te ydro plants and differs from (7) in tat te otpt of te ydro plants mst inclde te mandatory ydro injections. Ineqalities (9) and (1) declare tat te otpt of eac nit if it is online mst be above its tecnical imm. Ineqalities (11) and (12) take into consideration te provision of secondary down reserve wic implies tat te nit mst be in AGC mode and ensre tat te nit will enerate above its AGC imm. Constraints (13)-(15) state te reserve availability of te eneration nits. In particlar tey state tat te amont of eac reserve type tat te nit is able to provide mst not exceed its wic reflects te imm amont of eac reserve type tat te nit can provide. Note tat te primary and tertiary reserve can be provided if te nit is online wereas te provision of secondary reserve (bot p and down) ires te nit to be in AGC mode. Ineqality (16) declares tat a nit in order to be in AGC mode mst necessarily be online. Ineqality (17) declares te availability of eac nit. Constraints (18) and (19) represent te imm ptime and downtime constraints. In case of a nit startp (stdown) te eneration nit mst stay online (offline) for a certain nmber of time periods (ors). Note tat constraint (18) is mltiplied wit te availability of eac nit so tat it is dropped in case an otae occrs. Eqalities (2)-(25) define dependent variables and declare initial vales. Note tat constraints (18)-(22) are not linear. To sort ot for te nonlinearities tey can be replaced wit eqivalent ineqalities introdcin axiliary variables werever necessary. More specifically te definition of te stdown sinal (2) can be replaced by te followin ineqalities: V ST ST (26) 1 ST ST 1.1(1 V ).1 (27) 1 Te replacement of te definition of te variables (time conters) X and W needs te introdction of two
4 axiliary inteer (nonneative) variables K and L. Constraints (18) (19) and (21)-(22) can be replaced by te followin ineqalities: ( K X ST MU ( ST ST 1 avail 1 )) ST (28) K X (29) K M(1 ST ) X (3) 1 K M ST (31) W 1 W 1 ST MD( ST 1 ST ) L (32) 1 L W (33) L M(1 ST ) W (34) L M ST (35) were M is a bi nmber (e.. a vale of 1 is sfficient for tis stdy case). Te formlation tat reslts after te above replacements is a Mixed Inteer Linear Proram (MILP) problem tat can be modeled and solved wit any available MILP solver. Once te MILP problem is solved a Linear Proram (LP) problem is created by replacin constraints (16)-(25) wit te followin eqalities: ST ST (36) AGC AGC (37) V V (38) Parameters ST AGC V represent te optimal vales of binary variables ST AGC V tat are obtained after solvin te MILP problem. Te LP formlation allows for te calclation of clearin prices sin marinal pricin teory [2]. Te enery clearin price also called System Marinal Price (SMP) is detered by te sadow price of constraint (2). As for te reserve price varios pricin scemes can be sed. Te reserve can be priced at its marinal vale or at te iest bid accepted. A pay-as-bid sceme can also be sed for pricin reserve. Te sadow price approac is based on te dal variable of te reserve irement constraint. Under tis approac tere exists a stron interaction between enery and reserve as tese commodities ave interdependencies tat inflence clearin prices [3] especially te SMP wic deteres te larest volme of transactions tro te day-aead market. Te iest bid accepted sceme compensates all nits wit a niform price tat is set at te iest reserve bid tat is inclded in te DAS wile te pay-as-bid sceme sets different reserve prices wit eac prodcer bein paid is bid if cosen to provide reserve. A matter of controversy is weter reserve s sold be priced at all and if so weter tese s sold be inclded in te objective fnction or not. Te impacts of reserve s on clearin prices and te varios pricin scemes ave been exaed in detail in [4]. IV. NUMERICAL RESULTS In tis section we present some preliary nmerical reslts derived by solvin te DAS problem on an instance representin te Greek Electricity Market. Te inpt data to te DAS problem are listed in Tables I and II. Qantities are iven in MW and prices for enery and reserve bids in /MW. Te bids are considered to be te same for all 24 ors. Minimm ptimes/downtimes are iven in ors and stdown costs in. Te Eqivalent Demand Forced Otae Rate (EFOR D ) vales are iven as a percentae and provide a measre of te probability tat a eneration nit will not be available de to a forced otae. Units U1-U22 represent linite nits wic serve as base nits. Units U23 U26 are oil nits nits U27 U33 are combined cycle and as nits and nits U34 U4 are peaker nits. Peakers U37-U4 in particlar are not real bt are introdced in te problem in order to explore te need for te constrction of new peaker nits in Greece. Unit U41 is an areate representation of ydro plants. For te needs of or analysis we assmed a random scenario of otaes for te termal nits based on teir EFOR D. We sed te EXCEL random enerator to enerate random nmbers between and 1 for 366 days (year 28); if at a particlar day te correspondin random nmber was less tan te EFOR D ten te nit was considered to be navailable drin tat day. Unit Q Q AGC TABLE I UNITS DATA (DAS INPUT) AGC MU C MD EFOR D U1 274 165 32 24 8 6.4 U2 274 165 32 24 8 8.267 U3 283 165 32 24 8 6.968 U4 283 165 32 24 8 9.418 U5 342 2 3 24 8 3.516 U6 273 152 33 24 8 8.688 U7 273 152 33 24 8 8.617 U8 292 15 22 24 8 9.294 U9 275 153 29 24 8 11.959 U1 275 153 29 24 8 1.15 U11 3 16 29 24 8 13.463 U12 3 16 29 24 8 21.498 U13 64 5 35 24 8 26.288 U14 116 8 3 24 8 25.448 U15 116 8 3 24 8 24.941 U16 274 16 26 24 8 23.911 U17 3 2 35 24 8 6.42 U18 8 5 35 24 8 6.42 U19 113 66 35 24 8 24.348 U2 113 66 35 24 8 15.299 U21 27 195 32 24 8 22.933 U22 26 195 32 24 8 1.195 U23 144 6 57 12 8 3.194 U24 144 6 57 12 8 4.64 U25 143 5 58 8 8 15.929 U26 287 121 56 8 8 1.48 U27 173 71 7 16 8 9.687 U28 55 155 53 175 59 5 5 4.452 U29 377 24 357 26 5 3 3 5.91 U3 476 144 456 164 53 5 5 6.86 U31 151 65 74 16 8 5.454 U32 188 15 83 16 8 5.462 U33 389 24 65 3 3 5.975 U34 49 12 1.789 U35 49 12 1.789 U36 49 12 1.789 U37 15 12 1 U38 15 12 1 U39 15 12 1 U4 15 12 1 U41 3 295 5
5 Unit P TABLE II UNITS DATA (DAS INPUT) CONT. PR SRR pr TR P sr P tr P SDC U1 32.1 24 3 3.1 5.1 45 U2 32.2 24 3 3.2 5.2 45 U3 32.3 24 3 3.3 5.3 45 U4 32.4 24 3 3.4 5.4 45 U5 3.1 24 3 3.5 5.5 45 U6 33.1 24 3 3.6 5.6 45 U7 33.2 24 3 3.7 5.7 45 U8 22 24 45 3.8 5.8 45 U9 29.1 24 45 3.9 5.9 8 U1 29.2 24 45 3.1 5.1 8 U11 29.3 24 45 3.11 5.11 8 U12 29.4 24 45 3.12 5.12 8 U13 35.1 24 7 3.13 5.13 12 U14 3.2 24 12 3.14 5.14 17 U15 3.3 24 12 3.15 5.15 17 U16 26 24 3 3.16 5.16 48 U17 35.2 12 5.17 6 U18 35.3 12 5.18 2 U19 35.4 24 75 3.17 5.19 22 U2 35.5 24 75 3.18 5.2 22 U21 32.5 24 75 3.19 5.21 65 U22 32.6 24 75 3.2 5.22 65 U23 57.1 12 34 1 6.1 15 U24 57.2 12 34 11 6.2 15 U25 58 12 12 12 6.3 21 U26 56 24 19 13 6.4 4 U27 7 12 12 14 7.1 4 U28 59 36 2 18 15 59 7.2 24 U29 5 24 1 137 16 7 7.3 14 U3 53 3 28 18 17 53 7.4 16 U31 74 12 45 18 8.1 11 U32 83 12 45 19 8.2 22 U33 65 24 139 65 9 14 U34 149.9 49 9.51 1 5 U35 149.91 49 9.52 1 5 U36 149.92 49 9.53 1 5 U37 149.93 15 9.6 4 5 U38 149.94 15 9.7 4 5 U39 149.95 15 9.8 4 5 U4 149.96 15 9.9 4 5 U41 15 29 295 5 1 All nits are initially considered to be online so tat tey all ave a common startin point. Initial vales for te time conters tat are not sown are iven so tat tey will not affect te dispatcin. Te data for system load scedled net imports RES injections and mandatory injections from ydro plants were taken from te HTSO istorical database for te year 28 and are available in [5]. Te pmpin profile tat we sed is sown in Table III. Te primary reserve irements were fixed to 8 MW for all ors and days. Te secondary reserve irements were te same for all days and are presented in Table IV. Te tertiary reserve irements were calclated for every or as a percentae of te system load. Te actal amont was set at 5% of te system load. We also exaed tree oter scenarios wit increased irements for tertiary reserve p to 7.5% 1% and 12.5%. TABLE III PUMPING PROFILE 1 2 3 4 5 6 7 8 9 1 11 12 Pmp 22 47 47 47 47 47 47 11 13 14 15 16 17 18 19 2 21 22 23 24 Pmp TABLE IV SECONDARY RESERVE REQUIREMENTS 1 2 3 4 5 6 7 8 9 1 11 12 SRU 15 2 2 2 2 2 2 3 3 3 3 3 SRD 15 5 5 5 5 5 5 5 5 5 5 5 13 14 15 16 17 18 19 2 21 22 23 24 SRU 3 3 25 25 25 25 3 3 3 3 25 25 SRD 5 5 1 1 1 1 5 5 5 5 1 1 We sed te matematical proram lanae AMPL [6] to model te DAS problem and te ILOG CPLEX 9.1 optimization commercial solver to solve it. We assmed a pricin sceme for tertiary reserve tat is similar to te actal sceme for primary and secondary reserve i.e. a imm bid accepted sceme and we set te price cap at 1. Te cosen sceme may not be te best bt is te closest to crrent practice. Alternative scemes sc as marinal pricin cold be exaed to detere possible differences. We also assmed tat te ydro plants are ivin s tat are ier tan tose of te termal stations wic in or example are at te price cap for enery. We adopt te same practice also for te tertiary reserve so tat te ydro plants are te last to be dispatced for enery and tertiary reserve. Tis andlin of te s from te ydro plants in te day-aead market is jstified as follows. Hydro plants may be able to prodce at teir imm otpt for a certain period of time bt in te lon-term tey are bond by an enery constraint. As a conseqence in dry years as was te year sed in or example te dispatcin of ydro plants mst be very carefl and well desined. Terefore te ability of tose plants to provide tertiary reserve is limited especially wen te probability tat tis reserve will be converted to enery is expected to increase (as is te case wit te perspective of te introdction of a bi amont of RES in te system). Te DAS market problem was solved iteratively for 366 days. Te stats of eac nit at or 1 of eac day was set eqal to te stats of te same nit at or 24 of te previos day and te same was done wit te time conters X and W. If tere was an otae of a nit in a particlar day te imm ptime constraint for tat nit for tat day was dropped. Also becase of te random way tat te otaes were allocated troot te year tere were a limited nmber of days wit very i load and increased irements for tertiary reserve were te DAS problem ad no feasible soltion. To deal wit tis infeasibility we introdced a fictitios eneration nit to cover te deficit in system load and/or tertiary reserve. Neverteless te impact of tis nit in te yearly reslts for te peakers was neliible. In Tables V and VI we present te reslts of te
6 simlation. Specifically we present te overall yearly prodction of te peakers and teir calclated profits. Qantities are iven in MW and prices in millions of. As te peakers U34-U4 are ivin teir bids in a specific order from te ceapest to te most expensive one for bot enery and tertiary reserve te reslts reveal te nmber of peakers tat can fit in te market for eac scenario. For ease of exposition we refer to U34-U36 as Peaker 1 (tis nit is modeled as tree separate nits in te DAS market problem bt can be considered as one nit of 147 MW) and nits U37-U4 as Peakers 2-5 respectively. TABLE V YEARLY GENERATION OF PEAKERS Tertiary reserve irements Units 5% 7.5% 1% 12.5% Peaker 1 Peaker 2 12 Peaker 3 28 439 2 Peaker 4 91 222 9 967 28 Peaker 5 99 477 68 55 4 74 32 TABLE VI YEARLY PROFITS OF PEAKERS Tertiary reserve irements Units 5% 7.5% 1% 12.5% Peaker 1 4.966 6.87 8.112 8.933 Peaker 2 3.415 5.43 7.196 8.368 Peaker 3 2.591 3.841 5.833 7.437 Peaker 4 3.82 2.492 4.33 5.92 Peaker 5 3.45 2.98 2.961 4.514 For te scenario wit te lowest reserve irements (5%) we observe tat Peakers 1 and 2 are sed almost only for tertiary reserve. Peaker 3 is sed for bot enery and reserve wereas Peakers 4 and 5 are sed mainly for enery. As te tertiary reserve irements increase te nmber of peakers tat are sed almost exclsively for te provision of tertiary reserve increases p to 5 for te 12.5% scenario. Te profits of te peakers from bot te enery and te tertiary reserve market are considerably i. Let s focs on te profits from te tertiary reserve market. Peaker 1 is always dispatced to provide tertiary reserve and its profits vary between 5-9 million. Peaker 2 wic is also dispatced to provide tertiary reserve as profits tat rane between 3.4-8.3 million. Tese two can fit in te market even wit te 5% scenario. It is wort noticin tat in te scenario of 12.5% te profits for all te 5 peakers rane between 4.5-9 million. Weter te incentives for te installation of new peakers sold be tro te market desin or not is beyond te scope of or analysis. Or aim in tis paper was to provide some sefl insit of a potential market desin tat wold inclde te tertiary reserve as a separate commodity and explore ow sccessfl sc a desin wold be to attract investments in new peakers. Or preliary reslts do sow tat te profits from participatin in a tertiary reserve market can be rater i. Incentives can also be iven tro some kind of contract for a aranteed income for a certain period of time and/or tro te capacity availability tickets as is crrently te case in Greece. To pt te investment in peakers in perspective wit respect to te incentives for sc an investment we ive te followin crde fires. It is estimated tat te installation cost for a peaker of abot 15 MW is approximately 6 million. At te moment te capacity availability tickets for eac nit are 35 per installed MW per year and wit a recent law tis ticket will be dobled for peakers for a period of 4 years. Tis means tat a nit of 15 MW will receive 1.5 million for eac of te followin 4 years in case of fll availability and zero EFOR D. Hence 42 million ot of te 6 million oriinally invested can be recovered in 4 years wile te rest can be recovered tro participation in te market. In te near ftre te introdction of new RES in te system will create te need for more tertiary reserve and te probability tat tis reserve will be sed drin te day will increase. Terefore it is expected tat te need for peakers will be particlarly pressin. V. CONCLUDING REMARKS AND POSSIBLE EXTENSIONS In tis paper we presented a formlation of te Greek electricity market wit a tertiary reserve market add-on in order to explore te incentives for investin in new peakers. Te need for tese nits is stronly dependent to te increasin interest in RES to cover te ier irements for tertiary reserve; owever teir i installation costs discorae potential investors. We simlated te proposed market desin for one year sin istorical data of year 28 and exaed for possible scenarios reardin te tertiary reserve irements. Te reslts sowed tat tere are considerably i profits to be made wic is an indication tat a tertiary reserve market cold actally provide te adeqate incentives. However te reslts of tis analysis are only indicative. In order to be able to reac more precise conclsions most notably for te actal nmber of peakers tat cold be viable in te market many more scenarios need to be tested. For instance many different random allocations of otaes need to be exaed to provide a tit confidence interval on te reslts. Te same approac mst be followed for treatin data of te RES injections and mandatory ydros. In eneral as was already mentioned te treatment of ydro plants is a matter of reat importance tat can completely alter te reslts. Te DAS market problem can be extended to inclde te transmission constraints and zonal pricin sceme as was presented in [7] for bot enery and ancillary services. Te ramp p/down limits as well as a startp/stdown profile can also be introdced in te problem. Alternative pricin scemes for reserves sc as marinal pricin daily instead of orly bids for reserves may also be considered. Te ierarcical natre of different reserve types i.e. te sbstittability of lower qality ancillary services by ier qality ones can also be inclded in te DAS formlation. Lastly te biddin beavior of te market participants is also sinificant and can exert a sinificant impact on te final reslts. Terefore for a specific market desin te way tat te nits can respond to imize teir profits needs to be exaed. Mltiple blocks for enery s wit orly bids make tis isse rater complicated. An attempt to address tis problem is described in [8]; owever a lot of work still remains to be done.
7 VI. ACKNOWLEDGMENT Te ators wis to specially tank Professor M.C. Caramanis former Cairman of Greece s Relatory Atority for Enery for is idance and spport. VII. DISCLAIMER Please be advised tat te material contained in tis paper is for information edcation researc and academic prposes only. Any opinions proposals and positions expressed in tis paper are solely and exclsively of te ators and do not necessarily represent te views of te HTSO partially or nilaterally. VIII. REFERENCES [1] Relatory Atority for Enery Grid Control and Power Excane Code for Electricity Atens Greece 25. [2] F.C. Scweppe M.C. Caramanis R.D. Tabors R.E. Bon Spot pricin of electricity Klwer Academic Pblisers Boston MA 1988. [3] P. Andrianesis G. Liberopolos and G. Kozanidis Enery and Reserve Interaction in Greece s Electricity Market paper presented at te 6 t Mediterranean Conf. and Exibition on Power Generation Transmission Distribtion and Enery Conversion Tessaloniki Greece Nov. 28. [4] P. Andrianesis G. Liberopolos K. Sakellaris and A. Vlacos Impacts of reserves and fixed costs on Greece s day-aead scedlin problem paper presented at te Promiteas-2 International Black Sea Enery Policy Conference Enery Investments and Trade Opportnities Atens Greece Oct. 28. [5] Hellenic Transmission System Operator official website: www.desmie.r. [6] R. Forer D.M. Gay and B.W. Kernian AMPL: A modelin lanae for matematical proram Boyd & Fraser Danvers MA 1993. [7] P. Andrianesis G. Liberopolos and A. Papalexopolos Application of zonal pricin in Greece s electricity market paper presented at te EEM9 Conf. Leven Belim May 26 29 29. [8] P Andrianesis G. Liberopolos and G. Kozanidis Enery-reserve markets wit non-convexities: An empirical analysis paper presented at te IEEE/PES Power Tec 29 Conf. Bcarest Romania Jn. 28 Jl. 2 29. IX. BIOGRAPHIES Panaiotis E. Andrianesis radated from te Hellenic Military Academy (21) and received is B.Sc. (24) deree in economics from te National and Kapodistrian University of Atens Greece. Crrently e is prsin a P.D. deree in mecanical enineerin at te University of Tessaly Volos Greece. His researc interests inclde power system economics electricity markets operations researc and optimization. Geore Liberopolos received is B.S. (1985) and M.En. (1986) derees in mecanical enineerin from Cornell University and is P.D. (1993) deree in manfactrin enineerin from Boston University. Crrently e is Professor of Stocastic Metods in Prodction Manaement Head of te Prodction Manaement Laboratory and Cairman of te Department of Mecanical Enineerin at te University of Tessaly Volos Greece. Prior to joinin te University of Tessaly e was Lectrer at te Department of Manfactrin Enineerin at Boston University and Visitin Researcer in te Laboratoire d Informatiqe at te Université Paris IV France. He is Associate Editor of IIE Transactions and Co-editor of OR Spectrm. He as co-edited several collected volmes of books/jornals wit temes in te area of qantitative analysis of manfactrin systems. He as pblised nmeros scientific papers in IEEE INFORMS and oter jornals mostly in operations researc/manaement and atomatic control. His researc interests inclde applied probability operations researc and atomatic control models and metodoloies applied to prodction and operations control. Dr. Liberopolos is a member of INFORMS te Hellenic Operations Researc Society and te Tecnical Camber of Greece. Pandelis Biskas received is Dipl. En. Deree from te Department of Electrical Enineerin Aristotle University Tessaloniki in 1999 and is P.D in 23 from te same niversity. He also did is Post Doc researc from Marc 24 till Ast 25 in te same niversity. Crrently e is wit te Aristotle University of Tessaloniki in te Department of Electrical and Compter Enineerin as a Lectrer. His researc interests are in power system operation & control in electricity market operational and relatory isses and in transmission pricin.