Nodal Pricing: The Theory and Evidence of Indonesia Power System

Transcription

1 Internatona Journa of Energy Economcs and Pocy ISSN: avaabe at http: Internatona Journa of Energy Economcs and Pocy, 018, 8(6), Noda Prcng: The Theory and Evdence of Indonesa Power System Dzkr Frmansyah Hakam 1, 1 Centre for Energy, Petroeum and Mnera Law and Pocy. Unversty of Dundee, Scotand, Unted Kngdom, PT. PLN (Persero), Jakarta, Indonesa. Ema: d.f.hakam@dundee.ac.uk Receved: 11 June 018 Accepted: 3 September 018 DOI: ABSTRACT Ths research presents a stysed noda prcng mode of Indonesa power system wth engneerng-economc constrants. The modeng n ths research adopts the 8 nodes stysed mode for the Sumatra power system, by ncorporatng generaton, transmsson and power system stabty constrant. Noda prcng anayss s performed based on drect current optma power fow and margna cost cacuaton n each node. Ths research s the frst ever to estmate noda prces n the Indonesan eectrcty market. Noda prcng mode n ths paper provdes a proper nvestment sgnas for Indonesan stakehoder n performng generaton expanson pannng. Keywords: Noda Prcng, Sumatra Power System, Stysed Mode JEL Cassfcatons: C610, D410, D470, D INTRODUCTION Noda prcng provdes an economc sgna by smutaneousy cearng the market by ncorporatng generaton and demand functon as nputs. The noda prcng concept undernes the fundamenta theory n determnng an optma eectrcty prce to acheve wefare maxmsaton under specfc constrants. Noda market prcng refects the opportunty costs of servng a margna ncrease n demand whe compyng wth transmsson constrants. Noda prcng provdes severa advantages to the market agents: Increasng market wefare, provdng proper nvestment sgnas to the generator and gnorng bypass ssue,.e., the opportunty to eave the market f prce s not equa to margna cost (Green, 007). Schweppe et a. (1988) estabshed the concept of noda prcng. Noda prcng anayss s performed based on optma power fow (OPF) and margna cost cacuaton n each node. The bddng acton from the generatng frm s a snge shot game. The GenCos submt ther fxed suppy functon to the ISO by acknowedgng ther rvas bd functon. Ths one-shot gamng s for one bddng tme nterva; thus, the ISO cears the market after a the frms submt ther bds resutng n the market cearng prce. The market cearng prce s a numerc cacuaton from the ISO by consderng the generaton and transmsson network structure n the market. In other words, t s based on the eectrcty suppy-demand baance and wefare maxmsaton. The ISO pays the eectrcty prce to a generatng frms n the form of a bus-noda prce based on the amount of eectrcty power sod to the eectrcty poo, whe the consumer pays the eectrcty prce to the ISO based on the actve power oad receved. Macatangay (1998) mpemented the concept of noda prcng n Engand and Waes s market ncorporatng transmsson constrant and optmsed the eectrcty prce as a dua vaue n the programme. Green (007) ncuded transmsson osses n the OPF to appy the concept of noda prcng to the Engand and Waes market. In ths research, the transmsson oss s assumed to be reatvey sma and neggbe. Thus, the oad fow formuaton s approxmated by the drect current (DC) oad fow equaton. The Ths Journa s censed under a Creatve Commons Attrbuton 4.0 Internatona Lcense Internatona Journa of Energy Economcs and Pocy Vo 8 Issue

2 Hakam: Noda Prcng: The Theory and Evdence of Indonesa Power System equbrum structure n ths study was apped to cacuate Nash equbrum for a partcuar bd functon and eectrcty network. The power fow foows the rue of Krchhoff s aw whch states that eectrcty njecton n a partcuar node/bus s equa to the vector sum of the eectrcty nput and output. The Indonesa power system conssts of two prmary power systems,.e., Sumatra and Java-Ba power system. The current academc terature on Indonesa power system s mted to the technca aspect of Sumatra and Java-Ba power systems. Optma power fow studes the on Java-Ba power system was conducted by Wartana et a. (01) whe optma power fow studes on the Sumatra power system was conducted by Hakam et a. (011). Hakam and Asekomeh (018) provdes the overvew of Indonesa power system whe Hakam et a. (01) provdes the overvew of Sumatra power system and energy mx. Faza et a. (015) conducted the power system studes on the future nterconnecton of Sumatra-Java system through a HVDC transmsson system. The majorty of exstng research dscussed the technca aspect of the grd,.e. oad fow anayss, short crcut, and stabty anayss. For exampe, Hakam et a. (011) conducted a oad fow, short crcut and transent anayss of power system nterconnecton between the North System and Mdde-South System on Sumatra s 75 kv transmsson system. Athough current studes are mted n the engneerng aspect, these studes contrbuted to the scentfc modeng of ths research. For exampe, optma oad fow studes provded an nsght of the upper and ower boundares of transmsson constrant accordng to the oad fow and transent stabty anayss. The objectve n ths research s to deveop a stysed modeng of noda prcng n Indonesa power system. Ths research w smuate noda markets under perfect competton for two power system: (1) Four node power system () Sumatra power system. Ths research s the frst ever to estmate noda prces n the Indonesan eectrcty market. Ths noda prcng approach n Indonesa power system contrbutes to the current body of terature on eectrcty market modeng and market prcng study. The economc mode deveoped n ths research s a poneer one based on network subsystem boundares set up by the PLN TSOs. Ths research adopts the 8 nodes stysed mode for the Sumatra power system, by ncorporatng generaton, transmsson and power system stabty constrant. The stysed mode n ths research does not exacty represent the compexty of the Indonesa power system. However, t attempts to capture the mportant aspects of the techno-economc of Indonesa power system. Thus, stysed modeng, ncudng eectrcty demand and power generaton aocaton for each node, s a cruca one and shoud ncorporate a the demands and suppers n the system.. METHODOLOGY.1. DC Power and Law of Parae Crcut Assumng a power settng wth m transmsson nes and n nodes, X s a vector of reactance (m m). P F s a vector of DC power fow (m 1). M s the node-branch ncdence matrx for an ange phase vector matrx (n-1 m) excudng the reference node (sack bus),.e. a node wth phase ange s zero. P s the power njecton matrx (n-1 1). B s the susceptance matrx. Based on a DC power fow assumpton, the power njecton n node n s the dfference between power generaton producton q S and consumer demand q d. Thus, the power fow n the transmsson ne coud be denoted as a near functon between PTDF and q s -q d. PF = PTDF( qs qd) DC oad fow as a functon of PTDF 1 Any form of network aggregates changes the cabe mt and reactance. Thus, a transmsson ne ateraton from doube or snge ph to a snge crcut w nfuence the oop fow of the power system. Foowng the aw of parae crcut as n Hagspe et a. (014) and Hogan et a. (010), doubng the transmsson capacty w have the ne s reactance. X n j T 0 = X The aw of parae crcut j T 0 j n j 0 T j s the cabe mt at nta state whch connects node to j; whe X s the cabe reactance at nta state. 0 j.. Optmsaton Probem and Karush-Kuhn-Tucker Condton Fgure 1 shows the optmsaton probem structure. An optmsaton probem n mathematca programmng s a mathematca functon that has a purpose of maxmsng or mnmsng objectve functon subject to objectve constrants (equates and nequates). In the case of ocatona margna prcng (LMP), the objectve constrants usuay consst of generaton capacty and cabe therma mt nequates constrants. Fgure 1: Optmsaton probem structure Ths paper s structured nto 5 chapters as foows. Chapter one, of whch ths secton s a part, presents the ntroducton and overvew of ths research. Chapter two contans the methodoogy that expans the DC power fow, optmsaton formua, and demand-suppy functon. Chapter two aso appes the concept of noda prcng n a four-node power system. Chapter three s the stysed mode of Sumatra power system. Chapter four appes the noda prcng approach n Sumatra power system. Chapter fve s the concuson. 136 Internatona Journa of Energy Economcs and Pocy Vo 8 Issue 6 018

3 Hakam: Noda Prcng: The Theory and Evdence of Indonesa Power System Assumng that x R n s the varabe vector of optmsaton, f(x): R n R s the objectve functon, g(x): R n R e s the equaty constrant functon, h(x): R n R s the nequaty constrant functon. The souton x R n s feasbe f meetng the equaty, nequaty and bound constrants. The optma souton occurs ony f the feasbe regons are meetng the objectve functon. The optmsaton probem has a genera form as foows: Max or Mn x f(x) subject to: g(x)=0; h(x) 0. The optmsaton technque n an equbrum probem s deveoped as a vta too to mode and sove energy market probems under uncertanty condtons. The prncpa use of KTT condton s to fnd a unque prce and proft equbrum n the market. The unque equbrum resut mpes a convex feasbe regon for the generator, ISO and consumer s probem whch suggest that the oca optmum of the probem s aso the goba optmum. Thus, t s mportant to have a convex probem snce non-convex equatons w resut n non-exstence or mut-equbrum. In the eectrcty market, the convex probem combnes a market partcpants (supper, transporter, and consumer) through KTT combnaton; thus, the equaton s soved n the market cearng mechansm. Assumng F(x) s smooth and concave, and H(x) s convex, the objectve functon s as foows: Max x F(x) Subject to: (x) 0; x 0 Assume λ s Lagrange mutper for H constrant, genera FOC KTT condtons for a constraned optmsaton probem above s: F H 0 x λ 0 x x 0 λ H 0.3. Suppy-demand Functon and Consumer-producer Surpus The GenCos produce eectrcty energy based on the true generator cost whe the consumer provdes the demand functons whch refects the energy used. The nverse demand functon s a near functon wth negatve sope as foows: p (q d ) = a b q d ; =1,, I Inverse near demand 3 Where a and b are the oad demand coeffcent and q d s the actve oad demand at node. a > 0 and I s the number of the consumer. Tota generaton cost conssts of fxed (f ) and varabe costs C (q s ): 1 C ( qs ) = f cq s dq s ; = 1,, I Tota cost functon 4 1 C ( qs ) = cq s dq s ; = 1,, I Varabe cost functon 5 Generatng frm bd/margna cost functon MC (q s ) s a near functon snce the appcaton of constant margna cost s not fuy representatve of the true generaton cost n the eectrcty ndustry. Margna cost functon s the dervatve of the tota cost functon as foow: MC (q s )=c d q s ; =1,, I Margna cost functon 6 We defne consumer surpus as the net consumer beneft. Thus, the tota consumer surpus s the sum of each consumer surpus based on a partcuar prce. The consumer surpus for each regon coud dffer dependng on the eectrcty network structure, e.g. transmsson constrant, generator and consumer confguraton/ocaton. Assumng D (p ) s the eectrcty demand for consumer at prce (p ), the consumer surpus for nverse near demand functon s: 1 CS( p) = ( a p) D( p) ; = 1,, I Consumer surpus 7 Producer surpus s the generator net beneft receved from seng eectrcty demand to the power poo, defned as PS (p ): 1 PS( p) = ( p c) qs( p) ; = 1,, I Producer surpus 8 To cacuate the nverse near demand functon, t was assumed here that the reference pont for q d s the peak oad demand n node q 0. Ths study categorsed the LSE accordng to the dstncton of resdenta and non-resdenta (busness and ndustra) node. Prce data from the Indonesa Energy Mnstry provdes the prce reference p o. Based on the near nverse demand functon n Equaton 3, the demand functon s provded as foows: ( qd ) ; 1,, = a p = Lnear demand functon 9 qd I b b The demand ntercept a >0 and sope b >0, the eastcty of demand s cacuated as foows: ε ( ) ( ) 1 ( ) q p q p q p q q b q d d d = = d d d Eastcty of demand 10 1 p0 b = ; a = p0 bq Cacuatng near demand 0 11 ε q0 functon from p 0 and q 0 Usng b and a parameters, the nverse demand functon for each node s cacuated by assumng eastcty exogenousy. The demand cacuaton usng ths approach s mpemented n European noda prcng as n Leuthod et a. (01) and provdes fexbty to conduct an anayss usng varous eastcty assumptons as n Green (007). Internatona Journa of Energy Economcs and Pocy Vo 8 Issue

4 Hakam: Noda Prcng: The Theory and Evdence of Indonesa Power System.4. Wefare Maxmsaton by ISO: Perfect Competton The system operator manages the baancng mechansm of eectrcty suppy from generatng frms and the power demand from consumers. The ISO maxmse the tota wefare π (p) by choosng a snge prce for each bus (1,,I) n the mesh network by takng nto account network (generaton and transmsson) mts as nequaty and equaty constrants. Assumng that P (q d ) s the energy consumpton beneft; MC (q s ) s the tota cost of generators at node ; q s s the actve oad suppy from generator at bus and qs s the avaabe capacty of generator at node, the ISO wefare maxmsaton probem s formuated as foows: max qd Subject to/constrants: q s s P ( qd ) qd MC ( qs ) Wefare maxmsaton 1 q q = 0 d ( ) PTDF q q T s s d Eectrcty demand baance Transmsson constrant q Generaton constrant 15 qd>0 qs>0 Non negatvty 16 Defnton: Assume that λ, μ, µ - s Lagrange mutpers/dua varabes for eectrcty demand baance and transmsson constrant, respectvey. λ s the mesh network prce for a of the nodes n the power system. Proposton 1: In a congested transmsson ne, the noda prces are not unform due to noda prces dscrmnaton and creatng LMP. Proof: For the ISO wefare maxmsaton probem, a genera FOC KTT condtons for ISO mxed compementary probem (MCP) can be derved as foows: ( d ) λ ( ) = 0 P q PTDF µ µ q q = 0 s d ( ) 0 PTDF q q T 0 µ s d ( ) 0 PTDF q q T 0 µ s d FOC for ISO condton 17 The frst FOC KTT condton for MCP above yeds a genera ocatona prcng equaton: P ( qd ) =λ PTDF ( µ µ ) Genera LMP functon 18 Assume ( ) PTDF µ µ = γ a premum charge for transmsson congeston exsted n the network and formuated as the dfference between noda prce P (q d ) and system prce λ. A transmsson congeston charge depends on the avaabty of congeston and transmsson dua varabes μ, μ -. A frm receves payment γ when the frm njectng power to bus and pays a wthdrawa charge γ j when wthdrawng power at bus j, and the dfference of ths transmsson cost s defned as a wheeng charge γ -γ j. Consderng the sackness condton of μ and µ when the transmsson ne s uncongested μ =μ =0, the noda prce P (q d ) s equa to λ and the entre noda prce w be unform. When the transmsson ne s congested, μ 0; μ 0, the bus prces are not unform and depends on the transmsson ne. ( ) In the case where there s no transmsson constrant, the noda prce s equvaent for each bus. If there are any transmsson constrants and congeston n the mesh network, then the ISO w cacuate the market cearng prce for each node based on the wefare maxmsaton. The generaton frm coud bd a suppy functon other than ther true margna cost functon. Thus, the effect of the market prce coud dffer dependng on the generaton confguraton. Defnton: Load payment P( q ) d d payoff from the consumer at the noda prce whe generaton charge P( q ) q d s at the noda prce. q s the net consumpton s the energy producton pad to the generator Proposton : The dfference between oad payment and generaton charge s the tota congeston rent η. Proof: We cacuate the dfference between generaton charge and oad payment as foows: ( ) ( ) η= P q q P q q d d d s ( )[ ] η= P q q q d d s Observe that P ( qd ) qd and P( q ) Congeston rent 19 q d s condton. Thus, η s cacuated as foows: satsfes the KTT λ[ qd qs ] P( )[ ] qd qd qs [ qd qs ] PTDF ( µ µ ) = 0 η= λ [ qd qs ] [ qd qs ] PTDF ( µ µ ) η= λ [ qd qs ] [ qd qs ] PTDF ( µ µ ) Notce that for the KTT condton, λ [ qd qs ] = 0, and PTDF ( µ µ ) = γ s a premum charge for transmsson congeston; thus the congeston rent s premum charge tmes the mport/export of eectrcty n node [q d -q s ].( ) [ q ] d qs γ η= 138 Internatona Journa of Energy Economcs and Pocy Vo 8 Issue 6 018

5 Hakam: Noda Prcng: The Theory and Evdence of Indonesa Power System Defnton: If ( ) PTDF qs qd T where the transmsson constrant s bndng to the near functon, then the transmsson ne s congested; and vce versa. If PTDF ( q s qd ) T, then the transmsson ne s uncongested. Proposton 3: In the case where transmsson congestons occur n the power system, the ISO coects the surpus transmsson rent. Proof: Notce the KTT condton for transmsson constrant: ( )[ ] ( ) = 0 P q q q T PTDF µ µ d d s ( ) η= T PTDF µ µ Snce the sackness condton mpes T 0; μ 0; μ - 0 the congeston rent η 0. The congeston premum of the transmsson ne beng non-negatve ndcates that the oad payment s hgher than the generaton charge,.e. the dua varabes are bndng to the nequaty condton ( ). Defnton: For near demand and margna cost, consumer surpus 1 and producer proft s cacuated as CS( p) = ( a p) D( p) 1 and PS( p) = ( p c) qs( p), respectvey. Proposton 4: The congeston n transmsson ne resuts n rent transfer from consumers and producers to the ISO. Proof: We cacuate the tota wefare as foows: TW=CS (p )P S (p ) 1 1 TW = a p D p p c qs p ( ) ( ) ( ) ( ) 1 TW = ad p cq p pd p pq ( ( ) s( ) ( ( ) s) ) Tota wefare functon n LMP 0 peak oad power pant. A four-node eectrcty network s used here to determne the technca and economc nsghts of noda prcng n power system. Four-node market confguraton. Fgure presents the confguraton of the nterconnected four-node system. Assume a four-node nterconnected power system wth each node consstng of one power pant and one LSE. The four transmsson nes are dentca. Thus, they have smar admttance and resstance for each subsecton. The smuatons were conducted n two transmsson condtons whch are congested and uncongested transmsson nes. For the transmsson-constraned condton, ne 3 and ne 1 4 are mted to MW, whch defnes that power fow from node to node 3 and from node 3 to node s mted to MW. The transmsson ne characterstcs (resstance and admttance) and AC fow varabe (phase and votage ange) nfuence the congeston nomna. Tabe 1 shows the shft factor matrx for four-node power system wth unform nes. Snce DC oad fow s apped, ths smuaton consders PTDF as a functon of shft power factor and power njecton. Node no. 4 s chosen as a sack bus and assumng a zero-reference ange for ths node by deetng the row and coumn of sack bus mpedance matrx. DC oad fow cacuaton uses transmsson ne s shft factor matrx. The souton for shft factor matrx s as foows: Tabe provdes the near demand functon for each LSE, and suppy functon from each power pant. Exogenous eectrcty demand conssts of resdenta and ndustry demand. Hence, these eectrcty demands aggregated n power substaton as a near prce functon. The LSE for each node appes neastc demand. Generaton capacty constrans power pant output. The bggest supper of the system s GenCo3 wth a capacty of 10 MW whe the smaest payer s GenCo 1 wth a capacty of 30 MW. The varety of margna cost functon n ths mode represents the mx of generaton technooges n the rea power system. Fgure : Four nodes nterconnected power system Notce that for transmsson uncongested a D (p )-c q s (p )=0, and for transmsson congested a D (p )-c q s (p )=0. Acknowedgng proposton and 3, f the transmsson s congested, then the net market (producers and consumers) surpus s beow the tota market wefare. The ISO captures the dfference of ths wefare ( )..5. Sma Scae Noda Prcng In a rea power system topoogy, the power system conssts of mutpe sub-networks wth each node consstng of one LSE and snge or mut power pant technooges. Each power pant has a dstnctve near margna cost whch represents unque generaton technoogy, e.g. base, ntermedate and peakng PP. The generaton technoogy mx of a power system coud be dvded based on the abty of the PP to ramp-up and ramp-down to adjust the eectrcty demand fuctuaton from LSE aggregate. Rampng rate, ow fue cost and ong constructon tme are characterstcs of baseoad power pant. In contrast, hgh rampng rate, hgh fue cost, and reatvey ow constructon tme are characterstcs of Tabe 1: Shft factor matrx for four node power system wth unform nes Internatona Journa of Energy Economcs and Pocy Vo 8 Issue

6 Hakam: Noda Prcng: The Theory and Evdence of Indonesa Power System Fgure 3: Case 1. Unconstraned transmsson power system Tabe : Maxmum capacty, demand and margna cost n k (MW) a b c d Tabe 3: Output q s (MW), demand q d (MW) and prce ($/MWH) n Case 1 Case q s q d p q s q d p Fgure 4: Case. Constraned transmsson power system Tabe 4: DC oad fow (MW) From to j P j (DC) case 1 P j (DC) case (5.19) 1 4 (9.13) (7.11) (.00) the market prce to set quantty equbrum. Smar to case 1, the shadow prces - occurrng at node 1 and 4 - are bounded by the generaton constrant, athough at a hgher eve due to the addtona congeston from the cabe mts. Demand and suppy equbrum Q n the nterconnected power system s convergng at eve MW. The market smuaton was conducted by cacuatng two dfferent case studes. The frst case study s unconstraned transmsson whe the second case study s constraned transmsson. Load fow and prce-quantty equbrum from ths power system modeng s obtaned as can be seen n Tabes 3 and 4. The frst case study as n Fgure 3 s an unconstraned network n a perfect competton envronment. As shown n Tabe 3, the equbrum prce s unform across the regons wth a nomna 4.37 $/MWH. The unform prce occurs due to the uncongested transmsson nes. The eectrcty coud fow to any LSE based on the economc sgnas provded from rea margna cost and demand. Node one and four were utsng a of ts generaton snce GenCo 1 has the owest margna cost and GenCo 4 has a reatvey sma margna cost compared to GenCo and 3. The oad fow resut (Tabe 4) foows the Krchoff aw. Node 1 s a defct regon where GenCo 1 wth a capacty of 30 MW suppes a MW eectrcty demand. The oad-fow mechansm (PTDF) baances ths energy shortage by mportng 9.13 MW of eectrcty from node 4 and exportng 1.5 MW of power to node. Pease note that the arc and nomna of power fow s accordng to the Krchhoff aw,.e., the sum of energy fow for a nodes s equa to zero. By cacuatng the suppy bddng from each supper and the consumers demand from LSE s, the system operator cears Transmsson congeston affects the market equbrum snce the cabe mt bounded the mport and export of the eectrcty. As can be seen n Fgure 4, the prce n each node s vared. The hghest prce n the system s 50.8 $/MWH ocated at node refectng the hgh producton cost n regon. The nta power fow from node 1 to node 4 s mted to MW from 9.13 MW, prevousy, whch resuts n 1.53 MW oad fow. The transmsson ne that s connectng node and 3 s appyng the smar nomna constrant, whch affects the oad fow causng t to reduce from 7.11 MW to MW. 3. THE MODEL Sumatra and Java-Ba power system are two of the argest power system n Indonesa. The smuaton setup n ths research s mted to Sumatra power system. The smuaton was performed usng perfect competton wth norma operaton based on PLN power fow data for the year 015 to derve the economc sgnas. Each node was modeed as an ndvdua payer who represents one GenCo. Note that n the Sumatra system, one GenCo ony serves one LSE/subsystem. A of the power system data was coected from PLN accordng to the references n PLN (015), (P3BS 016a), and (P3BS 016b). These power system data are accessbe as pubcy avaabe datasets. PLN pubshed these reports for power system pannng, evauaton, and nvestment purposes. The st of data coectons are as foows: 1. Demand: Non-concdent peak oad for the Sumatra system.. Generaton: Maxmum and avaabe capacty, 1 generaton 1 Avaabe capacty s the power pant capacty based on the avaabty 140 Internatona Journa of Energy Economcs and Pocy Vo 8 Issue 6 018

7 Hakam: Noda Prcng: The Theory and Evdence of Indonesa Power System technoogy and fue type and fue cost. The near margna cost data was cabrated accordng to the reasaton of generaton cost and output from each power pant. 3. Transmsson confguraton and characterstc. We coect reactance and transfer mt to perform DC oad fow anayss. Ths research appes a stysed mode based on the actua network confguraton of the 150 kv Sumatra system. To acqure a precsey stysed mode, the mode was crosschecked wth the PLN TSOs. The reactance data s for a snge crcut. Thus, the reactance ratng for parae cabe foows the aw of parae-crcut. 4. Load fow reasaton. PLN TSOs conduct oad fow anayss usng power system software, e.g. Dgsent and PSSE. PLN provdes the Sumatra power system pannng n P3BS (016b). The base case scenaro was adjusted for constraned noda prcng based on power fow reasaton from P3BS (016a). The power pant avaabe capacty ncudes the generaton from PLN, IPP and other power pant,.e., renta PP and excess power (e.g. excess power from the Aumnum Pant n South Sumatra). The near demand functon was cacuated accordng to the actua peak oad for each node, assumng that the prce eastcty of demand s neastc for a areas. The generaton cost was derved usng power pant data, e.g. fue cost, fue consumpton rate and effcency. The ntercepts and sopes of margna cost were cabrated, assumng near margna cost curves, accordng to the generaton transacton cost of PLN n 015. The ntercept and sopes of the demand curves were cabrated, assumng eastcty reference 0.15 and prce reference $/MWH, accordng to the approach by Leuthod et a. (01). The prce reference s based on Mnstry of Energy and Mnera Resources Reguaton No regardng eectrc energy tarffs provded by PLN (Assumng 1$ = 13,799 IDR n 015). The eectrcty tarff n Indonesa s vared accordng to the type of usage,.e., resdenta, busness, ndustry, and soca. The eectrcty tarff aso vared accordng to the crcut breaker capacty. For exampe, n resdenta tarff, R-1/TR (up to 450 VA) has a tarff of 415 IDR/KWH, R-/TR (up to 900 VA) has a tarff of 605 IRD/KWH, whe R-1/TR (up to 1300 VA), R-/TR (up to 00 VA), R-3/TR ( VA), and R-4/TR (above 6600 VA) have a tarff of 135 IDR/KWH (equa to $/MWH). The eectrcty tarff of $/MWH s aso charged for the hghest type of busness consumer,.e. above 6600 VA. Thus, ths eectrcty tarff was chosen as a prce reference n our modeng snce the eectrcty tarff for R-1/ TR (up to 450 VA) and R-/TR (up to 900 VA) s subsdsed by the Indonesan government. The eectrcty tarffs for ndustry and soca customer are aso beow $/MWH snce the eectrcty tarff for both customers are ncentvsed and subsdsed, respectvey. For fu Indonesa eectrcty tarffs, see ESDM (015). The generaton data characterstcs presented are the avaabe capacty of a node, power pant aocaton and margna cost factor that represents the actua generaton capabty of a power pant nstaaton. Avaabty factor takes account of the rea curtament/outages at a partcuar power pant. Each power system has ts own TSO, Sumatra s TSO (P3BS) s responsbe for the power system operaton and transmsson assets n the Sumatra. for each node. Short run margna cost s constructed by takng nto account fue cost, heat rate and energy producton, and the avaabty factor (AF) of each unt pant. These data are avaabe from the 015 operatona reasaton data of P3BS. The generaton capacty of each GenCo s bounded not by maxmum capacty k but by avaabe capacty qs and outages at partcuar perod k = AFxqs where AF s the AF whch accommodates the machne de-ratng of the power pant. For power pants that operate n 015, the AF aso takes COD tme nto the cacuaton,.e. the actua tme when the power pant energes and suppes eectrcty to LSEs. Note that ths mode assumes that hydropower pant operates at maxmum capacty as n the wet season. The stysed mode n ths research s acqured by transformng the orgna network confguraton as n Fgures 5-8 by appyng the aw of parae crcut. The eectrcty market modeng n ths research appes the aw of parae crcut to acqure accurate oad fow anayss. Fgure 5 shows the snge ne dagram of the 150 kv North Sumatra subsystem. As shown n the Fgure 5, the North Sumatra subsystem s aready connected by a 75-kV subsystem through the connecton of a transformer connectng the Bnja 150 kv to the Bnja 75 kv substaton. The power generaton mx n north Sumatra s comprsed of coa, gas, hydro, and dese PP. Fgure 6 shows the snge ne dagram of the 150 kv md-sumatra subsystem. The Md Sumatra subsystem connects the North Sumatra wth the South Sumatra subsystem. Further, the Md Sumatra subsystem s abundant wth hydro energy sources whch has resuted n ower fue costs compared to the North and South subsystems. The South Sumatra subsystem s ustrated n Fgures 7 and 8, and conssts of a 150 kv and 70 kv transmsson network. The power system modeng n ths research defned the market boundares of Sumatra power system based on P3BS (016a). One bus n the Sumatra mode represents one payer n the network. The smuaton was performed based on non-concdent peak oad data for 015. The served peak oad on the North Sumatra power system s 1839 MW, take pace at Thursday, on the 3 rd September 015 at tme 19.30, whe the served peak oad n the Md-South Sumatra subsystem s 3,048 MW, hed at Tuesday, on the 18 August 015 at tme Fgure 9 shows the stysed network of the Sumatra power system whe Fgure 10 shows the Sumatra power system map. The Sumatra power system s dvded nto eght nodes accordng to PLN subsystem dvson; each node contans one GenCo and one LSE. The South Sumatra subsystem consst of two nodes,.e. Aceh and Sumut; the Md Sumatra subsystem conssts of two nodes,.e. Rau and Sumbar; the South Sumatra subsystem s dvded nto the Jamb, Sumse, Bengkuu and Lampung nodes. The oad and generaton aocaton for the Sumatra power system s shown n Tabe 5. Tabe 6 shows the transmsson characterstcs for the Sumatra power system, e.g., node connectons, type of confguratons, base reactance and therma mt. Note that the type of transmsson confguraton affects the reactance nomna. Hence, transmsson aggregaton s performed to transform the granuar power system nto a stysed mode. The cabe mts for a parae transmsson ne s equa wth the mts of the basc network. The reactance Internatona Journa of Energy Economcs and Pocy Vo 8 Issue

8 Hakam: Noda Prcng: The Theory and Evdence of Indonesa Power System Fgure 5: Snge ne dagram of 150 kv north Sumatra subsystem 015 (P3BS 016b) Fgure 6: Snge ne dagram of 150 kv md Sumatra subsystem 015 (P3BS 016b) of transmsson ne connectng node to j (X j ) s n per unt, cacuated by dvdng each reactance nomna to the hghest reactance n the partcuar power system. 4. RESULT Tabe 7 shows the shft factor matrx for the Sumatra power system. Subsystems n the Sumatra power system are connected by a rada transmsson confguraton. Thus, the cacuaton of Sumatra s oad transfer dstrbuton factor s trva to sove compared to a oop confguraton. Tabe 8 shows the noda demand, generaton and wefare for Sumatra noda prcng. The Sumatra power system s not a regona baance system snce some of the subsystems need energy mport from another subsystem to meet oca energy demand. Aceh, Rau, and Lampung are the defct regons accordng to ther energy nsuffcency to meet subsystem demand. Note that beng a surpus regon does not automatcay defne the regon as an energy exporter. Aceh and Jamb produced zero energy producton snce Aceh produce a hgher eectrcty prce compared to Sumut, whe Jamb produces a hgher eectrcty prce compared to South Sumatra. The tota demand s equa to the tota suppy (5,101.8 MW). 14 Internatona Journa of Energy Economcs and Pocy Vo 8 Issue 6 018

9 Hakam: Noda Prcng: The Theory and Evdence of Indonesa Power System Fgure 7: Snge ne dagram of 150 kv south Sumatra subsystem 015 (P3BS 016b) Fgure 8: Snge ne dagram of 150 kv south Sumatra subsystem (P3BS, 016b) The Sumbar and Bengkuu subsystems are domnated by hydro PP wth ow fue cost. Thus, these two subsystems produce cheaper generator capacty to meet oca and connected subsystem demands. Fgure 9: Stysed network of Sumatra power system Power system constrants consst of transmsson constrant that refects the fow of actve power and votage constrants whch determned the amount of reactve power fow. In the Sumatra mode, the votage constrants were transformed nto therma mts to make the modeng more reabe. Thus, the appcaton of the DC mode n ths research does not underestmate the votage constrants caused by reactve power transfer. The Sumatra power Internatona Journa of Energy Economcs and Pocy Vo 8 Issue

10 Hakam: Noda Prcng: The Theory and Evdence of Indonesa Power System Tabe 5: Load and generaton aocaton for Sumatra power system Node Subsystem q d (MW) q s (MW) Power pant (MW) mc ($/MWH) n s1 Aceh Nagan Raya KKA Lhokseumawe 0 Lueng Bata 31.9 Cot Trueng 1.5 Puau Psang 9.8 Sewa Aggreko 30 n s Sumut Beawan Labuhan Angn 105 Pangkaan Susu 400 Growth Sumatra 19 Growth Asa 0 PKS Rambutan 1.8 Harkat Sejahtera 10 Beawan 637 Beawan 90 Gugur 1 Paya Pasr 34 Ttkunng 16 Renta Paya Pasr 115 Renta Beawan (AKE) 65 Renta Beawan MFO 10 Sbayak 10 Spanspahoras 50 Lau Renun 80 Asahan 180 Inaum (Transfer) 90 Tersebar 5 Partan 7.5 Sau 7.5 Hutaraja 5 Kara 8.3 n s3 Rau Teuk Lembu Baa Pungut 34 Renta Teuk Lembu 1 Baa Pungut (ex Beawan) 40 Baa Pungut Rau Power 6 Koto Panjang 114 n s4 Sumbar Ombn Teuk Srh 00 Pauh Lmo 49.5 Mannjau 67.8 Batang Agam 10.5 Sngkarak Seo Kencana 7 n s5 Jamb Bomassa RSPL Batang Har 56.6 Payo Sencah 93.6 Se Geam (CNG) 89.5 Se Geam 1 Tanjung Jabung 7. Payo Sencah 16 n s6 Bengkuu Mus Tes 18.1 Tes extenson 4.3 Lebong 11.5 Lahat 9.9 n s7 Sumse Bukt Asam Smpang Bembng 7 Renta PTBA 6 Banjarsar 0 Baturaja 0 Keramasan 4.8 Taang Dukuh 68.6 LM Borang 11 (Contd...) 144 Internatona Journa of Energy Economcs and Pocy Vo 8 Issue 6 018

11 Hakam: Noda Prcng: The Theory and Evdence of Indonesa Power System Tabe 5: (Contnued) Node Subsystem q d (MW) q s (MW) Power pant (MW) mc ($/MWH) n s7 Sumse Bukt Asam Smpang Bembng 7 Renta PTBA 6 Banjarsar 0 Baturaja 0 Keramasan 4.8 Taang Dukuh 68.6 LM Borang 11 Borang 67. Jakabarng 50.5 Renta Keramasan 45 Renta Jamb 9.7 Prabumuh 11.6 Sako 11.8 Mus Rawas 8 Borang 150 Indraaya 10.5 Gunung Megang 110 Mus II 19.4 Keramasan 74 Sunga Juaro n s8 Lampung Tarahan Sebaang 89 Gunung Sugh 14 Peabuhan Tarahan 10 Tarahan 16 Tarahan 0.5 Teuk Betung 1.6 Tegneneng 18 Uubeu Besa 89.6 Batuteg 8 Tota Tabe 6: Transmsson characterstc for Sumatra power system Trans From To j X j (p.u) T (MW) Confguraton t s1 1 Langsa Pangkaan Brandan km; AC3 t s Kota Pnang 3 Bagan Batu km; snge Haw t s3 3 Koto Panjang 4 Payakumbuh km; Duck t s4 4 Kranjao 5 Muarabungo km; twn Zebra t s5 5 Bangko 7 Lubuk Lnggau km; twn Zebra t s6 7 Lubuk Lnggau 6 Pekaongan km; ACSR X340 mm t s7 7 Baturaja 8 (Umpu Kemunng) km; AC3 (Kemunng) Tabe 7: Shft factor matrx for Sumatra power system Fgure 10: Map of Sumatra power system system suffers severa power system constrants,.e. transmsson mt, sma-sgna stabty, transent stabty and subsystem nterconnecton. Sma-sgna stabty s the system constrant reated wth network stabty resutng from sma dsturbances that eads to power system oscaton. The power system s stabe f the oscaton can be suppressed and system devaton remans ow for a perod of tme. In contrast wth sma-sgna stabty, transent stabty s caused by sudden and sgnfcant outages n the eectrca network. The North Sumatra and Md Sumatra subsystems were nterconnected n 007 through the 150 kv T/L Internatona Journa of Energy Economcs and Pocy Vo 8 Issue

12 Hakam: Noda Prcng: The Theory and Evdence of Indonesa Power System Tabe 8: Equbrum of demand, generaton, prce and wefare for Sumatra power system Node Subsystem q d (MW) q s (MW) p PS CS TW n s1 Aceh ,96 100,96 n s Sumut 1,65.07, , , ,488 n s3 Rau ,375 18, ,049 n s4 Sumbar , ,80 186,191 n s5 Jamb ,954 9,954 n s6 Bengkuu ,896 45,671 5,567 n s7 Sumse , ,78 85,14 88,40 n s8 Lampung ,73 315,995 33,77 Tota 5, , ,803 1,739,887 1,797,690 Tabe 9: Power transfer for Sumatra noda prcng Trans From node To node j P j (DC) (MW) t s1 n s1 n s 98.4 t s n s n s t s3 n s3 n s t s4 n s4 n s t s5 n s5 n s t s6 n s6 n s6 1. t s7 n s7 n s Fgure 11: Load fow reasaton of Sumatra power system (P3BS, 016a) Bagan Batu - Kota Pnang. However, due to stabty ssue arsng from nterconnectng the two subsystems, the system remans separated. The ne connectng these two subsystems s operated n normay-open condton. Hakam et a. (011) for further expanaton regardng transent stabty and nterconnecton probems n the Sumatra power system. Sumse s domnated by Coa PP,.e. Bukt Asam PP (33.1 MW), Smpang Bembng PP (7 MW), Banjarsar PP (0 MW), Indraaya PP (10.5 MW), and Gunung Megang PP (110 MW). In contrast, generaton technoogy n Jamb s domnated by gas and o fue-based PP,.e., Batang Har PP (56.6 MW), Payo Sencah PP (93.6 MW), and Se Geam CNG PP (89.5 MW). Sumse provded a ow fue cost compared to Jamb and Lampung. The dfferences n fue cost between Sumse and Jamb as we as the energy defct n the Mdde Subsystem caused a sgnfcant oad fow from South (node 7) to Md Sumatra (node 6), especay on the peak oad condton. The sma sgna stabty mt caused a power transfer mtaton to 30 MW. Transmsson constrants T refects the cabe therma mt for a 150 kv overhead transmsson nes. Sma-sgna stabty constrant reduces the transfer mt of t s5 whch connects the South and Md Sumatra subsystems to 30 MW. Interconnecton constrants between the North and Md Sumatra subsystems mt the cabe mt of t s to 90 MW whch refects the actua demand n the nearest substaton. The therma constrant n t s7, and stabty constrants n t s5, are normay bndng and have an mpact on the prces snce the Sumse subsystem transports energy at ower prce compared to the mporter subsystems (Lampung and Jamb). Tabe 9 above shows the DC power transfer fow for the Sumatra power system. The arc of oad fow s nfuenced by the power njecton of each node (q s =q d ). The negatve sgn n t s1 and t s5 shows that oad fow has an opposte drecton to the anchor ponts,.e. oad fow for transmsson t s1 s from Pangkaan Brandan (Sumut subsystem) to Langsa (Aceh subsystem) whe oad fow for transmsson t s5 s from Lubuk Lnggau (Sumse subsystem) to Bangko (Jamb subsystem). Power transfer from Sumse to Lampung s 37 MW, whe the cabe mt of t s7 s MW. Thus, the transmsson t s7 does not meet the contngency N-1 crtera (coapse n one overhead cabe w cause the coapse of transmsson nes). Note that power transfer from Sumse to Jamb (10.5 MW) s beow the mt of sma sgna stabty constrant (30 MW). Therma constrant n t s5 s bound to the equbrum due to Lampung as a defct subsystem. Thus, Lampung needs eectrcty mported from the connected subsystem (South Sumatra). In contrast, Jamb s a surpus regon where the Jamb power pants can adequatey produce eectrcty for Jamb s LSE. However, Jamb has a hgher fue cost and eectrcty prce compared to South Sumatra. Instead of producng ts own eectrcty, Jamb mports a requred eectrca energy from South Sumatra. It can be seen from Tabe 9 that South Sumatra has a surpus eectrcty energy of 36 MW to transfer nto the Lampung and Jamb subsystems. The smuaton resut n Case 1 s smar wth the power system reasaton as n Fgure 11. As mentoned earer n Chapter 3, ths research assumes the fu avaabe capacty of Hydro PP. In addton, ths study gnores the TOP (Take or Pay) contract between IPP and prmary energy producers wth PLN. Therefore, as can be compared n Tabe 9 and Fgure 11, there are dfferences n power transfer, especay for the transmsson nes connectng the Sumbar and Bengkuu nodes 3. 3 Sumbar and Bengkuu are abundant wth hydro resources compared to other nodes. 146 Internatona Journa of Energy Economcs and Pocy Vo 8 Issue 6 018

13 Hakam: Noda Prcng: The Theory and Evdence of Indonesa Power System The stysed modes n ths research do not fuy accuratey represent the rea power system at the detaed eve of a ow votage power substaton. However, the modeng was based on the actua network topoogy of a 150-kV power network by usng a bottom-up approach. In the Sumatra system, cabe nes connect two power substaton (SS) at the end of each node,.e., transmsson ne t s1 s connectng the 150 kv Langsa SS at n s1 Aceh wth the 150 kv Pangkaan Brandan SS at n s Sumut. Thus, the mode w response n a smar way compared to the actua power system n respondng to any changes n generaton and demand. 5. CONCLUSIONS Based on the smuaton n the chapter 4, we found that the prce (p ) n each noda coud be dfferent f there s a network constrant n the eectrcty mesh network. In a non-constrants network, the noda prce s equvaent for each node athough there s a devaton of true margna cost between generatng frm. Transmsson and generaton system constrants affect the equbrum noda prces. The wefare was reduced when the transmsson has mted transfer capabty. Ths research presents a stysed economc mode of the Indonesa eectrcty market to cacuate noda prcng of Indonesa s power system wth engneerng constrants. Ths study s the frst study that anayses LMP usng perfect competton optmsaton n the Indonesa eectrcty market whch contrbutes to the current academc terature. Ths mode uses actua power system data from 015 that was acqured from PLN, an Indonesa state-owned eectrcty company. Ths noda prcng mode s based on a smpfed DC oad fow by appyng the PTDF to the equaton and comparng t wth the actua power transfer reasaton. The eectrcty stakehoder n Indonesa coud appy ths noda prcng regme rather than unform prce regme to ncrease the socety wefare. By usng ths mode, PLN coud mtgate the rsk of power generaton nvestment by nvest n power generaton effcenty accordng to the economc sgnas from noda prcng modeng. 6. ACKNOWLEDGMENT We acknowedged the support from Indonesa Endowment Fund for Educaton (LPDP). REFERENCES ESDM. (015), Mnstry of Energy and Mnera Resources Reguaton No. 31 Year 014 Regardng Eectrc Energy Tarff Provded by PLN. Indonesa. Faza, R., Muhammad, N., Rcky, F., Stephan, P. (015), Sumatra- Java HVDC Transmsson System Modeng And System Impact Anayss. Conference: 015 IEEE Endhoven PowerTech. Green, R. (007), Noda prcng of eectrcty: How much does t cost to get t wrong? Journa of Reguatory Economcs, 31(), Hagspe, S., Jagemann, C., Lndenberger, D., Brown, T., Cherevatsky, S., Troster, E. (014), Cost-optma power system extenson under fowbased market coupng. Energy, 66, Hakam, D.F., Ayodee, A. (018), Gas monetsaton ntrcaces: Evdence from Indonesa. Internatona Journa of Energy Economcs and Pocy, 8(), Hakam, D.F., Luqman, A., Taufq, F. (01), Sustanabe Energy Producton In Sumatra Power System. Ba, Indonesa: In IEEE Conference on Power Engneerng and Renewabe Energy 01, J. Hakam, D.F., Rzk, W., Eko, Y.P. (011), Swtchng Study for 75 kv Padang Sdempuan-Payakumbuh Transmsson Lne. Bandung, Indonesa: 011 Internatona Conference on Eectrca Engneerng and Informatcs. pe10-3. Hogan, W., Juan, R., Ingo, V. (010), Toward a combned merchantreguatory mechansm for eectrcty transmsson expanson. Journa of Reguatory Economcs, 38(), Leuthod, F.U., Hannes, W., Chrstan, H. (01), A arge-scae spata optmzaton mode of the European eectrcty market. Networks and Spata Economcs, 1(1), Macatangay, R. (1998), Space-tme prces of whoesae eectrcty n Engand and Waes. Uttes Pocy, 7(3), P3BS. (016a), Operaton Evauaton Year 015: Sumatra Power System. Pekanbaru, Indonesa. P3BS. (016b), Operaton Pannng Year 016: Sumatra Power System. Pekanbaru, Indonesa. PLN. (015), Rencana Umum Penyedaan Tenaga Lstrk (Natona Eectrcty Suppy Busness Pan) PT PLN (Persero) Jakarta, Indonesa. Avaabe from: RUPTL/RUPTL PLN pdf. Schweppe, F., Mchae, C., Rchard, T., Roger, B. (1988), Spot Prcng of Eectrcty. The Kuwer Internatona Seres n Engneerng and Computer Scence. 1 st ed. Massachusetts: Kuwer Academc Pubshers. Wartana, I.M., Sngh, J.G., Werakorn, O., N, P.A. (01), Optma Pacement of a Seres FACTS Controer n Java-Ba 4-Bus Indonesan System for Maxmzng System Loadabty by Evoutonary Optmzaton Technque. Proceedngs - 3 rd Internatona Conference on Integent Systems Modeng and Smuaton, ISMS. p Internatona Journa of Energy Economcs and Pocy Vo 8 Issue