Long Term Preventive Generation Maintenance Scheduling with Network Constraints Ali Badri a*, Ahmad Norozpour Niazi b, Seyyed Mehdi Hoseini c

Transcription

1 Avalable onlne at Energy Proceda 14 (2012) ICAEE2011 Long Term Preventve Generaton Mantenance Schedulng wth Network Constrants Al Badr a*, Ahmad Norozpour Naz b, Seyyed Mehd Hosen c a Department of Electrcal Engneerng, Shahd Rajaee Teacher Tranng Unversty,Tehran,Iran b Department of Electrcal Engneerng, Shahd Rajaee Teacher Tranng Unversty,Tehran,Iran c Department of Electrcal Engneerng, Nushravan Unversty,Babol,Iran Abstract The whole Electrc Power System ncludes three aspects: Generaton, Transmsson and Dstrbuton that all need mantenance to enhance system relablty and energy effcency. Most unt mantenance schedulng (MS) packages consder preventve mantenance schedulng for generatng unts over one or two years tme horzon n order to mnmze the total operaton costs whle satsfyng system energy requrements. The ncluson of network constrants n generatng unt mantenance wll ncrease the complexty of the problem. Ths paper manly consders the generaton mantenance schedulng takng nto account the system securty and relablty ndces. For more realstc study transmsson, securty constrants as well as crew constrants and system relablty ndces such as amount of not suppled energy are consdered for the proposed mantenance schedulng problem. General Algebrac Modellng System (GAMS) s the utlzed for solvng optmzaton problem. An IEEE 24 bus test system s employed for smulaton and show the accuracy of results Publshed by Elsever Ltd. Selecton and/or peer-revew under responsblty of the organzng commttee of nd Publshed Internatonal by Conference Elsever Ltd on Advances n Energy Engneerng (ICAEE). Open access under CC BY-NC-ND lcense. Keywords: Generaton mantenance schedulng, Network Constrant, Electrcty Market, System Relablty. 1. Introducton Snce 1980s, many countres have carred out the electrc power market reform. Its essental target was breakng the monopoly operaton pattern of tradton electrc power ndustry and buldng a compettve power ndustry. Therefore, t can reduce the electrc power producton cost and electrcty prce. In addton, t can mprove the power supply qualty and promote the healthy development of electrc power ndustry. Extra competton and growng complexty n power generatng systems as well as a need for hgh servce relablty and low producton costs provoked addtonal nterests n automatc schedulng * Correspondng author. Tel.: ; fax: E-mal address: A_Badr73@yahoo.com Publshed by Elsever Ltd. Selecton and/or peer-revew under responsblty of the organzng commttee of 2nd Internatonal Conference on Advances n Energy Engneerng (ICAEE). Open access under CC BY-NC-ND lcense. do: /j.egypro

2 1890 Al Badr et al.\ / Energy Proceda 14 (2012) technques for mantenance of generators, transmsson, and related equpment. The soluton methods can be categorzed as nteger programmng, decomposton methods [1], dynamc programmng, smulated annealng method [2], probablstc approach [3] and artfcal ntellgence method [4, 5]. In fact, Independent System Operator (ISO) s a neutral operator responsble for mantanng nstantaneous balance of the system. The ISO performs ts functon by controllng the dspatch of flexble plants and gves order to adjust or curtal loads to ensure that loads match avalable generatng resources n the system. Nomenclature X t Unt mantenance status, 0 f unt s off-lne for mantenance S Perod n whch mantenance of generatng unt starts e Earlest perod for mantenance of generatng unt to begn l Latest perod for mantenance of generatng unt to begn γ t Weekly unt mantenance cost penalty factor d Duraton of mantenance for generatng unt r Vector of dummy generators whch corresponds to energy not suppled at tme perod t f Maxmum lne flow capacty n matrx term f Actve power flow n vector term g max, t Mnmum power generaton for each unt at tme t g mn, t Maxmum power generaton for each unt at tme t g t Vector of power generaton for each unt at tme t d t Vector of the demand n every bus at tme t S Node-branch ncdence matrx ε Acceptable level of expected energy not suppled Generally, mantenance schedulng n a raw system may be dvded nto three stages of long-term, short-term and real-tme [6]. Long-term mantenance schedulng (LTS) consders the schedule of generatng unts on a horzon of 1 or 2 years n order to mnmze the total system operaton costs. The long-term schedulng problem tackles fuel allocaton, budgetng, emsson, producton and mantenance costng. The solutons obtaned from LTS can then be used as gudelnes and bases for addressng unt commtment and optmal power flow problems [7, 8, 9 and 10]. The objectve of short-term schedulng (STS) s to mnmze the cost of operaton over hourly, daly or weekly perods. Because dynamc economc dspatch s fundamental for real tme control of power systems, the STS brngs about a commtment strategy for real-tme economc dspatch to meet system requrements n an on-lne operaton. The dynamc economc dspatch s solved for short perods of tme n whch the system load condtons can be assumed constant. Ths paper represents a model for long-term preventve generaton mantenance schedulng problem. For more realstc study transmsson securty constrants as well as crew constrants and system relablty ndces such as amount of not suppled energy are consdered for the proposed mantenance schedulng problem. Due to dscrete nature of model, mxed nteger programmng (MIP) s appled to solve the problem. The paper s organzed as follows: Secton 2 represents the formulaton of proposed mantenance

3 Al Badr et al.\ / Energy Proceda 14 (2012) schedulng model and soluton methodology. In secton 3, a case study s presented to show the accuracy of the results and secton 4 provdes the concluson. 1. Problem Descrpton and Soluton Methodology The proposed long term MS problem s determnng the perod for whch generatng unts should be taken off lne, over one or two years plannng horzon n order to mnmze the total operaton cost, whle transmsson and relablty constrants are taken nto account. Leave out the network n mantenance schedulng may result n loss of nformaton on schedulng lmtatons. When network constrants are ncluded, the problem becomes consderably more realstc and complex that could be referred as an ntegrated mantenance schedulng. The methodology for the soluton of ths problem s dscussed n ths paper. The long-term generaton mantenance schedulng n the power market envronment s a large-scale optmzaton problem. Mathematcally, t can be formulated as follow: Mn Ct γ t (1 x t ) + t Subject to Mantenance Constrants ctgt (1) t for t e or t l + d xt = 1 (2) for S t S + d x = 0 (3) for t l xt = t e 0 or 1 (4) Seasonal lmtatons (5) Resources avalablty (6) Desrable schedule (7) Crew avalablty (8) System constrants t sf + g + r = d (9) t g (10) mn gt g max t r d (11) t f f. N (12) r ε (13) t Eq (1) corresponds to a mxed nteger-programmng problem snce X t s nteger varables and g t s contnuous. The objectve of (1) s to mnmze the total mantenance and producton costs over the operatonal plannng perod. The frst term of objectve functon s the mantenance cost of generators and the second s the energy producton cost. Constrants (2-4) represent the mantenance wndow stated n terms of mantenance varables (S ). The generaton mantenance may not be scheduled before ther earlest perod (e t ) or after latest perod, allowed for mantenance (l +d ). The set of constrants (5-8) represents crew resources avalablty,

4 1892 Al Badr et al.\ / Energy Proceda 14 (2012) seasonal lmtatons, desrable schedules, as well as other constrants such as fuel and emsson constrants. Seasonal lmtatons may be ncorporated n (e ) and (l ) values of constrant (2-4). If we consder that n each mantenance area, we have lmted resources and crew avalable, the set of constrants wll be stated as follows: A σ (1 x ) z (14) m t mt In the case of representng a resource constrant, Z mt would be the amount of m avalable resource n area A for each tme t and σ m would be a percentage of ths resource requred for unt. In the case of a crew constrant, Zmtwould be the number of mantenance crew n area w and σ would be a percentage of ths number requred for mantenance of unt. The set of system constrants (9)-(13), whch represent the peak load balance, transmsson flow lmts and allowable unserved energy, wll be checked by the ISO. To solve the problem, the mxed nteger problem solver of GAMS optmzaton software s employed. 2. Case Study The proposed method s appled to the 24 bus IEEE-RTS. Ths system s made of 32 generatng unts, 20 demand sdes, 24 buses and 38 transmsson lnes. A three months study perod of summer weeks, weeks 18-29, s consdered. Some generatons facltes n a partcular area need mantenance wthn the study perod. The coverage of mantenance area s from buses 1 through 10. Table 1 gves the generators placement and capacty data. Operatng characterstcs of the generatng unts are llustrated n Table 2. Fg. 1 depcts weekly peak loads as the percent of the annual peak load. As shown the maxmum peak load s n week 23. Subsequently, weekly penalty factors consdered for generators are provded n Table 3. It s assumed that durng the three months, crew constrant s up to two groups for generaton mantenance. Detaled system data for transmsson lnes, generators and loads can be found n [11]. Table 1: Generators data Unt Capacty (MW) Bus Table 2: Generatng unts operatng data Sze MW Fuel Ol #6 Ol #2 Hydro Coal Ol #6 Coal Ol #6 Coal Nuclear Cost (US$/MBtu) Heat rate (Btu/KWh) Here, two cases are studed for MS problem consderng transmsson relablty and securty constrants as follow: Case 1: Study generator mantenance problem consderng consumers energy not suppled ndex for each week. Case 2: Study generator mantenance problem consderng transmsson securty constrant. In Case 1, the ndex of not suppled energy s taken nto account as the sgnfcant factor from system operator whle mplementng mantenance schedulng. For ths purpose, energy not suppled n each week s lmted to the maxmum of 1%, 11% and 18 % of the total weekly load. Table. 4 represents generators operaton and mantenance data for three cases. As shown ncreasng n maxmum energy not suppled

5 Al Badr et al.\ / Energy Proceda 14 (2012) level results n decreases n operaton costs and system total costs as well. Although, system total cost s reduced, however, system relablty level wll be decreased. % Load Week Fg 1: Weekly peak load n percent of annual peak Table 3: Penalty Factor for generator unt mantenance cost week Penalty factor Subsequently, table 5 shows correspondng unt mantenance schedulng durng specfed 12 months. Table 4: Total operaton & mantenance cost for generatng unt (case 1) maxmum of Energy not served n each week Total Operaton & Mantenance cost Mantenance cost Operaton cost 1% of the total weekly load $ $ $ 11% of the total weekly load $ $ $ 18% of the total weekly load $ $ $ Table 5: Mantenance schedulng of generatng unt (case 1) Unt Week on mantenance (ε =1% of load) Week on mantenance (ε =11% of load) Week on mantenance (ε =18% of load) Case 2 studes the effect of transmsson securty on mantenance schedulng problem. For more clarfcatons, the mpact of transmsson securty lmts on MS problem s nvestgated. Two cases are consdered for ths study. In the former case, t s assumed that there s no lmt on transmsson capacty constrants whle n the latter case t s assumed that transmsson capacty of lne 7-8 s reduced to half. Smlarly, Table 6 llustrates the system operaton and mantenance costs. As t s appear, transmsson lmts lead to ncreases n system-aggregated costs. Subsequently, correspondng unt mantenance schedulng durng specfed 12 months are provded n Table. 7. Table 6: Total operaton & mantenance cost for generatng unt (case 2) Condton Total Operaton & Mantenance cost Mantenance cost Operaton cost wthout any lmts n transmsson $ $ $ transmsson lne capacty 7-8 has reduced to half $ $ $

6 1894 Al Badr et al.\ / Energy Proceda 14 (2012) Fnally, a comparson between unt mantenance schedulng costs s mplemented n Fg. 2. As depcted n ths fgure transmsson securty and relablty constrants may have profound effects on generators mantenance and operaton costs. Table 7: Mantenance schedulng of generatng unt (case 2) Unt Wthout any lmts n Transmsson lne capacty 7-8 has reduced to transmsson half Total Operaton & Mantenance cost Mantenance cost Operaton cost Total Operaton & Mantenance cost Mantenance cost Operaton cost 1% of the total weekly load 11% of the total weekly load 18% of the total weekly load Wthout any lmts n transmsson Transmsson lne capacty 7 8 has reduced to half Fg 2: Comparson of generators MS costs for cases 1, 2 3. Conclusons Ths paper presents generaton mantenance schedulng consderng network constrants. The test results demonstrate that lmts on energy not suppled and transmsson lne capacty affect the loadng ponts of unts and ncrease the generaton of expensve and neffcent unts, resultng n an ncrease n the overall cost of operaton. The extenson of generaton mantenance schedulng to nclude network constrants s sutable to the problem of mantenance wth probablstc data. Usng the proposed method, addtonal complex constrants can be mposed on the mantenance schedulng problem. References [1] Salvador Perez Canto. Applcaton of benders decomposton to power plant preventve mantenance schedulng; Scence drect 2008; [2] K. Suresh, N. Kumarappan. Combned genetc algorthm and smulated annealng for preventve unt mantenance schedulng n power system [3] M.K.C. Marwal, S.M. Shahdehpour. A probablstc approach to generaton mantenance scheduler wth network constrants; IEEE 1999; [4] R. Eshraghna, M.H. Modr Shanech, H. Rajab Mashhadl. A new approach for mantenance schedulng of generatng unts n power market; KTH 2006 [5] A.M. Lete Da Slva, L.A.F. Mansob, G.J. Anders. Evaluaton of generaton and transmsson mantenance strateges based on relablty worth; Scence drect 2004; [6] Y. Yare, G. K. Venayagamoorthy. A dfferental evoluton approach to optmal generator mantenance schedulng of the ngeran pwer system;ieee 2008.

7 Al Badr et al.\ / Energy Proceda 14 (2012) [7] M. Manbach, A.H. Parsaefard, M.R. Haghfam. A new soluton for mantenance schedulng usng mantenance market smulaton based on game theory;ieee [8] Hemantkumar Barot, Kankar Bhattacharya. Securty coordnated mantenance schedulng n deregulaton based on genco contrbuton to unserved energy;ieee [9] Dusmanta Kumar Mohantaa, Pradp Kumar Sadhub, R. Chakrabartc. Determnstc and stochastc approach for safety and relablty optmzaton of captve power plant mantenance schedulng usng ga/sa-based hybrd technques; Scence drect 2007; [10] Zhang Shuangme, Ge Ju. Study on generaton and transmsson mantenance schedulng under electrcty market [11] IEEE Relablty Test System; IEEE 1979