Problems of fast frequency variation control in interconnected power systems

Transcription

1 Problems of fast frequency varaton control n nterconnected power systems V. Chuvychn, A. Sauhats, R. Petrchenko, G. Bochkarjova Abstract Ths paper descrbes problem of fast frequency varaton control n large nterconnected power system. Interconnecton of the large power system wth dfferent phlosophy of frequency control can cause neffectve and sometmes nonselectve behavor of frequency automaton. Paper presets results of analyss of frequency control usng spnnng reserve. Optmal dstrbuton of the prmary reserve can be based on cooperatve game theory method. Severe system dsturbances can result n fast frequency drop, whch makes fast governor and boler response mpossble. If the governor acton cannot actvate spnnng reserve quckly enough to restore the system to ts normal operatng frequency, underfrequency load sheddng (UFLS) serves as a lastresort tool to prevent the system from collapse. Applcaton of smart meterng and communcaton can mprove effcency of emergency automaton durng underfrequency condton. A new load sheddng method s suggested. Smulaton of frequency behavor was conducted for exstng load sheddng system and a new one. Keywords Interconnected power systems, smart meterng system, cooperatve game theory, load sheddng. D I. INTRODUCTION URING the last years many nternatonal projects were devoted to the problem of nterconnecton of the large transmsson networks n Europe [1], [2]. There are many problems that complcate such nterconnecton: 1) Intal perod of nterconnected operaton of power systems usually s characterzed by relatvely weak ntersystem tes. There wll be problem of control the ntersystem tes. 2) Dfferent phlosophy of frequency and actve power control can cause oscllatons of power transmtted through the ntersystem te lne. 3) Methods of frequency control for normal and emergency operatonal condtons n dfferent systems are dfferent. Ths work has been supported by the European Socal Fund wthn the project «Support for the mplementaton of doctoral studes at Rga Techncal Unversty». V.Chuvychn s wth the Faculty of Power and Electrcal Engneerng, Rga Techncal Unversty, Rga, LV-1010 Latva (e-mal: chuvychn@eef.rtu.lv). A. Sauhats s wth the Faculty of Power and Electrcal Engneerng, Rga Techncal Unversty, Rga, LV-1010 Latva (e-mal: sauhatas@eef.rtu.lv). R. Petrchenko s wth the Faculty of Power and Electrcal Engneerng, Rga Techncal Unversty, Rga, LV-1010 Latva (e-mal: romans.petrcenko@rtu.lv). G. Bochkarjova s wth the Faculty of Power and Electrcal Engneerng, Rga Techncal Unversty, Rga, LV-1010 Latva (e-mal: galna@eef.rtu.lv). Compatblty problem of operaton durng normal and emergency condtons of nterconnected power systems s aroused. Durng emergency stuaton n the power system caused by generatng power defcency frequency declne takes place. Dynamcs of frequency behavor can have very dfferent character. It depends on the value of dsturbance, response of emergency automaton, governor system and reasons of emergency stuaton. If governor acton cannot actvate spnnng reserve fast enough to restore the system to ts normal operatng frequency, frequency actuated automatc load sheddng serves as a last-resort tool to prevent the system from the collapse [3], [4]. Interconnecton of large power systems wth dfferent underfrequency automaton and systems' parameters can cause neffectve behavor of frequency n the nterconnecton. For small frequency devaton prmary and secondary frequency control form a resource to mprove effcency of power system. Many papers are dedcated to the qualty mprovement of frequency control. The ntroducton of market condtons n the task of the power system operaton changes formulaton and solvng of energy cost mnmzaton problem n the transent condtons. For fast and deep frequency declne load sheddng automaton systems wll be appled to restore a normal frequency. Smart meterng and communcaton systems should be used for creaton of new soluton. Paper descrbes possble approach to use smart technology for control fast frequency varaton n the nterconnected power system. II. POSSIBLE APPROACH FOR FREQUENCY CONTROL USING SPINNING RESERVE Ths chapter llustrates example of frequency control usng spnnng reserve operaton. For ntegrated power system dstrbuton of spnnng reserves s mportant problem. The operaton of several power companes (players), whch smplfed scheme s presented n Fg.1, s observed. Let us suppose that each company contans several power plants, whch are operatng n market condtons. At the same tme the cooperaton between power plants s possble and the support of generated and consumed energy balance at the nomnal frequency and costs mnmzaton are the general targets of ths cooperaton. ISBN:

2 To provde these lmtatons the fulfllment of condtons (3) s requred: P, p A, j j (3) where A s the doman of the allowed states of the power system. The task for selecton of spnnng reserve can be formulated as: N n * * Pj,pj argmn (Pj Cj E j(cr )) t j, j1 1 (4) Fg.1 smplfed dagram of nterconnected power plants nvolved n prmary frequency control Each power plant s equpped wth smart meterng system nformaton from whch s suppled to the central processng block "Operator". Let us consder two possble approaches for spnnng reserve dstrbuton: 1 - Fve plants operate ndependently, supportng specfed part (reserve) of the planned power (classcal approach); 2 All fve plants strve to most proftable operaton and costs mnmzaton. Let s suppose, that fve power plants are operatng A, B, C, D, E (Fg.1). The power capactes of each plant are - P 1, P 2, P 3, P 4, P 5. Producton cost values - C 1, C 2, C 3, C 4, C 5 (EUR/MWh). Each plant provdes a reserve - p 1, p 2, p 3, p 4, p 5, wth producton costs c 1, c 2, c 3, c 4, c 5 (EUR/MWh). The resultng costs RC j for each Δt j (where j =1,...N) are RC j 5 ( P C p c ) t, (1) 1 Assume, that generators must provde a planned power P Σ of nterconnected power system, as well as possble random devatons of power p Σ for all of C <C +1 and c <c +1. Suppose that densty of power devatons ƒ j (p j ) s known for each tme nterval Δt. The goal s to provde balance of powers: 5 1 ( P p ) P p, (2) wth mnmal RC j n the nterval Δt. Selectng generators powers P and reserves p t s necessary to take nto account many techncal lmtatons, whch depend on: thermal and electrcal loads of consumers, operatonal condton of thermal and electrcal network, water levels n reservor of hydro power plants. wth (2) and (3) condtons fulfllment n equaton (4) E j (RC j ) s mathematcal expectaton of the spnnng reserve expenses -th power plant at j-th tme nterval of Δt. p j max E ( RC ) p c f ( p )dp (5) j j j j j j j, 0 P * j and p * j - optmal capacty of consdered plants (provdng the mnmum cost). Redstrbuton of network prmary power reserve must be done by nformaton and control center operator. Takng nto account complexty of processes, operator has to be realzed by IT technologes. Let s descrbe steps of nformaton and control center s algorthm: a) algorthm ranks all players (power dstrcts) by value of proftablty, from the least to the most proftable (as example see Table 1, 2nd column); b) algorthm solves all possble combnatons of the prmary reserve allocaton among the players and calculates the most proftable varant. Durng analyss of varants the ntal data for creaton coalton of players can be receved. c) algorthm calculates the sharng of beneft among the players; d) operator generates control sgnals to change the settngs of the prmary reserve control equpment for each players. III. BENEFIT SHARING APPROACH The mentoned above algorthm's pont c) s most nterestng, because beneft sharng s one of the man power market ssues. In the case of the coalton formed by two players, proft sharng s equally dstrbuted between the players (50%/50%). However, n the case of three, four or fve players coalton, the problem of equtable proft sharng appears. So, f the optmal varant s a coalton of more than two players, proft sharng s done by usng the Shapley vector [5]. Let s suppose that the -th player gets beneft equal to the average value of ths player contrbutons to all coaltons [5]: ISBN:

3 ( S 1)!( n S )! ( v ) ( v( S ) v( S \ { })) (9) SN n! The number ν(s)-ν(s \ {}) s the contrbuton of player when he s jonng the coalton S \ {}, but the weght factor ( S 1)!( n S )! n!, can be nterpreted as the probablty of the coalton S \ {} formng. Shapley s value of cooperatve game s a vector [5]: T (10) ( v ) ( 1 ( v ),..., n( v )) IV. CASE STUDY Let s consder operaton of proposed automaton usng a specfc example. Suppose that ntal capactes of power plants are known, as well as prces, reserves, etc. Frequency control reserve s stated by system operator equal to 4% of maxmal generated power P Max. The ntal parameters requred for calculatons are shown n the Table 1. The power system operates n normal condton, when the generated and consumed actve powers are equal (ntal stuaton): P Generated = P Load = 6.2 (p. u.). Table 1 Parameters of Players Before Optmzaton Fg.2 players proftablty n dfferent stuatons Table 2 Players Proftablty n Dfferent Stuatons and Results of Proft s sharng A. Example of Possble Annual Proft Calculaton The vsual demonstraton of advantage to use suggested method s practcal calculaton of addtonal proft usng denomnated quanttes of power system parameters. When average value of addtonal proft s equal (see Fg.2) to 0.2% for power system wth average capacty equal to 1000 MW and average prce of energy s 50EUR/MWh. Addtonal proft s EUR. Such addtonal beneft wll cover the cost of creaton the approprate control system. C cost prce producton cost value of produced power (p. u.). Let s assume that the sellng prce of electrcty s equal to 1.1 (p. u.). Algorthm calculates proftablty of all possble combnatons of players. In our case, usng the basc parameters lsted n Table 1 and usng equatons (9) and (10), the algorthm calculates (Fg.2 stuaton 1) that the best varant would be a coalton of all players ( A+B+C+D+E ) wth a total addtonal proft equal to (p. u.). As a result of proposed optmzaton, the prmary reserve control functon s redstrbuted from more effectve power staton to a less one (wth a hgher producton cost). The total value of the prmary reserve sn't changed. The maxmum addtonal proft arses n case when the prmary reserve was not actvated n specfc tme nterval. Usng equatons (9) and (10), t s possble to calculate the sharng of proft among the players. The results of the sharng are presented n Table. 2 V. APPLICATION OF SMART LOAD SHEDDING SYSTEM Severe system dsturbances can result n fast frequency drop, whch makes fast governor and boler response mpossble. If the governor acton cannot actvate spnnng reserve quckly enough to restore the system to ts normal operatng frequency, underfrequency load sheddng (UFLS) serves as a last-resort tool to prevent the system from collapse. In most power systems up-to-date automatc load sheddng systems practcally foresee dsconnecton of the load at underfrequency wthout tme delay or wth small delay [6]-[8]. The numbers of load sheddng steps and the value of load to be shed vary for dfferent power systems. Some power systems use rate-of-change of frequency as addtonal factor to shed a load [9]. Fg. 3 presents an example of a frequency varaton durng operaton of the frequency actuated load sheddng system. The pont f < f nom corresponds to the moment of actve power defcency appearance n the power system. From ths moment the frequency drop starts. When the power system frequency reaches the level of a frst load sheddng settng f set1, the frst part of the load s dsconnected. The next part of the ISBN:

4 load wll be dsconnected when frequency reaches second load sheddng settng f set2. Wth each next load sheddng step rate-of-change of frequency declne s cut down and after a certan moment the ncrease of frequency takes place. Such logcs of the load sheddng operaton apply to most of the power system utltes [9]. f nom f set1 f set2 f set3 f set4 f set5 f [Hz] load sheddng 2 1 t [s] Fg.3 power system frequency varatons durng the UFLS operaton UFLS schemes can be categorzed nto three groups [3]: 1. the tradtonal UFLS schemes; 2. the sem-adaptve UFLS schemes; 3. the adaptve UFLS schemes. Exstng UFLS automaton has drawbacks, whch lmt the adaptablty of emergency stuaton control to a change of underfrequency stuaton n a power system. UFLS trppng frequency settngs are selected for some specfc emergency stuaton, whch s consdered as more probable for a specfc power system. It s not possble to foresee all stuatons that can occur n the power system. UFLS operaton wll be secure effectve only for the pre-calculated emergency cases. Problems related to the value of a load to be shed are very topcal. Redundantly trpped load can create overfrequency stuatons, whch sometmes s more dangerous than underfrequency. Mentoned stuaton s presented n Fg.4. The overfrequency stuaton after UFLS operaton happens because the total dsconnected load of the steps of UFLS s two tmes as large as the defcency of actve power n the network [9]. A. Analyss of Frequency Behavor for Dfferent Algorthms of UFLS Authors nvestgated frequency behavor durng sever dsturbances n unted ENTSO-E power system and IPS/UPS power system of Russa. Approprate model was developed wth the real settngs of load sheddng automaton. Dfferent varants of frequency behavor were nvestgated [9]. Fg.4 frequency behavor for 10.8% of actve power defcency As an example frequency behavor at a load defcency of 20% and an teratve defcency of the actve power of 6.7% s shown n Fg.5. Durng cascaded event the stablzaton of frequency can occur at a dangerously low level wthout beng notced by UFLS. It can happen f the actve power dsconnected by UFLS s not suffcent to return frequency back to permssble range. In the consdered case the frequency stays at 48.9 Hz. The reason for that s dfferent load sheddng phlosophy of two nterconnected power systems. Fg.5 frequency behavor for 26.7% of actve power defcency B. Applcaton of Smart Technology Approach to the Load Sheddng System Let us return to example of Fg.1 and explan behavor of load sheddng automaton. Each dstrct s equpped wth an nteractve power measurng devce (a smart meter) [10]-[14]. The nformaton center Operator receves full nformaton about the current condton of the consumpton of actve power n each power dstrct, about the locaton and value of the power defcency that has arsen. In that way, an nteractve nformaton system between the dstrcts load and emergency automaton s set up. Usng descrbed meterng system the automaton operatng process (let us call t smart underfrequency load sheddng system SUFLS) can be presented by few calculaton cycles: 1. Determnaton of the value of the defcency. Transformed rotor swng equaton can be used for calculaton of ths defcency [6], [7]: ISBN:

5 df f P TJ f kload, (11) dt k where T J rotor's nerta constant; k gov governor speed droop; k Load load-dampng constant; f frequency. 2. Memorzaton of the value of defcency and ts locaton; 3. Calculaton of the number of substatons to compensate the defcency; 4. Calculaton the optmum varant for load dsconnecton. To compare the results of the operaton of UFLS and SUFLS automaton, a mathematcal model has been constructed by usng Matlab/Smulnk software. Fg.6 llustrates frequency behavor n the case of emergency stuaton for exstng UFLS and smart SUFLS automaton. gov Fg.6 frequency behavor n the case of emergency stuaton The more effectve operaton of SUFLS automaton s obvous. The more effectve approach s to actvate load sheddng automaton n dstrcts, where power defcency takes place. VI. CONCLUSION Smart meterng systems can be used for both small frequency varatons and deep frequency declne operaton. Optmal dstrbuton of the prmary reserve can be based on cooperatve game theory method. Proposed algorthm, whch takes nto account techncal lmtatons and economc aspects of the prmary frequency control partcpants, has to be used. Results of practcal calculaton of addtonal proft prove feasblty of applcaton of game theory method for optmal dstrbuton of frequency control reserves. The mathematcal model of the power system wth proposed calculaton algorthm can be performed usng computer program Matlab Smulnk. Integraton of large power systems wth dfferent phlosophy of underfrequency load sheddng systems can cause neffectve and sometmes not selectve frequency control. A new load sheddng method s suggested. Smulaton of frequency behavor was conducted for exstng load sheddng system and a new one. ACKNOWLEDGMENT Ths work has been supported by the European Socal Fund wthn the project «Support for the mplementaton of doctoral studes at Rga Techncal Unversty». REFERENCES [1] Seventh Framework Programme, Energy.7.2, Collaboratve Project - Large-scale ntegrated project PEGASE: Pan European Grd Advanced Smulaton and state Estmaton, Grant agreement no.: [2] Collaboratve Project ICOEUR: Intellgent Coordnaton of Operaton and Emergency Control of EU and Russan Power Grds, Grant Agreement no.: [3] U. Rudez, R. Mhalc, Analyss of Underfrequency Load Sheddng Usng a Frequency Gradent, IEEE Trans. Power Syst., vol. 26, no. 2, pp , Apr., [4] V. Chuvychn, N. Gurov, S. Venkata, R. Brown, An adaptve approach to load sheddng and spnnng reserve control durng underfrequency condtons, IEEE Trans. Power Syst., vol. 11, no. 4, pp , Nov [5] Lloyd S. Shapley. A value for n-person games. In contrbuton to the Theory of Games, Volume II, by H.W. Kuhn and A.W.Tucker, edtors. Annals of Mathematcal Studes v.28, pp Prnceton Unversty Press. [6] V. Chuvychn, N. Gurov, S. Kene, Applcaton of New Emergency Control Prncple n Power Systems, Proceedngs of IEEE PowerTech2009 Conference, Romana, Bucharest, 29 June 2 July, 2009 pp 1-6. [7] L. Sternnson, Transent processes n power systems durng operaton of frequency and power, Moscow: Energy, 1975 (n Russan). [8] A. Barzam, System s automaton, Energoatomzdat, 1986 (n Russan). [9] A. Sauhats, V. Chuvychn, V. Strelkovs, R. Petrchenko, E. Antonov, Underfrequency Load Sheddng n Large Interconnecton, PowerTech2013 conference, Grenoble, France, June [10] M. Larson, C. Rehtanz, Predectve frequency stablty control based on wde-area phasor measurements, n Proc IEEE Power Eng. Soc. Summer Meetng, Chcago, IL, Jul. 2002, vol. 1, pp [11] H. Seyed, M. Sanaye-Pasand, New centralzed adaptve load-sheddng algorthms to mtgate power system blackouts, IET Gener. Transm. Dstrb., vol. 3, no. 1, pp , Jan [12] F. Plo, G. Psano, G. G. Soma, Advanced DMS to Manage Actve Dstrbuton Networks, IEEE Bucharest Power Tech Conference, Bucharest, Romana, 29 June 2 July, [13] H. Lemmen, SMART Transmsson System presentaton, IEEE Smart Grd World Forum, Brussels, Belgum, 2-3 december, [14] O. Samuelsson, M. Hemmngsson, A. H. Nelsen, K. O. H. Pedersen, J. Rasmussen, Montorng of power system events at transmsson and dstrbuton level, IEEE Trans. Power Syst., vol. 21, no. 2, pp , May, Vladmr Chuvychn (M'1979, SM'1990) was born n Russa on January 17, He receved dploma engneer degree n 1965, Canddate of Techncal Scence degree (Ph.D) n 1975 and Dr.habl.sc. degree n 1997 from Rga Techncal Unversty, Rga, Latva. Snce 1965 he s wth Rga Techncal Unversty, Faculty of Electrcal and Power Engneerng where s currently Professor. He was a vstng Fulbrght Scholar at the Unversty of Texas, Arlngton, n 1990 and at the Unversty of Washngton, Seattle n Hs research nterests nclude ISBN:

6 protectve relayng and power system automaton and control. He holds many patents n ths area. Antans Sauhats receved Dpl. Eng., Cand.Techn.Sc., and Dr.hab.sc.eng. degrees from Rga Techncal Unversty (former Rga Polytechnc Insttute) n 1970, 1976, and 1991 respectvely. Snce 1991 he s Professor at Electrc Power Systems. Snce 1996 he s the Drector of the Power Engneerng Insttute of Rga Techncal Unversty. Area of research actvty: power system automaton and optmsaton, protectve relayng, development of methods and means for control of normal and emergency condtons n electrc power systems, HV Transmsson lnes fault locaton. He has more than 200 scentfc publcatons. Prncpal nvestgator of the Latvan Scentfc Councl grants; Prncpal nvestgator, projects wth Latva, Lthuana, Estona, Russa power systems and utltes. Developed and mplemented dsturbance recordng systems, fault locaton system, protectve relayng, out of step condton system for kv transmsson lnes and power plants. Developed hardware and software wdely used power systems n Latva, Lthuana, Estona, and Russa. Total number of mplementaton more than 500 protectve relayng and automaton termnals. Correspondng Member, Latvan Academy of Scences, Awarded wth Year Prze of the Latvan Academy of Scences and Publc Jont Stock Company "Latvenergo" for lfe contrbuton n energetcs and engneerng, Roman Petrchenko receved B.Sc. degree n electrcal engneerng from the Rga Techncal Unversty (RTU) n 2006 and 2008, from 2010 he s PhD student at RTU, Latva. Hs research nterests nclude power system automaton and optmsaton, smart grd technologes. He s author of more than 15 papers. He s IEEE student member. Galna Bochkarjova receved Electrcal engneer degree from the RTU n At present she s Research Engneer at Faculty of Power and Electrcal Engneerng. Area of research actvty: protectve relayng, power system automaton, development of methods and means for control of emergency condtons n electrc power systems, HV Transmsson lnes fault locaton. She has 37 scentfc publcatons. ISBN: