Variable Energy Resource Induced Power System Imbalances: Mitigation by Increased System Flexibility, Spinning Reserves and Regulation

Transcription

1 Varable Energy Resource Induced Power System Imbalances: Mtgaton by Increased System Flexblty, Spnnng Reserves and Regulaton Aramazd Muzhkyan Masdar Insttute of Scence and Technology P.O. Box Abu Dhab, UAE Emal: Amro M. Fard Masdar Insttute of Scence and Technology P.O. Box Abu Dhab, UAE Emal: Kamal Youcef-Toum Massachusetts Insttute of Technology Cambrdge, MA, USA Emal: Abstract The mpact of varable energy resources (VER) on power system reserve and regulaton requrements has been a subject of extensve research n recent years. However, the conclusons about the scale of the mpact dverge, snce most of the results are obtaned from specfc case studes. Ths paper proposes a generalzed approach to the assessment of power system reserve and regulaton requrements. It uses a power system enterprse model that conssts of three layers: the physcal grd, resource schedulng and balancng operatons. Resource schedulng s modeled as a securty-constraned untcommtment (SCUC) problem. The balancng layer conssts of three components, namely the regulaton servce, the real-tme market and operator manual actons. The real-tme market s mplemented as a securty-constraned economc dspatch (SCED) problem. The IEEE RTS96 relablty test system s used for the physcal layer. Three man resources contrbutng to the balancng of power system are studed: reserves, regulaton and generator rampng rates. Ther mpacts on power system mbalance mtgaton n the presence of VER s studed. Index Terms Power system mbalances, wnd ntegraton I. INTRODUCTION The mpact of varable energy resource (VER) penetraton on power system operatons and plannng has been the subject of extensve research n recent years [1], [2]. Here, sgnfcant wnd and solar penetraton adds new levels of varablty and uncertanty to power systems; thus mpedng balanced operatons as defned by the North Amercan Electrc Relablty Corporaton (NERC) s control performance standards (CPS). The man concluson of such studes s that renewable energy ntegraton ncreases power system reserves and regulaton requrements. However, most of these results are based on specfc case studes that commedsafnt on the suffcency of a power system s flexblty n relaton to VER penetraton [3], [4], thus lmtng ther generalzablty [5]. Consequently, a recent revew has motvated the need for holstc assessment methods [6]. To that effect, ths paper seeks to develop a methodology of power system mbalance mtgaton characterzaton n the case of VER penetraton. It draws as nspraton the concept of ntegrated enterprse control n whch both physcal as well as enterprse processes are modeled to gan an understandng of the holstc system behavor [7]. In such a way, the varablty of renewable energy resources can be vewed as an nput dsturbance whch the (enterprse) power system systematcally manages to delver attenuated power system mbalances. The power system s modeled n terms of ts key characterstcs: flexblty (rampng capabltes of conventonal generaton unts), power spnnng reserve and regulaton capacty. Furthermore, the enterprse power system modelng ncludes three layers, namely resource schedulng, balancng actons, and a physcal layer whch represents the buses, transmsson lnes, loads and generators. Whle t s not possble wthn the scope of ths paper to model all enterprse power system processes, the ones most relevant to power system mbalances are captured: unt commtment, regulaton servce, real-tme market, and operator manual actons. Ths paper s the second of two to systematcally study power system mbalances. The frst paper [8] addresses the development of an assessment methodology for power system mbalances. Such an approach facltates makng casendependent conclusons. Ths paper focuses on assessng a power system s mbalance mtgaton capabltes based upon ther physcal as well as enterprse control characterstcs. Imbalances are presented n terms of the correspondng levels of CPS complance. A. Power System Imbalances II. BACKGROUND Power system relable operatons requre the balance of generaton and consumpton. The msmatch between generaton and consumpton also creates a msmatch between mechancal and electrcal torques appled on generators. If generaton exceeds consumpton, generators accelerate, and the system frequency ncreases. Lkewse, f consumpton exceeds generaton, generators decelerate, and the system frequency decreases. Devatons of the system frequency from the rated value may have a damagng effect on power system equpment, that s desgned for workng at a specfc frequency.

2 The nstantaneous power balance n the system wth F frequency devaton s gven by the followng equaton [9]: P m P D = D F F + dw k (1) dt where P m and P D are changes n mechancal power and electrcal power demand correspondngly, D F s the aggregated dampng parameter of generators, and dw k /dt s the rate of change n knetc energy. For a specfc control area power generaton and consumpton mbalance s called area control error (ACE). ACE s a crucal parameter of power system mbalance montorng and performance assessment. For steady-state smulatons, dw k /dt s zero, and Equaton (1) takes the followng form: ACE = D F F (2) Equaton (2) s used to measure ACE n real power systems. Frst, the devaton of the system frequency from rated value s measured, then ACE s calculated from (2). Measurements Unt Commtment Fg. 1. Regulaton Conventonal Generaton VER Generaton Spnnng Reserves Resource Schedulng Unt schedulng Balancng Actons RT Market Physcal Grd Transmsson System Reserve schedulng Manual Actons Resdental Loads Commercal Loads Industral Loads Conceptual model of power system operatons Commtment Decsons Dspatch Decsons B. Imbalance Mtgaton Power systems always experence mbalances due to forecastng uncertantes, generator outages, equpment falures and other contngences. Any mbalance trggers a counter acton from the power system to mtgate t. Tradtonally, the power system dynamcs are classfed as a herarchy of dynamcs: prmary, secondary and tertary [?]. Prmary dynamcs and control address transent stablty phenomena n the range of Hz [?]. These represents the nertal response of generators and loads and may be controlled by generator output adjustments mplemented by automatc generaton control and automatc voltage control [9]. Secondary and tertary control are managed by ndependent system operators and balancng authortes and are the man focus of ths work. Detaled descrptons of these technques are presented n the next secton. C. Control Performance Standards Implementaton of all balancng actons, however, does not elmnate mbalances completely. As a matter of fact, deal generaton and consumpton balance s not requred. The North Amercan Electrc Relablty Corporaton (NERC) defnes the requrements of mbalance mtgaton, called control performance standard (CPS) [10]. Accordng to Standard BAL a, each balancng authorty shall operate such that ts average ACE for at least 90% of clock-ten-mnute perods (6 non-overlappng perods per hour) durng a calendar month s wthn a specfc lmt L 10. The value of L 10 s obtaned from the system propertes. Thus, ths standard provdes a tool for balancng performance assessment. III. METHODOLOGY AND SIMULATION SETUP The power system enterprse model, used n ths study, s comprsed of three nterconnected layers: the physcal grd, resource schedulng and balancng actons. The conceptual dagram of the model s presented n Fg. 1. Resource schedulng occurs pror to the operatng day. At ths pont, the optmal set of generaton unts for the operatng day s defned, along wth ther schedules. Also, durng the resource schedulng, the requred reserve amount s scheduled. The scheduled resources are then managed n the real-tme to mantan power system balance. Three balancng actons, namely regulaton, real-tme market and operator manual actons, contrbute to the balancng of the system. Separate components of the model are descrbed n detal n the followng subsectons. A. Resources Schedulng Relable operatons of the power system start one day pror the operatng day wth the schedulng of necessary resources. Accordng to Fg. 1, t accomplshes three parallel actons, namely: unt commtment, unt schedulng and reserve schedulng. 1) Generaton Commtment and Schedulng: The ultmate goal of the schedulng layer s to choose the rght set of generaton unts, that are able to meet real-tme demand requrements wth mnmum cost, and defne ther schedules. Ths procedure s mplemented va a software-based optmzaton tool, called Securty-Constraned Unt commtment (SCUC), whch has a mathematcal formulaton n the form of a lnear mxed-nteger

3 program [11]: mn s.t. 24 t=1 (w,t C F + C G + w u,tc U + w d,tc D ) (3) P,t G = Pt D (4) T apple P,t G P,t G 1 apple T (5) w,t P G,mn apple P,t G apple w,t (6) w,t = w,t 1 + w,t u w,t d (7) w,t P,t G P res (8) where the followng notatons are used: C F,C G,C U,C D P D t,p G,mn T w,t w,t,w u,t d P res fxed, generaton (fuel), startup and shutdown costs of generator power output of generator at tme t total demand at tme t max/mn power lmts of generator maxmum rampng rate of generator schedulng tme step, normally, 1 hour number of generators ON/OFF state of the generator startup/shutdown ndcators of generator system reserve requrements The objectve of ths model s to fnd the optmal set of generators and ther schedules, that wll meet the demand wth mnmal total operatng cost of the system. Constrant (4) s the power balance equaton, requred to keep the system balanced. The other two constrants, (5) and (6), are the physcal lmtatons on generators rampng rates and power outputs respectvely. The soluton to ths mathematcal program yelds optmal w,t commtment and P,t G generaton schedules for each generator. 2) Adequacy of the Scheduled Generaton: The unt commtment model takes power system characterstcs, such as generaton levels, rampng lmts, and demand forecasts as nputs. Later, the schedulng of resources s performed to satsfy those requrements. However, the scheduled resources rarely match the real-tme requrements due to a seres of objectve lmtatons. Three man factors can be dstngushed as potentally causng ths msmatch. Frst, the nformaton about real-tme power demand s lmted to the day-ahead demand forecast. Even the best possble forecastng technques are able to provde the demand forecast wth lmted accuracy.. The resultng forecast error s the power msmatch that emerges durng real-tme operatons. The second factor s the dfference between resource schedulng and real-tme balancng tme steps. Normally, the unt commtment model schedules resources wth a 1 hour tme resoluton. Durng these 1 hour ntervals, the generaton schedules does not change, whch makes t practcally mpossble to match the scheduled generaton to the varatons of the load. Ideally, the presence of a perfect forecast may only guarantee matchng the scheduled generaton and the average demand for 1 hour ntervals. The devatons of the actual demand from ts average value on 1 hour ntervals are referred to as ntra-hour varatons. The thrd factor s lossless transmsson assumpton reflected n constrant (4). Generally, as a drect consequence of the power flow equatons, the power balance conssts of three components, namely generaton, demand and transmsson lne losses [12]: P,t G = Pt D + P LOSS t (9) where Pt LOSS s the total transmsson loss n the system. Excluson of the loss component s necessary to mantan a lnear SCUC model. As a result, some porton of the scheduled generaton s consumed by transmsson lnes and does not fll the demand. 3) Reserve Schedulng: The dscusson of the prevous secton establshes the objectve lmtatons that make matchng the scheduled generaton and the actual demand mpossble. Normally, ths ssue s easly solved by the procurement of reserve generaton n addton to the scheduled generaton. The reserve requrements are settled based on the exstng standards and the experence of power system operators. Constrant (8) n the unt commtment model reflects the reserve procurement, where P res s the reserve requrement. One mportant fact s neglected n the formulaton of the unt commtment problem, however. Generally, the schedulng process has two components. The schedulng of generaton and the schedulng of the necessary rampng capabltes are reflected n the constrants (4) and (5) respectvely. The symmetry of these two components becomes more obvous f constrant (5) s splt nto two separate constrants as follows: R G,t = 1 / T (10) R G,mn apple R G,t apple (11) where R,t G s the dfference between consecutve values of scheduled generaton and can be referred to as scheduled rampng. As a result, the dfferences between day-ahead generaton schedulng and real-tme requrements are also reflected n the rampng doman. Real-tme rampng requrements tend to be always dfferent from the day-ahead scheduled value. However, the provson of rampng reserves s not ncluded n the unt commtment model. In the case of rampng schedulng, t s assumed that the commtted generaton unts have enough rampng capabltes to follow varatons of the load. Ths assumpton s vald for tradtonal power systems wth slowly changng load. However, the ntegraton of VER s lkely to ncrease the varablty of the net load. As a result, the system may chroncally suffer from not enough rampng capabltes.

4 To overcome ths problem, the provson of rampng reserves s ncorporated nto an enhanced unt commtment model for use n ths study. After the proposed modfcatons, the unt commtment model transforms nto the followng form: mn s.t. 24 t=1 w,t C F + C G + w u,tc U + w d,tc D (12) P,t G = Pt D (13) R G,t = 1 / T (14) w,t P G,mn apple P,t G apple w,t (15) apple R,t G apple (16) w,t = w,t 1 + w,t u w,t d (17) w,t P,t G P res (18) w,t w,t P G,mn R G,t w,t + R,t G P res (19) R res (20) R res (21) where constrants (20) and (21) are for rampng up and rampng down reserves respectvely. Also, the ntroducton of rampng reserve requrements gves a lever by whch to vary the flexblty of the system. Ths s mportant for testng the mpact of system flexblty on mbalance mtgaton. 4) Reserve Requrements and Scheduled Reserves: The reserve procurement va constrant (8) n the unt commtment guarantees that the actual amount of reserves s greater or equal to the reserve requrement P res. The left-hand-sde s the dfference between the maxmum avalable generaton capacty and the total scheduled generaton, whch s, by defnton, the actual amount of avalable reserves. Snce the left-hand-sde of constrant (8) can only change dscretely wth the commtment of new generaton unts, the actual reserves usually exceed the mnmum requred amount. The dfference between the avalable reserves and the mnmum requrement depends on dfferent factors, such as the generaton portfolo of the system, the demand level, etc. Generally, ths dfference may have any value n a range from 0 to the largest generaton unt capacty. Ths knd of stuaton may sgnfcantly affect the results of VER ntegraton studes. For example, n power systems wth mostly large generaton unts, avalable reserve capacty s lkely to exceed mnmum reserve requrement sgnfcantly. Ths creates the possblty of VER ntegraton nto the power system wthout any addtonal requrements. As a result, one may conclude, that the ntegraton of VERs nto the system brngs no addtonal requrements on reserves, whch s not generally true. To ncrease generalzaton capabltes of the results n the current study, t s assumed that the actually avalable reserves and the mnmum requrements always match. In other words, there s no extra reserve capacty n addton to the requrement. Obvously, ths s the worst case scenaro, that shows how bg the mbalances can be and how low CPS can drop. Ths approach reduces the case-dependency of results. The quantty of actual reserves s lmted by manpulatng the maxmum generaton levels of scheduled generators after the SCUC program has completed. Frst, the scalng factor s defned as a rato of the requred reserve to the actually scheduled reserves: P t = NP G P res w,t (22) Equaton (22) shows that the scalng factor depends on tme. Ths s because the unt commtment schedules a dfferent amount of reserves for dfferent tme ntervals. The scalng factor s then used to change the maxmum output of generators, so that the avalable reserves equal the reserve requrements:,t = + P t (23) Note, that the maxmum outputs of generators,t used n the real-tme market change over tme. Whle ths s a nonphyscal stuaton, t s requred to demonstrate the mpact of ncreased reserves on the system mbalances. Falng to do so would cause the quantty of actual reserves dependng on the nequalty n the SCUC program. The same reasonng apples to the schedulng of rampng reserves. The scalng factor for rampng reserves s defned smlar to generaton reserves: R t = NP G R res w,t R G,t (24) Accordngly, the adjusted maxmum rampng rates of the generators are gven by the followng equaton:,t = R,t G + t R R,t G (25) Unlke generaton and rampng reserves, the provson of regulaton reserves s market based. The requrements are announced n the market, and the generators wth the cheapest offers are chosen. As a result, the actually avalable regulaton reserves always match the requrement. B. Balancng Actons Day-ahead resource schedulng defnes the set of generators avalable for the operatng day. Durng the operatng day, scheduled resources are managed n real-tme to ensure balancng of the system at any moment. Fg. 1 shows that balancng actons consst of three components operatng at dfferent tmescales: regulaton, real-tme market and operator manual actons. Each component s descrbed n detal n further subsectons.

5 1) Measurement of the Power System State and Imbalances: Effectve balancng of the power system requres up-to-date nformaton about the system state. In real power systems, ths functon s carred out by the state estmator, whch provdes nformaton about bus voltages and angles, generaton and consumpton levels, and the ACE. Rather than ntroducng the complexty and errors assocated wth state estmaton mplementaton, power flow analyss s used to estmate the state of the system. The values of bus voltages and angles are obtaned at every smulaton step for the gven levels of generaton and consumpton. Ths approach s equvalent to perfect state estmaton. Referrng to Equaton 2, the ACE estmaton requres measurement of the system frequency devaton. However, for steady-state smulaton the concept of frequency s not applcable. Instead, a desgnated slack generator consumes the msmatch of generaton and consumpton to make steadystate power flow equatons solvable. Therefore, for steady state smulatons the power system mbalances are measured as the output slack generator [9]. 2) Regulaton Servce: Generally, the regulaton servce s represented by a dynamc model [12]. However, for steady state smulatons a smplfed model s mplemented. At a tme scale slower than 1 mnute, the effectve transfer functon smplfes to a gan wth saturaton lmts. Thus, n the current study, the regulaton s mplemented as follows. At each smulaton step, the regulaton servce responds to the mbalances by movng ts output to the opposte drecton. The regulaton output changes untl mbalance mtgaton or regulaton servce saturaton. 3) Real-Tme Market: The real-tme market operates n parallel to the regulaton servce to suppress mbalances, but n a slower tme-scale. It moves avalable generator outputs to new setponts n the most cost-effcent way. In ts orgnal formulaton, generaton re-dspatch s mplemented as a non-lnear optmzaton model, called AC optmal power flow (ACOPF) [13]. Due to problems wth convergence and computatonal complexty [11], most of the US ndependent system operators (ISO) moved from ACOPF to lnear optmzaton models. The most commonly used models s Securty-Constraned Economc Dspatch (SCED), formulated as an ncremental lnear optmzaton program [14]: mn s.t. (b P,t G +2c P,t G P,t) G (26) N B (1,t)( P,t G P,t) L =0 (27) N B a l,,t ( P L,t) apple F max l F l,t (28) R G t apple apple R G t (29) P G,mn apple apple (30) where the followng notatons are used: b,c generator offer curve lnear and quadratc coeffcents, P L,t bus ncremental generaton and load F l,t,f max l N B,t a l,,t t lne l power flow level and flow lmt number of buses bus ncremental transmsson loss factor bus generaton shft dstrbuton factor to lne l real-tme market tme step, normally, 5 mnutes. The use of ncremental values for generaton and load allows the ncorporaton of senstvty factors and the lnearzaton of the program. Senstvty factors establsh the lnear connectons between changes n power njectons on buses and state-related parameters of the system [15]. Two senstvty factors are used n the current model, namely the ncremental transmsson loss factor (ITLF) and the generaton shft dstrbuton factor (GSDF). ITLF for bus shows how much the total system losses wll change, f the njecton on bus ncreases by a unt [16]:,t t where Pt LOSS s total system loss at moment t, P,t s the power njecton at bus. The ncorporaton of ITLF nto the model results n a lnearzed power balance constrant (27). The GSDF shows how much the power flow n lne l changes f the njecton on bus ncreases by a unt [16], [17]: a l,,t l,t The ncorporaton of the GDSF nto the model results n a lnearzed lne flow lmt constrant (28). The other two constrants (29) and (30) are the physcal lmts of the generator rampng rates and outputs. The objectve functon n Equaton (26) s the mnmzaton of total generaton cost. The values for b and c are the lnear and quadratc coeffcents of generator offer curves submtted for day-ahead schedulng. Observaton of the model shows that some nput parameters, such as P,t G, F l,t,,t, a l,,t, depend on the current state of the system. These parameters are calculated before each SCED teraton based on a full AC power flow analyss n what are called hot start models n the lterature [14]. 4) Operator Manual Actons: In the normal operatng mode of the power system, the regulaton servce and the real-tme market are able to mantan the generaton and consumpton balance effectvely. The avalable generaton reserves and the generator rampng rates are enough to follow the slowly changng load. However, n the case of contngences, the stuaton changes. A sudden outage of a major generaton unt creates large mbalances that cannot be mtgated by realtme markets and regulaton servces. Onlne generaton unts do not have enough reserve capacty and enough rampng

6 rates to fll the gap quckly. These knds of stuatons requre system operator manual actons, n the form of deployment of contngency reserves, decsons on the locaton of actvated reserves, etc. Manual actons, unlke the other two components of balancng, have no dedcated operatons tmescale and are used as necessary. In the absence of operator models n the lterature, ther actons are mplemented as follows. The system mbalances are montored durng the nterval of smulatons. The trgger of operator manual nterventon nto the balancng procedure works, when the actual mbalances exceed 80% of the largest generaton unt. The actons of the operators nclude balancng of the system manually and brngng new generaton unts onlne to suppress mbalances. IV. CASE STUDY In ths case study, the mtgaton of power system mbalances by manpulatons of generaton, rampng and regulaton reserves s tested. Three dfferent mtgaton scenaros are smulated. Durng each scenaro, one of three reserves are changed n a reasonable range, whle others stay constant. Ths approach allows the study of the mpact of a partcular reserve on mtgaton of the mbalances. The performance of balancng actons s evaluated by complance to CPS s. Before these three scenaros, a tradtonal power grd wthout wnd ntegraton s smulated to determne the amount of reserves and regulaton that the tradtonal system needs to mantan the balance. Ths approach makes sure that any mbalances are the results of wnd ntegraton. The scenaros are mplemented as steady-state smulatons n the Matlab envronment. The smulatons run for one week perod wth a tme step of 1 mnute that corresponds to the regulaton tmescale. It s assumed that faster dynamcs are mtgated by system nerta and load response. A. Test System The IEEE RTS-96 relablty test system s used as the physcal grd model [18]. It s composed of three nearly dentcal control areas, wth a total of 73 buses and 99 generators. The yearly peak load s 8550MW. B. Incorporaton of Wnd and Load Data Wnd and load data from Bonnevlle Power Admnstraton repostores [19] are used for the current case study. The best avalable data has 5-mnute resoluton, whch does not satsfy the requrement of smulatons wth 1-mnute tme step. Ths dffculty s overcome by up-samplng the avalable data to 1- mnute resoluton. The up-samplng process s performed wth the use of snc functons to not ntroduce dstortons nto the power spectrum and not change the spectral wdth [20]. VER ntegraton nto the actual physcal system requres an allocaton to the system buses. Ths choce defnes potental congestons occurrences n the system, whch may sgnfcantly alter the results of mbalances for the same scenaro. Snce congestons are outsde the score of the research, the VER capacty dstrbuton s mplemented n a way that mnmzes congeston probabltes. Wnd generaton on each bus s proportonal to the system load on that bus. As a frst analyss, the temporal power generaton profle of each wnd turbne across the topology s assumed to be entrely correlated spatally. Future work can generalze ths model to nvestgate systematc approaches to partal geographc correlatons to confrm recent emprcal evdence n ths regard [21]. Due to the correlaton of temporal profles, all VER generaton sources have the same varablty and forecast error, whch allows these parameters to be addressed n aggregate. C. Smulaton Scenaros Three smulaton scenaros are developed to study the mtgaton of mbalances by enhancng three characterstcs of power system: generaton, rampng and regulaton reserves. Each smulaton scenaro s conducted by varyng a sngle parameter whle holdng all other parameters equal. Bref descrptons of each scenaro are presented n the followng paragraphs. Scenaro 0: Balancng of the tradtonal power system. The tradtonal power system s smulated wthout wnd ntegraton. Ths s the base case scenaro, and defnes reserve requrements for balancng the tradtonal system. Scenaro 1: Imbalance mtgaton by ncreasng generaton reserves ceters parbus. Scenaro 2: Imbalance mtgaton by ncreasng regulaton reserves ceters parbus. Scenaro 3: Imbalance mtgaton by ncreasng rampng reserves ceters parbus. V. SIMULATION RESULTS AND DISCUSSION VI. CONCLUSION REFERENCES [1] E. Ela, B. Krby, E. Lannoye, M. Mllgan, D. Flynn, B. Zavadl, and M. O Malley, Evoluton of operatng reserve determnaton n wnd power ntegraton studes, n Power and Energy Socety General Meetng, 2010 IEEE, 2010, pp [2] H. Holttnen, M. Mllgan, E. Ela, N. Menemenls, J. Dobschnsk, B. Rawn, R. J. Bessa, D. Flynn, E. Gomez-Lazaro, and N. K. Detlefsen, Methodologes to Determne Operatng Reserves Due to Increased Wnd Power, Sustanable Energy, IEEE Transactons on, vol. 3, no. 4, pp , [3] A. Robtalle, I. Kamwa, A. Oussedk, M. de Montgny, N. Menemenls, M. Huneault, A. Forcone, R. Malhot, J. Bourret, and L. Berner, Prelmnary Impacts of Wnd Power Integraton n the Hydro-Quebec System, Wnd Engneerng, vol. 36, no. 1, pp , Feb [Onlne]. Avalable: [4] T. Agner, S. Jaehnert, G. L. Doorman, and T. Gjengedal, The Effect of Large-Scale Wnd Power on System Balancng n Northern Europe, Sustanable Energy, IEEE Transactons on, vol. 3, no. 4, pp , [5] M. H. Albad and E. F. El-Saadany, Comparatve study on mpacts of wnd profles on thermal unts schedulng costs, pp , [6] A. M. Fard, The Need for Holstc Assessment Methods for the Future Electrcty Grd, n GCC CIGRE Power 2013, Abu Dhab, UAE, 2013, pp [7] P. Martn, The Need for Enterprse Control, InTech, vol. Nov/Dec, pp. 1 5, [8] A. Muzhkyan, A. M. Fard, and K. Youcef-Toum, Varable Energy Resource Induced Power System Imbalances : A Generalzed Assessment Approach, n IEEE Conference on Technologes for Sustanablty, Portland, Oregon, 2013, pp. 1 8.

7 [9] A. Gómez-Expósto, A. J. Conejo, and C. Cañzares, Electrc Energy Systems: Analyss and Operaton. Boca Raton, Fla: CRC, [10] North Amercan Electrc Relablty Corporaton, Relablty Standards for the Bulk Electrc Systems of North Amerca, NERC Relablty Standards Complete Set, pp. 1 10, [Onlne]. Avalable: Standards Complete Set 1Dec08.pdf [11] S. Frank and S. Rebennack, A Prmer on Optmal Power Flow: Theory, Formulaton, and Practcal Examples, Colorado School of Mnes, Tech. Rep. October, [12] F. Mlano, Power system modellng and scrptng, 1st ed. New York: Sprnger, [13] J. Carpenter, Contrbuton to the economc dspatch problem, Bull. Soc. Franc. Electr., vol. 3, no. 8, pp , [14] B. Stott, J. Jardm, and O. Alsaç, DC Power Flow Revsted, Power Systems, IEEE Transactons on, vol. 24, no. 3, pp , [15] A. J. Wood and B. F. Wollenberg, Power generaton, operaton, and control, 2nd ed. New York: J. Wley & Sons, [Onlne]. Avalable: [16] K.-S. Km, L.-H. Jung, S.-C. Lee, and U.-C. Moon, Securty Constraned Economc Dspatch Usng Interor Pont Method, n Power System Technology, PowerCon Internatonal Conference on, 2006, pp [17] L. S. Vargas, V. H. Quntana, and A. Vannell, A Tutoral Descrpton of an Interor Pont Method and Its Applcatons to Securty-Constraned Economc Dspatch, Power Systems, IEEE Transactons on, vol. 8, no. 3, pp , [18] C. Grgg, P. Wong, P. Albrecht, R. Allan, M. Bhavaraju, R. Bllnton, Q. Chen, C. Fong, S. Haddad, S. Kuruganty, W. L, R. Mukerj, D. Patton, N. Rau, D. Reppen, A. Schneder, M. Shahdehpour, and C. Sngh, The IEEE Relablty Test System A report prepared by the Relablty Test System Task Force of the Applcaton of Probablty Methods Subcommttee, Power Systems, IEEE Transactons on, vol. 14, no. 3, pp , [19] Bonnevlle Power Admnstraton, WIND GENERATION & Total Load n The BPA Balancng Authorty. [Onlne]. Avalable: [20] A. V. Oppenhem and R. W. Schafer, Dscrete-Tme Sgnal Processng, 3rd ed. Prentce Hall, [21] M. Mllgan, K. Porter, E. DeMeo, P. Denholm, H. Holttnen, B. Krby, N. Mller, A. Mlls, M. O Malley, M. Schuerger, and L. Soder, Wnd power myths debunked, Power and Energy Magazne, IEEE, vol. 7, no. 6, pp , [Onlne]. Avalable: