Optimal Sizing and Siting of Multi-purpose Utility-scale Shared Energy Storage Systems
Narayan Bhusal, Mukesh Gautam, Mohammed Benidris, Sushil J. Louis
OOptimal Sizing and Siting of Multi-purposeUtility-scale Shared Energy Storage Systems
Narayan Bhusal*,
Student Member, IEEE , Mukesh Gautam*,
Student Member, IEEE ,Mohammed Benidris*,
Member, IEEE , and Sushil J. Louis**,
Member, IEEE ,*Department of Electrical and Biomedical Engineering,**Department of Computer Science and Engineering,University of Nevada, Reno, NV 89557, USAEmails: [email protected], [email protected],[email protected], and [email protected]
Abstract —This paper proposes a nondominated sorting geneticalgorithm II (NSGA-II) based approach to determine optimalor near-optimal sizing and siting of multi-purpose (e.g., voltageregulation and loss minimization), community-based, utility-scaleshared energy storage in distribution systems with high penetra-tion of solar photovoltaic energy systems. Small-scale behind-the-meter (BTM) batteries are expensive, not fully utilized, and theirnet value is difficult to generalize and to control for grid services.On the other hand, utility-scale shared energy storage (USSES)systems have the potential to provide primary (e.g., demand-side management, deferral of system upgrade, and demandcharge reduction) as well as secondary (e.g., frequency regulation,resource adequacy, and energy arbitrage) grid services. Under theexisting cost structure, storage deployed only for primary purposecannot justify the economic benefit to owners. However, deliveryof storage for primary service utilizes only 1-50% of total batterylifetime capacity. In the proposed approach, for each candidateset of locations and sizes, the contribution of USSES systems togrid voltage deviation and power loss are evaluated and diversePareto-optimal front is created. USSES systems are dispersedthrough a new chromosome representation approach. From thelist of Pareto-optimal front, distribution system planners will havethe opportunity to select appropriate locations based on desiredobjectives. The proposed approach is demonstrated on the IEEE123-node distribution test feeder with utility-scale PV and USSESsystems.
Index Terms —Multi-use battery storage; NSGA-II; photo-voltaic; power loss; utility-scale shared energy storage.
I. INTRODUCTIONEnergy storage systems have become essential componentsto modern distribution systems to overcome technical andoperation challenges introduced by renewable energy sources(RESs). Also, falling prices of batteries and incentive programshave led to their wide adaption [1]. California Public UtilitiesCommission has approved a new target—1.3 gigawatt (GW),out of which 200 megawatt (MW) on the customer-side needsto be installed in the state by 2020—for energy storage systems[2]. Large utilities, such as Southern California Edition (SCE)and Pacific Gas and Electric (PG & E), have received approvalfrom California public utility service to build storage facilities(195 MW for SCE and 567.5 MW for PG&E) [1]. Also,Arizona Public Service (APS) is planning to install 850 MWstorage by 2025 [1]. The economic justification is vital forthese new installations of energy storages to be sustainable. Nevertheless, under the existing cost structure, economicbenefits of storage systems that are deployed for primarypurposes only (e.g., demand side management, deferral ofsystem upgrade, and demand charge reduction) cannot bejustified as these storages utilize only – of total batterylifetime capacity [3].Behind-the-meter (BTM) energy storage systems are ex-pensive and not fully utilized due to the single-use andbeing dispersed on large geographical areas [3]. Also, itis difficult to coordinate between a large number of BTMenergy storage systems to provide grid services. On the otherhand, utility-scale shared energy storage (USSES) systemshave the potential to provide grid services and increase theutilization of photovoltaic (PV) systems and other RESs. Theyalso provide opportunities for owners of distributed energyresources (DERs) to lease parts of the storage instead ofbuying individual energy storage systems.Development of multi-use strategies and models for commu-nity based USSES has become critical for both customers andutilities. Also, new multi-use business models which utilizebattery storage for both primary and secondary (e.g., frequencyregulation, resource adequacy, and energy arbitrage) servicesare needed to increase economic benefits for both USSESowners and investors [3], [4]. Furthermore, determining op-timal locations and sizes of USSES is an important factor toprovide grid services. Optimal locations and sizes of energystorage system (ESS) can be used for several grid services suchas improving power factor and serving demand for peak timeperiod, improving voltage profiles and reducing the power loss,controlling high energy imbalance charges, and improvingpower quality and system reliability [5].Several studies have been conducted to assess the impor-tance of storage systems in providing grid services. In [6],energy storage systems have been used for peak shaving. Theauthors of [7] have utilized batteries for frequency regulationand peak shaving through a joint optimization framework. In[8], a battery storage system has been used to increase theutilization of RESs. A method to quantify the required energystorage to firm up wind generation has been proposed in [9]. In[10], the authors have used battery storage to defer upgradationof distribution feeders. Optimal sizes and locations of battery a r X i v : . [ ee ss . S Y ] A ug torage units have been determined for voltage regulation in[11]. Analysis of potential grid services that can be providedby USSES systems and potential technical and operationchallenges have been discussed in [3]. The economic viabilityof several battery storage technologies has been presented in[12], [13]. Benefits, applications, and technologies associatedwith utility-scale energy storage system have been provided in[14]. Calculation of stacked revenue and technical benefits ofa grid-connected energy storage systems has been presentedin [15].Numerous approaches have been proposed in the literaturefor optimal sizing and siting of distributed generation (DG)and energy storage systems (ESS). A review of models,methods, and future directions for distributed generation place-ment in distribution systems have been provided in [16]. Atwo-stage sequential Monte Carlo simulation (MCS)-basedstochastic strategy has been proposed in [17] to determinethe minimum size of movable energy resources (MERs) forservice restoration and reliability enhancement. Genetic algo-rithm (GA) and particle swarm optimization (PSO) have beencombined together in [18] for the optimal sizing and siting ofDGs to minimize power losses, improve voltage regulation,and enhance voltage stability. A GA based framework hasbeen presented in [19] for the optimal placement of ESS in ahigh wind integrated systems to minimize the operational cost.In [20], multi-objective combined PSO and non-dominatedsorting genetic algorithm (NSGA-II) based approach has beenproposed for optimal siting of ESS to minimize the operationalcost and improve the voltage profile. Authors of [21] haveproposed a GA based approach for optimal planning of ESSin smart grids to consider the cost sustained by asset ownersthrough total planning period.Although several approaches have been proposed for sizingand siting of ESSs, they focus on very large scale ESSfor transmission level system and for DG placement, whichhas different operational characteristics than USSES systems.Also, most of the literature have combined all the objectivefunctions to develop a single objective with various arbitrarychosen weights. Therefore, the proposed work is novel in thesense that it determines the optimal sizes and sites of multi-purpose USSES with leasing opportunities at the distributionlevel. Also, this paper emphasizes on the concept of multi-purpose community based USSES for the economic justifica-tion. In terms of the GA representation, the approach presentedin this paper disperses the resources rather than placing themin a single location. Dispersing resources have several benefitsspecially during the time of extreme disasters (natural disastersas well as man-made attacks) [22].This paper proposes an NSGA-II based approach to deter-mine the optimal or near optimal sizes and locations of USSESfor multiple purposes (voltage deviation and loss minimiza-tion). For each candidate set of locations, the contribution ofUSSES systems to grid voltage deviations and power lossesare evaluated and diverse Pareto-optimal front is created. Fromthe list of Pareto-optimal front, distribution system plannerswill have opportunity to select the appropriate sizes and sites based on the desired objectives. The proposed approach isdemonstrated on the IEEE 123-node distribution test feederwith utility-scale PV and USSES systems.The rest of the paper is organized as follows. Section IIdiscusses the problem of optimal sizes and sites of energystorage systems and their importance to grid services. SectionIII presents the development of the proposed approach. SectionIV examines the proposed approach through case studieson IEEE-123 node test system. Finally section V providesconcluding remarks.II. P ROBLEM F ORMULATION
Determining optimal sizes and locations of USSES systemscan be determined based on several factors such as minimiza-tion of operating costs, power losses and voltage deviations,improvement of power quality and reliability, and frequencyregulation. Power loss and voltage deviation minimization aretwo important operational measures which have significantimpact on both technical and economic aspects of distributionsystem operation. Therefore, proper consideration should begiven for minimizing power losses and voltage deviationswhile determining optimal sizes and sites of community basedUSSES.Power loss between bus j and k with photovoltaic (PV) atnode k can be computed as follows [23]. P Loss = R jk ( P j + Q j ) V j + R jk V j ( P P V + Q P V − P j P P V − Q j Q P V ) (cid:18) GL (cid:19) , (1)where P j and Q j are the injected real and reactive powerat bus j ; R ij is resistance of line segment i to j ; V j is thevoltage at node j ; P P V and Q P V are, respectively, the realand reactive power produced by a PV system at bus k ; G isthe distance from a bus with a PV system ( k ) to the source;and L is the total feeder length from the source to bus j .The total power loss of the feeder at the i th hour can becalculated as follows. P iT,Loss = N (cid:88) j =1 P iLoss ( j, j + 1) , (2)where N is the total number of line segments.Since the work presented here is for a planning purpose, theaverage of total power losses for a year is considered, whichcan be expressed as follows. P avg,Loss = (cid:80) i =1 P iT,Loss , (3)Similar to average power losses, voltage deviations can alsobe averaged over a year as follows. ∆ V avg = (cid:80) i =1 (cid:16) V imax − V imin (cid:17) , (4)where V imax and V imin are, respectively, the maximum andminimum system voltage at hour i .herefore, the objective function considered in this papercan be expressed as follows. F Minimize P T,Loss , (5) F Minimize ∆ V avg , (6)Subject to: (cid:88) S G,i − (cid:88) S L,i − T L s = 0 , (7) S minG ≤ S G ≤ S maxG , (8) S ij ≤ S maxij , (9) V mink ≤ V k ≤ V maxk , (10) B minSOC ≤ B SOC ≤ B maxSOC , (11)where (7) denotes the power balance equation ( S G,i , S L,i , and
T L s represents the generation, load, and transmission loss,respectively); (8) refers to the generation limits constraint; (10)represents voltage limits constraint; and (11) denotes batterystate-of-charge limits constraint.Since community based USSES systems are installed at thedistribution level, which have several candidate locations, theproblem becomes very challenging. For larger systems, thenumber of possible scenarios becomes dramatically very largeand more specifically when multiple community based USSESunits are needed. The complexity of the problem with fixedsizes and variable locations can be demonstrated using (12). S = L c ! n B !( L c − n B )! , (12)where L c is total number of candidate nodes in a feeder and n B is the number of USSES systems to be installed at thefeeder.In this work, we have considered both optimal sizing andsiting of USSES systems, which makes the problem more com-plicated. For example, to place , , or USSES systems in possible locations, the total number of combinations are,respectively, more than . × , . × , and . × .As this is a planning problem, to properly incorporate theload changing scenario, all simulation are run for a year withone hour time steps. Therefore, exhaustive search could takemonths find the optimal and complete solution. Therefore,in this work, GA has been adopted to find the optimal ornear optimal locations and sizes for community-based USSESsystems to minimize the power losses and voltage deviations.Several methods in the literature have used arbitrary chosenweighting factors to convert multi-objective problems into asingle objective function. Also, the problem must be solvedfor every change in the desired priority in weight relatedconsiderations. To deal with these problems, non-dominatedsorting genetic algorithm has been proposed in [24] whichprovides pareto-optimal fronts rather than one solution. Thesolution can be chosen from the pareto-optimal solutions basedon desired objectives. NSGA-II has been proven to be mosteffective for multi-objective optimization on a number ofpower system benchmark problems [25]–[27]. In [25], NSGA-II has been adopted to solve generation expansion planning problems. Electric distribution service restoration (minimizeout-of service area, minimize switching operation, minimizepower loss) has been performed using NSGA-II in [26].Authors of [27] have adopted NSGA-II for optimal DG sitingand sizing with storage systems and feeder reconfigurationeffects. III. S OLUTION A PPROACH
A. Unbalanced Power Flow and Simulation Environment
In this work, MATLAB and OpenDSS are integrated toperform the proposed work. In MATLAB, all the controlcommands and GA functions are performed which callsOpenDSS engine to perform unbalanced three-phase powerflow. OpenDSS provides all the monitored information back tothe MATLAB to perform the remaining tasks. The OpenDSSis an open source power system simulation environment fordistribution system simulation, which is developed by ElectricPower Research Institute (EPRI) [28]. The OpenDSS calcu-lates unbalanced power flow using Newton’s Method (notethat Newton’s Method implemented in OpenDSS is differentfrom the Newton-Raphson method).
B. Representation of the Proposed Approach
In the proposed problem, the USSES need to be dispersedin different locations rather than placing them in one singlelocation; thus, conventional representation techniques can notperform this task. Therefore, we propose a novel representationtechnique as shown in Fig. 1, where for N possible nodes, G , G , G .... G i ... G n − and G n are node numbers from to N . The locations obtained from these selections maynot be unique; therefore, unique locations are obtained usingGA numbers ( G , G , G .... G i ... G n − , G n ) in termsof L , L , L , L , ... L i , .., L n − , and L n . The uniquelocations are created based on the distance between two GAnumbers. In this process, all newly created unique numbersare checked for similarity and if they are similar, the distancebetween a particular number and the next number in clockwisedirection is determined. This process is repeated until thechromosome becomes completely unique. For example, if [ x , x , x , x , x ] = [ G , G , G , G , G ] is a chromosomeobtained after performing selection. The unique location canbe obtained as follows. The first term is obtained as x (cid:48) = G ,and rest of the terms are obtained from the distance betweenthem using the technique shown in Fig. 1. For example,second term x (cid:48) is distance between the number x and x . x (cid:48) is compared with all the elements of the newly createdchromosome (for x (cid:48) , here it has to compare with x (cid:48) ) andif it is similar the distance between x and x , the distancebetween x and x are added together to create the unique thenumber. This process is repeated until all the numbers in thenewly created chromosome are unique. The limitation of thisrepresentation is when chromosome is [ G n , G n , G n , G n , G n ] ,it becomes never ending loop, therefore, for this case a randomnumber is generator (less than G n ) and one of the number ofthe chromosome is replaced with the random number. L n L L L L L L L L L L i . . . . .. . . G n-1 G G G G G G G G G G i G n . . .. .. . Fig. 1. Representation of chromosome to avoid multiple USSES systems ina single location
150 149
34 8 13
20 19 1821
33 31
32 250
135 35 36
47 49 5051 151 300
152 52 53
54 565594
195 95 93 92 91
87 8690 88
160 68610 35073 74
69 70 716798 99 100
450 451111110106
102 103
72 7877
80 79
82 85 Fig. 2. IEEE 123 node distribution test system
IV. C
ASE S TUDIES AND R ESULTS
In order to validate the proposed approach, all simulationshave been performed on the IEEE-123 bus radial distributiontest feeder [29] using MATLAB and OpenDSS integratedenvironment. The IEEE-123 node test feeder, as shown in Fig.2, is characterized by having overhead and underground lines,four voltage regulators, four shunt capacitor banks, multiplesectionalizing and tie-switches, and unbalanced loading withconstant current, power, and impedance models. The total realand reactive loads of this system are, respectively, kWand kVar. Network data of the IEEE 123-node test feederare given in [30]. Ten PVs each of sized kVA ( kWat maximum power point tracking) are placed at nodes , , , , , , , , , and .In this paper, it is assumed that community based USSES Fig. 3. USSES profile developed using the combine profile of PV output andload demand. units are fully charged and discharged in every hours (asin (13)) to ensure maximum utilization of batteries. (cid:88) t =1 P tBatt = 0 , (13)where P Batt is charging (negative) or discharging (positive)power of a battery at any hour t of a day. Charging anddischarging profiles of community based USSES is drivedbased on expected energy produced from solar PV and loaddemand as shown in Fig. 3.The optimal sizes and sites of community based USSESare searched on the IEEE-123 feeder. For siting of USSES,out of nodes, only three phase nodes are selected(all three phase nodes are represented by black dot in Fig.2). For sizing of USSES, sizes ranging from to kWh are considered. The parameters used for performing theNSGA-II are as follows: two point mutation (one for sizeand one for site), two point crossover, binary representationwith total string length of ( for site and for sizing), population size, and generations (population size andnumber of generations are determined after some trial). Theobtained Pareto-optimal solutions are as shown in Table I andthe plot between two objectives (Voltage deviation and Powerloss) is as shown in Fig.4. The simulations are run for times (this number can be any number, we just checked for times, every time result were similar, therefore, we didn’tcheck further), in every simulation runs the sizes and sitesobtained for each cases are similar. From this we can drawthe conclusion that the obtained solution is good set of Pareto-optimal solution. V. C ONCLUSION
This paper has proposed an NSGA-II based approach todetermine optimal or near-optimal sizing and siting of com-munity based USSES system in solar photovoltaic integrated
ABLE IS
IZE AND SIZING OF
USSES
AND RESPECTIVE POWER LOSS AND VOLTAGE DEVIATION . Location 1 Location 2 Location 3 Location 4 Location 5 Size 1 Size 2 Size 3 Size 4 Size 5 Power Loss Voltage deviation
55 72 66 21 1 104.3988 230.2053 944.5748 102.6393 100 78.73578975 0.05071933355 72 66 21 80 104.3988 230.2053 944.5748 102.6393 100 85.87852289 0.05063234929 53 67 21 80 104.3988 230.2053 944.5748 102.6393 100 87.02726953 0.05060912155 72 66 21 1 104.3988 230.2053 944.5748 102.6393 100 78.73578976 0.05071933329 53 67 21 80 104.3988 230.2053 944.5748 102.6393 100 87.02726953 0.05060912156 76 66 21 7 104.3988 230.2053 946.3343 102.6393 100 81.26472466 0.05071907355 72 66 21 80 104.3988 230.2053 944.5748 102.6393 100 85.87852289 0.05063234955 72 66 21 1 104.3988 230.2053 944.5748 102.6393 100 78.73578975 0.05071933355 72 66 21 1 104.3988 230.2053 944.5748 102.6393 100 78.73578976 0.05071933329 52 66 21 83 104.3988 230.2053 944.5748 101.7595 100 87.24845573 0.05046052155 72 66 21 80 104.3988 230.2053 944.5748 102.6393 100 85.87852289 0.05063234955 72 66 21 1 104.3988 230.2053 944.5748 102.6393 100 78.73578976 0.05071933329 52 66 21 83 104.3988 230.2053 944.5748 102.6393 103.5191 87.95235634 0.0504674729 52 66 21 83 104.3988 230.2053 944.5748 102.6393 103.5191 87.95235634 0.0504674729 52 66 21 83 104.3988 230.2053 944.5748 102.6393 103.5191 87.95235634 0.0504674729 52 66 21 83 104.3988 230.2053 944.5748 102.6393 103.5191 87.95235634 0.0504674729 52 66 21 83 104.3988 230.2053 944.5748 102.6393 103.5191 87.95235634 0.0504674729 52 66 21 83 104.3988 230.2053 944.5748 102.6393 103.5191 87.95235634 0.0504674729 52 66 21 83 104.3988 230.2053 944.5748 102.6393 100 87.36678385 0.05046822629 52 66 21 83 104.3988 230.2053 944.5748 102.6393 103.5191 87.95235634 0.0504674755 72 66 21 80 104.3988 230.2053 944.5748 102.6393 100 85.87852289 0.05063234929 52 66 21 83 104.3988 230.2053 944.5748 102.6393 100 87.36678385 0.05046822655 72 66 21 80 104.3988 230.2053 944.5748 102.6393 100 85.87852289 0.05063234955 72 66 21 1 104.3988 230.2053 944.5748 102.6393 100 78.73578976 0.05071933355 72 66 21 1 104.3988 230.2053 944.5748 102.6393 100 78.73578976 0.05071933329 53 67 21 80 104.3988 230.2053 944.5748 102.6393 100 87.02726953 0.05060912129 53 67 21 80 104.3988 230.2053 944.5748 102.6393 100 87.02726953 0.05060912155 72 66 21 1 104.3988 230.2053 944.5748 102.6393 100 78.73578976 0.05071933355 72 66 21 1 104.3988 230.2053 944.5748 102.6393 100 78.73578976 0.05071933355 72 66 21 1 104.3988 230.2053 944.5748 102.6393 100 78.73578976 0.05071933355 72 66 21 1 104.3988 230.2053 944.5748 102.6393 100 78.73578976 0.05071933329 52 66 21 83 104.3988 230.2053 944.5748 102.6393 100 87.36678385 0.05046822655 72 66 21 80 104.3988 230.2053 944.5748 102.6393 100 85.87852289 0.05063234929 53 67 21 80 104.3988 230.2053 944.5748 102.6393 100 87.02726953 0.05060912129 52 66 21 83 104.3988 230.2053 944.5748 102.6393 103.5191 87.95235634 0.0504674755 72 66 21 80 104.3988 230.2053 944.5748 102.6393 100 85.87852289 0.05063234929 52 66 21 83 104.3988 230.2053 944.5748 102.6393 100 87.36678385 0.05046822629 52 66 21 83 104.3988 230.2053 944.5748 102.6393 100 87.36678385 0.05046822629 52 66 21 83 104.3988 230.2053 944.5748 102.6393 103.5191 87.95235634 0.0504674729 52 66 21 83 104.3988 230.2053 944.5748 102.6393 103.5191 87.95235634 0.0504674755 72 66 21 1 104.3988 230.2053 944.5748 102.6393 100 78.73578976 0.05071933355 72 66 21 1 104.3988 230.2053 944.5748 102.6393 100 78.73578976 0.05071933355 72 66 21 80 104.3988 230.2053 944.5748 102.6393 100 85.87852289 0.05063234929 53 67 21 80 104.3988 230.2053 944.5748 102.6393 100 87.02726953 0.05060912129 52 66 21 83 104.3988 230.2053 944.5748 102.6393 100 87.36678385 0.05046822655 72 66 21 80 104.3988 230.2053 944.5748 102.6393 100 85.87852289 0.05063234929 52 66 21 83 104.3988 230.2053 944.5748 102.6393 103.5191 87.95235634 0.0504674729 52 66 21 83 104.3988 230.2053 944.5748 102.6393 100 87.36678385 0.05046822629 52 66 21 83 104.3988 230.2053 944.5748 102.6393 103.5191 87.95235634 0.0504674729 52 66 21 83 104.3988 230.2053 944.5748 102.6393 103.5191 87.95235634 0.05046747 distribution system for multiple-purpose (voltage deviationminimization, and loss minimization). For each candidate setof locations and sizes, the contribution of USSES systemsto grid voltage deviation and power loss were evaluated anddiverse Pareto-optimal front were created. From the list ofPareto-optimal front, distribution system planners will haveopportunity to select the appropriate location based on desiredobjective. The proposed approach was demonstrated on theIEEE-123 node distribution test feeder with utility-scale PVsystems. Similar results in multiple runs validated the opti-mality of the obtained Pareto-optimal solution. R \ K912/78912194.PDF[3] G. Fitzgerald, J. Mandel, J. Morris, and H. Touati, “The economicsof battery energy storage,” Rocky Mountain Institute, Tech. Rep., Oct.2015.[4] N. Bhusal, M. Abdelmalak, and M. Benidris, “Optimum locationsof utility-scale shared energy storage systems,” in , Jaipur, India, Dec 2019, pp. 1–6.ig. 4. Pareto optimal solution, respective sizes and cites are as shown inTable I[5] Q. Fu, L. F. Montoya, A. Solanki, A. Nasiri, V. Bhavaraju, T. Abdallah,and D. C. Yu, “Microgrid generation capacity design with renewablesand energy storage addressing power quality and surety,”
IEEE Trans-actions on Smart Grid , vol. 3, no. 4, pp. 2019–2027, Dec 2012.[6] T. H. Mehr, M. A. S. Masoum, and N. Jabalameli, “Grid-connectedlithium-ion battery energy storage system for load leveling and peakshaving,” in , Sep. 2013, pp. 1–6.[7] Y. Shi, B. Xu, D. Wang, and B. Zhang, “Using battery storage forpeak shaving and frequency regulation: Joint optimization for superlineargains,”
IEEE Transactions on Power Systems , vol. 33, no. 3, pp. 2882–2894, May 2018.[8] Y. Wang, X. Lin, and M. Pedram, “Adaptive control for energy storagesystems in households with photovoltaic modules,”
IEEE Transactionson Smart Grid , vol. 5, no. 2, pp. 992–1001, March 2014.[9] S. Sulaeman, Y. Tian, M. Benidris, and J. Mitra, “Quantification ofstorage necessary to firm up wind generation,”
IEEE Transactions onIndustry Applications , vol. 53, no. 4, pp. 3228–3236, July 2017.[10] T. Zhang, A. E. Emanuel, and J. A. Orr, “Distribution feeder upgradedeferral through use of energy storage systems,” in
IEEE Power andEnergy Society General Meeting (PESGM) , Boston, MA, USA, July2016, pp. 1–5.[11] H. Nazaripouya, Y. Wang, P. Chu, H. R. Pota, and R. Gadh, “Optimalsizing and placement of battery energy storage in distribution systembased on solar size for voltage regulation,” in
IEEE Power EnergySociety General Meeting , Denver, CO, USA, July 2015, pp. 1–5.[12] K. Divya and J. Østergaard, “Battery energy storage technology forpower systems: An overview,”
Electric Power Systems Research , vol. 79,no. 4, pp. 511–520, 2009.[13] B. Dunn, H. Kamath, and J.-M. Tarascon, “Electrical energy storage forthe grid: a battery of choices,”
Science , vol. 334, no. 6058, pp. 928–935,2011.[14] R.Carnegie, D. Gotham, D. Nderitu, and P. Preckel, “Utility ScaleEnergy Storage Systems Benefits, Applications, and Technologies,”Tech. Rep., June 2013.[15] Y. Tian, A. Bera, M. Benidris, and J. Mitra, “Stacked revenue andtechnical benefits of a grid-connected energy storage system,”
IEEETransactions on Industry Applications , vol. 54, no. 4, pp. 3034–3043,July 2018.[16] P. S. Georgilakis and N. D. Hatziargyriou, “Optimal distributed gener-ation placement in power distribution networks: Models, methods, andfuture research,”
IEEE Transactions on Power Systems , vol. 28, no. 3,pp. 3420–3428, Aug 2013. [17] N. Bhusal, M. Gautam, and M. Benidris, “Sizing of movable energyresources for service restoration and reliability enhancement,” in , Montreal, QC,Canada, Aug 2020, pp. 1–5.[18] M. Moradi and M. Abedini, “A combination of genetic algorithmand particle swarm optimization for optimal dg location and sizingin distribution systems,”
International Journal of Electrical Power &Energy Systems , vol. 34, no. 1, pp. 66 – –74, 2012.[19] M. Ghofrani, A. Arabali, M. Etezadi-Amoli, and M. S. Fadali, “Aframework for optimal placement of energy storage units within a powersystem with high wind penetration,”
IEEE Transactions on SustainableEnergy , vol. 4, no. 2, pp. 434–442, April 2013.[20] S. Wen, H. Lan, Q. Fu, D. C. Yu, and L. Zhang, “Economic allocationfor energy storage system considering wind power distribution,”
IEEETransactions on Power Systems , vol. 30, no. 2, pp. 644–652, March2015.[21] G. Carpinelli, G. Celli, S. Mocci, F. Mottola, F. Pilo, and D. Proto,“Optimal integration of distributed energy storage devices in smartgrids,”
IEEE Transactions on Smart Grid , vol. 4, no. 2, pp. 985–995,June 2013.[22] N. Bhusal, M. Abdelmalak, M. Kamruzzaman, and M. Benidris, “Powersystem resilience: Current practices, challenges, and future directions,”
IEEE Access , vol. 8, pp. 18 064–18 086, 2020.[23] R. S. Rao, K. Ravindra, K. Satish, and S. V. L. Narasimham, “Powerloss minimization in distribution system using network reconfigurationin the presence of distributed generation,”
IEEE Transactions on PowerSystems , vol. 28, no. 1, pp. 317–325, Feb 2013.[24] K. Deb, A. Pratap, S. Agarwal, and T. Meyarivan, “A fast and elitistmultiobjective genetic algorithm: NSGA-II,”
IEEE Transactions onEvolutionary Computation , vol. 6, no. 2, pp. 182–197, April 2002.[25] P. Murugan, S. Kannan, and S. Baskar, “Application of NSGA-II algo-rithm to single-objective transmission constrained generation expansionplanning,”
IEEE Transactions on Power Systems , vol. 24, no. 4, pp.1790–1797, Nov 2009.[26] Y. Kumar, B. Das, and J. Sharma, “Multiobjective, multiconstraintservice restoration of electric power distribution system with prioritycustomers,”
IEEE Transactions on Power Delivery , vol. 23, no. 1, pp.261–270, Jan 2008.[27] F. Abbasi and S. M. Hosseini, “Optimal dg allocation and sizing inpresence of storage systems considering network configuration effectsin distribution systems,”
IET Generation, Transmission Distribution ,vol. 10, no. 3, pp. 617–624, 2016.[28] EPRI. Open distribution system simulator. [Online]. Available:https://sourceforge.net/projects/electricdss/[29] S. Elsaiah, M. Benidris, and J. Mitra, “Reliability improvement ofpower distribution system through feeder reconfiguration,” in
Interna-tional Conference on Probabilistic Methods Applied to Power Systems(PMAPS) , Durham, UK, July 2014, pp. 1–6.[30] Distribution System Analysis Subcommittee, “1992 test feedercases,” IEEE, PES, Tech. Rep., 1992. [Online]. Available: http://sites.ieee.org/pestestfeeders/resources/, Durham, UK, July 2014, pp. 1–6.[30] Distribution System Analysis Subcommittee, “1992 test feedercases,” IEEE, PES, Tech. Rep., 1992. [Online]. Available: http://sites.ieee.org/pestestfeeders/resources/