lable at ScienceDirect

Journal of Cleaner Production 203 (2018) 1187e1200

Contents lists avai

Journal of Cleaner Production

journal homepage: www.elsevier .com/locate/ jc lepro

Sustainable planning of hybrid microgrid towards minimizingenvironmental pollution, operational cost and frequency fluctuations

Vahid Sohrabi Tabar, Mehdi Ahmadi Jirdehi*, Reza HemmatiDepartment of Electrical Engineering, Kermanshah University of Technology, Kermanshah, Iran

a r t i c l e i n f o

Article history:Received 6 November 2017Received in revised form30 March 2018Accepted 7 May 2018Available online 10 September 2018

Keywords:Carbon capture and storageHydrogen gas stationMicrogridRenewable energyTransient stabilityStochastic programming

* Corresponding author. Tel.: þ98 83 38305000; faE-mail address: m.ahmadi@kut.ac.ir (M.A. Jirdehi)

https://doi.org/10.1016/j.jclepro.2018.05.0590959-6526/© 2018 Elsevier Ltd. All rights reserved.

a b s t r a c t

Microgrids mainly use conventional and renewable energy resources at the same time. Conventionalenergy resources produce environmental pollution and need high cost for operation. In recent years,penetration of renewable resources such as photovoltaic and wind turbine has been rapidly grown inmicrogrids. Reduction of power losses and pollution are the main advantages of integrating renewableresources into networks. But, the renewable energy resources comprise low inertia and stability of thenetwork integrated with such units is low. As a result, the environmental pollution and stability ofmicrogrid are considered as the main problems and a new modeling of microgrid energy management isproposed by this paper to tackle such drawbacks. In this regard, environmental pollution is reduced byincluding hydrogen gas station and carbon capture-storage system. As well, the virtual synchronousgenerator is used to provide sufficient inertia and improving transient stability. The uncertain parametersare incorporated in the planning and stochastic programming is applied to tackle such uncertainties.Problem is mathematically expressed as a stochastic mixed integer linear programming and solved bythe augmented Epsilon-constraint method. Finally, a comprehensive sensitivity analysis is carried out toevaluate the results. Based on the simulation results, by installing carbon capture-storage system,operational cost of microgrid is reduced from 64.998 $ to 56.043 $ and 1791.75 kg of carbon dioxide isstored. The revenue equal to 24 $ in one day is achieved by H2 station without any pollution. The stabilityof microgrid is also significantly improved by installing virtual synchronous generator. The resultsdemonstrate the viability and effectiveness of the proposed method to minimize environmental pollu-tion, operation cost and frequency fluctuations in microgrid energy management.

© 2018 Elsevier Ltd. All rights reserved.

1. Introduction

Microgrids usually include distributed generations (DGs) andlocal loads that can be isolated or connected to the main grid. Theenvironmental pollution, global warming, and limits of fossil fuelresources are the main reasons for utilizing renewable resources.These resources can be suitably integrated into small networkssuch as microgrids. In real microgrids, because of limitations on therenewable energy resources such as stochastic generation, it is notpossible to supply the demand only by renewable energy resourcesand the operator has to apply conventional energy resources inaddition to the renewable ones. Application of renewable andconventional energy resources causes several problems in micro-grids. Conventional energy resources produce environmental

x: þ98 83 38305006..

pollution and the network operator is often willing to reduce theconventional energy resources and increase the renewable energyunits. On the other hand, high penetration level of renewable re-sources reduces the inertia of the microgrid and leads to stabilityproblems such as fast frequency deviations and stability collapse.Since the inertia of renewable energy resources is less than theinertia of conventional generators. Thus, if all energy resources arechosen as renewable, the microgrid inertia will be significantlyreduced and it is not possible to control the frequency of themicrogrid. Therefore, low inertia problem and environmentalpollution are occurred by renewable and conventional generators,respectively. As a result, it is required to tackle these two problems(i.e., environmental pollution related to the conventional units andstability problem associated with renewable units) in microgridscheduling and planning.

Different aspects of microgrids have been investigated by re-searchers so far. Control and energy management are two impor-tant aspects of microgrid studies. Frequency and voltage control

Nomenclature

Symbols, indexes and parametersA Wind generator blade area (m2)CM-CHP, COP-CHP CHP maintenance cost ($) and operation cost

($/kWh)COP-WT, CCONS-WT WT operation cost ($/kWh) and constant cost

($)COP-PV, CCONS-PV PV operation cost ($/kWh) and constant cost ($)CM-MT, COP-MT MT maintenance cost ($) and operation cost

($/kWh)CM-ES, COP-ES ESmaintenance cost ($) and operation cost ($/kWh)CBuy, CSell Constant price of buying and selling energy ($/kWh)CM-H2, CM-G H2 and gasoline stations maintenance costs ($)CM-VSG VSG maintenance cost ($/kWh)CPH2, CSH2 Cost of producing ($/kWh) and selling H2 ($/Lit)CSG, CPG Cost of selling and producing gasoline ($/Lit)CFuel Cost of fuel ($/kWh)CBuy (t), CSell (t) Cost of buying and selling energy ($)CG (t) Cost of gasoline station ($)CCCS (t) Cost of CCS system ($)CH2 (t) Cost of H2 station ($)CCHP (t) Total cost of CHP ($)CPV (t) Total cost of PV ($)CMT (t) Total cost of MT ($)CWind (t) Total cost of WT ($)CES (t) Total cost of ES ($)CVSG (t) Cost of VSG ($)df and dt Frequency (Hz) and time (s) variationsEFi Emission factor for ith generator (kg/kWh)EFCHP Emission factor of CHP (kg/kWh)EFMT Emission factor of MT (kg/kWh)EFVSG Emission factor of VSG (kg/kWh)EFMG Emission factor of main grid (kg/kWh)EFG Emission factor of gasoline (kg/Lit)ESmax, ES

min Maximum and minimum energy of ES (kWh)ELD (t) Electrical load demand (kW)ES (t), ES (0) ES energy and Initial state of charge (kWh)EMG (t) Emission of gasoline station (kg)EMCHP (t) Emission of CHP (kg)EMMT (t) Emission of MT (kg)EMVSG (t) Emission of VSG (kg)EMMG (t) Emission of microgrid (kg)EH2(t) Energy for producing H2 (kWh/Lit)ei Constrained objective functionsf, f* Microgrid frequency and references frequencyfp(x) Objective functionsGTSCT, GTNOCT Solar radiation for STC and NOCT (kW/m2)GT (t) Solar radiation on tilted module plane (kW/m2)H Inertia (kg.m2)kvi, kr Virtual inertia (kg.m2) and constant coefficientkvd, kk Damping and constant coefficientNOCT Normal operating cell temperature (�C)NPVs, NPVp Number of series and parallel cells in PV modulePmax

E-dech, Pmax

E-ch ES maximum discharge and charge rate (kW)PMTmax, PCHP

max Maximum powers of MT and CHP (kW)PLine Line transfer power limit (kW)

PPV, STC Maximum test power for the STC (kW)PG and PL Total produced and demanded powers (kW)PWT (t) WT power (kW)PVSG(t) VSG output (kW)PVSGi (t), PVSG

d (t) Inertia response and damping power (kW)PPV (t) PV power (kW)PCHP (t) CHP power (kW)PMT (t) MT power (kW)PBuy(t), PSell (t) Buy and sell powers (kW)PES (t) ES power (kW)PGi (t) Produced power by ith generator (kW)Rm (t) Reserve margin of the microgridRmax

m (t), Rminm (t) Maximum andminimum reservemargin (%)

RH2 (t) Revenue by selling H2 ($)RG (t) Revenue by selling gasoline ($)RCCS (t) Revenue of CCS system ($)Rp, STP Revenue and store prices by CCS system ($/kg)ri Range of ith objective functionsi Feasible region of ith objective functionTamp, Tjstc Environmental and reference temperature of PV (�C)T Last time intervalt Time (h)Tj (t) Cell temperature of PV (�C)V(t) Wind speed (m/s)VOLH2 (t), VOLG (t) Volume of H2 and gasoline (Lit)Vnom Nominal wind speed (m/s)Vcut-in, Vcut-out Minimum and Maximum wind speed (m/s)x Vector of decision variablesr and W Air density (kg/m3) and Time intervalhCHP CHP generator electrical efficiencyhMT MT generator electrical efficiencyhE C, h

ED ES charge and discharge efficiency coefficients

hw Wind generator power coefficientAbbreviationsANN Artificial neural networkBFA Bellman ford algorithmCCS Carbon capture and storageCHP Combined heat and powerCA Clustering algorithmCO2 Carbon dioxideDG Distributed generationDVR Dynamic voltage restorerESS Energy storage systemHHA Hyper heuristic algorithmMBFO Modified bacterial foraging algorithmMPABC Multi period artificial bee colonyMPGSA Multi period gravitational search algorithmMILP Mixed integer linear programmingMINLP Mixed integer non-linear programmingMIQP Mixed integer quadratic programmingMT Micro turbineMPP Maximum power pointPSO Particle swarm optimizationPV PhotovoltaicPDF Probability distribution functionSQP Sequential quadratic programmingVSG Virtual synchronous generatorWT Wind turbine

V.S. Tabar et al. / Journal of Cleaner Production 203 (2018) 1187e12001188

V.S. Tabar et al. / Journal of Cleaner Production 203 (2018) 1187e1200 1189

(Bahrani et al., 2013; Brabandere et al., 2007; Farahani et al., 2017)are the main control strategies in microgrids. In the other side,microgrid energy management is studied from different perspec-tives (Tabar et al., 2017; Su et al., 2014; Wang et al., 2015;Schoonenberg and Farid, 2017; Rezvani et al., 2015). In a recentresearch microgrid energy management by considering dynamicvoltage restorer (DVR) is investigated and effect of DVR on theconnection line between the microgrid and the main grid is fullystudied (Jirdehi et al., 2017). Energy management in a small scalemicrogrid such as a building (Marzband et al., 2017; Le�si�c et al.,2017; Hemmati, 2017), considering demand response program(Amrollahi and Bathaee, 2017; Nikmehr et al., 2017; Korkas et al.,2016; Aliasghari et al., 2018), management of industrial microgridby virtual power plant and implementation the proposed methodon a real case study (Hooshmand et al., 2018), investigation optimalrenewable DGs sizing and location (Dawoud et al., 2017), optimallocation, size and operation of storage (Hemmati, 2018) and pro-posing microgrid energy management by considering assets life-time (Choobineh and Mohagheghi, 2016) are another new aspectsof microgrid energymanagement that are considered a lot in recentstudies.

In order to cope with the pollution problems, several methodsare suggested such as increasing renewable resource and equippingconventional resource with carbon capture-storage (CCS) system(Partridge and Gamkhar, 2012). In recent years, CCS system hasbeen developed rapidly. This system includes three stages as carboncapture, transfer, and storage (Leung et al., 2014; Sreenivasulu et al.,2015). In order to tackle the inertial problems, different techniquesare proposed such as improving transient stability by virtual inertia(Soni et al., 2013), application of robust control (Hossain et al.,2015), and providing frequency control service based on the vir-tual inertia concept (Rezaei and Kalantar, 2015). As a result, it isnecessary to include these issues in microgrid planning and energymanagement.

Microgrid energy management is usually subjected to technicalconstraints that will be optimized by various methods. This con-strained optimization problem can be solved using heuristic opti-mization techniques or mathematical approaches. For instance,following heuristic techniques have been applied to solve con-strained optimization problem: artificial neural network (ANN) andmodified bacterial foraging algorithm (MBFO) (Motevasel and Seifi,2014), hyper heuristic algorithm (HHA) (Mallol-Poyato et al., 2015),multi period artificial bee colony (MPABC) combined with markovchain (Marzband et al., 2015), particle swarm optimization (PSO)(Lu et al., 2017), bellman ford algorithm (BFA) (Tai et al., 2017),clustering algorithm (CA) (Amini et al., 2017) and multi periodgravitational search algorithm (MPGSA) (Marzband et al., 2014). Aswell, themathematical methods are used to solve the problem suchas CPLEX (Igualada et al., 2014) and GUROBI (Mazzola et al., 2015).

In this paper, a multi objective stochastic programming is pre-sented for microgrid energy management. The environmentalpollution is one of the main reasons for utilizing renewable re-sources and achieving a clean production. So, the proposed prob-lem aims at minimizing total pollution and cost simultaneously. Inorder to tackle the pollution problems, hydrogen gas station andCCS system are included in the planning. Hydrogen gas station isutilized rather than conventional resource such as gasoline stationsand is compared to them. Also, emission of microgrid will beresolved by considering CCS system. Microgrid low inertia problembased on energy management is one of the significant issues thathas never been studied. In order to overcome low inertia problem,virtual synchronous generator (VSG) is used to improve transientstability. Important parameters such as wind speed, solar radiation,and electrical load are considered as stochastic parameters. Also,the problem is expressed as a multi objective stochastic mixed

integer linear programming (MILP) and solved using theaugmented Epsilon-constraint method in the last version of GAMS/CPLEX. Simulation results demonstrate the efficiency of the pro-posed planning. As a result, the key novelty and originality aspectsof the proposed paper can be defined as follows:

� Resolving environmental emission of microgrid by consideringhydrogen gas station and CCS.

� Utilization of renewable resources for producing pure fuel asnew solution to reduce environmental pollution.

� Considering various power resources (conventional andrenewable) and technologies (CCS, VSG and H2 station) simul-taneously in microgrid energy management and study theircorrelation at the same time.

� Improving microgrid low inertia problem and transient stabilityby utilizing VSG. In order to overcome low inertia problem thesize of VSG is optimized by the proposed program.

� Proposing a multi objective stochastic programming and usingaugmented Epsilon-constraint method to solve the problem.

Finally, the rest parts are classified as follows: modeling ofmicrogrid is addressed in section 2, formulation of the problem ispresented in section 3, the solution and test system are given insections 4 and 5, respectively. Simulation results and sensitivityanalysis are investigated sections 6e7, and eventually section 8 isassigned to the conclusions and remarks.

2. Problem modeling

Accurate mathematical modeling of microgrid componentsmakes great impacts on the final results. Various modeling can beutilized in the problems, but an appropriate selection depends onmany factors such as complexity of the problem. In this regard,mathematical modeling including MILP (Craparo et al., 2017),mixed integer non-linear programming (MINLP) (Oliveira et al.,2017), mixed integer quadratic programming (MIQP) (Zachar andDaoutidis, 2017) and sequential quadratic programming (SQP)(Xiao and Ai, 2017) are the most popular modeling in microgridmanagement. In this study, mathematical modeling of the micro-grid components is proposed based on MILP. Equation simplifica-tion, lower simulation time, and achieving global optimal solutionare the main features of this method.

2.1. Hydrogen gas station

Hydrogen gas stations mainly consume the produced electricalenergy by renewable energy resources and then produce hydrogenas a powerful and pure fuel. The produced hydrogen is derived fromwater. Hydrogen is a powerful, clean energy, and far more efficientthan other sources that can be used as the fuel of machines such ashybrid vehicles. Hydrogen burnt vehicles are rapidly full charged(3e5min) and can drive for long distances about 480 kme640 km(Calculate and compensate for your CO2). In order to produce1000 L H2 about 50 kWh energy is needed and this energy is mainlymet by wind turbine (WT) or photovoltaic (PV) (BalancingMechnism Reporting System). In order to show the effectivenessof hydrogen burnt vehicles, two cases are modelled and comparedin this paper. First, the vehicles which consume hydrogen andsecond, the vehicles which consume conventional fuels such asgasoline. Modeling of H2 and gasoline burnt vehicles are giventhrough (1) to (5). These equations show the obtained revenue byselling hydrogen, hydrogen cost, gasoline cost, gasoline revenue,and the amount of emitted pollution by gasoline, respectively.

V.S. Tabar et al. / Journal of Cleaner Production 203 (2018) 1187e12001190

RH2ðtÞ ¼XTt¼1

ðVOLH2ðtÞ$CSH2Þ (1)

CH2ðtÞ ¼XTt¼1

ðEH2ðtÞ$CPH2$VOLH2ðtÞÞ þ CM (2)

CGðtÞ ¼XTt¼1

ðVOLGðtÞ$PGÞ þ CM�G (3)

RGðtÞ ¼XTt¼1

ðVOLGðtÞ$CSGÞ (4)

EMGðtÞ ¼XTt¼1

ðVOLGðtÞ$EFGÞ (5)

2.2. Carbon capture and storage system

CCS system has been introduced as the best system for collect-ing pollution based on environmental protection agency (EPA)standards (US Environmental Protection Agency). This systemcaptures the emitted carbon dioxide (CO2) and transports it to theproper storage place. The stored CO2 can be used for enhanced oilrecovery (Center for Climate and Energy Solutions; Saboori andHemmati, 2016). As a result, CCS modeling is proposed as follows,cost and revenue by CCS are defined by (6) and (7), respectively(Saboori and Hemmati, 2016). It is assumed that the consumedenergy by CCS is provided by the generating units. In other words,

PWT ¼�

0 ct : Vcutin � VðtÞ and Vcutout � VðtÞ0:5:r:A:hW :minðVðtÞ; VnomÞ3 ct : Vcutin � VðtÞ � Vcutout (15)

each generating unit is equipped with a CCS and the consumedenergy by this CCS is supplied by its own generating unit.

CCCSðtÞ ¼XTt¼1

ðPGiðtÞ:EFi:STP :qÞ (6)

RCCSðtÞ ¼XTt¼1

ðPGiðtÞ:EFi:RP :qÞ (7)

2.3. Virtual synchronous generator

VSG is a simple synchronous generator with inertia anddamping features including energy storage, inverter, and controlsystem (Bevrani et al., 2014). One of the suggested solution to tacklethe frequency deviations is to produce virtual inertia during highfrequency variation. This purpose can be achieved by adding VSG tothe microgrid for producing virtual inertia. The required amount ofVSG can be determined based on the standard frequency deviation.Thus, VSG modeling is defined by (8) to (14) (Lopes, 2013).

PVSGðtÞ ¼ PiVSGðtÞ þ PdVSGðtÞ (8)

PiVSGðtÞ ¼ �kvi$k2r $f $

dfdt

(9)

PAVSGðtÞ ¼ kvd:k2k :�f * � f

�(10)

dfdt

¼ PG � PL2H2p

(11)

dfdt

� 1 (12)

CVSGðtÞ ¼XTt¼1

CFuel:PVSGðtÞ:qþ CM�VSG (13)

EMVSGðtÞ ¼XTt¼1

PVSGðtÞ:EFVSG:q (14)

Where, the produced power by VSG, inertia response, and dampingpower are defined by (8) to (10), respectively. Also, frequency de-viation, frequency constraint, cost, and pollution of VSG can begiven by (11) to (14), respectively.

2.4. Wind turbine and photovoltaic modeling

WT output power is given as (15) (Villanueva and Feij�oo, 2010).PV output power depends on cells temperature and solar irradiancein maximum power point (MPP) that can be formulated as (17)(Riffonneau et al., 2011). Cell temperature of PV is calculated by (16)(Riffonneau et al., 2011) and then output power of PV at each timecan be achieved by (17). It should be noted that wind speed andsolar radiation are uncertain parameters.

TjðtÞ ¼ Tamp þ GT ðtÞGTSTC

� ðNOCT � 20Þ (16)

PPV ðtÞ ¼�PPV ;STC � GT ðtÞ

GTSTC��1� g�

�TjðtÞ � TjSTC

���� NPVs

� NPVp

(17)

2.5. Energy storage system

In order to optimize the operation and improving the flexibilityof microgrid, energy storage systems (ESSs) are included andmodelled through (18) to (21) (Motevasel and Seifi, 2014). Theoutput power of ESS is shown by (18) and (19) specifies theremained energy in ESS after charging-discharging to preventreducing batteries lifetime. The limitations for charging-discharging power at each hour are given by (20). The initial andfinal energy of ESS must be equal as defined by (21).

PESðtÞ ¼ ESðtÞ � ESðt � 1Þ (18)

EminS � ESðtÞ � Emax

S (19)

V.S. Tabar et al. / Journal of Cleaner Production 203 (2018) 1187e1200 1191

(EminS � ESð0Þ �

XTt¼1

PESðtÞ � EmaxS � ESð0Þ (20)

ESð0Þ ¼ ESðTÞ (21)

3. Problem formulation

In the proposed problem, two objective functions are definedandminimized at the same time. Total operational cost of microgridand environmental pollution are the objective functions. It shouldbe noted that the problem includes H2 station, VSG, CCS, ESS, andconventional-renewable resources.

3.1. Total operational cost of microgrid

Total operational cost of microgrid is given by (22).

8>>>>><>>>>>:

FðCostÞ ¼XTt¼1

ðCCHPðtÞ þ CWindðtÞþCPV ðtÞ þ CBuyðtÞ � CSellðtÞþ

CESðtÞ þ CH2ðtÞ þ CMTðtÞ þ CCCSðtÞþCVSGðtÞ � RCCSðtÞ � RH2ðtÞ � RGðtÞÞ

(22)

Different parts of (22) are defined as follows:

CCHPðtÞ ¼XTt¼1

CFuel:PCHPðtÞ:q

hCHPþ COP�CHP :PCHPðtÞ:q

þ CM�CHP

(23)

CWindðtÞ ¼XTt¼1

COP�WT :PWT ðtÞ:qþ CCONS�WT (24)

CPV ðtÞ ¼XTt¼1

COP�PV :PPV ðtÞ:qþ CCONS�PV (25)

CMTðtÞ ¼XTt¼1

CFuel:PMTðtÞ:q

hMTþ COP�MT :PMTðtÞ:q

þ CM�MT

(26)

CBuyðtÞ ¼XTt¼1

CBuy:PBuyðtÞ:q (27)

CSellðtÞ ¼XTt¼1

CSell:PSellðtÞ:q (28)

CESðtÞ ¼XTt¼1

COP�ES:PESðtÞ:qþ CM�ES (29)

Cost of combined heat and power (CHP), WT, PV and microturbine (MT) are presented by (23) to (26). First term of (23) showsfuel cost, second, and third terms represent operational andmaintenance cost, respectively. The operational and fixed costs ofWT and PV are given by (24) and (25), respectively. First term of(26) indicates cost of generation for MT, and the second and thirdterms specify the operational and maintenance cost, respectively.Cost of buying and selling power at each time are defined by (27)and (28), respectively. Furthermore, operational and maintenancecosts of the electrical battery are given by (29).

3.2. Environmental pollution

Environmental pollution caused by microgrid and main gridgeneration is defined as (30).

8><>: FðEmissionÞ ¼

XTt¼1

ðEMCHPðtÞ þ EMMTðtÞþEMMGðtÞ þ EMVSGðtÞ þ EMGðtÞÞ

(30)

Different parts of (30) are defined as follows:

EMCHPðtÞ ¼XTt¼1

PCHPðtÞ:EFCHP :q (31)

EMMTðtÞ ¼XTt¼1

PMTðtÞ:EFMT :q (32)

EMMGðtÞ ¼XTt¼1

PBuyðtÞ:EFMG:q (33)

Where, (31) to (33) indicate the produced pollution by CHP, MT,and main grid, respectively. Also, produced pollution by VSG andgasoline station are presented in (30) and defined in previoussections.

3.3. Problem constraints

Microgrid energy management includes many constraintswhich limit the operation of microgrid. As a result, microgridshould operate under following technical constraints:

8><>:

ELDðtÞ ¼XTt¼1

ðPWT ðtÞ þ PPV ðtÞ þ PMTðtÞþ

PCHPðtÞ þ PESðtÞ þ PBuyðtÞ � PSellðt�� (34)

PCHP � PmaxCHP (35)

PMT � PmaxMT (36)

�PBuyðtÞ or PSellðtÞ

� � PLine (37)

(PESðtÞ

.hED � Pmax

E�dech for dischðPESðtÞ>0Þ�hEC :PESðtÞ � Pmax

E�ch for chðPESðtÞ<0Þ(38)

RmðtÞ ¼Total capacityðtÞ � Peak loadðtÞ

Peak loadðtÞ� 100% (39)

Rminm ðtÞ � RmðtÞ � Rmax

m ðtÞ (40)

Constraint (34) confirms power equilibrium in the microgrid.Constraints (35) and (36) show CHP and MT power limitations,respectively. Line capacity and Charging-discharging powers arelimited by (37) and (38). Also, reserve margin and permitted levelsare specified by (39) and (40).

4. Solving technique

This paper proposes a stochastic multi objective programming.Total cost and environmental pollution are the objective functionsof the proposed method. Wind speed, solar radiation, and load areregarded as stochastic parameters. In order to solve such multi

V.S. Tabar et al. / Journal of Cleaner Production 203 (2018) 1187e12001192

objective programming, an accurate method is utilized as theaugmented Epsilon-constraint mathematical model. Scenariogeneration and reduction are one of most important parts of theproposed stochastic problem.

Probabilitydensity Main value

30% Higher30% Lower

Fig. 1. The probability density function of stochastic parameters.

4.1. Augmented Epsilon-constraint method

In order to solve the proposed energy management program,advanced model of the Epsilon-constraint method (augmentedEpsilon-constraint) is used to get better results. This method hasseveral advantages such as changing the original feasible region,generating non-extreme efficient solutions, and it is independentto the objective functions scales (Mavrotas, 2009). Briefly, a multiobjective problem with augmented Epsilon-constraint can bedefined as follows (Mavrotas, 2009):

Assume the following multi objective problem should be opti-mized as follows:

max�f1ðxÞ; f2ðxÞ; :::; fpðxÞ

�subject tox2s; i2f1;2;3; :::;pg

(41)

Hence, in Epsilon-constraint method one of the objectivefunctions will be optimized using other objective functions asconstraints.

max f1ðxÞsubject tof2ðxÞ � e2;f3ðxÞ � e3;:::fpðxÞ � ep;x2s;

(42)

In order to force the problem to find efficient and optimal resultssubject to the defined constraints, a new relationship is defined as(43).

max�f1ðxÞ þ eps� �

s1 þ :::þ sp��

subject tof2ðxÞ � s2 ¼ e2f3ðxÞ � s3 ¼ e3:::fpðxÞ � sp ¼ epx2s and si2Rþ

(43)

Formulation (43) only produces efficient solutions and theproposed planning has alternative optimal solutions that one ofthem (exhibited by x0) dominates the other one. As a result, thiscase is defined by (44).8>><>>:

e2 þ s2 � e2 þ s02;e3 þ s3 � e3 þ s03;:::ep þ sp � ep þ s0p

(44)

Based on (44) and considering at least one strict inequality, (45)is achieved.

Xpi¼2

si <Xpi¼2

s0i (45)

It is suggested to replace si by si/ri to avoid any scaling problems.Hence, final objective function will be defined by (46).

max�f1ðxÞ þ eps� �

s1=r2 þ :::þ sp�rp��

(46)

The final equation is given by (46) and used to solve the multiobjective problem. Also, it should be noted that eps is a smallamount (10�3 to 10�6).

4.2. Stochastic programming

Wind speed, solar radiation, and load are modelled as stochasticparameters. In order to model the stochastic programming, sce-nario generation and reduction techniques are applied. The un-certain parameters are assumed to have a continuous probabilitydistribution function (PDF) with 30% standard deviation that isshown in Fig. 1. Then, the continuous PDF is estimated by discretePDF including Nn steps. If there are Mm uncertain parameters, andeach parameter is estimated by Nn steps, therefore, there are Nn

Ti*Mm

scenarios. Where, Ti shows the time intervals of next 24-h (e.g., sixtime intervals and each one including 4 h). After producing allscenarios and the probability related to each scenario, the mostprobable scenarios with the highest possibility (50 scenarios) ofoccurrence are selected. This approach results in a trivial error atthe outputs, but it significantly reduces the simulation time.

5. Test system

The test casemicrogrid includes electrical load,WT, PV, CHP, MT,ESS, VSG, CCS, H2, and gasoline stations. In order to reduce scenariogeneration, energy management is carried out for 24-h that isdivided into six four-hour intervals. In general, in microgrid energymanagement two different models can be used including nodalmicrogrid (Tabar et al., 2017; Jirdehi et al., 2017) and network basedmicrogrid (Vergara et al., 2017; Haddadian and Noroozian, 2017a,2017b). In this study, a nodal microgrid without power losses isconsidered for analyzing. In such structures, grid lines will beignored and it is assumed that all components are installed into onenode (nodal system). So, the constraints such as power losses andvoltage limit are ignored. CFuel is equal to 0.027 $/kWh, Pline is30 kW, and reserve margin is considered between 10 and 20percent. Transportation and storage cost for CCS is 0.015 $/kg. Itshould be noted that obtained revenue from storing CO2 is equal to0.01 $/kg and 0.02 $/kg based on the storage technologies(enhanced oil recovery and deep saline formations, respectively).Also, the maintenance cost of VSG is about 0.002 $. All other pa-rameters are summarized in Figs. 2 and 3 and Tables 1e6. A simplemodel of the proposed microgrid is illustrated in Fig. 4 and flow-chart of the proposed method is shown in Fig. 5.

1 2 3 4 5 60

0.02

0.04

0.06

Solarradiation(kW/m

2 )

1 2 3 4 5 6

Time interval

0

10

20

30

Windspeed(m/s)

Fig. 2. Wind speed and solar radiation in deterministic state.

V.S. Tabar et al. / Journal of Cleaner Production 203 (2018) 1187e1200 1193

6. Simulation results

Simulation results are given in several sections and all aspects ofthe problem are thoroughly analyzed.

1 2 30

0.1

0.2

0.3

0.4

Cost($/kW

h)

Selling priceBuying price

1 2 3Time

80

100

120

140

Power(kW)

Fig. 3. Price of exchanging energy and electrical lo

6.1. Results of the proposed planning

The results of the proposed stochastic MILP program are pre-sented in Table 7. Also, the amount of demanded inertia is

4 5 6

4 5 6intervalad (Balancing Mechnism Reporting System).

Table 1Emission factors (Motevasel and Seifi, 2014).

Emission type Emission factors (kg/kWh)

MT CHP Grid VSG

CO2 0.724 0.822 0.922 0.724

Table 2Batteries factors and efficiency factors (Hawkes and Leach, 2009).

hCHP 0.35 PmaxE-dech (kW) 20

hMT 0.3 PmaxE-ch (kW) 20

hw 0.59 ESmin (kWh) 2hE

C¼ hED 0.95 ESmax (kWh) 40

Table 3PV and WT characteristics (Villanueva and Feij�oo, 2010; Riffonneau et al., 2011;Hawkes and Leach, 2009).

Tamp (�c) 20 Vcut-out (m/s) 25Tjstc (�c) 25 NPVs 70GTSCT (kW/m2) 1 NPVp 30NOCT (�c) 45.5 g 0.043%r (kg/m3) 1.23 A (m2) 100Vnom (m/s) 12 PPV, STC (kW) 0.165Vcut-in (m/s) 5 GTNOCT (kW/m2) 0.8

Table 4Maintenance and operational costs (Hawkes and Leach, 2009).

Components Cost

Maintenance or constant ($) Operation ($/kWh)

CHP 0.002 0.005MT 0.001 0.004WT 0.002 0.005PV 0.001 0.003ES 0.001 0.004

Table 5Characteristics of the generating units.

Components Limitation

Power (kW) Number Inertia (kg.m2)

CHP 90 1 2MT 40 1 0.88WT 80 1 1.4PV 25 4 0

Table 6Parameters of gasoline and H2 units.

Component Fuels

H2 Gasoline

Selling cost per Lit ($/Lit) 4.730 1.8Producing cost per Lit ($/Lit) 4.250 1.62Pollution per Lit (kg/Lit) 0 14Maintenance cost ($) 0.002 0.002

V.S. Tabar et al. / Journal of Cleaner Production 203 (2018) 1187e12001194

presented in Table 8. Finally, cost and revenue by CCS system andH2 station are shown in Tables 9 and 10, respectively. As seen inTable 7, because of low-cost off-peak during first time interval(0e4), most of demanded power is met by CHP (80.14 kW) and WT(62.7 kW). Also, 30 kW is sold to themain grid to obtain revenue. Aswell, ESS is charged in this condition. During second time interval(4e8), load is decreased as well WT power is decreased from62.7 kW to 53.51 kW. As a result, CHP and MT increase their power.The extra generated power is stored in ESS and send to the maingrid to obtain revenue. During third time interval (8e12), load isincreased and the generated power by PV is increased from zero to60.49 kW. Therefore, the generated power by CHP is decreased andMT power is also increased to supply the load by less pollutantresources. In order to obtain revenue, transferred power to themain grid is still 30 kW. As well, ESS is discharged to meet power.Because of reducing WT output power from 34.81 kW to zero,fourth time interval is so critical. The generated power by CHP, MT,and PV are increased and ESS is discharged tomeet the load. In next

time interval (16e20), load demand is increased to the higher valueand total generated power byWT and PV are decreased. As a result,produced power by CHP and MT are increased, power is receivedfrom the main grid, and ESS is discharged. In last time interval(20e24), powers of WT, CHP and MT are increased and the extrapower is sold to themain grid. It is clear that ESS works on chargingstate during low-cost off-peak hours and stores energy. While, itworks on discharging state during high-cost on-peak hours.

Based on Table 8, VSG size is determined (for the worst scenariowith more inertia problem and islanding operation mode) to solvethe stability problem during all time intervals. The maximumrequired inertia is installed at fourth time interval due to lack ofmain inertia producers such as MT and WT. As well, because ofexisting sufficient inertia at second time interval, the needed inertiais zero.

Table 9 shows CCS characteristics. As shown in Tables 7 and 9,the produced power by conventional resources (i.e., CHP and MT) isminimum at third time interval. Hence, the stored CO2 by CCS islower than other times and the obtained revenue by CCS is reduced.On the other side, the generated power by CHP and MT are veryhigh at last time interval and the produced pollution will beincreased and the gained revenue by CCS is higher than otherstimes. Eventually, the CCS total revenue is achieved to 8.996 $ at theend of the scheduling.

The effects of considering H2 fuel stations in the microgrid aredemonstrated through comparing H2 to conventional fuel (gaso-line). As shown in Table 10, with burning 50 L gasoline and H2 in themicrogrid, 700 kg CO2 is released by gasoline while the amount ofreleased pollution by H2 is zero. Also, the amount of achievedrevenue by H2 is more than gasoline fuel. By installing H2 stationand selling produced H2 more revenue excluding environmentalpollution is achieved.

6.2. Impact of uncertainty on the planning

In order to show the impact of uncertainty on the planning aswell as demonstrating the advantages of the proposed stochasticplanning, results under deterministic state are presented inTable 11. Results show that the planning cost of the stochasticmethod is more than deterministic state by 317.3%. Also, along withincreasing cost, pollution is increased by 118.5%. As shown inTables 7 and 11, in the deterministic state, the produced power byMT is zero under four first time intervals. But, because of uncer-tainty in the stochastic state, the generated power by MT isincreased and the pollutionwill be increased. It is worth remarkingthat in fifth time interval, the produced power by MT and CHP inthe stochastic planning are less than the deterministic one. Thisissue is due to increasing the generated power by WT. As well, byincreasing the generated power by conventional generators such asCHP at first to fourth time intervals, the amount of pollution isincreased from 1511.05 kg to 1791.7 kg. In order to show therobustness and superiority of the stochastic programming, WToutput power is reduced by 10% and reserve margin for both thestochastic and deterministic cases is depicted in Fig. 6. It is clear

Circuit Breaker

0.4 kV

20 KV Main Grid

Transformer

Pure Generators (WT, PV and H2Station)

Carbon Capture and Storage UnitEnergy Storage Unit

CO2 Generators (CHP, VSG, MT andGasoline Station)

Local Loads

CO2

MicrogridCentral Unit

Fig. 4. Model of the proposed microgrid.

V.S. Tabar et al. / Journal of Cleaner Production 203 (2018) 1187e1200 1195

that under deterministic state, the reserve margin constraint isviolated while the stochastic planning can successfully tackle theuncertainty and cope with 10% changing the power of WT. Thus,stochastic programming is more robust and even can support morereduction in the generation or growth in the demand. Hence, highplanning cost of stochastic planning is acceptable and justifiable.

6.3. Impact of VSG on the planning

Planning objective functions and final results excluding VSG(i.e., without constraint (12)) are presented in Table 12 and Fig. 7. Asshown in Table 12, VSG does not show a great effect on the oper-ation of MT or CHP. Also, considering VSG in microgrid increases

Fig. 5. Flowchart of the proposed modeling.

Table 8VSG size at each time interval.

Time interval 1 2 3 4 5 6

Virtual inertia (kg.m2) 0.652 0 1.18 2.37 1.610 0.891

Table 9Cost and revenue of CCS.

Time interval Cost and revenue ($)

Operation cost ($) Obtained cost ($) Revenue ($)

1 4.244 5.658 1.4142 3.948 5.264 1.3153 3.750 4.999 1.2494 4.580 6.106 1.5265 4.891 6.521 1.6306 5.587 7.449 1.862

Table 10Revenue and pollution of H2 and gasoline per 50 L.

Type of station Data

Revenue ($) Pollution (kg)

H2 station 24 0Gasoline station 9 700

V.S. Tabar et al. / Journal of Cleaner Production 203 (2018) 1187e12001196

cost and pollution from 55.752 $ and 1787.647 kg to 56.043 $ and1791.7 kg, respectively. But, VSG is required to support the stabilityof the network. In order to provide a stable network with suitableperformance, df/dt should be lower than 1 or at least equal to 1. Asshown in Fig. 7, in the network without VSG, df/dt at some intervalsis more than standard value and microgrid stability is not passed.However, by installing VSG and injecting the inertia to the micro-grid, df/dt is decreased and network stability is approved. As ex-pected, by injecting obtained inertia, frequency deviation will befixed on bondmargin (i.e., 1). The main reason of this case is that allconstraint will be optimized by considering cost and pollutionobjective functions. So, this case is validated trust of simulationresults.

Table 7Results of the proposed stochastic planning.

Time interval PCHP (kW) PMT (kW) PWT (kW)

1 80.14 6.86 62.72 73.69 7.18 53.513 66.58 9.4 34.814 80.66 12.58 05 85 13.34 23.516 90 26.38 60.55Total planning cost ($)Total planning pollution (kg)

6.4. Impact of CCS on the planning

By removing CCS in the microgrid, planning cost is increasedfrom 56.043 $ to 64.998 $ as shown in Table 13. In fact, the obtainedrevenue (given in Table 9) from storing and selling CO2 is removed.Without CCS, 1791.75 kg of CO2 will be released into the atmo-sphere that this case is not acceptable. Therefore, CCS not only leadsto revenue, but also reduces the environmental pollution.

6.5. Impact of H2 station on the planning

In order to provide a comparative study, two storage tanks withmaximum capacity of 50 L are considered for both H2 and gasolinestations in the microgrid. It should be noted that 2.5 kWh energy isneeded for water decomposition and producing 50 L of H2 fuel. Thisenergy is supplied by renewable resources. The amount of emittedpollution by H2 station is zero. It is assumed that all stored fuels inboth two stations should be consumed during 24 h. Effect of H2station on microgrid is presented in Table 14. By comparingTables 14 and 7, H2 Station increases revenue without producingpollution. But, without H2 station, the cost is increased from 56.043$ to 79.923 $. In order to provide a better comparison, this fuel iscompared to gasoline in Table 10. By utilizing gasoline, the emissionis increased from 1791.647 kg to 2491.647 kg. Also, by utilizing H2,the cost is decreased from 79.923 $ to 56.043 $.

PPV (kW) PES (kW) PSell (kW) PBuy (kW)

0 5 30 00 0.075 30 060.49 �0.3343 30 072.92 �1.6157 30 015.40 �1.125 0 1.8550 3 20.975 056.0431791.7

Table 11Microgrid generation without uncertainty.

Time interval PCHP (kW) PMT (kW) PWT (kW) PPV (kW) PES (kW) PSell (kW) PBuy (kW)

1 74.22 0 62.7 0 5 30 02 58.20 0 62.7 0 3 30 03 45.35 0 46.19 68.15 �3 30 04 71.78 0 0 82.16 �3 30 05 90 16.77 0 17.35 0 0 06 90 17.19 62.7 0 3 30 0Total planning cost ($) 17.660Total planning pollution (kg) 1511.050

1 2 3 4 5 6Time interval

80

90

100

110

120

130

140

150

160

Power(kW)

Reserve marginStochasticDeterministicPeak load

Fig. 6. Reserve margin for stochastic and deterministic planning following 10% reduction in output power of WT.

Table 12Microgrid generation without VSG.

Time interval PCHP (kW) PMT (kW) PWT (kW) PPV (kW) PES (kW) PSell (kW) PBuy (kW)

1 80.14 6.86 62.7 0 5 30 02 73.69 7.18 53.51 0 0.075 30 03 66.58 9.4 34.81 60.49 �0.3343 30 04 80.66 12.58 0 72.92 �1.6157 30 05 85 13.34 23.51 15.40 �1.125 0 1.8556 90 26.38 60.55 0 3 20.975 0Total planning cost without VSG ($) 55.752Total planning pollution without VSG (kg) 1787.647

V.S. Tabar et al. / Journal of Cleaner Production 203 (2018) 1187e1200 1197

6.6. Sensitivity analysis

Sensitivity analysis on some parameters is presented in Table 15.Reserve margin constraint is an important parameter in the plan-ning. As shown in Table 15, by increasing minimum reserve, theamount of generated power will be increased. Thus, increasing inminimum reserve margin increases the cost and pollution. The costand pollution are increased from 56.043 $ and 1791.7 kg to 59.47 $and 1803.913 kg, respectively. As well, change the initial energy ofESS make an effect on pollution and cost. This issue is due toreducing the generated power by the component such as CHP andMT. In other words, when the initial energy of the batteries is

increased, the amount of generated power by conventional gener-ators will be decreased. As a result, cost and pollution will bedecreased from 56.043 $ and 1791.7 kg to 53.282 $ and 1783.115 kg,respectively. By increasing line limitation, the selling power to themain grid will be increased. Hence, the cost is decreased from56.043 $ to 15.676 $ but pollution is increased from 1791.7 kg to1905.909 kg.

7. Conclusions

This paper addresses a stochastic microgrid energy manage-ment including new features such as VSG, CCS system, and H2 fuel

1 2 3 4 5 6Time interval

0.8

1

1.2

1.4

1.6

1.8

2

2.2

Frequencydeviation(Hz/s)

Standard valueWithout VSGWith VSG

Fig. 7. Effect of VSG on microgrid stability.

Table 13Microgrid generation without CCS.

Time interval PCHP (kW) PMT (kW) PWT (kW) PPV (kW) PES (kW) PSell (kW) PBuy (kW)

1 80.14 6.86 62.7 0 5 30 02 73.69 7.18 53.51 0 0.075 30 03 66.58 9.4 34.81 60.49 �0.3343 30 04 80.66 12.58 0 72.92 �1.6157 30 05 85 13.34 23.51 15.40 �1.125 0 1.8556 90 26.38 60.55 0 3 20.975 0Total planning cost without CCS ($) 64.998Total planning pollution without CCS (kg) 1791.75

Table 14Microgrid generation without H2.

Time interval PCHP (kW) PMT (kW) PWT (kW) PPV (kW) PES (kW) PSell (kW) PBuy (kW)

1 80.14 6.86 62.7 0 5 30 02 73.69 7.18 53.51 0 0.075 30 03 66.58 9.4 34.81 60.49 �0.3343 30 04 80.66 12.58 0 72.92 �1.6157 30 05 85 13.34 23.51 15.40 �1.125 0 1.8556 90 26.38 60.55 0 3 20.975 0Total planning cost without H2 ($) 79.923Total planning pollution without H2 (kg) 1791.647

Table 15Sensitivity analysis in the planning.

Case Specifications Objective functions

Cost ($) Pollution (kg)

Nominal case 56.043 1791.75% Increasing minimum reserve margin 59.470 1803.91350% Increasing ESS initial energy 53.282 1783.11520% Increasing transfer line limitation 15.676 1905.909

V.S. Tabar et al. / Journal of Cleaner Production 203 (2018) 1187e12001198

station. As it was referred, environmental pollution and low inertiaproblem are the main issues that are dealt with by this paper. The

proposed problem optimizes pollution and cost at the same timesubject to several constraints. Wind speed, solar radiation, and loadare modelled as uncertain parameters and stochastic method hasbeen used to tackle the uncertainties. The proposed stochasticplanning is solved using CPLEX in GAMS software. The resultsdemonstrate that considering uncertainties in the planning in-creases the cost and pollution by 317.3% and 118.5%, respectively.But, the proposed stochastic planning is more robust underdifferent changes such as wind speed variation and can successfullyovercome the fluctuations. The pollution problem is moreoversolved by CCS and H2 station. The CCS system not only removes CO2

V.S. Tabar et al. / Journal of Cleaner Production 203 (2018) 1187e1200 1199

emission but also makes revenue for the network. In this regard,1791.75 kg of CO2 is stored by CCS and a revenue equal to 8.996 $ isachieved. Furthermore, new fuel station (H2 station) is compared tothe gasoline station in the microgrid. Installing H2 fuel station re-duces the cost by 70.12% and does not increase the pollution. As aresult, the pollution of microgrid is resolved by combination of CCSsystem and H2 station. The results also indicate that consideringVSG significantly increases the transient stability of the network. Asseen, considering VSG in microgrid increases cost and pollutionfrom 55.752 $ and 1787.647 kg to 56.043 $ and 1791.7 kg, respec-tively. But, VSG is required to support the stability of the network.As shown, the maximum required inertia is installed at fourth timeinterval due to lack of inertia in the power producers such as MTand WT. As well, because of existing sufficient inertia at secondtime interval, the required inertia is zero. As future work, followingitems can be suggested; (I) considering different microgrid struc-tures and technologies including new pure generators and effect ofthem on environmental issues, (II) combining short term planningwith long term planning, (III) investigating the effects of differentcomponents for long time interval, (IV) considering the consumedenergy of CCS as a separate variable.

References

Aliasghari, P., Mohammadi-Ivatloo, B., Alipour, M., Abapour, M., Zare, K., 2018.Optimal scheduling of plug-in electric vehicles and renewable micro-grid inenergy and reserve markets considering demand response program. J. Clean.Prod. 186, 293e303, 2018/06/10/.

Amini, M.H., Boroojeni, K.G., Dragi�cevi�c, T., Nejadpak, A., Iyengar, S.S., Blaabjerg, F.,2017. A comprehensive cloud-based real-time simulation framework foroblivious power routing in clusters of DC microgrids. In: 2017 IEEE SecondInternational Conference on DC Microgrids (ICDCM), pp. 270e273.

Amrollahi, M.H., Bathaee, S.M.T., 2017. Techno-economic optimization of hybridphotovoltaic/wind generation together with energy storage system in a stand-alone micro-grid subjected to demand response. Appl. Energy 202, 66e77, 9/15/.

Bahrani, B., Saeedifard, M., Karimi, A., Rufer, A., 2013. A multivariable designmethodology for voltage control of a single-DG-unit microgrid. IEEE Trans. Ind.Inf. 9 (no. 2), 589e599.

Balancing Mechnism Reporting System. Available: http://www.bmreports.com/.Bevrani, H., Ise, T., Miura, Y., 2014. Virtual synchronous generators: a survey and

new perspectives. Int. J. Electr. Power Energy Syst. 54, 244e254, 1//.Brabandere, K.D., Bolsens, B., Keybus, J.V.d., Woyte, A., Driesen, J., Belmans, R., 2007.

A voltage and frequency droop control method for parallel inverters. IEEE Trans.Power Electron. 22 (no. 4), 1107e1115.

Calculate and compensate for your CO2. Available: http://www.myclimate.org/.Center for Climate and Energy Solutions. Available: http://www.c2es.org/

technology/factsheet/CCS.Choobineh, M., Mohagheghi, S., 2016. A multi-objective optimization framework for

energy and asset management in an industrial Microgrid. J. Clean. Prod. 139 (no.Suppl. C), 1326e1338, 2016/12/15/.

Craparo, E., Karatas, M., Singham, D.I., 2017. A robust optimization approach tohybrid microgrid operation using ensemble weather forecasts. Appl. Energy201, 135e147, 2017/09/01/.

Dawoud, S.M., Lin, X., Okba, M.I., 2017. Optimal placement of different types of RDGsbased on maximization of microgrid loadability. J. Clean. Prod. 168 (no.Suppl. C), 63e73, 2017/12/01/.

H. F. Farahani, A. Rabiee, and M. Khalili, "Plug-in electric vehicles as a harmoniccompensator into microgrids," J. Clean. Prod., vol. 159, no. Suppl. C, pp.388e396, 2017/08/15/ 2017.

Haddadian, H., Noroozian, R., 2017. Multi-microgrids approach for design andoperation of future distribution networks based on novel technical indices.Appl. Energy 185, 650e663, 2017/01/01/.

Haddadian, H., Noroozian, R., 2017. Optimal operation of active distribution systemsbased on microgrid structure. Renew. Energy 104, 197e210, 2017/04/01/.

Hawkes, A.D., Leach, M.A., 2009. Modelling high level system design and unitcommitment for a microgrid. Appl. Energy 86 (no. 7e8), 1253e1265, 7//.

Hemmati, R., 2017. Technical and economic analysis of home energy managementsystem incorporating small-scale wind turbine and battery energy storagesystem. J. Clean. Prod. 159, 106e118.

Hemmati, R., 2018. Optimal design and operation of energy storage systems andgenerators in the network installed with wind turbines considering practicalcharacteristics of storage units as design variable. J. Clean. Prod. 185, 680e693,2018/06/01/.

Hooshmand, R.-A., Nosratabadi, S.M., Gholipour, E., 2018. Event-based scheduling ofindustrial technical virtual power plant considering wind and market pricesstochastic behaviors - a case study in Iran. J. Clean. Prod. 172, 1748e1764, 2018/

01/20/.Hossain, M.J., Pota, H.R., Mahmud, M.A., Aldeen, M., 2015. Robust control for power

sharing in microgrids with low-inertia wind and PV generators. IEEE Trans.Sustain. Energy 6 (no. 3), 1067e1077.

Igualada, L., Corchero, C., Cruz-Zambrano, M., Heredia, F.J., 2014. Optimal energymanagement for a residential microgrid including a vehicle-to-grid system.IEEE Trans. Smart Grid 5 (no. 4), 2163e2172.

Jirdehi, M.A., Tabar, V.S., Hemmati, R., Siano, P., 2017. Multi objective stochasticmicrogrid scheduling incorporating dynamic voltage restorer. Int. J. Electr. Po-wer Energy Syst. 93, 316e327, 12//.

Korkas, C.D., Baldi, S., Michailidis, I., Kosmatopoulos, E.B., 2016. Occupancy-baseddemand response and thermal comfort optimization in microgrids withrenewable energy sources and energy storage. Appl. Energy 163, 93e104, 2016/02/01/.

Le�si�c, V., Martin�cevi�c, A., Va�sak, M., 2017. Modular energy cost optimization forbuildings with integrated microgrid. Appl. Energy 197, 14e28, 7/1/.

Leung, D.Y.C., Caramanna, G., Maroto-Valer, M.M., 2014. An overview of currentstatus of carbon dioxide capture and storage technologies. Renew. Sustain.Energy Rev. 39, 426e443.

Lopes, M.T. a. L., 2013. A virtual synchronous machine to support dynamic fre-quency control in a mini-grid that operates in frequency droop mode. EnergyPower Eng. 5 (no. 3), 259e265.

Lu, X., Zhou, K., Yang, S., 2017. Multi-objective optimal dispatch of microgrid con-taining electric vehicles. J. Clean. Prod. 165 (no. Suppl. C), 1572e1581, 2017/11/01/.

Mallol-Poyato, R., Salcedo-Sanz, S., Jim�enez-Fern�andez, S., Díaz-Villar, P., 2015.Optimal discharge scheduling of energy storage systems in MicroGrids based onhyper-heuristics. Renew. Energy 83, 13e24, 11//.

Marzband, M., Ghadimi, M., Sumper, A., Domínguez-García, J.L., 2014. Experimentalvalidation of a real-time energy management system using multi-periodgravitational search algorithm for microgrids in islanded mode. Appl. Energy128, 164e174, 9/1/.

Marzband, M., Azarinejadian, F., Savaghebi, M., Guerrero, J.M., 2015. An optimalenergy management system for islanded microgrids based on multiperiodartificial bee colony combined with Markov chain. IEEE Syst. J. PP (no. 99), 1e11.

Marzband, M., Alavi, H., Ghazimirsaeid, S.S., Uppal, H., Fernando, T., 2017. Optimalenergy management system based on stochastic approach for a home Micro-grid with integrated responsive load demand and energy storage. Sustain. CitiesSoc. 28, 256e264, 1//.

Mavrotas, G., 2009. Effective implementation of the ε-constraint method in multi-objective mathematical programming problems. Appl. Math. Comput. 213 (no.2), 455e465, 7/15/.

Mazzola, S., Astolfi, M., Macchi, E., 2015. A detailed model for the optimal man-agement of a multigood microgrid. Appl. Energy 154, 862e873, 9/15/.

Motevasel, M., Seifi, A.R., 2014. Expert energy management of a micro-gridconsidering wind energy uncertainty. Energy Convers. Manag. 83, 58e72,2014/07/01/.

Nikmehr, N., Najafi-Ravadanegh, S., Khodaei, A., 2017. Probabilistic optimal sched-uling of networked microgrids considering time-based demand response pro-grams under uncertainty. Appl. Energy 198, 267e279, 7/15/.

Oliveira, D.Q., Zambroni de Souza, A.C., Santos, M.V., Almeida, A.B., Lopes, B.I.L.,Saavedra, O.R., 2017. A fuzzy-based approach for microgrids islanded operation.Electr. Power Syst. Res. 149, 178e189, 2017/08/01/.

Partridge, I., Gamkhar, S., 2012. A methodology for estimating health benefits ofelectricity generation using renewable technologies. Environ. Int. 39 (no. 1),103e110, 2//.

Rezaei, N., Kalantar, M., 2015. Smart microgrid hierarchical frequency controlancillary service provision based on virtual inertia concept: an integrated de-mand response and droop controlled distributed generation framework. EnergyConvers. Manag. 92, 287e301, 3/1/.

Rezvani, A., Gandomkar, M., Izadbakhsh, M., Ahmadi, A., 2015. Environmental/economic scheduling of a micro-grid with renewable energy resources. J. Clean.Prod. 87, 216e226, 2015/01/15/.

Riffonneau, Y., Bacha, S., Barruel, F., Ploix, S., 2011. Optimal power flowmanagementfor grid connected PV systems with batteries. IEEE Trans. Sustain. Energy 2 (no.3), 309e320.

Saboori, H., Hemmati, R., 2016. Considering carbon capture and storage in electricitygeneration expansion planning. IEEE Trans. Sustain. Energy 7 (no. 4),1371e1378.

Schoonenberg, W.C.H., Farid, A.M., 2017. A dynamic model for the energy man-agement of microgrid-enabled production systems. J. Clean. Prod. 164 (no.Suppl. C), 816e830, 2017/10/15/.

Soni, N., Doolla, S., Chandorkar, M.C., 2013. Improvement of transient response inmicrogrids using virtual inertia. IEEE Trans. Power Deliv. 28 (no. 3), 1830e1838.

Sreenivasulu, B., Gayatri, D.V., Sreedhar, I., Raghavan, K.V., 2015. A journey into theprocess and engineering aspects of carbon capture technologies. Renew. Sus-tain. Energy Rev. 41, 1324e1350.

Su, W., Wang, J., Roh, J., 2014. Stochastic energy scheduling in microgrids withintermittent renewable energy resources. IEEE Trans. Smart Grid 5 (no. 4),1876e1883.

Tabar, V.S., Jirdehi, M.A., Hemmati, R., 2017. Energy management in microgrid basedon the multi objective stochastic programming incorporating portable renew-able energy resource as demand response option. Energy 118, 827e839.

Tai, X., Sun, H., Guo, Q., Li, Z., 2017. A distributed power routing method betweenregional markets based on Bellman-Ford algorithm. In: 2017 IEEE Conference

V.S. Tabar et al. / Journal of Cleaner Production 203 (2018) 1187e12001200

on Energy Internet and Energy System Integration (EI2), pp. 1e5.US Environmental Protection Agency. Available: www.epa.gov.Vergara, P.P., L�opez, J.C., da Silva, L.C.P., Rider, M.J., 2017. Security-constrained

optimal energy management system for three-phase residential microgrids.Electr. Power Syst. Res. 146, 371e382, 2017/05/01/.

Villanueva, D., Feij�oo, A., 2010. Wind power distributions: a review of their appli-cations. Renew. Sustain. Energy Rev. 14 (no. 5), 1490e1495, 2010/06/01.

Wang, Z., Chen, B., Wang, J., Begovic, M.M., Chen, C., 2015. Coordinated energy

management of networked microgrids in distribution systems. IEEE Trans.Smart Grid 6 (no. 1), 45e53.

Xiao, F., Ai, Q., 2017. New modeling framework considering economy, uncertainty,and security for estimating the dynamic interchange capability of multi-microgrids. Electr. Power Syst. Res. 152, 237e248, 2017/11/01/.

Zachar, M., Daoutidis, P., 2017. Scheduling and supervisory control for cost effectiveload shaping of microgrids with flexible demands. J. Process Control. https://doi.org/10.1016/j.jprocont.2017.06.004.