You are on page 1of 24

Journal of Energy Storage 29 (2020) 101301

Contents lists available at ScienceDirect

Journal of Energy Storage


journal homepage: www.elsevier.com/locate/est

Day-ahead energy management and feeder reconfiguration for microgrids T


with CCHP and energy storage systems
Seyed Saeed Fazlhashemia, Mostafa Sedighizadeha, , Mohammad E. Khodayarb

a
Faculty of Electrical Engineering, Shahid Beheshti University, Evin, Tehran, Iran
b
Electrical and Computer Engineering Department, Southern Methodist University, 6251 Airline Rd, Dallas, TX 75205, United States

ARTICLE INFO ABSTRACT

Keywords: Energy management featuring distribution feeder reconfiguration (DFR) and reactive power control, improves
Combined cooling, heating, and power (CCHP) the technical and economic efficiency of microgrids. The present work proposes a framework that leverages
Distribution feeder reconfiguration (DFR) scenarios to jointly manage the real and reactive power dispatches of the controllable generation resources as
Distributed generation (DG) well as the topology of the distribution feeder. Multiple operation measures are optimized including the op-
Energy storage system (ESS)
eration cost, real power loss, the voltage stability index (VSI), and the greenhouse gas emissions of the microgrid.
Hybrid Big Bang- Big Crunch (HBB-BC) algorithm is used to solve the formulated optimization problem. Non-
dispatchable and dispatchable distributed generation units (DGs), as well as the battery and thermal energy
storage systems (BESS and TESS), are considered as a hybrid energy system. Combined cooling, heating, and
power (CCHP) units are considered as dispatchable DGs and wind and solar photovoltaic generations are con-
sidered as non-dispatchable DGs. The efficiency of the proposed model and solution algorithm is investigated
using a 33-bus microgrid, and the simulation outcomes are discussed.

1. Introduction optimal DFR and day-ahead power dispatch of generation resources to


minimize the operation cost of the microgrid. In [8], a stochastic model
1.1. Motivation and literature review for optimal day-ahead energy management and DFR is proposed, con-
sidering the unit commitment in the upstream power market. While the
Energy management systems improve the key operational factors in variability and uncertainty in wind generation and electric demand is
microgrids such as real power loss, voltage stability and emissions captured using Monte-Carlo simulation, the energy storage systems
[1–4]. DFR practices adjust the state of the sectionalizers and tie such as BESS and TESS are not addressed in the aforementioned re-
switches to affect the power flow in the distribution network and fur- search works. In [9] a two-stage algorithm is proposed for optimal day-
ther improve the operational metrics. Earlier research addressed the ahead energy management and DFR, considering non-dispatchable DGs
DFR and energy management in the distribution networks. A multi- in microgrids; however, the uncertainties in the demand and generation
objective fuzzy framework is presented in [2] to simultaneously opti- resources are not addressed. In [1], a multi-objective optimization
mize the distribution feeder topology and perform energy management problem is formulated for optimal energy management and DFR to
in distribution networks. Similar to [3], a Multi-Objective Hybrid Big minimize the real power loss, annual operation costs, and emissions
Bang-Big Crunch algorithm (MOHBB-BC) is applied to minimize the considering a forecasted pattern for wind speed, solar irradiance, and
power loss, emission, and operation cost while the voltage stability electric demand. In [10], the optimal energy management and DFR are
index is maximized, and the uncertainty in demand is captured con- simultaneously performed by using Hong's 2 m point estimation method
sidering the Triangular Fuzzy Number. While the proposed approaches based on an exchange market algorithm (EMA), to minimize the op-
in [2,3] address the energy management and demand uncertainty, the erational costs, and to improve the reliability and resilience of micro-
variation in demand profile is ignored. Moreover, the uncertainty in the grids, considering the uncertainty in the generation profile. In [11], the
generation of non-dispatchable DGs was not considered. The un- optimal generation schedule and DFR are determined by formulating a
certainty in the renewable generation resources has a dramatic impact multi-objective optimization problem that minimizes the real power
on strategic energy management decisions in microgrids [5,6]. The loss and phase unbalance and improves the voltage profile in the dis-
multi-objective stochastic model proposed in [7], determines the tribution network. In [12], DFR and energy management are performed


Corresponding author.
E-mail address: m_sedighi@sbu.ac.ir (M. Sedighizadeh).

https://doi.org/10.1016/j.est.2020.101301
Received 14 July 2019; Received in revised form 22 January 2020; Accepted 16 February 2020
2352-152X/ © 2020 Elsevier Ltd. All rights reserved.
S.S. Fazlhashemi, et al. Journal of Energy Storage 29 (2020) 101301

Nomenclature s (1 for discharging mode; otherwise 0).


fi(t,s) Value of objective function p at hour t in scenario s.
Sets fpmin Nadir value for objective function p.
fpmax The ideal value for objective function p.
T Set of hours in the operation period. ap(t,s) Fuzzy membership for objective function p at hour t in
NBR Set of branches. scenario s.
NHES Set of hybrid energy systems (HES). PG(z,t,s) Generated real power at bus z at hour t in scenario s
NCB Set of capacitor banks. (pu).
Nbus Set of buses. PD(z,t,s) Real power demand at bus z at hour t in scenario s
S Set of scenarios. (pu).
Np Set of populations. |V(z,t,s)| The absolute value of voltage magnitude at bus z bus
NDec Set of decision variables. at hour t in scenario s (pu).
Niter Set of iterations. δ(z,t,s) Voltage phase at bus z at hour t in scenario s.
D Set of distances between clusters. |Y(z,r)| The magnitude of admittance between buses z and r
(pu).
Variables φ(z,r,t,s) The phase of admittance between buses z and r (pu).
QG(z,t,s) Generated reactive power at bus z at hour t in scenario
Ploss(t,s) Real power loss at hour t in scenario s (kW). s (pu).
I(k,t,s) Current in the branch k at hour t in scenario s (A). QD(z,t,s) Demanded reactive power at bus z at hour t in scenario
VSIr(t,s) Voltage stability index for bus r at hour t in scenario s s (pu).
(pu). |I(z,r,t,s)| The magnitude of current of between buses z and r at
Vz(t,s) Voltage of bus z at hour t in scenario s (pu). hour t in scenario s (pu).
Pzr(t,s) Real power flow between buses z and r at hour t in ch
PBESS (i , t , s ) Real power charged by BESS in HES i at hour t in
scenario s (pu). scenario s (kW).
Qzr(t,s) Reactive power flow between buses z and r at hour t in dis
PBESS (i , t , s ) Real power discharged by BESS in HES i at hour t in
scenario s (pu). scenario s (kW).
Cgrid(t,s) Cost of energy exchanged with the upstream grid at EBESS(i,t,s) Energy stored in BESS in HES i at hour t in scenario s
hour t in scenario s ($/h). (kWh).
CWT(i,t,s) Cost of real power generated by WT of HES i at hour t
ch
HTESS (i , t , s ) Thermal energy stored in TESS in HES i at hour t in
in scenario s ($/h). scenario s (kW).
CPV(i,t,s) Cost of real power generated by PV of HES i at hour t
dis
HTESS (i , t , s ) Thermal energy withdrawn by TESS in HES i at hour t
in scenario s ($/h). in scenario s (kW).
CCCHP(i,t,s) Cost of real power generated by CCHP unit in HES i at ETESS(i,t,s) Stored thermal energy in TESS in HES i at hour t in
hour t in scenario s ($/h). scenario s (kWh).
CTESS(i,t,s) Cost of thermal energy exchanged with TESS in HES i Hab(i,t,s) Thermal power generated by auxiliary boiler in HES
at hour t and scenario s ($/h). i at hour t in scenario s (kW).
CBESS(i,t,s) Cost of electric energy exchanged with BESS in HES i gτ Fitness value for candidate solution τ.
at hour t and scenario s ($/h). X (h, ) Candidate solution θ for particle τ in iteration h.
Pgrid(t,s) Real power exchanged with the upstream grid at hour X c (h) Center of mass θ in iteration h.
t (kW). X gbest (h) Global best position all of the candidate solution found
PWT(i,t,s) Real power generated by WT of HES i at hour t in up to iteration h.
scenario s (kW). X pbest (h, )
Best previous experience for particle τ in iteration h.
PWT(i,t,s) Real power generated by PV of HES i at hour t in
scenario s (kW). Parameters
PCCHP(i,t,s) Real power generated by CCHP unit in HES i at hour t
in scenario s (kW). R(k) Resistance of kth branch (Ω).
HCCHP(i,t,s) Thermal power generated by CCHP unit in HES i at Rzr Resistance between buses z and r (pu).
hour t in scenario s (kW). Xzr Reactance between buses z and r (pu).
HAB(i,t,s) Thermal power generated by an auxiliary boiler in ρgrid(t) Price of electricity at hour t ($/kWh).
HES i at hour t in scenario s (kW). aWT(i) Fixed investment cost of WT in HES i ($/h).
CCB(l,t,s) Cost of reactive power generated by capacitor bank l aPV(i) Fixed investment cost of PV in HES i ($/h).
at hour t in scenario s ($/h). bWT(i) Variable cost of WT in HES i ($/h).
QCB(l,t,s) Reactive power generated by capacitor bank l at hour t bPV(i) Variable cost of PV in HES i ($/h).
in scenario s (kVAr). aCCHP(i) Fixed investment cost of CCHP unit in HES i ($/h).
PBESS(i,t,s) Real power exchanged by BESS in HES i at hour t in bCCHP(i) Variable cost of CCHP unit in HES i ($/h).
scenario s (kW). ηele Electrical efficiency of CCHP.
ch
UBESS (i , t , s ) Charging status of BESS in HES i at hour t in scenario s
DG
CostCapital Capital cost of DG ($/kW).
(1 for charging mode; otherwise 0). DG
PCapacity Capacity of DG (kW).
dis
UBESS (i , t , s ) Discharging status of BESS in HES i at hour t in sce Gr Annual discount rate.
nario s (1 for discharging mode; otherwise 0). CFDG(i) Capacity factor of DG in HES i.
HTESS(i,t,s) Thermal power exchanged by TESS in HES i at hour t ηthe Thermal efficiency of auxiliary boilers.
in scenario s (kW). TLife Lifetime of DG, capacitor bank, BESS, and TESS (year).
ch
UTESS (i , t , s ) Storing status of TESS in HES i at hour t in scenario s (1
O &M
CostCCHP Operation and maintenance cost of CCHP($/kWh).
for charging mode; otherwise 0).
Fuel
CostCCHP Fuel cost of CCHP ($/kWh).
dis
UTESS (i , t , s ) Withdrawal status of TESS in HES i at hour t in scenario CostWTO& M
Cost of operation and maintenance of WT ($/kWh).

2
S.S. Fazlhashemi, et al. Journal of Energy Storage 29 (2020) 101301

aCB(l) Fixed investment cost of capacitor bank l ($/h). max


ETESS (i ) Maximum stored thermal energy in TESS in HES i
bCB(l) Variable cost of capacitor bank l ($/h). (kWh).
CB
CostCapital Capital cost of capacitor bank ($/kVAr). ηEH Rate of thermal power generated by CCHP units.
CB
PCapacity Capacity of capacitor bank (kVAr). rτ Random number following Gaussian distribution.
CFCB(l) Capacity factor of capacitor bank l. β1 Parameter to limit the size of search space.
O &M
CostCB Cost of operation and maintenance of capacitor bank X max Upper limit of the θth decision variable.
($/kVArh). X min Lower limit of the θth decision variable.
aBESS(i) Fixed investment cost of BESS in HEs i ($/h). β2 Adjusting parameter for considering the effect of
bBESS(i) Variable cost of BESS in HEs i ($/h). global optimum on new position of candidates.
aTESS(i) Fixed investment cost of TESS in HES i ($/h). β3 Adjusting parameter for considering the effect of local
bTESS(i) Variable cost of TESS in HES i ($/h). optimum on new position of candidates.
ρgas Price of natural gas in the upstream market ($/m3). si Solar irradiance (kW/m2).
βgas Rate of exchanging natural gas to electricity (m3/ fb(si) Beta PDF of si.
kWh). αs Parameter of the Beta PDF.
BESS
CostCapital Capital cost of BESS ($/kW). βs Parameter of the Beta PDF.
BESS
PCapacity Capacity of BESS (kW). μs Mean of forecasted solar irradiance (kW/m2).
TESS
CostCapital Capital cost of TESS ($/kW). σs Standard deviation of forecasted solar irradiance (kW/
TESS
PCapacity Capacity of TESS (kW). m2).
CFBESS(i) Capacity factor of BESS in HES i. ηpv Efficiency of PV module.
CFTESS(i) Capacity factor of TESS in HES i. Spv Area of PV module (m2).
ER(i) Emission rate of DG in HES i (kg/kWh). Prated Rated output power of WT (kW).
ERgrid Emission rate of grid (kg/kWh). vr Rated wind speed (m/s).
Ikmax Permitted current of branch k (pu). v Wind speed (m/s).
Vmin Maximum allowable voltage (pu). vct Cut-in wind speed (m/s).
Vmax Minimum allowable voltage (pu). vco Cut-out wind speed (m/s).
min
PCCHP (i ) Minimum allowable real power generated by CCHP μd Mean of forecasted electrical demand (kW).
unit in HES i (kW). σd Standard deviation of forecasted electrical demand
max
PCCHP (i ) Maximum allowable real power generated by CCHP (kW).
unit in HES i (kW). ∂ Vector of random variables between zero and one.
min
HCCHP (i ) Maximum allowable thermal power generated by K Number of clusters.
CCHP unit in HES i (kW). Dheating(t,s) Thermal power demand by consumers at hour t in
max
HCCHP (i ) Minimum allowable thermal power generated by scenario s (kW).
CCHP unit in HES i (kW). Dcooling(t,s) Cooling power demand by consumers at hour t in
min
HAB (i ) Maximum allowable thermal power generated by scenario s (kW).
auxiliary boiler in HES i (kW). BESS Battery Energy Storage System.
max
HAB (i ) Minimum allowable thermal power generated by BB-BC Big Bang-Big Crunch.
auxiliary boiler in HES i (kW). BM Biomass.
min
QCB (l ) Minimum allowable reactive power generated by ca- CCHP Combined Cooling, Heating and Power.
pacitor bank l (kVAr). CHP Combined Heat and Power.
max
QCB (l ) Maximum allowable reactive power generated by ca- DER Distributed Energy Resource.
pacitor bank l (kVAr). DFR Distribution Feeder Reconfiguration.
ch max
PBESS (i ) Maximum real power charged by BESS in HES i (kW). DG Distributed Generation.
dis max
PBESS (i ) Maximum real power discharged by BESS in HES i FC Fuel Cell.
(kW). GT Geothermal.
ch
BESS Charging efficiency of BESS. HBB-BC Hybrid Big Bang Big Crunch.
dis
BESS Discharging efficiency of BESS. HES Hybrid Energy System.
min
EBESS (i ) Minimum energy in BESS in HES i (kWh). MGO Microgrid Operator.
max
EBESS (i ) Maximum energy in BESS in HES i (kWh). MT Micro-turbine.
ch max
PTESS (i ) Maximum thermal power charged by TESS in HES PSO Particle Swarm Optimization.
i (kW). PV Photovoltaic
dis max
PTESS (i ) Maximum thermal power discharged by TESS in HES i PDF Probability Distribution Function.
(kW). SH Small Hydro.
ch
TESS Storage efficiency of TESS. TESS Thermal Energy Storage System.
dis
TESS Withdrawal efficiency of TESS. VSI Voltage Stability Index.
min
ETESS (i ) Minimum stored thermal energy in TESS in HES i WT Wind Turbine.
(kWh).

by using a stochastic min-max problem, that minimizes the number of (CSA), to improve the power loss and voltage stability indices. In [15],
switching and the real electric power loss while stabilizes the bus vol- an expansion planning problem is formulated considering the network
tages . In [13], an alternative multi-objective optimization problem is reconfiguration and energy management to minimize the investment
proposed for optimal DFR and energy management, which maximizes and operation costs of DGs as well as demand-side management costs.
the DG owner's profit, and minimizes the cost of the distribution net- In [16], the dedicated search teaching-learning based optimization
work operator. In [14], a multi-objective optimization problem for DFR (DSTLBO) algorithm is proposed for allocating the distributed genera-
and energy management is solved using a cuckoo search algorithm tion capacity to minimize the energy loss and to improve the voltage

3
S.S. Fazlhashemi, et al.

Table 1
Recent research works addressing DFR and energy management in microgrids.
Reference No. Type of formulation Objective function Uncertainty Solution method Reactive power DER
control
Demand Generation Mathematical Heuristic

[1] Deterministic Real power loss in the network, annual operation costs – – – Strength Pareto Evolutionary Algorithm – WT, PV
and emissions 2 (SPEA2)
[2] Deterministic Operation cost, real power loss in the network, emission, and – – – HBB-BC – WT, PV, MT, FC
voltage stability index
[3] Stochastic Operation cost, real power loss in the network, emission, and ✓ – – Pareto based HBB-BC – WT, PV, MT, FC
voltage stability index
[7] Stochastic Operation cost, real power loss in the network, reliability, and – ✓ MISOCP (GAMS) – – WT, PV, BM, MT, SH, FC,
number of switching GT, Electric Vehicle
[8] Stochastic Benefit of MGO ✓ ✓ – PSO – WT, MT, BESS
[9] Deterministic Cost of energy – – – PSO – WT, PV

4
[10] Stochastic Operation cost, resiliency, and reliability ✓ ✓ – EMA – WT, PV, CHP, BESS
[11] Deterministic Phase balancing, real power loss in the network, improving the – – – Bacterial Foraging with Spiral Dynamic – General form of DER
voltage profile, and operation cost algorithm
[12] Stochastic Real power loss in the network, voltage stability, and the ✓ – – Knee point driven evolutionary – SH
number of switching algorithm
[13] Stochastic DG owner's profit and the distribution company's costs ✓ ✓ MINLP (GAMS) – – WT
[14] Deterministic Real power loss in the network and voltage stability index – – – CSA – General form of DER
[15] Deterministic Operation cost and demand side cost – – – Differential evolution algorithm – General form of DER
[16] Deterministic energy loss and voltage profile – – – DSTLBO ✓ General form of DER
[17] Deterministic Costs of line upgrades, energy loss, switching operations, – – - Non-dominated sorting genetic – General form of DER
emission, and captial, operation and maintenance cost of DGs algorithm (NDSGA)–
[18] Deterministic Real power loss and voltage profile in the network – – – Harmony Search Algorithm (HSA) – General form of DER
[19] Deterministic Real power loss and voltage profile in the network – – – integrating PSO and ABC algorithm ✓ General form of DER
with HSA
[20] Deterministic Real power loss and voltage profile in the network – – – bacterial foraging optimization (BFO) ✓ General form of DER
[21] Deterministic Operation cost, power quality and reliability – – – NSGA-II ✓ General form of DER
Current paper Stochastic Operation cost, real power loss in the network, emission, and ✓ ✓ – HBB-BC ✓ WT, PV, CCHP, TESS, BESS
voltage stability index
Journal of Energy Storage 29 (2020) 101301
S.S. Fazlhashemi, et al. Journal of Energy Storage 29 (2020) 101301

profile in the distribution network. In [17], network reconfiguration is technique.


leveraged as a tool to improve the expansion planning practices in The main contributions of this paper are as follows:
distribution networks considering the environmental and economic
measures, including investment and operation costs of DGs and network • A stochastic model is proposed for microgrid operation with coupled
components, as well as the emission of DG and utility grid. In [18], the DFR, day-ahead energy management, and hourly reactive power
metaheuristic harmony search algorithm combined with sensitivity control for improving the voltage profile and the VSI by switchable
analysis is presented to minimize the power loss using network re- capacitor banks.
configuration. • An integrated model is proposed for microgrids, considering the
There are few pieces of research that addressed the optimal network HESs to reduce operation cost, real power loss, and emissions and to
reconfiguration and energy management with the allocating capacitor improve the VSI.
banks. In [19], particle artificial bee colony and harmony search al- • A scenario-based method is proposed for addressing the uncertainty
gorithms are integrated to procure the network reconfiguration and in wind speed, solar irradiance, and electrical demand.
capacitor bank expansion strategies that minimize the real power loss • A sensitivity analysis is performed to assess the technical, economic,
and improve the nodal voltage magnitude in the distribution networks. and environmental aspects of the day-ahead operation of microgrid
In [20], a fuzzy bacterial foraging optimization (BFO) method is de- considering the variation in thermal demand.
veloped for network reconfiguration as well as capacitor bank and DG
capacity allocation to minimize the real power loss in the distribution The rest of this paper is organized as follows. In Section 2, the multi-
network. In [21], a non-dominated sorting genetic algorithm (NSGA-II) objective optimization problem is formulated. The uncertainties in
is used for capacitor bank and renewable energy resource allocation as generation and demand profiles are addressed in Section 3. The solution
well as network reconfiguration considering the voltage stability in- algorithm is presented in Section 4. Section 5 presents and discusses the
dices, emission, as well as the operation, and capacity expansion costs. simulation results. Section 6 addresses the conclusions of this research.
The studies carried out for coordinated DFR and energy manage-
ment in microgrids are classified based on the following criteria: 1) the 2. HES architecture and problem formulation
type of problem (determinestic/stochastic), 2) the objective, 3) the type
of the DERs considered, and 4) the proposed solution methodology. The typical system under study is a MG equipped by two HESs
These studies are classified considering these criteria in Table 1. which include solar PV, wind generation, CCHP, auxiliary boiler (AB),
BESS, and TESS. The architecture of the proposed HES is illustrated in
1.2. Contributions and organization Fig. 1. The CCHP is composed of several components: the power gen-
eration unit (PGU), heat recovery unit (HRU), and absorption chiller
This paper addresses the optimization of day-ahead energy man- (AC). The CCHP operates in following hybrid load mode [22]. The
agement and DFR in a microgrid equipped with hybrid energy systems under syudy system considers three types of energy demands, including
(HES) and capacitor banks. Each HES includes one dispatchable DG electrical, thermal, and cooling demand. In this paper, it is assumed
capable of generating electricity and heat (i.e., also referred to as the that the MGO has access to all the necessary data. The energy man-
CCHP unit), two non-dispatchable DGs (wind turbine and PV genera- agement is carried out for 24 hours with 1-hour time step. Details of the
tion unit), two energy storage systems (BESS and TESS), and one aux- thermodynamic modelling of the CCHP, TESS, and BESS can be found
iliary boiler. A multi-objective function is formulated to determine the in [23,24], and [25], respectively. However, in this work, similar to
hourly configuration of the microgrid network, and the hourly dispatch [10], [22] and [26], the details of the thermodynamic models of the
of the thermal and electrical generation of CCHP units, thermal gen- these assets are not considered As pointed out in [23], the efficiency of
eration of the auxiliary boiler, and charging/discharging power of BESS HES components is a function of the generated/charged/discharged
and TESS. The real power loss, operation cost, and emissions are electrical or thermal power; However, similar to [10,22] and [26], the
minimized and the VSI index is maximized. The HBB-BC algorithm efficiency is a fix parameter in the presented formulation.
solves the proposed multi-objective optimization problem in a fuzzy
framework. The uncertainty in the wind and solar PV generation and 2.1. Problem formulation
electric demand is taken into account using a number of the scenarios
and the representive scenarios are selected by a scenario reduction This subsection describes the problem formulation including the

Fig. 1. The architecture of HES.

5
S.S. Fazlhashemi, et al. Journal of Energy Storage 29 (2020) 101301

Fig. 2. The Beta probability distribution of solar irradiation at 12:00 pm


[36,37].
Fig. 3. Mean and standard deviation of solar irradiance [36,37].

objective function and the network and generation units’ constraints.

2.2.1. Objective functions

• Minimization of real power loss


Here, the real power loss is formulated as follows [27]:

f1 (t , s ) = Ploss = R ( k ) × I (k , t , s ) 2 t T, s S
k NBR (1)

• Maximization of VSI Fig. 4. Hourly mean wind speed [37,38].


In order to ensure the stability of microgrid, VSI is maximized. For a
typical radial feeder with two end buses of z and r, the VSI for bus r is
expressed as follows [2,28]:

VSIr (t , s ) = [Vz (t , s )]4 4 × [Pzr (t , s ) × Xzr Qzr (t , s ) × Rzr ]2


4 × [Vz (t , s )]2 × [Pzr (t , s ) × Rzr Qzr (t , s ) × Xzr ]
t T, s S, r, z Nbus forz r (2)
Here, the voltage of all buses and the currents in all branches are
calculated by running the power flow analysis, By running the power
flow, the real and reactive power of each line (Pzr and Qzr) are calcu-
lated and consequently, the VSIs of the buses are determined. The bus
with the minimum VSI value that has the maximum sensitivity to vol-
tage instability is referred to as the critical bus. The critical bus features
the largest VSI deviation from the unity and is prone to voltage collapse.
In order to improve the weakest network bus with highest VSI devia-
tion, as one of the objectives, the largest VSI deviation is minimized as
[2,28]:
f2 (t , s ) = max{1 VSIr (t , s )} t T, r Nbus
r (3) Fig. 5. The Weibull probability distribution of wind speeds at 4:00 am [37,39].

• Minimization of the total operational cost


The total operational cost of microgrid includes the cost of energy
exchanged with the upstream grid, BESS, TESS, the cost of real power
supplied by DGs, and the cost of reactive power provided by the ca-
pacitor banks as shown in (4) [29]. The total operation cost is mini-
mized within the operation horizon.
f3 (t , s )=
Cgrid (t , s ) + i NHES
CWT (i , t , s ) + i NHES
CPV (i, t , s )

+ i NHES CCCHP (i , t , s ) + l NCB CCB (l, t , s )+

i NHES
CTESS (i, t , s ) + i NHES
CBESS (i, t , s ) t T, s S

(4) Fig. 6. Normal probability distribution of electrical demand at 12:00 pm


[37,40].
The cost of energy exchanged with the upstream grid is given by
[29]:
Cgrid (t , s ) = Pgrid (t , s ) × grid (t ) t T, s S (5)

6
S.S. Fazlhashemi, et al. Journal of Energy Storage 29 (2020) 101301

The cost of energy (electric and thermal power) generated by CCHPs CB


CostCapital CB
× PCapacity × Gr
consists of two parts: fixed investment costs (cost of equipment, infra- aCB (l) = l NCB
TLife × 365 × 24 × CFCB (l) (12)
structure, commissioning, etc.) and the variable costs (cost of operation,
maintenance and fuel). For solar PV and WT generation units, the cost O &M
bCB (l) = CostCB l NCB (13)
of power generation is given by [24]:
The cost of energy exchanged with the BESS and TESS is computed
CWT (i , t , s ) = aWT (i) + bWT (i) × PWT (i , t , s ) t T, s S, i NHES as follows:
(6) For BESS:

CPV (i , t , s ) = aPV (i) + bPV (i) × PPV (i , t , s ) t T, s S, i NHES (7) CBESS (i, t , s ) = aBESS (i) + bBESS (i ) × PBESS (i, t , s )

For CCHPs, the cost of energy generation is given by [24]: + grid (t ) × PBESS (i , t , s ) t T, s S, i NHES
(14)
PCCHP (i , t , s ) Hab (i , t , s )
CCCHP (i, t , s ) = aCCHP (i ) + bCCHP (i) × + For TESS:
ele the

t T, s S, i NHES (8) CTESS (i, t , s ) = aTESS (i ) + bTESS (i ) × HTESS (i , t , s )


+ gas (t ) × HTESS (i, t , s ) t T, s S, i NHES
where
(15)
DG DG
CostCapital × PCapacity × Gr
aCCHP (i ) = aWT (i) = aPV (i ) = i NHES where
TLife × 365 × 24 × CFDG (i )
BESS BESS
CostCapital × PCapacity × Gr
(9) aBESS (i) = i NHES
TLife × 365 × 24 × CFBESS (i) (16)
O&M Fuel O&M
bCCHP (i ) = CostCCHP + CostCCHP , bWT (i) = CostWT , bPV (i) TESS TESS
CostCapital × PCapacity × Gr
O& M
= CostPV Fuel
CostCCHP = × i NHES (10) aTESS (i) = i NHES
(17)
gas gas
TLife × 365 × 24 × CFTESS (i )
The cost of reactive power generated by the capacitor banks is
bBESS (i ) = O &M
CostBESS , bTESS (i ) = O&M
CostTESS i NHES (18)
formulated as follows [30]:

CCB (l, t , s ) = aCB (l) + bCB (l) × QCB (l, t , s ) t T, s S, • Minimization of the total emission
l NCB (11)
The total emission of the upstream grid and the CCHP units in HES
where for the generating electrical and thermal energy is formulated as

Fig. 7. HBB-BC algorithm to solve multi-objective optimization problem.

7
S.S. Fazlhashemi, et al. Journal of Energy Storage 29 (2020) 101301

follows: cloumn is removed to procure a square branch-to-node matrix “A”. The


determinant of "A" conveys the readiality and connectivity of the net-
f4 (t , s ) = Pgrid (t , s ) × LF × ERgrid work.. If the determinant of “A” is 1 or −1, the radiality and con-
PCCHP (i , t , s ) nectivity are guaranteed; however, if the determinant of "A" is zero, the
+ × CFDG (i) × ER (i ) radiality and connectivity constraint is not satisfied
i NHES ele

+
Hab (i , t , s )
× CFDG (i) × ER (i ) t T, s S
• Branch current limit
the (19)
In order to ensure that the currents in distribution cables do not
exceed the limits, the following constraint is considered [2]:
2.2.2. Aggregating objective functions by a fuzzy operator I (k , t , s ) Ikmax t T, s S, k NBR (25)
Since the formulated objective functions (1), (3), (4), (19) have
different characteristics and scales, they should properly be scaled in
order to find a feasible optimal solution. This paper uses a Fuzzy op- • Bus voltage limits
erator for such scaling as shown below [31,32]:
The voltage at each bus should be within the minimum and max-
1 fp (t , s ) < fpmin imum limits as shown below [2]:
f pmax f p (t , s ) Vmin V (z , t , s ) Vmax t T, s S, z Nbus (26)
a p (t , s ) = fpmin f p (t , s ) fpmax t T, s S,
f pmax f pmin

0 f p (t , s ) fpmax
• Generation limits for CCHPs
p {1, 2, 3, 4} (20)
The produced electrical and thermal power by CCHPs in HES should
To combine the scaled objective functions, the “max geometric satisfy the following limits [2,22]:
mean” operator is used as shown below [31,32]: min
PCCHP (i ) PCCHP (i , t , s ) max
PCCHP (i ) t T, s S, i NHES (27)
µ (t , s ) = [ 1 (t , s ) × 2 (t , s) × 3 (t , s) × 4 (t , s )]1/4 t T, s S
min max
HCCHP (i ) HCCHP (i , t , s ) HCCHP (i ) t T, s S, i NHES
(21)
(28)
The overall fitness function is written as:

Max F = µ (t , s )
(22)
• Auxiliary boiler operation limits
s S t T
The thermal power generated by auxiliary boilers of HES should be
within the minimum and maximum limits as shown below [22]:
2.2.3. Network and generation units’ constraints
The following constraints should be considered for the proposed min
HAB (i ) HAB (i, t , s ) max
HAB (i ) t T, s S, i NHES (29)
multi-objective optimization problem.
• Reactive power limits for the capacitor banks
• Power flow equations
The output reactive power of capacitor banks should meet the fol-
Real and reactive electrical power balance at each bus of microgrid lowing conditions [34]:
should satify the following conditions [27]:
min
QCB (l) QCB (l, t , s ) max
QCB (l ) t T, s S, l NCB (30)
PG (z , t , s ) PD (z , t , s ) = V (z , t , s ) × V (r , t , s ) × Y ( z , j )
r Nbus

× cos( (z , t , s ) (r , t , s ) (z, r , t , s ))
• BESS operation constraints
t T, s S, z Nbus for z r (23) The following constraints are considered for the operation of BESS
in HES [22,35]:
QG (z, t , s ) QD (z, t , s ) = V (z , t , s ) × V (r , t , s ) × Y (z , r ) ch ch max Ch
r Nbus
0 PBESS (i , t , s ) PBESS (i) × U BESS (i , t , s ) t T, s S,

× sin( (z, t , s ) (r , t , s ) (z , r , t , s )) i NHES (31)

t T, s S, z Nbus for z r (24) 0 dis


PBESS (i , t , s ) dis max
PBESS dis
(i) × UBESS (i , t , s ) t T, s S,
(32)
• Network radiality and connectivity
i NHES
dis
UBESS Ch
(i , t , s ) + U BESS (i , t , s ) 1 t T, s S, i NHES (33)
The preservation of the radiality while ensuring the concectivity is
an important constraint in the DFR [32]. In order to validate this EBESS (i, t , s ) = EBESS (i , t 1, s ) dis
PBESS (i , t , s ) × dis
BESS
constraint, the bus incidence matrix is formed based on the Kirchhoff ch
algebraic method [31,33]. This matrix consists of one row and one PBESS (i , t , s )
+ ch
t T , t > 1, s S, i NHES
column for each node, respectively. The following rule is considered in BESS
arranging the elements of the bus incidence matrix: (34)

•a (35)
min max
k,z = 0 if branch k is not connected to node z EBESS (i ) EBESS (i , t , s ) EBESS (i ) t T, s S, i NHES
•a k,z= 1 if branch k is directed away from node z
•a k,z = −1 if branch k is directed toward node z • TESS operation constraints
One node is taken into account as the reference and associated The following constraints are considered for the operation of TESS

8
S.S. Fazlhashemi, et al. Journal of Energy Storage 29 (2020) 101301

in HES [22,35]:
ch ch max Ch
0 HTESS (i, t , s ) HTESS (i) × UTESS (i , t , s ) t T, s S, i NHES
(36)
dis dis max dis
0 HTESS (i , t , s ) HTESS (i) × UTESS (i , t , s ) t T, s S, i NHES
(37)
dis
UTESS Ch
(i, t , s ) + UTESS (i , t , s ) 1 t T, s S, i NHES (38)

dis dis
ETESS (i, t , s ) = ETESS (i, t 1, s ) HTESS (i , t , s ) × TESS
ch
HTESS (i , t , s) Fig. 9. The forecasted hourly electrical demand of microgrid [7].
t T , t > 1, s S, i NHES
(39)
ch
TESS

3.1. PV generation output


min
ETESS (i ) ETESS (i, t , s ) max
ETESS (i ) t T, s S, i NHES (40)
The bimodal distribution represents the probability distribution of
• Heating and cooling demands’ constraints the PV generation at each hour. Here, the bimodal distribution is ex-
pressed as a linear combination of two unimodal distributions. Each
The heating and cooling demands are served by the thermal energy unimodal distribution is presented by a he Beta PDF shown in Fig. 2
generated by CCHPs, auxiliary boilers, and TESSs. It is assumed the [36,37]:
heating and cooling demands are located close to the CCHPs. The ( s + s)
heating and cooling demands should satisfy the following constraints × si ( s 1) × (1 si )( s 1) 0 si 1, s 0, s 0
fb (si) = ( s ) ( s)
pertaining to thermal energy balance [22,35]: 0 otherwise
PCCHP (i , t , s ) (43)
HCCHP (i , t , s ) = t T, s S, i NHES
EH (41)
µs × (1 + µs )
s = (1 µs ) × 2
1
dis
HCCHP (i , t , s ) + Hab (i , t , s ) + HTESS (i , t , s ) ch
HTESS (i , t , s ) s (44)

= D heating (t , s) + D cooling (t , s) t T, s S, i NHES (42) µs × s


s =
(1 µs ) (45)

3. Uncertainty in generation and demand The mean and standard deviation of the solar irradiance are cal-
culated using the historical data received from the meteorology stations
This section describes the models used to represent the uncertainty [37,38].
in the wind and solar PV generation and the electricity demand in the It is assumed that the PV units are in the location where μs and σs of
microgrid. the solar irradiation in a day-ahead operation horizon, are similar to

Fig. 8. The 33-bus microgrid.

9
S.S. Fazlhashemi, et al. Journal of Energy Storage 29 (2020) 101301

Fig. 10. The participation factor of demand on distribution network buses [7].

Table 2
Parameters of HES and capacitor bank*.
CCHP [22] Auxiliary Boiler [22] Wind Generation [37]

Parameter Value Parameter Value Parameter Value

min
PDG 0 min
HAB 0 DG
CostCapital 1500
D
max
PDGD 633 max
HAB 350 DG
PCapacity 400
min
HDG 0 ηthe 0.8 Gr 0.136
D
max
HDG 700 BESS [49] CFDG 0.2
D
ρgas 0.157 Parameter Value TLife 20
βgas 0.09043 BESS
CostCapital 1775 O& M
CostDG 0.05
ND
ηele 0.3 BESS
PCapacity 1200 Prated 400
DG
CostCapital 3674 ch max
PBESS 200 vr 12
DG
PCapacity 633 dis max
PBESS 200 vco 25
Gr 0.136 CFBESS 0.25 vct 3.5
Fig. 11. The thermal and cooling demand [22]. CFDG 0.2 min
EBESS 120 PV [37]
TLife 10 max
EBESS 1200 Parameter Value
O& M
CostDG 0.0039 ch 0.85 DG
CostCapital 6675
D BESS
ηEH 0.9 dis 0.95 DG
PCapacity 400
BESS
ER 14.447 TLife 25 Gr 0.136
TESS [49] Gr 0.136 CFDG 0.25
Parameter Value O& M
CostBESS 0.05 TLife 20
TESS
CostCapital 1800 Capacitor Bank [50] O& M
CostDG 0.05
ND
TESS
PCapacity 1200 Parameter Value ηpv 18.6
ch 0.95 CB
CostCapital 9 Spv 40
TESS
dis 0.95 CB
PCapacity 400 Grid [3]
TESS
min
ETESS 120 Gr 0.136 Vmin 0.95
max
ETESS 1200 CFCB 0.2 Vmax 1.05
CFTESS 0.25 TLife 25 ERgrid 5.46
Fig. 12. The day-ahead electricity price for microgrid [22].
ch max
PTESS 200 O& M
CostCB 0.05
dis max
PTESS 200
those in Fig. 3. TLife 25
The output power of the PV units is determined by incorporating the O& M
CostTESS 0.05
efficiency of the PV cell, the area of the cell, and the available solar Gr 0.136
irradiance. The output power of the PV unit is computed as follows
[36,37]:

Units of quantities have been expressed in nomenclature.

(46)
Ppv (si ) = pv × S pv × si
Table 3
The initial parameters selected for HBB-BC algorithm.

3.2. Wind generation output Parameters Values Parameters Values

β1,β2 1, 0.4 Pop Size 50


The uncertainty in wind speed is represented by the Weibull prob- β3 0.8 Max Iteration 100
ability distribution function with two parameters: the scale factor, a, Pm 0.2 Max Trial 30
and shape factor, b. In this paper, these parameters are set
to a = vmean/0.9 and b = 2 . The forecasted mean of wind speed, vmean, is
shown in Fig. 4 and the Weibull probability distribution function is
given by [37,38]:
b
fw (v ) = ba bv b 1e ( ax ) (47)

10
S.S. Fazlhashemi, et al. Journal of Energy Storage 29 (2020) 101301

Table 4
The network reconfiguration decisions and values of objective functions at each hour for Case 1.
Hour The expected values considering all scenarios The outcomes for the most probable scenario

Emission (Ton) Cost ($) VSI Power Loss (kW) Open Switches Emission (Ton) Cost ($) VSI Power Loss (kW) Open Switches

1 2.210 142.4 0.905 22.8 S7,S10,S34,S28,S36 2.210 160.6 0.901 24.1 S7,S9,S14,S37,S32
2 2.251 108.2 0.883 34.6 S7,S21,S11,S5,S36 2.250 127.1 0.890 32.3 S33,S9,S8,S28,S32
3 2.185 95.9 0.878 40.5 S6,S9,S34,S37,S36 2.178 111.9 0.896 27.2 S7,S9,S14,S37,S32
4 2.186 84.8 0.876 41.17 S6,S9,S14,S37,S32 2.194 101.3 0.858 54.7 S33,S9,S8,S28,S32
5 2.211 88.2 0.850 65.61 S33,S9,S34,S28,S32 2.206 104.8 0.861 56.2 S7,S9,S14,S37,S32
6 2.271 135.6 0.832 79.70 S7,S11,S13,S26,S32 2.277 154.6 0.825 90.8 S7,S9,S14,S37,S32
7 2.537 153.2 0.824 93.06 S7,S9,S14,S28,S32 2.538 182.2 0.821 95.4 S7,S9,S14,S37,S32
8 2.590 249.6 0.791 120.3 S7,S9,S34,S37,S32 2.481 278.5 0.814 104.2 S7,S9,S14,S37,S32
9 2.625 830.0 0.797 125.2 S7,S9,S13,S28,S32 2.631 863.6 0.790 134.7 S7,S9,S14,S37,S32
10 2.675 2241. 0.699 201.7 S33,S35,S34,S37,S36 2.653 2255 0.771 163.1 S7,S9,S14,S37,S32
11 2.731 2297 0.646 294.1 S33,S35,S34,S37,S36 2.679 2277 0.748 203.1 S7,S9,S14,S37,S32
12 2.762 2321 0.621 345.6 S33,S35,S34,S37,S36 2.684 2274 0.738 207.7 S33,S9,S8,S28,S32
13 2.705 883.2 0.696 277.6 S7,S33,S14,S28,S36 2.695 910.8 0.718 260.1 S7,S9,S14,S37,S32
14 2.718 2342 0.631 323.7 S33,S35,S34,S37,S36 2.667 2321 0.724 233.6 S33,S9,S8,S28,S32
15 2.588 1124 0.800 110.5 S7,S13,S11,S37,S32 2.594 1157 0.802 119.6 S7,S9,S14,S37,S32
16 2.564 1086 0.835 77.8 S6,S9,S14,S28,S36 2.565 1117 0.832 78.8 S33,S9,S8,S28,S32
17 2.552 352.5 0.855 60.8 S6,S9,S13,S28,S36 2.551 382.6 0.852 60.1 S33,S9,S8,S28,S32
18 2.563 250.6 0.829 80.2 S33,S8,S11,S28,S36 2.559 280.3 0.842 73.1 S33,S9,S34,S28,S32
19 2.568 227 0.82 96.9 S7,S11,S14,S28,S32 2.573 257.4 0.813 104.9 S7,S9,S14,S37,S32
20 2.614 246 0.78 136.8 S7,S8,S13,S28,S36 2.615 276.9 0.789 137.4 S7,S9,S14,S37,S32
21 2.617 697 0.79 119.7 S7,S9,S14,S26,S32 2.628 732.1 0.787 139.4 S7,S9,S14,S37,S32
22 2.594 291.6 0.84 70.4 S7,S11,S34,S27,S32 2.595 323.7 0.838 77.1 S7,S9,S14,S37,S32
23 2.237 194.3 0.88 35.6 S33,S11,S14,S27,S32 2.240 213.0 0.880 40.9 S7,S9,S14,S37,S32
24 2.197 165.1 0.91 21.6 S7,S11,S14,S27,S32 2.197 182.1 0.908 20.8 S7,S9,S14,S37,S32

The output power of the WT is computed as follows [37,39]:

00 v vct
(v vct )
Prated × vct v vr
Pw (v ) = (v vct )
Prated vr v vco
0 vc o v (48)

Fig. 5 shows the considered Weibull probability distribution for


wind speed.

3.3. Uncertainty in the electric demand

The uncertainty in the forecasted electric demand is captured by the


Fig. 13. The output thermal power of auxiliary boilers (ABs).
normal probability distribution function formulated below [37,40]:
(l µd )2
1 2 d2
fd (l) = ×e
d× 2
l= × d + µd (49)

Fig. 6 shows the considered normal probability distribution for the


forecast error of demand at 12:00 pm.

3.4. State selection

To capture the uncertainty in the output power of wind and solar PV


generation as well as the electrical demand, Monte Carlo Simulation
(MCS) is used to generate 3000 scenarios using the probability dis-
tribution functions. Each scenario includes 3 × 24 = 72 components,
each of which represents the PV module output, WT generation, and
Electrical demand at each hour [37,41].

3.5. Scenario reduction technique

In order to reduce the computational burden, the following scenario


reduction technique is used [42]. First, to minimize the intra-cluster
Fig. 14. The output thermal power of TESSs.
spread, the K-means algorithm is used for clustering [43]. Next, an
optimization problem is formulated and solved using genetic algorithm
(GA) to cluster the data and form the reduced scenarios.
The Davies-Bouldin (DB) validity index is used for the automatic

11
S.S. Fazlhashemi, et al. Journal of Energy Storage 29 (2020) 101301

Fig. 15. The stored energy of TESSs.

Finally, the DB validity index is calculated as follows [43,44]:


K
1
DB (K ) = Ru, qe
K u=1 (53)
A GA is used to yield the smallest DB(K) validity. By solving the
optimization problem minimizing (53), first, a specific number of cen-
troids is selected and allocated to the nearest scenario in the main
scenario set, and the centroids are updated accordingly. Then redis-
tribute the probabilities by adding the probabilities of unselected sce-
narios to the recently updated centroids in every cluster. Therefore, the
reduced scenario set is provided by the final selected scenarios with
associated probabilities.

4. HBB-BC solution algorithm

One of the efficient ways to improve the exploration capability of


optimization algorithm is hybridization. Using this feature, the HBB-BC
algorithm is formed by integrating the particle swarm optimization
algorithm [46] into the original BB-BC algorithm [45]. The HBB-BC has
two main stages, the Big Bang and the Big Crunch. The first stage
randomly generates the candidate solutions (particles) over the search
space of the problem. The second stage integrates the particles into a
Fig. 16. The voltage of buses (a: the expected value, b: in the most probable point referred to as the center of mass which is computed as follows:
scenario). [32]:
1
Np g
× X (h , )

clustering algorithm [44]. For the computation of the DB index, the


c (h )
X = 1
NDec , h Niter
ratio of the sums of within-cluster scatter to between-cluster separations Np g (54)
is calculated. The index uses both cluster and their corresponding
The solutions (particles) are updated as shown below, using three
sample means. To begin, the distance within uth cluster and the dis-
parameters i.e. the center of mass, the best position of each solution
tance between uth and vth clusters are denoted as follows [43,44]:
( X pbest (h, ) ), and the best global position ( X gbest (h) ) [32]:
1/ q
1 X (h + 1, ) = × X c (k ) + (1 2) ×( × X gbest (h) + (1 3) × X pbest (h, ) )
Su, q = X¯ mu q q 1 2 3
Nu X¯ Cu (50) r × × (X max
X min
)
1
+ NDec , Np, h Niter
1/e
h+1
duw, e = mu,a m w,a e = mu mw e e 1 (55)
a D (51) A mutation operator is used to guarantee that HBB-BC is not trapped
in a local optimal solution. This operator is expressed as follows [32]:
where Suq and duw.e are the distance within uth cluster and the distance
between uth and wth clusters, respectively; mu is the uth cluster center; X (h + 1, ) = X min + rand × (X max X min) if rand
Nu is the number of elements in the uth cluster Cu; q (which is an in-
< Pm NDec , Np, h Niter (56)
teger) and e is arbitrarily selected.
Next, Ru. qe is given as [43,44]: where, Pm is the mutation probability, which is between [0, 1]. As the
mutation probability gets closer to 1, the number of particles with
Su, q + Sw, q
Ru, qe = max w = 1, …, K w u mutation increases. Here, rand is a random number uniformly gener-
d uv, e (52) ated between 0 and 1.

12
S.S. Fazlhashemi, et al. Journal of Energy Storage 29 (2020) 101301

Table 5
The reactive power dispatch of capacitor banks and values of objective functions at each hour for Case 2.
Hour Expected value of scenarios The outcomes for the most probable scenario

Emission (kg) Cost ($) VSI Power Loss Reactive power generated by Emission (kg) Cost ($) VSI Power Loss Reactive power generated by CBs
(kW) CBs (kVAr) (kW) (kVAr)

1 2.127 178.30 0.900 23.17 372,400,400 2.065 162.66 0.880 27.97 0,57,346
2 2.168 135.50 0.894 26.78 400,400,400 2.232 132.72 0.881 35.09 400,400,400
3 2.119 133.93 0.873 37.72 400,400,400 2.260 127.49 0.886 32.15 400,400,400
4 1.924 115.35 0.876 33.76 400,400,400 2.072 107.06 0.854 49.57 400,400,400
5 1.807 101.99 0.853 47.47 400,400,400 2.016 98.51 0.851 50.59 400,400,400
6 1.974 140.17 0.835 62.51 400,400,400 2.165 133.59 0.806 93.71 400,400,400
7 2.074 135.91 0.826 66.69 400,400,400 2.123 118.94 0.804 94.03 400,400,400
8 2.326 266.44 0.823 67.88 383,400,392 2.053 194.59 0.801 95.11 400,400,400
9 2.428 614.81 0.774 110.60 392,0,400 2.579 649.42 0.816 67.26 400,400,400
10 2.512 1071.4 0.825 62.52 165,279,400 2.551 1204.2 0.817 70.30 400,400,400
11 2.557 1009.1 0.750 149.85 400,129,400 2.257 900.56 0.811 92.95 400,60,400
12 2.499 912.65 0.741 176.11 165,0,400 2.406 1006.2 0.810 84.56 400,190,400
13 2.397 508.74 0.726 178.70 332,204,400 2.326 545.45 0.779 106.88 400,400,400
14 2.261 967.57 0.790 99.76 400,0,400 2.221 1113.3 0.795 90.23 400,400,400
15 1.973 758.50 0.788 92.15 400,350,400 2.057 888.78 0.812 74.31 400,400,400
16 2.008 945.58 0.841 55.91 400,400,400 1.598 826.14 0.852 43.98 400,400,400
17 2.064 327.46 0.857 45.39 400,400,400 2.236 366.11 0.841 62.80 400,400,400
18 2.148 231.40 0.832 65.33 400,400,400 1.904 218.77 0.842 54.37 400,400,400
19 1.994 166.12 0.830 60.50 400,400,400 1.612 116.10 0.828 59.57 400,400,400
20 2.531 284.02 0.825 60.39 190,325,400 2.463 286.66 0.830 59.28 400,400,400
21 2.534 484.11 0.788 92.13 400,400,400 2.605 520.13 0.825 62.07 400,400,400
22 2.395 264.84 0.844 56.56 400,400,400 2.314 265.15 0.824 74.60 400,400,400
23 2.212 239.63 0.871 38.14 200,400,400 2.255 226.25 0.862 47.63 400,400,400
24 2.163 206.03 0.897 23.33 400,390,0 2.320 202.25 0.881 28.46 0,290,300

Fig. 17. The output electrical power of BESSs (a: expected value of scenarios, b: in the most probable scenario).

In this optimization problem, the decision variables include the Y (t ) = [Tie ( , t ), Pgrid (t ), PCCHP (i, t ), Hab (j, t ), PBESS (i , t ),
hourly status of open switches in each loop (integer variables), the
HTESS (i, t ), QCB (l, t )]
output (electrical and thermal) power of CCHPs, the output thermal
power of auxiliary boilers, the exchanged power with upstream grid,
NTie , i NHES , l NCB , t T (57)
the exchanged power with BESSs and TESSs, and the reactive power It is worth noting that in the proposed optimization model, four
generated by capacitor banks at each hour (continuous variables). objective functions are presented as (1), (3), (4), and (19) which are
Consequently, the vector of decision variables is as follows: aggregated as (21) subjected to constraints (23) -(42) and decision
variables (57).

13
S.S. Fazlhashemi, et al. Journal of Energy Storage 29 (2020) 101301

Fig. 18. The SoC of BESSs (a: expected value of scenarios, b: in the most probable scenario).

Fig. 19. The output thermal power of TESSs (a: expected value of scenarios, b: in the most probable scenario).

The steps taken in the utilized algorithm are given below: Step 3: Check the connectivity and radiality constraint.
Step 4: Carry out load flow analysis using the direct approach
Step 1: Define the input data including the network and algorithm proposed in [47] and calculate the value of the fitness function.
data. The network data includes the base distribution network Step 5: Calculate the "best" and "global" positions of the each par-
configuration, bus and branch data, the fundamental loops and their ticle as well as the center of mass using (54).
switches. The algorithm data consists of the number of population, Step 6: Update the candidates using (55) and apply the mutation
parameters pertaining to limit the size of the search space, ad- operation (56).
justable parameters, mutation probability (Pm), and the number of Step 7: Repeat Steps 3–6 until the termination criterion is met. Here,
iterations. the number of iterations is considered as the stopping criterion.
Step 2: Randomly generating the initial population.

14
S.S. Fazlhashemi, et al. Journal of Energy Storage 29 (2020) 101301

Fig. 20. The energy stored in TESSs (a: expected value of scenarios, b: in the most probable scenario).

Fig. 21. The output electrical power of CCHPs (a: expected value of scenarios, b: most probable scenario).

Fig. 7 illustrates the flowchart of the proposed algorithm. objective optimization problem is applied to a microgrid with 33 buses
[48]. The voltage level and nominal real and reactive power con-
5. Simulation results sumptions are 12.66 kV, 3715 kW, and 2300 kVAr respectively. The
numbers of ties (normally open) and sectionalizing (normally closed)
The presented MOHBB-BC algorithm to solve the formulated multi- switches in this radial distribution network are 5 and 32, respectively

15
S.S. Fazlhashemi, et al. Journal of Energy Storage 29 (2020) 101301

Fig. 22. The output thermal power of CCHPs (a: expected value of scenarios, b: in the most probable scenario).

Fig. 23. The bus voltages (a: expected value of scenarios, b: in the most probable scenario).

16
S.S. Fazlhashemi, et al. Journal of Energy Storage 29 (2020) 101301

Table 6
The network reconfiguration decisions, reactive power dispatch of capacitor banks and values of objective functions at each hour for Case 3.
Hour The expected value of scenarios The outcomes for the most probable scenario

Emission (kg) Cost ($) VSI Power Loss Open Switches and Reactive power Emission (kg) Cost ($) VSI Power Loss Open Switches and Reactive power
(kW) generated by CBs(kVAr) (kW) generated by CBs (kVAr)

1 2.052 164.22 0.908 20.28 S7, S10, S34, S28, S36, 1.984 156.12 0.923 16.55 S6, S10, S34, S37, S32
78,39,105 ,320,320,400
2 2.167 132.90 0.905 24.99 S7, S35, S10, S37, S32, 2.222 131.61 0.903 27.47 S6, S8, S9, S37, S36
380,190,110 ,300,325,400
3 2.127 131.89 0.891 34.01 S7, S21, S13, S28, S36, 2.255 126.11 0.901 23.71 S7, S9, S14, S37, S32
40,220,380 ,350,206,400
4 1.954 114.56 0.882 36.42 S7, S35, S10, S25, S32, 2.102 108.21 0.818 40.42 S7, S11, S34, S37, S32
220,120,270 ,260,310,400
5 1.963 109.97 0.862 54.24 S7, S35, S13, S27, S32, 2.112 106.29 0.879 49.85 S7, S8, S9, S28, S32
260,175,340 ,400,400,400
6 2.064 146.39 0.840 69.81 S6, S9, S34, S25, S36, 2.272 145.23 0.847 82.51 S7, S9, S14, S28, S32,380,215,400
125,45,388
7 1.970 110.17 0.842 68.01 S7, S8, S13, S37, S32, 2.253 123.33 0.852 76.73 S6, S9, S14, S37, S32
328,394,295 ,400,400,400
8 1.970 173.55 0.839 76.17 S7, S9, S14, S28, S31, 2.041 192.96 0.847 73.94 S6, S9, S14, S37, S32
320,195,200 ,400,400,400
9 2.835 515.22 0.832 58.64 S3, S11, S13, S38, S8, 2.856 702.98 0.862 73.47 S6, S10, S14, S4, S34
214,231,382 ,350,200,400
10 2.737 846.55 0.866 63.44 S7, S13, S35, S37, S31 2.667 1104.4 0.883 53.99 S33, S12, S9, S4, S8
,0,150,115 ,400,400,400
11 2.696 804.11 0.873 67.71 S7, S8, S11, S37, S31 2.355 796.06 0.897 60.54 S33, S35, S10, S4, S31
,400,125,400 ,400,400,400
12 2.616 722.91 0.829 105.78 S19, S21, S13, S26, 2.478 914.38 0.891 57.29 S33, S10, S13, S4, S30
S30,125,340,318 ,400,400,400
13 2.857 569.63 0.823 111.61 S33, S9, S13, S27, S30 2.468 652.25 0.817 112.65 S3, S10, S13, S28, S31
,118,250,180 ,400,400,400
14 2.611 756.57 0.812 107.81 S33, S35, S13, S37, S30,400,0,25 2.442 967.97 0.880 64.53 S7, S11, S12, S37, S30
,262,400,400
15 2.607 519.61 0.895 36.41 S20, S33, S10, S37, 2.470 620.46 0.905 41.56 S6, S11, S12, S4, S31
S31,290,375,320 ,250,350,400
16 2.723 582.7 0.892 48.99 S7, S9, S13, S24, S34 2.226 530.73 0.920 21.51 S6, S35, S9, S37, S30
,52,165,80 ,150,400,400
17 2.993 389.53 0.894 30.51 S7, S33, S10, S28, S31 2.514 385.94 0.882 37.18 S7, S10, S14, S37, S32
,0,400,0 ,40,50,400
18 2.872 314.89 0.875 44.27 S7, S11, S34, S27, S31 1.906 217.91 0.877 43.88 S6, S10, S14, S37, S32
,0,132,185 ,364,340,400
19 2.201 182.30 0.819 91.56 S7, S11, S35, S25, S32 1.848 156.83 0.842 74.22 S6, S34, S9, S4, S32
,200,137,145 ,400,400,400
20 2.983 287.59 0.880 73.40 S6, S9, S13, S3, S34 1.977 167.12 0.827 88.34 S7, S33, S34, S37, S32
,306,230,125 ,190,350,400
21 2.870 417.66 0.887 58.05 S33, S10, S14, S28, 2.722 480.84 0.888 47.91 S33, S13, S9, S7, S37
S34,58,190,330 ,175,390,400
22 3.053 304.23 0.896 31.82 S33, S35, S12, S26, 2.308 244.94 0.873 48.06 S6, S11, S34, S37, S32
S34,350,400,230 ,400,320,400
23 2.091 209.02 0.898 29.43 S7, S11, S35, S28, S32 2.133 203.41 0.819 33.18 S7, S14, S9, S37, S32
,370,100,195 ,0,0,400
24 2.046 186.61 0.913 17.91 S7, S9, S14, S37, S32 2.202 183.14 0.912 21.29 S7, S14, S9, S28, S32
,62,0,145 ,0,0,0

[2]. The single-line diagram of the system is shown in Fig. 8. Two HES algorithm are presented in Table 3. Due to the utilization of the random
that were introduced in Section 2, are installed on buses 4, and 14. operators in (54) and (55), the outcome of the HBBC optimization al-
Three capacitor banks are considered at buses 7, 13, and 29 for con- gorithm is not fixed. Therefore, the simulation runs for 30 times to
trolling the reactive power and adjusting the voltage profile. The de- obtain a set of solutions. The average value of the procured solutions is
mand profile of the microgrid is given in Figs. 9 and 10. Fig. 9 displays considered as the final solution, i.e., ground truth, for the algorithm.
the maximum electrical demand at each bus and Fig. 10 illustrates the Three different cases are considered as follows:
participation factor of the maximum electrical demand at each hour for
each bus. Fig. 9 shows the mean of the hourly forecasted electrical Case 1: Energy management with DFR without capacitor banks
demand which are further represented by the normal distribution considering HESs equipped with an auxiliary boiler and TESS.
function. The standard deviation of electrical demand, σd is 1%. It is Case 2: Energy management without DFR with capacitor banks
assumed that the thermal and cooling demands are located at buses 4, considering HESs equipped with all of the components.
and 14. Fig. 11 shows the thermal and cooling demand profiles. The Case 3: Energy management with DFR and capacitor banks con-
day-ahead forecasted electricity price for microgrid is shown in Fig. 12. sidering HESs equipped with all of the components.
The simulation is performed using the MATLAB (R2014b) on a PC with
Intel Core i7, 2.5 GHz CPU with 12GB of RAM. The simulation results in the above-mentioned case studies are
The characteristics of the HES and capacitor banks are represented compared to the base case. The base case is similar to Case 1, without
in Table 2. DFR.
The initial parameters selected for the HBB-BC optimization

17
S.S. Fazlhashemi, et al. Journal of Energy Storage 29 (2020) 101301

Fig. 24. The output electrical power of BESSs (a: expected value of scenarios, b: in most probable scenario).

Fig. 25. The stored energy in BESSs (a: expected value of scenarios, b: in most probable scenario).

5.1. Case 1 – energy management with DFR without capacitor banks auxiliary boilers and TESSs, while the electrical demand is supplied by
considering HESs equipped with an auxiliary boiler and TESS the upstream main grid. It is shown in Table 4 that for the most
probable scenario, the maximum real power loss and minimum VSI
Table 4 lists the DFR decisions and the values of the objective occurred at hour 13:00. Furthermore, in most probable scenario, the
function at each hour. Given the uncertainties in the renewable gen- emission will significantly increase at hour 13:00. Similar observations
eration and demand, the simulation results represent the most probable were made at hour 12:00 for the expected emission, VSI and real power
scenario and the expected outcomes of the simulation. In this case, only loss due to the increased heating, cooling, and electrical demand.
auxiliary boiler and TESS in the HES are considered, and other assets The operation cost reached its maximum at hour 14:00 due to the
(i.e., PGU, WT, PV, and BESS) are not taken into account. Therefore, the high electricity price (Fig. 12), and charging the TESSs by auxiliary
cooling and heating demands on buses 4 and 14 are supplied by the boilers at this hour. Figs. 13 and 14 show the output thermal power of

18
S.S. Fazlhashemi, et al. Journal of Energy Storage 29 (2020) 101301

Fig. 26. The output thermal power of TESSs (a: expected value of scenarios, b: in most probable scenario).

Fig. 27. The stored energy in TESSs (a: expected value of scenarios, b: most probable scenario).

auxiliary boilers and TESSs, in both scenarios, respectively. Since the cooling demands. Fig. 14 shows the charging/discharging modes of
uncertainty in thermal demand is not modeled and distribution gen- TESS. It is shown that TESSs discharge when the thermal and cooling
eration is not rconsidered in this case, Figs. 13 and 14 are valid for both demands reach their peak level and charge in off-peak hours. Similar
scenarios. It is seen from Fig. 13 that the auxiliary boilers operate at results are observed in Table 4 for the most probable scenario.
nearly 70% of their capacity from 7:00 to 22:00 to meet the heating and Fig. 15 shows that the stored energy in TESSs remains within the

19
S.S. Fazlhashemi, et al. Journal of Energy Storage 29 (2020) 101301

Fig. 28. The output electrical power of CCHPs (a: expected value of scenarios, b: most probable scenario).

power loss and minimum VSI occurred at hour 13:00, when the elec-
trical demand is at its peak. As the electrical demand increases, the
generation of CCHPs increases as shown in Fig. 21. Here, the excess
thermal energy will be stored in TESSs as shown in Fig. 19. The gen-
erated thermal power by CCHP units is shown in Fig. 22. It is worth
noting that as CCHPs meet the thermal and cooling demand, the aux-
iliary boilers do not operate in this case. Fig. 23 depicts the voltage
profiles for all buses in maximum and minimum electrical demand. As
shown in this figure only a few buses will suffer form undervoltage
during peak hours. It is concluded that DGs reduce the electrical de-
mand supplied by the upstream grid and therefore, the voltage drop is
less than that in Case 1. As shown in Table 5, the maximum operation
cost occurs at hour 10:00 due to high electricity prices, as well as large
Fig. 29. The output thermal power of auxiliary boilers (expected value of electrical, heating, and cooling demands. Figs. 17, and 19 show the
scenarios). operation schedule of BESS and TESSs respectively. Here, the maximum
emission occurs at hour 11:00 when the CCHPs provide maximum
acceptable range. As the electric distributed generation resources were thermal energy to satisfy the cooling and heating demands. Similar
not used in the microgrid, the voltage drop along the feeder could be outcomes are observed for the most probable scenario in Table 5.
significant. Therefore, to solve the proposed optimization model ig-
noring the distributed electricity resources, it is assumed that the lower
5.3. Case 3 – energy management with DFR and capacitor banks
range in (26) is set to 0.9 (p.u). Fig. 16 shows the voltage profile in
considering HESs equipped with all of the components
maximium and minimum electrical demand. It is seen that the voltage
of buses at the end of the feeders is less than the lower limits during
This case considers the DFR and generation scheduling of DGs and
peak hours.
capacitor banks simultaneously. Table 6 shows the reactive power of
capacitor banks, the status of switches, the power loss, emission, vol-
5.2. Case 2 – energy management without DFR with capacitor banks tage stability index and the operation cost at each hour. Figs. 24, and 25
considering HESs equipped with all of the components show the output electrical power and the energy stored in BESSs at each
hour, respectively. Figs. 26, and 27 illustrate the output thermal power
Table 5 presents the reactive power supply by capacitor banks and and the stored energy in TESSs at each hour, respectively. Similar to the
the power loss, voltage stability index, emission, and operation cost at previous cases, the maximum real power loss and minimum VSI occur
each hour. Since network reconfiguration is not considered in this case, at hour 13:00, when the electrical demand reaches its peak. Similar to
the states of tie switches are not changed as shown in Fig. 7. Figs. 17, Case 2, when the electrical demand increases, the generation of CCHPs
and 18 show the output electrical power and stored energy in BESSs at increases as shown in Fig. 28 to meet electrical demand. Therefore, the
each hour, respectively. For both scenarios, Figs. 19, and 20 illustrate thermal power generated by auxiliary boilers (shown in Fig. 29), and
the output thermal power and stored energy in TESSs at each hour, CCHPs (shown in Fig. 30) satisfy the cooling and heating demands and
respectively. As shown in Figs. 18 and 20, the energy stored in BESS and the excess thermal power is stored in TESSs. The output power of TESS
TESS is within an acceptable range. In this case, the maximum real is shown in Fig. 26. As shown in Figs. 25, and 27 the energy stored in

20
S.S. Fazlhashemi, et al. Journal of Energy Storage 29 (2020) 101301

Fig. 30. The output thermal power of CCHPs (a: expected value of scenarios, b: most probable scenario).

Fig. 31. The bus voltages (a: expected value of scenarios, b: most probable scenario).

BESSs, and TESSs are within the acceptable range. Fig. 31 shows that real power losses are reduced. Similar to the previous case, the max-
the voltage at all buses is within admissible ranges. Dispatching DGs imum operation cost occurs at hour 10:00, because of high electricity
will reduce the flow in the feeders and therefore, the voltage drops and price, and large electrical, heating and cooling demand. Furthermore,

21
S.S. Fazlhashemi, et al. Journal of Energy Storage 29 (2020) 101301

Fig. 32. Objective values with changes in thermal demand in Case 3.

Table 7
The comparison of results for different case studies.
Case study Most probable scenario The outcomes for the most probable scenario

Total power loss Average VSI Total operational cost Total emission Total power loss Averaged VSI Total operational cost Total emission
(kWh) ($) (ton) (kWh) ($) (ton)

Base Case 3.475e+03 0.755 1.704e+04 60.104 3.62e+03 0.749 1.706e+04 60.189
Case 1 2.87e+03 0.804 1.661e+04 59.763 2.54e+03 0.821 1.704e+04 59.573
Case 2 1.733e+3 0.827 1.0219e+4 53.206 1.55e+03 0.8331 1.041e+04 52.702
Case 3 1.361e+3 0.869 0.869e+4 59.013 1.27e+3 0.879 0.941e+4 54.974

Therefore, the increase in the generated electric power in PGUs would


lead to an increase in supplied thermal power by HRUs. This could lead
to significant changes in individual objectives as shown in Fig. 32.
Fig. 32.a shows the real power loss of the microgrid network in three
conditions. As nominal thermal demand (values used in Case 3) is
shown as Situation 1, the increase in thermal demand to 1.5 of the
nominal thermal demand is shown in Situation 2. Situation 3 features
the thermal demand equal to twice the thermal demand in Case 3. As
shown in Fig. 32.a the increase in thermal demand contributed to the
overall reduction in the network loss in the operation horizon. More-
over, the increase in the thermal demand would lead to an increase in
the emission as shown in Fig. 32.d.

5.5. Comparative analysis

This section provides a comparative analysis between the Cases 1–3


Fig. 33. The value of fitness function in cases 1–3. and Base Case. Here, the Base Case is the IEEE-33 bus system with no
HES, no DFR, and no capacitor bank. In the Base Case, the demand is
the maximum emission occurred at hour 13:00 when the CCHPs reach served by auxiliary boilers. Table 7 compares the operation indices in
their maximum generation during the operation horizon. As shown in the most probable scenario as well as the expected operation indices. As
Table 6, similar observations are made in the simulation with the most shown in this table, the most favorable outcomes for total real power
probable scenario. loss, operation cost, and average VSI are achieved in Case 3. Here, the
least total emission occurred in Case 2. Fig. 33 displays the convergence
5.4. Sensitivity analysis of individual objective functions in Case 3 to of the fitness function in Cases 1–3. As shown in this figure, the algo-
thermal (heating & cooling) demand variation rithm converges after a large number of iterations in Case 3 due to the
complexity of formulation.
In this case, the effect of variation in thermal demand is analyzed in
Case 3. It is noted that increasing the thermal demand could lead to an 6. Conclusions
increase in the electric power generation in PGUs since the auxiliary
boilers cannot supply the total thermal demand. Here, the size of the In this paper, the HBB-BC algorithm is used for simultaneous DFR,
auxiliary boilers is not enough to serve the increased thermal demand. and the generation scheduling of DGs and capacitor banks. The

22
S.S. Fazlhashemi, et al. Journal of Energy Storage 29 (2020) 101301

proposed formulation captures multiple objectives, including real [14] T.T. Nguyen, A.V. Truong, and T.A. Phung, "A novel method based on adaptive
power loss, VSI index, operational cost, and emission considering the cuckoo search for optimal network reconfiguration and distributed generation al-
location in distribution network," Int. J. Electr. Power Energy Syst., vol. 78, pp.
uncertainties in wind generation, solar PV generation, and electrical 801–815, 2016/06/01/ 2016.
demand. The uncertainty in generation and demand is represented [15] S. Zhang, H. Cheng, D. Wang, L. Zhang, F. Li, and L. Yao, "Distributed generation
using scenarios. Scenario reduction technique is used to limit the planning in active distribution network considering demand side management and
network reconfiguration," Appl. Energy, vol. 228, pp. 1921–1936, 2018/10/15/
number of considered scenarios in the operation horizon. To capture 2018.
the uncertainty in the operation horizon, the expected values of the [16] N. Kanwar, N. Gupta, K.R. Niazi, and A. Swarnkar, "An integrated approach for
operation indices as well as the the operation indices in the most distributed resource allocation and network reconfiguration considering load di-
versity among customers," Sustain. Energy Grids Netw., vol. 7, pp. 37–46, 2016/09/
probable scenario are taken into account. To evaluate the proposed 01/ 2016.
model three case studies and a Base Case are considered. Case 1 ad- [17] A. Zidan, M.F. Shaaban, E.F. El-Saadany, Long-term multi-objective distribution
dresses the optimal operation of microgrid with DFR and HESs with network planning by DG allocation and feeders’ reconfiguration, Electr. Power Syst.
Res. 105 (2013) 95–104.
auxiliary boiler and TESS, without capacitor banks. Case 2 addresses
[18] R.S. Rao, K. Ravindra, K. Satish, S. Narasimham, Power loss minimization in dis-
the optimal operation of microgrid without DFR with capacitor banks tribution system using network reconfiguration in the presence of distributed
considering the HESs equipped with all components. Case 3 is Case 2 generation, IEEE Trans. Power Syst. 28 (1) (2013) 317–325.
with DFR. The Base Case features no DFR, no HES and no capacitor [19] M. K and J. S, "Integrated approach of network reconfiguration with distributed
generation and shunt capacitors placement for power loss minimization in radial
bank. The following observations were made using the expected values distribution networks," Appl. Soft Comput., vol. 52, pp. 1262–1284, 2017/03/01/
of the operation indices. First, compared to the Base Case, the real 2017.
power loss is reduced by 60%, 50%, and 17% in Cases 3, 2, and 1, [20] M. Mohammadi, A.M. Rozbahani, S. Bahmanyar, Power loss reduction of dis-
tribution systems using BFO based optimal reconfiguration along with DG and
respectively. Second, compared to Base Case, the VSI index is increased shunt capacitor placement simultaneously in fuzzy framework, J. Cent. South Univ.
by 15.1%, 9.5%, and 6.4% in Cases 3, 2, and 1, respectively. Third, 24 (1) (2017) 90–103.
compared to Base Case, the total operational costs are reduced by [21] S.N. Ravadanegh, M.R.J. Oskuee, M. Karimi, Multi-objective planning model for
simultaneous reconfiguration of power distribution network and allocation of re-
49.0%, 40.0%, and 2.5% in Cases 3, 2, and 1, respectively. And Fourth, newable energy resources and capacitors with considering uncertainties, J. Cent.
compared to Base Case, the emissions are decreased by 1.8%, 11.4%, South Univ. 24 (8) (2017) 1837–1849.
and 0.56% in Cases 3, 2, and 1, respectively. A similar trend is observed [22] M. Sedighizadeh, M. Esmaili, N. Mohammadkhani, Stochastic multi-objective en-
ergy management in residential microgrids with combined cooling, heating, and
in the simulation results for the most probable scenario. power units considering battery energy storage systems and plug-in hybrid electric
vehicles, J. Clean. Prod. 195 (2018) 301–317.
Declaration of Competing Interest [23] Z. Bai, T. Liu, Q. Liu, J. Lei, L. Gong, H. Jin, Thermodynamic analysis of a CCHP
system integrated a chemical recuperation process of methanol decomposition,
Energy Procedia 142 (2017) 1582–1588.
None. [24] S. Hameer, J.L. Van Niekerk, Thermodynamic modelling of thermal energy storage
systems, Energy Procedia 93 (2016) 25–30.
[25] J.A. Osara, M.D. Bryant, A thermodynamic model for lithium-ion battery de-
References gradation: application of the degradation-entropy generation theorem, Inventions 4
(2) (2019) 23.
[1] I. Ben Hamida, S.B. Salah, F. Msahli, and M.F. Mimouni, "Optimal network re- [26] W. Gu, et al., Modeling, planning and optimal energy management of combined
configuration and renewable DG integration considering time sequence variation in cooling, heating and power microgrid: a review, Int. J. Electr. Power Energy Syst.
load and DGs," Renew. Energy, vol. 121, pp. 66–80, 2018/06/01/ 2018. 54 (2014) 26–37.
[2] M. Esmaeili, M. Sedighizadeh, and M. Esmaili, "Multi-objective optimal re- [27] M. Sedighizadeh, M. Dakhem, M. Sarvi, H.H. Kordkheili, Optimal reconfiguration
configuration and DG (Distributed generation) power allocation in distribution and capacitor placement for power loss reduction of distribution system using im-
networks using big bang-big crunch algorithm considering load uncertainty," proved binary particle swarm optimization, Int. J. Energy Environ. Eng. 5 (1)
Energy, vol. 103, pp. 86–99, 2016/05/15/ 2016. (2014) 3.
[3] M. Sedighizadeh, M. Esmaili, and M. Esmaeili, "Application of the hybrid big bang- [28] H. Nasiraghdam, S. Jadid, Optimal hybrid PV/WT/FC sizing and distribution
big crunch algorithm to optimal reconfiguration and distributed generation power system reconfiguration using multi-objective artificial bee colony (MOABC) algo-
allocation in distribution systems," Energy, vol. 76, pp. 920–930, 2014/11/01/ rithm, Sol. Energy 86 (10) (2012) 3057–3071.
2014. [29] T. Niknam, A.K. Fard, A. Seifi, Distribution feeder reconfiguration considering fuel
[4] J.C. Leite, I.P. Abril, M.S.S. Azevedo, Capacitor and passive filter placement in cell/wind/photovoltaic power plants, Renew. Energy 37 (1) (2012) 213–225.
distribution systems by nondominated sorting genetic algorithm-II, Electr. Power [30] J. Olamaei, T. Niknam, G. Gharehpetian, Application of particle swarm optimiza-
Syst. Res. 143 (2017) 482–489. tion for distribution feeder reconfiguration considering distributed generators,
[5] G. Ferruzzi, G. Cervone, L. Delle Monache, G. Graditi, F. Jacobone, Optimal bidding Appl. Math. Comput. 201 (1–2) (2008) 575–586.
in a day-ahead energy market for micro grid under uncertainty in renewable energy [31] M. Sedighizadeh, M. Ghalambor, A. Rezazadeh, Reconfiguration of radial dis-
production, Energy 106 (2016) 194–202. tribution systems with fuzzy multi-objective approach using modified big bang-big
[6] M. Di Somma, G. Graditi, E. Heydarian-Forushani, M. Shafie-Khah, P. Siano, crunch algorithm, Arab. J. Sci. Eng. 39 (8) (2014) 6287–6296.
Stochastic optimal scheduling of distributed energy resources with renewables [32] M. Sedighizadeh, S. Ahmadi, M. Sarvi, An efficient hybrid big bang–big crunch
considering economic and environmental aspects, Renew. Energy 116 (2018) algorithm for multi-objective reconfiguration of balanced and unbalanced dis-
272–287. tribution systems in fuzzy framework, Electr. Power Components Syst. 41 (1)
[7] M. Sedighizadeh, G. Shaghaghi-shahr, M. Esmaili, and M.R. Aghamohammadi, (2013) 75–99.
"Optimal distribution feeder reconfiguration and generation scheduling for micro- [33] A.Y. Abdelaziz, F. Mohammed, S. Mekhamer, M. Badr, Distribution systems re-
grid day-ahead operation in the presence of electric vehicles considering un- configuration using a modified particle swarm optimization algorithm, Electr.
certainties," J. Energy Storage, vol. 21, pp. 58–71, 2019/02/01/ 2019. Power Syst. Res. 79 (11) (2009) 1521–1530.
[8] R. Jabbari-Sabet, S.-M. Moghaddas-Tafreshi, S.-S. Mirhoseini, Microgrid operation [34] N. Ghaffarzadeh, H. Sadeghi, A new efficient BBO based method for simultaneous
and management using probabilistic reconfiguration and unit commitment, Int. J. placement of inverter-based DG units and capacitors considering harmonic limits,
Electr. Power Energy Syst. 75 (2016) 328–336. Int. J. Electr. Power Energy Syst. 80 (2016) 37–45.
[9] G. Gutiérrez-Alcaraz, E. Galván, N. González-Cabrera, M. Javadi, Renewable energy [35] N. Mohammadkhani, M. Sedighizadeh, M. Esmaili, Energy and emission manage-
resources short-term scheduling and dynamic network reconfiguration for sustain- ment of CCHPs with electric and thermal energy storage and electric vehicle,
able energy consumption, Renew. Sustain. Energy Rev. 52 (2015) 256–264. Therm. Sci. Eng. Progr. 8 (2018) 494–508.
[10] F.S. Gazijahani, J. Salehi, Integrated DR and reconfiguration scheduling for optimal [36] A. Zakariazadeh, S. Jadid, P. Siano, Smart microgrid energy and reserve scheduling
operation of microgrids using Hong's point estimate method, Int. J. Electr. Power with demand response using stochastic optimization, Int. J. Electr. Power Energy
Energy Syst. 99 (2018) 481–492. Syst. 63 (2014) 523–533.
[11] M.R. Kaveh, R.-.A. Hooshmand, S.M. Madani, Simultaneous optimization of re- [37] M. Sedighizadeh, M. Esmaili, A. Jamshidi, M.-.H. Ghaderi, Stochastic multi-objec-
phasing, reconfiguration and DG placement in distribution networks using BF-SD tive economic-environmental energy and reserve scheduling of microgrids con-
algorithm, Appl. Soft Comput. 62 (2018) 1044–1055. sidering battery energy storage system, Int. J. Electr. Power Energy Syst. 106 (2019)
[12] J. Shukla, B. Das, V. Pant, Stability constrained optimal distribution system re- 1–16.
configuration considering uncertainties in correlated loads and distributed gen- [38] S. Talari, M. Yazdaninejad, M.-.R. Haghifam, Stochastic-based scheduling of the
erations, Int. J. Electr. Power Energy Syst. 99 (2018) 121–133. microgrid operation including wind turbines, photovoltaic cells, energy storages
[13] E. kianmehr, S. Nikkhah, and A. Rabiee, "Multi-objective stochastic model for joint and responsive loads, IET Gen. Transm. Distrib. 9 (12) (2015) 1498–1509.
optimal allocation of DG units and network reconfiguration from DG owner's and [39] M. Motevasel, A.R. Seifi, Expert energy management of a micro-grid considering
DisCo's perspectives," Renew. Energy, vol. 132, pp. 471–485, 2019/03/01/ 2019. wind energy uncertainty, Energy Convers. Manage. 83 (2014) 58–72.
[40] A. Zakariazadeh, S. Jadid, P. Siano, Economic-environmental energy and reserve

23
S.S. Fazlhashemi, et al. Journal of Energy Storage 29 (2020) 101301

scheduling of smart distribution systems: a multiobjective mathematical program- Softw. 37 (2) (2006) 106–111.
ming approach, Energy Convers. Manage. 78 (2014) 151–164. [46] J. Kennedy, Particle swarm optimization, Encycl. Mach. Learn. (2010) 760–766.
[41] M. Petrollese, L. Valverde, D. Cocco, G. Cau, J. Guerra, Real-time integration of [47] J.-.H. Teng, A direct approach for distribution system load flow solutions, IEEE
optimal generation scheduling with MPC for the energy management of a renew- Trans. Power Delivery 18 (3) (2003) 882–887.
able hydrogen-based microgrid, Appl. Energy 166 (2016) 96–106. [48] M.E. Baran, F.F. Wu, Network reconfiguration in distribution systems for loss re-
[42] W. Su, J. Wang, J. Roh, Stochastic energy scheduling in microgrids with inter- duction and load balancing, IEEE Trans. Power Delivery 4 (2) (1989) 1401–1407.
mittent renewable energy resources, IEEE Trans. Smart Grid 5 (4) (2014) [49] T. Weitzel, M. Schneider, C.H. Glock, F. Löber, S. Rinderknecht, Operating a sto-
1876–1883. rage-augmented hybrid microgrid considering battery aging costs, J. Clean. Prod.
[43] J. Sachs, O. Sawodny, Multi-objective three stage design optimization for island 188 (2018) 638–654.
microgrids, Appl. Energy 165 (2016) 789–800. [50] A.A. El-Fergany, A.Y. Abdelaziz, Capacitor allocations in radial distribution net-
[44] A. Jose-Garcia, W. Gómez-Flores, Automatic clustering using nature-inspired me- works using cuckoo search algorithm, IET Gen. Transm. Distrib. 8 (2) (2014)
taheuristics: a survey, Appl. Soft Comput. 41 (2016) 192–213. 223–232.
[45] O.K. Erol, I. Eksin, A new optimization method: big bang–big crunch, Adv. Eng.

24

You might also like