You are on page 1of 15

Energy 282 (2023) 128909

Contents lists available at ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

Analysis of microgrid configuration with optimal power injection from grid


using point estimate method embedded fuzzy-particle swarm optimization
Hemanth Chaduvula ∗, Debapriya Das
Department of Electrical Engineering, Indian Institute of Technology Kharagpur, Kharagpur, 721302, India

ARTICLE INFO ABSTRACT

Keywords: This paper aims to analyze the effect of integration of combined heat and power (CHP) units, renewables, shunt
Optimal allocation capacitors, and a heat boiler in a grid connected microgrid (GMG) system. In addition to that, a comparative
Zero bus concept based microgrid study is conducted between the GMG and zero bus concept based microgrid (ZBMG) systems. In ZBMG, a
Fuzzy
predefined amount of power is injected from the grid. Therefore, one of the dispatchable distributed energy
Particle swarm optimization
resources (DERs) in the ZBMG maintains the power balance and the corresponding DER unit is entitled as
Point estimate method
zero bus DER. In this paper, eight different microgrid configurations are formed by several combinations of
renewable and non renewable sources, which are categorized under GMG and ZBMG systems. The multiple
objectives such as total cost, emission, and energy loss are combined and solved using fuzzy technique and
particle swarm optimization (PSO) algorithm. Also, point estimate method (PEM) is implemented to handle the
uncertainty in renewables. The effect of grid, dispatchable units, renewables, and heat boiler on the objectives
in each microgrid configuration is inspected. The results show that the microgrid configuration with non
renewables is attained minimum cost and has a minimum emission with renewables. The integration of all
types of sources in a microgrid configuration leads to a decrement in all the objectives. The proposed method
is tested on a 33-bus grid connected microgrid system considering electrical and heat loads.

1. Introduction of DER units are determined based on the objective considered. In [23],
the optimal location of DERs is determined from the maximum power
The combined heat and power (CHP) units have been recognized as stability index and the optimum size of DERs is obtained by minimizing
a great resource for electrical and heat demand applications. The CHP the active power loss. The optimal location, power factor, and capacity
units also called cogeneration units are very efficient for simultaneous of DER units are determined using a combination of genetic algorithm
production of electricity and heat [1–4]. Further, the operation of
(GA) and an analytical approach for minimizing system loss [24].
these units lowers the primary energy usage and reduces the green
Aman et al. [25] determined the optimal placement and sizing of
gas emissions [5]. The District Heating Systems (DHS) with CHP units
multiple DER units to maximize the loadability of the system using
and heat boilers are established for space heating and domestic hot
water in residential and industrial dwellings [6–9]. To minimize CO2 a Hybrid particle swarm optimization (HPSO) algorithm. In [26], the
emissions led to the integration of renewable based solar thermal power optimal allocation of DER units is determined by minimizing the cost
plant in DHS in spite of its high investment. The thermal energy of unsupplied energy, the installation cost of DERs, and power losses,
storage is also integrated with solar field to mitigate the uncertainties. improving voltage profile using fuzzy integrated ant lion optimizer
The configurations and sizes of various components in DHS are opti- (ALO) algorithm.
mized for minimizing economic cost and/or environmental emissions Whereas, integration of intermittent renewables in the system is
[10–19]. However, the works presented in [10–19] have not considered challenging as the power injection from the photovoltaic (PV) and wind
the optimization of electrical energy. turbine (WT) units varies in each instant. Hence, for optimal microgrid
The optimal allocation of distributed energy resources (DERs) pro- scheduling, the uncertainties in PV and WT units have to be considered.
vides various technical benefits in the system such as minimum loss, im-
In [27], the optimal siting and sizing of DER units are determined by
proving the voltage profile, and alleviate the system upgrade
considering the uncertainty in renewables using point estimate method
[20,21]. Otherwise, inappropriate allocation of sources causes a detri-
(PEM). In [28], probabilistic power flow is implemented in a 33-node
mental impact on the system [22]. The optimal location and capacity

∗ Corresponding author.
E-mail addresses: hemanthped@gmail.com (H. Chaduvula), ddas@ee.iitkgp.ernet.in (D. Das).

https://doi.org/10.1016/j.energy.2023.128909
Received 10 August 2022; Received in revised form 20 June 2023; Accepted 24 August 2023
Available online 26 August 2023
0360-5442/© 2023 Elsevier Ltd. All rights reserved.
H. Chaduvula and D. Das Energy 282 (2023) 128909

and DM [32] optimized location, capacity, and power factor of dis-


Nomenclature patchable and renewable sources using Grid-connected Multi-objective
𝑁PVmod Number of PV modules in a PV array harmony search algorithm (GrMHS) to minimize energy loss, voltage
𝐹𝐹 Fill factor deviation, and cost of integration of sources. Kayal and Chanda [33]
determined the optimal mix of solar and wind units by minimizing
𝑉o , 𝐼o Output voltage and current of PV module
annual power loss, maximizing voltage stability, and improving the
𝑓𝑠 (𝑠(𝑡)) Beta probability density function of solar
network security using weighted aggregation PSO. Abdelaziz et al. [34]
irradiation
optimized fuel cost, switching operations, and loadability objectives in
𝛤 Gamma function
droop controlled islanded microgrid (IMG) system while uncertainties
𝑃WTo Output power of wind turbine in renewables and load demand are taken into account. However, the
𝑣cin , 𝑣r , 𝑣cout Cut-in, rated and cut out speed of wind heat demand was not covered in the above works.
turbine Bornapour et al. [35] developed a modified hybrid Bird Mating
𝑣max Maximum wind velocity Optimization Differential Evolution (BMO-DE) algorithm for optimal
𝑃r,WT Power rating of wind turbine planning of renewable energy sources (RES) and CHP units while
𝑓𝑣 (𝑣(𝑡)) Weibull probability density function of modeling the uncertainties. The operation and size of combined cool-
wind velocity ing heat and power (CCHP) units, boilers, chillers, and batteries are
𝑃𝑍𝐵𝐷𝐸𝑅 (𝑡), 𝑄𝑍𝐵𝐷𝐸𝑅 (𝑡) Active and reactive power output from determined to minimize lifetime cost while uncertainties are modeled
zero bus DER at 𝑡th hour by a Markovian process [36]. A mixed-integer linear programming
𝑃𝑐𝑎𝑙,𝑖 (𝑡), 𝑄𝑐𝑎𝑙,𝑖 (𝑡) Calculated active and reactive power (MILP) approach is proposed for minimizing the total cost of energy
entering at bus ‘𝑖’ in 𝑡th hour networks and a small neighborhood level in the presence of power
and heat demand [37]. In [38], the optimal scheduling of energy hub
𝑃𝐿,𝑍𝐵 (𝑡), 𝑄𝐿,𝑍𝐵 (𝑡) Active and reactive power load at zero
based microgrid is proposed to minimize its operation cost consider-
bus in 𝑡th hour
ing uncertainties in renewables, electrical and thermal loads. Zhang
𝑃𝑙𝑜𝑠𝑠 (𝑡), 𝑄𝑙𝑜𝑠𝑠 (𝑡) Active and reactive power loss in the
et al. [39] proposed a stochastic model predictive control strategy
system at 𝑡th hour
(SMPC) for optimal scheduling of CHP microgrid with renewables.
𝑃𝑔𝑟𝑖𝑑 (𝑡), 𝑄𝑔𝑟𝑖𝑑 (𝑡) Active and reactive power output from
In [40], the number of CHP units and renewables are planned in a smart
the grid in 𝑡th hour
cpt district for minimizing the installation and operation costs. Moradi
𝐶𝑖 , 𝐶𝑖m Capital and maintenance cost of 𝑖th DER et al. [41] determined the optimal allocation of sources and developed
unit the operational strategy for energy management in microgrid with
𝐶ifuel Fuel cost of 𝑖th CHP unit heat load considering cost, emission, and reliability objectives. Gab-
𝐶𝑅𝐹𝑖 Capital recovery factor of 𝑖th DER unit bar and Zidan [42] minimized emission and operation cost objectives
𝑌𝑖,𝑡 Yearly operating hours of 𝑖th DER unit in interconnected multi microgrids (MMGs) by optimally scheduling
𝑃𝑖𝑟 Active power rating of 𝑖th DER the cogeneration units, RES, and heat boiler. However, the uncer-
𝑃𝑖 (𝑡) Active power output of 𝑖th DER at time 𝑡 tainty in renewables was omitted. In [43], emission and operation
𝜂𝑖𝑃 , 𝜂𝑖𝐻 Power and heat efficiency of 𝑖th CHP unit cost objectives are optimized using a fuzzy PSO approach considering
𝑄𝑐𝑎𝑝 (𝑡) Discretized reactive power output from uncertainty in renewables and heat demand. In [44], fuzzy integrated
shunt capacitors PSO technique is used for solving a multi objective problem based
on energy loss, operation cost, and emission in IMG system. Further,
𝐶 buy (𝑡), 𝐶 sell (𝑡) Cost of power purchased from and sell to
the uncertainty in renewables is handled using a possibilistic approach
the main grid at time 𝑡
buy sell (𝑡) and a probabilistic approach is employed to model the uncertainties
𝑃grid (𝑡), 𝑃grid Amount of active power purchased from
in electrical and heat load. The works presented in literature [35–44]
and sell to the main grid at time 𝑡
heat , 𝐶 𝑚
were not considered either the uncertainty in renewables or optimal
𝐶bo bo
Heat price and maintenance cost of allocation of sources.
boiler Recently, a few literature, the optimal scheduling is conducted in
𝐻bo (𝑡) Heat output of boiler at time 𝑡 ZBMG where the power from the grid is set to a predefined value
𝐾𝑖𝑒 Emission coefficient of 𝑖th CHP unit in [45–48]. A predefined power from the grid can lower the emission
kgCO2 /MWh in the microgrid system. Further, the electricity price is less if lim-
𝑒
𝐾grid Emission coefficient of grid in ited power is drawn from the utility during peak periods. In [45],
kgCO2 /MWh economic load dispatch for fuel cost minimization is formulated in a
𝐾bo𝑒 Emission coefficient of boiler in microgrid having zero bus DER and renewable energy sources. In [46],
kgCO2 /MWh a combined analytical approach and symbiotic organism search (SOS)
are proposed to minimize power loss over different power injections
from the utility in the ZBMG system. Barik and Das [47] determined
the optimal allocation of DER units and shunt capacitors under un-
distribution system considering uncertainties in load demand, renew- certainties using PEM embedded mixed discrete PSO algorithm, but
ables, and location of DER units. Mohammadi et al. [29] proposed heat demand was omitted. In [48], energy loss, operation cost, and
stochastic energy management in microgrid in which uncertainties in emission objectives are optimized in the ZBMG system with electrical
load forecast, renewable generation, and the market price are consid- and heat load using a fuzzy PSO approach. However, the authors did
not consider uncertainties present in renewables.
ered to minimize the cost using adaptive modified firefly algorithm
The summary of the above literature and proposed work is listed
(AMFA). Bazmohammadi et al. [30] proposed hierarchical stochastic
in Table 1. Most of the literature considered either the optimization
energy management of interconnected renewable based microgrids and of heat energy in District Heating Systems or the optimization of
the operation of each microgrid with uncertainties is handled using electrical energy in microgrid system. Further, the above works focused
chance-constrained model predictive control (CCMPC). In [31], optimal on minimizing the cost and/or emission objectives. To the best of the
allocation of renewable energy sources (PV, WT, BIO) is determined us- author’s knowledge, none of the works deal with the optimization of
ing the analytical approach to minimize losses in the system. Kayalvizhi microgrid system with electrical and heat demand while considering

2
H. Chaduvula and D. Das Energy 282 (2023) 128909

Table 1
Summary of literature review and contribution of proposed work.
Ref GMG/MMG/ Optimal Uncertainty of Heat Objective Approach
IMG/ZBMG allocation renewables demand
of sources
[25] GMG Yes No No Maximum loadability Hybrid PSO
[26] GMG Yes No No Cost, Fuzzy + ALO
power loss,
Voltage profile,
Reliability
[32] GMG Yes Yes No Cost, Multi objective
Energy loss, harmony
Voltage deviation search algorithm
[33] GMG Yes Yes No Power loss, Weighted
Voltage stability index, aggregation PSO +
Network security index Probabilistic approach
[35] GMG No Yes Yes Maximum expected profit BMO-DE + PEM

[39] GMG No Yes Yes Operation cost MILP + SMPC


[41] GMG Yes No Yes Cost, Fuzzy + PSO +
Emission, quadratic programming
Reliability
[42] MMG No No Yes Operation cost, MOGA
Emission
[43] IMG No Yes Yes Operation cost, Fuzzy + PSO +
Emission probabilistic approach
[44] IMG No Yes Yes Operation cost, Fuzzy + PSO +
Emission, hybrid possibilistic-
Power loss probabilistic approach
[46] ZBMG Yes No No Power loss SOS + Analytical
[47] ZBMG Yes Yes No Operation cost, MDPSO + PEM
Power loss,
Voltage deviation
[48] ZBMG No No Yes Operation cost, Fuzzy + PSO
Emission,
Cost of energy loss
Proposed GMG and Yes Yes Yes Total cost, Fuzzy + PSO + PEM
work ZBMG Emission,
Energy loss

the uncertainties in renewables. This paper bridges the gap by opti- problem. Section 7 presents the analysis of results for each case. Finally,
mizing the power and heat outputs from CHP units, heat boilers, and the conclusions are outlined in Section 8.
renewable sources in different operating modes of microgrid systems
and, further, minimizing the cost, emission, and energy loss objectives. 2. System description
However, the heat generation of the sources exceeds the heat demand
during a few hours of the day. So, the limitation of the proposed work is In the present study, a 33-bus grid connected distribution system
the wastage of excess heat. In addition, the proposed fuzzy-PSO method with branch data and bus data as available in [49] is considered.
provides a single solution to the multi objective problem. The system has a total peak active power demand (𝑃𝐿𝑝𝑒𝑎𝑘 ) of 3.715
In light of the above, the main contributions of this paper are MW and total peak reactive power demand (𝑄𝑝𝑒𝑎𝑘 𝐿
) of 2.3 MVAr. In
presented below: addition to the electrical load, the system also has a heat load with a
maximum heat demand (𝐻𝐿𝑝𝑒𝑎𝑘 ) of 6 MW. The dotted line in Figs. 1
1. The optimal location and size of sources in microgrid with
and 2 represents the heat channel. The system shown in Fig. 1 meets
electrical and heat demand are determined by considering the
a majority of its power demand from distributed energy resources
objectives such as energy loss, emission, and total cost.
and then receives deficit power from the grid. Whereas the system
2. Fuzzy integrated PSO approach is used for solving the multi
shown in Fig. 2, the power from slack bus is set and the deficit power
objective problem, while uncertainty in renewables is addressed
demand is met by the natural gas turbine (NGT) unit. In order to
using the point estimate method.
meet the electrical and thermal loads, the system is integrated with
3. The microgrid with various combinations of renewables and
dispatchable CHP units, non-dispatchable renewable energy sources,
non renewables are analyzed in both GMG and ZBMG systems.
shunt capacitors and a heat boiler. The dispatchable CHP units include
Moreover, a comparative study between both the systems is
a NGT unit and a biomass (BIO) unit. In case of ZBMG system, either
illustrated.
the NGT or BIO unit fulfill the remaining power demand. The CHP units
4. The proposed method is validated with fuzzy-GA and fuzzy-
produce electrical output as well as heat output. The non-dispatchable
JAYA algorithms.
renewable energy sources are a photovoltaic (PV) unit and a wind
The rest of the paper is organized as follows: A brief description of turbine (WT) generator, both of which provide only active power to
system model and load flow computation with zero bus DER is given the system. The fixed and switched type shunt capacitors are connected
in Section 2. Section 3 presents the objective function and constraints. to the system depending on the reactive power support. The remaining
The method for solving the problem is reported in Section 4. Section 5 heat demand is met by the heat boiler when inadequate heat is supplied
details the case study. Section 6 explains the solution strategy of the from the CHP units in the system.

3
H. Chaduvula and D. Das Energy 282 (2023) 128909

Fig. 1. 33-bus grid connected microgrid system (Case 3.1).

Fig. 2. 33-bus grid connected microgrid system with NGT unit as zero bus DER (Case 3.2).

4
H. Chaduvula and D. Das Energy 282 (2023) 128909

wind velocity corresponding to the summer season and WT parameters


are taken from [33]. The mean output power of the wind turbine during
mean (𝑡)) is calculated using Eqs. (3)–(6). The hourly per unit
𝑡th hour (𝑃WT
output power of PV and WT units is shown in Fig. S1.
𝑣max
mean
𝑃WT (𝑡) = 𝑃WTo ⋅ 𝑓𝑣 (𝑣(𝑡)) (3)
∫0

⎧ 0 𝑣 < 𝑣cin or 𝑣 > 𝑣cout



𝑃WTo = ⎨ 𝑎 ⋅ 𝑣3 − 𝑏 ⋅ 𝑃r,WT 𝑣cin ≤ 𝑣 ≤ 𝑣r (4)
⎪ 𝑃r,WT 𝑣r ≤ 𝑣 ≤ 𝑣cout

𝑃r,WT 𝑣3cin
𝑎= ;𝑏 = (5)
𝑣3r − 𝑣3cin 𝑣3r − 𝑣3cin
( ( )
𝑘(𝑡) ( 𝑣(𝑡) )𝑘(𝑡)−1 𝑣(𝑡) )𝑘(𝑡)
𝑓𝑣 (𝑣(𝑡)) = ⋅ ⋅ exp − (6)
𝑐(𝑡) 𝑐(𝑡) 𝑐(𝑡)

2.2. Output of shunt capacitors

The shunt capacitors provide reactive power support and thereby


Fig. 3. Per unit electrical and heat load pattern over 24 h. helps the system in maintaining its voltage limits. In this work, fixed
and switched type shunt capacitors are connected to the system. The
capacitor sizes are chosen in steps of 50 kVAr. The fixed capacitor is
always connected to the system and the switched shunt capacitors are
connected in multiples of 50 kVAr. The total reactive power injected by
Fig. 3 shows a 24 h electrical and heat load pattern and it is taken
𝐸 (𝑡)) occurs at 12th fixed and switched shunt capacitors should lie within the limits shown
from [50]. The maximum p.u. electrical load (𝐿𝑜𝑎𝑑𝑝𝑢
𝐻 (𝑡)) occurs at 18th hour as in Eq. (20).
hour and maximum p.u. heat load (𝐿𝑜𝑎𝑑𝑝𝑢
shown in Fig. 3. The electrical and heat load at each hour is calculated
from following Eq. (1). 2.3. Zero bus DER

𝑃𝐿 (𝑡) 𝐸 (𝑡) ⎫
= 𝑃𝐿𝑝𝑒𝑎𝑘 × 𝐿𝑜𝑎𝑑𝑝𝑢 ⎪ In load flow study, each node of the 33-bus system is associated
= 𝑄𝑝𝑒𝑎𝑘 𝐸 (𝑡) ⎪ (1)
𝑄𝐿 (𝑡) 𝐿
× 𝐿𝑜𝑎𝑑𝑝𝑢 ⎬ with four quantities such as active power, reactive power, voltage
⎪ magnitude, and voltage angle. The voltage magnitude is known at bus
𝐻𝐿 (𝑡) = 𝐻𝐿𝑝𝑒𝑎𝑘 × 𝐿𝑜𝑎𝑑𝑝𝑢
𝐻 (𝑡)

⎭ 1 and its angle is taken as reference for the remaining bus voltage
angles. So, any two out of four quantities are prespecified at each
2.1. Mean PV and wind turbine power outputs node depending upon its type. The other two unknown quantities are
determined using the conventional Newton–Raphson load flow. In the
A PV array consists of a certain number of PV modules to achieve system shown in Fig. 2, the active and reactive power from bus 1
the desired power output. The output power of the PV array depends are also prespecified in addition to the voltage magnitude and angle,
on the number of PV modules, module parameters, solar irradiation, designating bus 1 as a 𝑃 𝑄𝑉 𝛿 bus [47]. In order to conserve the number
and ambient temperature. The randomness in solar irradiation causes of unknowns as the conventional load flow, no quantity is prespecified
uncertainty in its output power. Therefore, a probability density func- at one of the 33 buses in the system. The bus at which no quantity is
tion (PDF) is used for modeling the randomness in solar irradiation prespecified is termed as Zero Bus (ZB) and the DER unit located at ZB
‘𝑠’. The historical data on solar irradiation follows a beta PDF and is called as Zero Bus DER (ZBDER). The NGT unit at bus 3 in Fig. 2
its parameters (𝛼(𝑡), 𝛽(𝑡)) are calculated from given mean and standard act as a zero bus DER. The NGT unit is used for balancing the real
deviation for a particular hour. In this work, a 24 h mean and standard and reactive power demand in the system even if the power injection
deviation of solar irradiance corresponding to the summer season and from the grid is small. The mismatch in real and reactive powers are
PV module characteristics are taken from [33]. The mean output power calculated at all the buses except at zero bus. The mismatches are not
of PV array during 𝑡th hour (𝑃PV mean (𝑡)) is calculated using Eq. (2). defined at zero bus since no quantity is prespecified. At the beginning
of load flow, real and reactive power from the grid is predefined. An
mean 1 ⎫ optimum amount of power (5% of load power) is supplied from the
𝑃PV (𝑡) = ∫0 𝑃PVo ⋅ 𝑓𝑠 (𝑠(𝑡)) ⎪ grid in case of ZBMG system as shown in Eq. (7). At the end of the

mod ⎪ load flow, the calculation of real and reactive power output from the
𝑃PVo = 𝑁PV ∗ FF ∗ 𝑉o ∗ 𝐼o ⎬ zero bus DER is shown for an hour ‘𝑡’ in Eq. (8).

( ) ( ) ( ) }
(𝛼(𝑡)+𝛽(𝑡))
𝑓𝑠 (𝑠(𝑡)) = 𝛤𝛤(𝛼(𝑡))⋅𝛤 ∗ (𝑠(𝑡)) 𝛼(𝑡)−1 ∗ (1 − 𝑠(𝑡))𝛽(𝑡)−1 ; 𝛼(𝑡) > 0, 𝛽(𝑡) > 0⎪ 𝑃𝑔𝑟𝑖𝑑 (𝑡) = 0.05 × 𝐿𝑜𝑎𝑑𝑝𝑢𝐸 (𝑡) ∗ 𝑃 𝑝𝑒𝑎𝑘
(𝛽(𝑡)) ⎪ 𝐿
⎭ ( 𝐸 (𝑡) ∗ 𝑄𝑝𝑒𝑎𝑘
) (7)
𝑄𝑔𝑟𝑖𝑑 (𝑡) = 0.05 × 𝐿𝑜𝑎𝑑𝑝𝑢 𝐿
(2)
( ) ⎫
The output power of the wind turbine generator depends on the ∑
𝑃𝑍𝐵𝐷𝐸𝑅 (𝑡) = − 𝑛𝑖=2 𝑃𝑐𝑎𝑙,𝑖 (𝑡) + 𝑃𝐿,𝑍𝐵 (𝑡) + 𝑃𝑙𝑜𝑠𝑠 (𝑡) − 𝑃𝑔𝑟𝑖𝑑 (𝑡) ⎪
number of wind turbines and wind velocity. A Weibull probability ≠𝑍𝐵 ⎪
( ) ⎬
density function models the stochastic nature of wind velocity. The ∑ ⎪
𝑄𝑍𝐵𝐷𝐸𝑅 (𝑡) = − 𝑛𝑖=2 𝑄𝑐𝑎𝑙,𝑖 (𝑡) + 𝑄𝐿,𝑍𝐵 (𝑡) + 𝑄𝑙𝑜𝑠𝑠 (𝑡) − 𝑄𝑔𝑟𝑖𝑑 (𝑡)
Weibull PDF is chosen based on historical data and its parameters ⎪
(𝑘(𝑡), 𝑐(𝑡)) are calculated from given mean and standard deviation for
≠𝑍𝐵

a particular hour. In this work, a 24 h mean and standard deviation of (8)

5
H. Chaduvula and D. Das Energy 282 (2023) 128909

3. Objective function and constraints


∑ buy
𝑄𝑖 (𝑡) + 𝑄cap (𝑡) + 𝑄grid (𝑡) − 𝑄sell
grid
(𝑡) = 𝑄𝑝𝑒𝑎𝑘
𝐿
𝐸
⋅ 𝐿𝑜𝑎𝑑𝑝𝑢 (𝑡) + 𝑄loss (𝑡) (14)
Unlike previous studies which tried to minimize only the operation 𝑖∈CHP
cost [38,39], we aimed at minimizing the total cost, emission, and ∑
𝐻𝑖 (𝑡) + 𝐻𝑏𝑜 (𝑡) ≥ 𝐻𝐿𝑝𝑒𝑎𝑘 ⋅ 𝐿𝑜𝑎𝑑𝑝𝑢
𝐻
(𝑡) (15)
energy loss simultaneously, thus making our problem a multi-objective 𝑖∈CHP
optimization problem. The total cost, emission and energy loss are 𝜂𝑖𝐻
computed over 24 h (9)–(12). All the parameters used for calculating 𝐻𝑖 (𝑡) = 𝑃𝑖 (𝑡) ⋅ ; ∀𝑡 = 1, … , 𝑇 , 𝑖 ∈ CHP (16)
𝜂𝑖𝑃
the objectives are shown in Table S1 [42,51].
The heat output of CHP unit (𝐻𝑖 ) in (16) is determined from its
3.1. Total cost per day power output, power and heat efficiencies.
• Minimum and maximum limits of output power and power factor
The total cost includes the capital, operation and maintenance costs of CHP units
of CHP units, capital and maintenance costs of PV and WT units, capital
and maintenance costs of fixed and switched shunt capacitors, cost 0.1 × 𝑃𝐿𝑝𝑒𝑎𝑘 ⋅ 𝐿𝑜𝑎𝑑𝑝𝑢
𝐸
(𝑡) ≤ 𝑃𝑖 (𝑡) ≤ 0.6 × 𝑃𝐿𝑝𝑒𝑎𝑘 ⋅ 𝐿𝑜𝑎𝑑𝑝𝑢
𝐸
(𝑡) (17)
incurred from exchange of power between microgrid and grid, and the
boiler’s heat and maintenance cost in the system. There are no fuel costs
𝑝𝑓𝑖,min ≤ 𝑝𝑓𝑖 (𝑡) ≤ 𝑝𝑓𝑖,max (18)
for PV unit, WT generator and shunt capacitors and hence not included
in Eq. (9). • The capacity limits of PV and WT units
{ ( )
∑𝑇 ∑ 𝐶𝑖cpt 𝐶 fuel }
𝑓1 = ⋅ 𝐶𝑅𝐹𝑖 ⋅ 𝑃𝑖𝑟 + 𝑖 𝑃 ⋅ 𝑃𝑖 (𝑡) + 𝐶𝑖m ⋅ 𝑃𝑖 (𝑡) mod
𝑁PV,min mod ≤ 𝑁 mod,max
≤ 𝑁PV
𝑡=1 𝑖∈CHP
𝑌𝑖,𝑡 𝜂𝑖 PV (19)
( cpt ) 𝑁WT,min ≤ 𝑁WT ≤ 𝑁WT,max
∑ 𝐶𝑖
+ ⋅ 𝐶𝑅𝐹𝑖 ⋅ 𝑃𝑖𝑟 + 𝐶𝑖m ⋅ 𝑃𝑖 (𝑡) • The reactive power limits of shunt capacitor
𝑖∈PV,WT
𝑌𝑖,𝑡
( cpt )
∑ 𝐶𝑖 (9)
+ ⋅ 𝐶𝑅𝐹𝑖 ⋅ 𝑄𝑖 (𝑡) + 𝐶𝑖m ⋅ 𝑄𝑖 (𝑡) 𝑄min ≤ 𝑄cap ≤ 𝑄max (20)
𝑖∈fcap,swcap
𝑌 𝑖,𝑡
buy • Node voltage limits
+ 𝐶 buy (𝑡) ⋅ 𝑃grid (𝑡) − 𝐶 sell (𝑡) ⋅ 𝑃grid
sell
(𝑡)
} 𝑉min ≤ 𝑉𝑛 (𝑡) ≤ 𝑉max (21)
( heat 𝑚
)
+ 𝐶bo + 𝐶bo ⋅ 𝐻bo (𝑡) ∀𝑡 = 1, … , 𝑇
• Branch current limits
max
The formula for capital recovery factor of each source is given below 𝐼br (𝑡) ≤ 𝐼br (22)
(10), where 𝐿𝑖 represents life period of each source. In this work the
interest rate (𝑟) is chosen as 10%. 4. Methodology of solving multi objective problem

𝑟(1 + 𝑟)𝐿𝑖
𝐶𝑅𝐹𝑖 = (10) 4.1. Fuzzy-PSO
(1 + 𝑟)𝐿𝑖 − 1
The particle swarm optimization (PSO) is a meta-heuristic algorithm
3.2. Emission per day
resembling the social behavior of a bird flock [52]. PSO is an iterative
technique where the particle’s position and velocity are updated during
Emission from CHP units, grid, and the boiler is computed over 24 h
each iteration based on its own best position and group best position.
as shown in Eq. (11). The non dispatchable renewables such as WT
In every iteration, the velocity and position of 𝑖th particle is updated
units and PV generators, and shunt capacitors are zero pollutant sources
as shown in (23)–(24).
and therefore do not contribute to emission. The emission from the grid ( iter )
is omitted if the microgrid sells power to the grid. 𝑉iditer+1 = 𝜔𝑖 ⋅ 𝑉iditer + 𝑐1 ∗ rand() ∗ 𝑃id iter
− 𝑋id
( ) ( )
( ) + 𝑐2 ∗ rand() ∗ 𝐺diter − 𝑋id iter
(23)

𝑇 ∑ buy
𝑒 𝑒 𝑒
𝑓2 = 𝐾𝑖 ⋅𝑃𝑖 (𝑡) +𝐾grid ⋅𝑃grid (𝑡)+𝐾bo ⋅𝐻bo (𝑡) ; ∀𝑡 = 1, … , 𝑇
𝑡=1 𝑖∈𝐶𝐻𝑃
iter+1 iter
(11) 𝑋id = 𝑋id + 𝑉iditer+1 (24)

where 𝜔 is the inertia weight, 𝑐1 and 𝑐2 are acceleration coefficients.


3.3. Energy loss per day These PSO parameters are varied in each iteration to facilitate the
attainment of global optimum [53].
The hourly real power loss is added over 24 h to obtain the per day In order to solve a multi objective problem using PSO algorithm,
energy loss shown in Eq. (12). all the objectives must be combined to form a single objective using

𝑇 weighted sum technique. This technique requires the individual ob-
𝑓3 = 𝑃𝑙𝑜𝑠𝑠 (𝑡); ∀𝑡 = 1, … , 𝑇 (12) jectives to vary within the same range of values. As the objectives
𝑡=1 considered in this study have different ranges, the first step is to
Constraints bring all the objectives to a common range. This is achieved in fuzzy
The system has to satisfy the following constraints (13)–(22) along approach by transforming each objective into its membership value
with objectives. using (25). A trapezoidal membership function is considered for each
• Power and heat balance constraints objective as shown in Fig. S2.
( )
∑ ⎧ 1− 𝑓𝑖 −𝑓𝑖min
buy sell
𝑃𝑖 (𝑡) + 𝑃PV (𝑡) + 𝑃WT (𝑡) + 𝑃grid (𝑡) − 𝑃grid (𝑡) = 𝑃𝐿𝑝𝑒𝑎𝑘 ⋅ 𝐿𝑜𝑎𝑑𝑝𝑢
𝐸
(𝑡) + 𝑃loss (𝑡) 𝑓𝑖min < 𝑓𝑖 < 𝑓𝑖max
⎪ max
𝑓𝑖 −𝑓𝑖min
𝑖∈CHP 𝜇𝑓̃𝑖 = ⎨ (25)
1 𝑓𝑖 ≤ 𝑓𝑖min

(13) ⎩ 0 𝑓𝑖 ≥ 𝑓𝑖max

6
H. Chaduvula and D. Das Energy 282 (2023) 128909

The combined objective for PSO algorithm can be formulated as the 𝑘th concentration of input random variable 𝑝𝑙 and all other input
random variables set to be their respective mean values as given in (31).
𝑀𝑎𝑥 𝑓 = 𝑊1 ⋅ 𝜇𝑓̃1 + 𝑊2 ⋅ 𝜇𝑓̃2 + 𝑊3 ⋅ 𝜇𝑓̃3 (26)

where 𝜇𝑓̃𝑖 and 𝑊𝑖 are the membership value and weight assigned to 𝑍1,𝑘 = 𝐹 (𝑝1,𝑘 , 𝜇𝑝2 , … , 𝜇𝑝𝑚 ) ⎪
the ith objective respectively. The weights 𝑊𝑖 must satisfy the condition 𝑍2,𝑘 = 𝐹 (𝜇𝑝1 , 𝑝2,𝑘 , … , 𝜇𝑝𝑚 ) ⎪
⎬ ∀ 𝑘 = 1, 2, 3 (31)
∑3 ⋮ ⎪
𝑊𝑖 = 1. In this work, equal priority is given to all the objectives and 𝑍𝑚,𝑘 = 𝐹 (𝜇𝑝1 , 𝜇𝑝2 , … , 𝑝𝑚,𝑘 ) ⎪
𝑖=1 ⎭
1
henceforth 𝑊1 = 𝑊2 = 𝑊3 = .
The maximization of 𝑓 in (26) is
3 It is worth to note that 𝑍1,3 = 𝑍2,3 = ⋯ = 𝑍𝑚,3 as the function 𝐹 is
equivalent to minimization of all the objectives. evaluated using the mean values of all the input random variables as
shown in (32).
4.2. Handling uncertainties using point estimate method
𝑍1,3 = 𝑍2,3 = ⋯ = 𝑍𝑚,3 = 𝐹 (𝜇𝑝1 , 𝜇𝑝2 , … , 𝜇𝑝𝑚 ) (32)
In our system, the power from PV and WT units depend on the
solar irradiation and wind velocity respectively. The stochastic nature As a result, only (2𝑚 + 1) simulations are sufficient to obtain the set of
of solar irradiation and wind velocity causes uncertainty in the power values 𝑍𝑙,𝑘 . This is the reason why this scheme is called as the (2𝑚 + 1)
injected from these sources. This in turn renders uncertainty in total PEM method.
cost, emission, and energy loss objectives, which are computed using
5. Case study
PV and WT power as inputs to the load flow. This probabilistic nature
of the objective function is handled using the Point Estimate Method
In this work, the operation of the microgrid is examined by consid-
(PEM), which was first introduced by Emilio Rosenblueth in 1975 [54]
ering the following eight cases.
and later developed by many researchers [55]. In a nutshell, the PEM
Case 0 (Base case): GMG system with shunt capacitors and a heat
computes the statistical moments of a function of random variables
boiler
directly from the first few central moments of these variables. The
Case 1.1: GMG system with NGT unit, shunt capacitors and a boiler
central moments of any input random variable can be computed from
Case 1.2: ZBMG system with NGT unit, shunt capacitors and a boiler
its probability density function which is known a priori. In our system,
Case 2.1: GMG system with biomass, PV, and WT units, shunt capaci-
the total cost, emission and energy loss are the output random variables
tors and a boiler
which are functions of the power injected from PV and WT units which
Case 2.2: ZBMG system with biomass, PV, and WT units, shunt capac-
are the input random variables. The power output from the PV and
itors and a boiler
WT units follow beta and Weibull distributions respectively, whose
Case 3.1: GMG system with NGT, biomass, PV, and WT units, shunt
mathematical expressions are given in Eqs. (2) and (6) respectively.
capacitors and a boiler
In the Hong’s (2𝑚 + 1) scheme, the moments of the output random
Case 3.2: The system employs sources same as in case 3.1. The NGT
variable are obtained by performing only 2𝑚 + 1 simulations, where
unit in this case acts as a zero bus DER
𝑚 is the number of input random variables. This scheme has proved to
Case 3.3: In this case, the system is equivalent to case 3.1, but the
be efficient by providing results of reasonable accuracy at a minimal
deficit power demand is met by the BIO unit.
computational cost.
In all the cases, the deficit heat demand is met by the heat boiler.
In the (2𝑚 + 1) scheme, each input random variable, denoted by 𝑝𝑙
A detailed information regarding each case is presented in supporting
(𝑙 = 1, 2, … , 𝑚), is represented by three concentrations. Each concen-
information.
tration of input random variable 𝑝𝑙 is an ordered pair of location and
weight (𝑝𝑙,𝑘 , 𝑤𝑙,𝑘 ) where 𝑘 = 1, 2, 3. The locations 𝑝𝑙,𝑘 and weights 𝑤𝑙,𝑘
6. Solution of the proposed multi objective problem
can be computed using (27)–(28):

𝑝𝑙,𝑘 = 𝜇𝑝𝑙 + 𝜉𝑙,𝑘 ⋅ 𝜎𝑝𝑙 (27) The complete procedure of solving the optimization problem is
given in two parts. In the first part, the optimal locations of different
√ ⎫
𝜆𝑙,3 3 types of sources are determined at a peak electrical load hour. A
𝜉𝑙,𝑘 = 2
+ (−1)3−𝑘 𝜆𝑙,4 − 4
⋅ 𝜆2𝑙,3 ⎪
⎪ detailed procedure for first part is given in Algorithm 1.
for 𝑘 = 1, 2 ⎪
𝜉𝑙,3 =0 ⎪
(−1)3−𝑘 ⎬ (28) In the second part, the ratings of sources are determined by solving
𝑤𝑙,𝑘 = 𝜉𝑙,𝑘 (𝜉𝑙,1 −𝜉𝑙,2 )
for 𝑘 = 1, 2 ⎪ the multi objective problem at each hour using the point estimate

⎪ method embedded fuzzy PSO. Fig. 4 depicts the flowchart for optimiza-
1 ∑𝐾
𝑤𝑙,3 = − 𝑤𝑙,𝑘 ⎪ tion in microgrid employed with renewable and non renewable sources
𝑚 𝑘=1 ⎭
in which NGT unit as a zero bus DER (case 3.2).
where 𝜇𝑝𝑙 is the mean, 𝜎𝑝𝑙 is the standard deviation of the PDF followed
by the 𝑙th random variable. In Eq. (29), 𝜆𝑙,𝑗 is the 𝑗th standard central 7. Results and discussions
moment of the input random variable 𝑝𝑙 defined as

∫−∞ (𝑝𝑙 − 𝜇𝑝𝑙 )𝑗 𝑓𝑝𝑙 𝑑𝑝𝑙 All the simulations are carried out using MATLAB (R2019b) soft-
𝜆𝑙,𝑗 = 𝑗 = 1, 2, 3, 4 (29) ware built on a system with Intel Core i5 processor of 1.70 GHz, with an
(𝜎𝑝𝑙 )𝑗
8 GB RAM environment. In this work, three meta heuristic algorithms
where 𝑓𝑝𝑙 is the probability density function of 𝑝𝑙 . such as PSO, genetic algorithm (GA), and JAYA are used for solving the
The expected value and 𝑗th moment of output random variable 𝑍 objective and a population size of 300 and 500 maximum iterations are
are calculated by using Eq. (30). chosen for each technique.
∑ ∑𝐾 ⎫ The optimal locations and ratings of sources which minimize the
𝐸(𝑍) = 𝑚 𝑘=1 𝑤𝑙,𝑘 ⋅ 𝑍𝑙,𝑘
𝑙=1 ⎪ total cost, emission and energy loss for each case are shown in Table 2.
⎬ (30)
∑𝑚 ∑𝐾 In the base case, the system has shunt capacitors at bus 16 and bus
𝐸(𝑍 ) = 𝑙=1 𝑘=1 𝑤𝑙,𝑘 ⋅ (𝑍𝑙,𝑘 ) ⎪
𝑗 𝑗
⎭ 30 with a capacity of 550 kVAr and 1200 kVAr respectively, whereas
where 𝑤𝑙,𝑘 are the weights defined in Eq. (28) and 𝑍𝑙,𝑘 is the value all other cases have shunt capacitors only at a single bus. The rating
of output random variable obtained by evaluating the function 𝐹 with of NGT unit is found to be maximum in case 1.2 with a capacity

7
H. Chaduvula and D. Das Energy 282 (2023) 128909

Fig. 4. Flow chart of the PEM embedded Fuzzy-PSO approach for the system with NGT unit as zero bus DER.

of 3606.7 kW, 0.8542 lag p.f. The BIO unit has a maximum rating evident from these results that using only boiler for heat generation is
of 2999.5 kW, 0.8089 lag p.f. in case 2.2. The PV unit also has its not an efficient way of handling the system’s heat demand. Therefore,
maximum capacity in case 2.2 with a peak power rating of 660 kW, upf. reducing heat requirement from the boiler by using alternative sources
The rating of WT unit is obtained as 250 kW, upf in all cases wherever for heat generation may decrease the total cost and emission. Em-
it is present. The optimal power injection from the grid with respect to ploying sources which can simultaneously produce heat and electricity
the total load is obtained as 5% in case 1.2, 20% in case 2.2, 5% in case drastically reduce the total cost and emission as demonstrated in the
3.2 and 5% in case 3.3. Power injection from the grid is maximum in cases that follow.
case 2.2 as the cost of renewables is significantly higher compared to
that of the grid. Case 1.1:

Case 0: Base case In this case, the system is powered by a NGT unit in addition to the
grid, shunt capacitors, and boiler. The electrical demand is satisfied
This case incurs a total cost of 14472 $/day, emission of 35373 by the grid, shunt capacitors, and the NGT unit. Heat generation
kgCO2 /day and energy loss of 2508.44 kWh/day as reported in Table 3. from the boiler and NGT unit fulfills the heat demand. In this system
Emission from the energy sources is found to be maximum in this case configuration, the deficit power demand is met by the grid. As shown
compared to all other cases due to high emission characteristics of in Table 3, this case results in a total cost of 10384 $/day, emission of
the boiler, which is solely responsible for fulfilling the system’s heat 25845 kgCO2 /day, and energy loss of 910.22 kWh/day.
demand as shown in Fig. 6. Energy loss is also maximum in this case The inclusion of the NGT unit in this case reduces total cost, emis-
due to the absence of output from distributed energy resources as listed sion, and energy loss compared to that of the base case. The electrical
in Fig. 5. This case is moderate in terms of total cost because it does output from the grid and boiler’s heat output are reduced as the NGT
not include any type of generators which are associated with capital & unit generates both electricity and heat as shown in Figs. 5 and 6
maintenance costs in addition to the operation cost. respectively. As shown in Fig. 6, the heat output from the boiler in this
It is to be noted from Table 4 that out of the total cost of 14472 case is reduced to half of the total heat demand as the NGT produces
$/day, 3418 $/day (i.e. 23.62%) is spent on the boiler which is the the remaining half as opposed to the base case where the boiler alone
only heat source in this case. Also, emission from the boiler amounts satisfies the complete heat demand. As a result, cost and emission from
to nearly 68% of the total emission in this case as shown in Table 5. It is the boiler are also halved compared with those in the base case as

8
H. Chaduvula and D. Das Energy 282 (2023) 128909

Table 2
Optimal locations and ratings of sources corresponding to each configuration of microgrid.
Case Optimal NGT BIO WT PV Shunt Percent of
allocation Capacitors optimum power
from grid w.r.t
load
location – – – – 16 30
Base Case –
size – – – – 550 1200
kVAr kVAr
location 6 – – – 32
Case 1.1 –
size 2229 kW, – – – 600 kVAr
0.8490 lag p.f.
location 6 – – – 30
Case 1.2 5%
size 3606.7 kW, – – – 50 kVAr
0.8542 lag p.f.
location – 6 18 33 30
Case 2.1 -
size – 2216.3 kW, 250 kW, 132 kW, 700 kVAr
0.9 lag p.f. Upf Upf
location – 6 25 16 30
Case 2.2 20%
size – 2999.5 kW, 250 kW, 660 kW, 500 kVAr
0.8089 lag p.f. Upf Upf
location 3 30 16 18 14
Case 3.1 -
size 2229 kW, 1408.9 kW, 250 kW, 253 kW, 300 kVAr
0.9 lag p.f. 0.8 lag p.f. Upf Upf
location 3 30 18 16 14
Case 3.2 5%
size 2381.7 kW, 1578.1 kW, 250 kW, 132 kW, 300 kVAr
0.8873 lag p.f. 0.8033 lag p.f. Upf Upf
location 6 24 32 18 30
Case 3.3 5%
size 2216.3 kW, 1579.2 kW, 250 kW, 132 kW, 550 kVAr
0.8838 lag p.f. 0.9 lag p.f. Upf Upf

Algorithm 1 Finding locations of sources at peak load hour


Table 3
Per day objective values in each case. 1: Collect input data regarding load, PV, WT, and shunt capacitors.
Case Total cost Emission Energy loss 2: for 𝑡 = 12 (peak load hour) do
($/day) (kgCO2 /day) (kWh/day)
3: Specify lower and upper bounds of location and capacity of
Base case 14472 35373 2508.44
sources. In addition to that, real power from the grid is considered
Case 1.1 10384 25845 910.22
Case 1.2 𝟖𝟎𝟗𝟖 20206 1396.86 as an optimization variable in case of ZBMG.
Case 2.1 18730 7585 771.75 4: for i=1:𝑁𝑜𝑏𝑗 do
Case 2.2 21443 𝟐𝟒𝟎𝟎 869.39 5: Generate population using PSO.
Case 3.1 13469 9747 𝟔𝟎𝟓.𝟕𝟗 6: Run load flow, calculate objectives (9)-(12) and check
Case 3.2 14174 9049 655.95
constraints (13)-(22) for each particle.
Case 3.3 14276 8792 652.11
7: Update position and velocity of particle using Eqns (23)-(24)
based on fitness values.
8: Calculate minimum and maximum limits of objective
function.
listed in Tables 4 and 5. As shown in Table 4, the combined cost of
power generation from the NGT unit and grid is less compared to the 9: end for
cost of generating the power from the grid alone in the base case since 10: Repeat steps from 5 to 7 and calculate membership values of all
the NGT unit has lower capital, operation & maintenance costs. Also, objectives using Eq (25).
the cost of the boiler is 1717.50 $/day (i.e. 16.54%), which is nearly 11: Solve combined objective (26) using fuzzy-PSO.
half of the base case. From Table 5, it is evident that this case results 12: The locations of all sources and capacities of RES are
in lower emission than the base case due to a significant reduction
determined.
in emission from the boiler (24092 kgCO2 /day to 12107 kgCO2 /day)
13: end for
despite considerable NGT unit’s emission (i.e. 35.75%). From Table 6,
the cost, emission, and energy loss are reduced by 28.25%, 26.94%, and 14: Repeat above procedure for 25 simulations. The best out of 25
63.71% respectively in this case compared to the base case. Although simulations are chosen as optimal locations.
all the objectives are better in comparison with the base case, this case
is not optimal as the emission and energy loss can be further reduced
with the inclusion of renewable energy sources.
In this case, the NGT unit delivers maximum power and heat due
to limited power from the grid in addition to the absence of renew-
Case 1.2: ables. Consequently, total cost and emission are significantly reduced
compared to case 1.1, but energy loss is increased due to limited power
This case is same as case 1.1 except that the NGT unit satisfies the from the grid. So, a further reduction in cost and emission is observed
deficit power demand. The power from the grid is fixed as 5% of compared to the base case. The cost of the boiler is reduced to 673.83
load power. From Table 3, this case incurs a total cost of 8098 $/day, $/day and emission from the boiler is reduced to 4749.80 kgCO2 /day
emission of 20206 kgCO2 /day, and energy loss of 1396.86 kWh/day.

9
H. Chaduvula and D. Das Energy 282 (2023) 128909

Table 4
Per day cost of each source in each microgrid configuration.
Case Grid NGT BIO WT PV Shunt boiler 𝑓1
($ per day) ($ per day) ($ per day) ($ per day) ($ per day) capacitors ($ per day) ($ per day)
($ per day)

Base case 10745 (74.25%) 0 0 0 0 309.08 (2.14%) 3417.90 (23.62%) 14472

Case 1.1 4284.40 (41.26%) 4266.80 (41.09%) 0 0 0 115.56 (1.11%) 1717.50 (16.54%) 10384
Case 1.2 519.51 (6.42%) 6892.60 (85.11%) 0 0 0 12.11 (0.15%) 673.83 (8.32%) 8098

Case 2.1 3963 (21.16%) 0 13633 (72.79%) 169.89 (0.91%) 345.48 (1.84%) 150.01 (0.80%) 468.69 (2.50%) 18730
Case 2.2 2078.10 (9.69%) 0 17386 (81.08%) 169.89 (0.79%) 1727.40 (8.06%) 75.12 (0.35%) 6.19 (0.03%) 21443

Case 3.1 268.25 (1.99%) 4266.80 (31.68%) 8019.40 (59.54%) 169.89 (1.26%) 662.16 (4.92%) 60.73 (0.45%) 21.60 (0.16%) 13469
Case 3.2 519.51 (3.67%) 4120.30 (29.07%) 8954.70 (63.18%) 169.89 (1.20%) 345.48 (2.44%) 55.73 (0.39%) 8.21 (0.06%) 14174
Case 3.3 519.51 (3.64%) 3943.60 (27.62%) 9189.70 (64.37%) 169.89 (1.19%) 345.48 (2.42%) 107.34 (0.75%) 0.00 (0%) 14276

Table 5
Per day emission of each source in each microgrid configuration.
Case Grid NGT BIO boiler 𝑓2
(kgCO2 /day) (kgCO2 /day) (kgCO2 /day) (kgCO2 /day) (kgCO2 /day)
Base case 11281 0 0 24092 35373
(31.89%) (68.11%)
Case 1.1 4499.10 9238.90 0 12107 25845
(17.41%) (35.75%) (46.84%)
Case 1.2 546.12 14910 0 4749.80 20206
(2.70%) (73.79%) (23.51%)
Case 2.1 4144 (54.63%) 0 137.33 3303.80 7585
(1.81%) (43.56%)
Case 2.2 2184.50 0 172.16 43.69 (1.82%) 2400
(91.01%) (7.17%)
Case 3.1 276.79 9238.90 78.97 (0.81%) 152.28 9747
(2.84%) (94.79%) (1.56%)
Case 3.2 546.12 8357.30 88.09 (0.97%) 57.89 (0.64%) 9049
(6.03%) (92.35%)
Case 3.3 546.12 8154.50 91.10 (1.04%) 0 (0%) 8792
(6.21%) (92.75%)

Fig. 5. Per day active power output of sources in each case.

10
H. Chaduvula and D. Das Energy 282 (2023) 128909

Fig. 6. Per day heat output of sources in each case.

Table 6 other hand, renewables have fewer emissions, and also decrement in
Percentage deviation from base case objective values.
emission from both grid and boiler causes less emission compared to
Case Total cost Emission Energy loss the base case as shown in Table 5. So, it is concluded from Table 6
Case 1.1 −28.25% −26.94% −63.71% that the cost is increased by 29.42% compared to the base case as
Case 1.2 −44.04% −42.88% −44.31% renewables are expensive sources. Also, the presence of renewables
Case 2.1 +29.42% −78.56% −69.23% in the system decreases the emission and energy loss by 78.56% and
Case 2.2 +48.17% −93.21% −65.34% 69.23% in comparison with the base case respectively. This case is not
Case 3.1 −6.93% −72.45% −75.85% preferred because of the higher total cost.
Case 3.2 −2.06% −74.42% −73.85%
Case 2.2:
Case 3.3 −1.36% −75.15% −74.00%
The system in case 2.2 is the same as case 2.1 except that the biomass
unit is responsible for satisfying the deficit power demand. The opti-
as shown in Tables 4 and 5 respectively. The cost, emission, and energy mum power injection from the grid is obtained as 20% of load power.
loss are reduced by 44.04%, 42.88%, and 44.31% respectively in this This case results in a maximum output from renewables compared to
case compared to the base case as shown in Table 6. It is worth noting other cases as shown in Fig. 5. It is observed from Fig. 6 that the heat
that this case results in minimum cost compared to all other cases. This generation of biomass unit exceeds heat demand because of its output
case is also not optimal due to higher emission and energy loss than power and high heat efficiency. A minimum heat is produced from the
other cases. boiler since heat supplied from the biomass is not enough to meet the
heat demand in a few hours of a day. From Table 3, a total cost of
Case 2.1: 21443 $/day, emission of 2400 kgCO2 /day, and energy loss of 869.39
kWh/day are attained in this case.
In this case, the system is supported with renewable energy sources
such as biomass, PV, and WT units in addition to the main grid and The biomass unit delivers maximum power and heat in this case
shunt capacitors. The electrical power is generated from renewables, compared to other cases. From Table 4, the cost of power and heat
grid, and shunt capacitors. The non dispatchable renewables such as generation from biomass is 17386 $/day (81.08%), which is higher
PV and WT units are operated at unity power factor and deliver only than that of case 2.1. This case results in a maximum total cost as the
active power. The grid is responsible for fulfilling the remaining power cost incurred from renewables is enormous. As shown in Table 5, a
demand. The heat demand is met by the biomass and boiler sources. significant reduction in emission from the grid and boiler are noticed
In this case, a total cost of 18730 $/day, emission of 7585 kgCO2 /day, in this case compared to case 2.1. As a result, the total emission is less in
and energy loss of 771.75 kWh/day are obtained as shown in Table 3. this case despite a slight increase in biomass emission. Here, the energy
The inclusion of renewables in this case reduces emission and loss is more than that of case 2.1. From Table 6, the emission and
energy loss considerably but increases the total cost as compared to energy loss decrease by 93.21% and 65.34% respectively, but increases
that of the base case. As shown in Table 4, the cost incurred from the cost by 48.17% compared to the base case. This case yields the
biomass corresponding to its power and heat generation is 13633 highest total cost and lowest emission compared to all other cases.
$/day (i.e. 72.79%), which is more than the cost incurred from the Case 3.1:
grid in the base case (10745 $/day). Besides, the cost incurred from
non dispatchable renewables such as PV and WT units, grid, shunt In this case, the system includes the NGT unit and renewables in
capacitors, and the boiler is added to the biomass cost. Therefore, the addition to those present in the base case configuration. The electrical
total cost is more in this case compared to the base case despite an load is met by the grid, renewables, NGT unit, and shunt capacitors,
appreciable reduction in the boiler’s cost as noticed in Table 4. On the whereas heat is produced from the NGT, biomass, and the boiler. In this

11
H. Chaduvula and D. Das Energy 282 (2023) 128909

Fig. 7. The percentage of operation cost of each source in all cases.

case, the majority of the power demand is supplied by the distributed to base case values. This case is also recommended as all the objectives
energy resources and the grid maintains power balance in the system are reduced to a satisfactory level.
by either injecting deficit power or absorbing excess power generated
Case 3.3:
within the microgrid. As shown in Table 3, this case results in a total
cost of 13469 $/day, emission of 9747 kgCO2 /day, and energy loss of This case resembles case 3.2, the only difference being the biomass
605.79 kWh/day. generator responsible for satisfying the power balance instead of NGT
unit as in case 3.2. In this case, the grid delivers an optimum power
The integration of NGT and renewables in this case results in a
of 5% of load power. A total cost of 14276 $/day, emission of 8792
lower cost compared to that of the base case as shown in Table 4.
kgCO2 /day, and energy loss of 652.11 kWh/day are obtained from
This is mainly due to a significant reduction in boiler’s cost (21.60
Table 3. Compared to case 3.1, the cost of biomass unit is more and
$/day) and cost incurred from grid (268.25 $/day). A minimum cost
emission from the NGT unit is low; thus, the total cost is higher and
is incurred from the grid in this case compared to other cases. It
emission is lower as shown in Tables 4 and 5, respectively. It is noticed
is also observed that the total cost in this case is lower than the
that the boiler does not furnish any cost and emission in this case.
case with only renewables (case 2.1) but higher than the case with
From Table 6, the cost of the system is slightly decreased by 1.36%, the
only NGT unit (case 1.1). From Table 5, it is observed that the total
emission, and energy loss are reduced by 75.15% and 74% respectively.
emission in this case (9747 kgCO2 /day) is less than that of the base case
This case is also chosen for satisfactory reduction in all the objectives.
excluding the boiler’s emission (11281 kgCO2 /day). If boiler emission
is considered then the emission is still higher in the base case (35373 Results of case 3.1, case 3.2, and case 3.3:
kgCO2 /day). From Table 6, the cost, emission, and energy loss are The pie chart for operation cost and emission of each source in
decreased by 6.93%, 72.45%, and 75.85% respectively in comparison all cases is shown in Figs. 7 and 8 respectively. From Figs. 7 and 8,
with the base case. This case results in minimum energy loss compared the boiler has achieved maximum percentage of operation cost and
to all other cases. As all the objectives are reduced satisfactorily, this emission in base case and its share is insignificant in cases 3.1, 3.2,
case is preferred. and 3.3. It is noticed from Fig. 7 that the larger portion of NGT unit in
case 1.2 results in minimum total cost and major share of renewables in
Case 3.2: case 2.2 results in maximum total cost. From Fig. 8, a larger percentage
of boiler in base case has achieved maximum emissions compared to
The system configuration in this case is similar to case 3.1, but the other cases. A significant percentage of NGT unit’s operation cost and
remaining power demand is met by the NGT unit. The grid injects 5% negligible contribution from the boiler’s emission leads to satisfactory
of the load power for which the objectives are optimal. Similar to case cost and emission objectives for cases 3.1, 3.2, and 3.3.
3.1, the electrical power is delivered by the grid, NGT unit, renewables, Fig. 9 shows box and whisker plot for output power and/or output
and shunt capacitors. But the unmet electrical demand is fulfilled by the heat of each source over 15 runs for case 3.1, case 3.2, and case 3.3
NGT unit. The NGT unit, biomass, and boiler are heat generating units. respectively. A slight variation of output power from the grid is noticed
As shown in Table 3, a total cost of 14174 $/day, emission of 9049 in case of grid connected microgrid (Case 3.1) and a constant maximum
kgCO2 /day, and energy loss of 655.95 kWh/day are obtained. output power from the grid is observed in cases 3.2, and 3.3. The
As shown in Table 4, the cost incurred from the biomass unit and minimum variation in power and heat outputs of NGT unit as well
grid are more in this case compared to case 3.1. As a result, this case as BIO unit are noticed in case 3.1. The per day output power and
results in higher cost compared to case 3.1 despite a slight decrement in heat values of NGT unit lies between median and 75th percentile and
costs of PV and NGT units. On the other hand, the total emission is less of the whiskers in case 3.2, whereas, the per day output power and
than case 3.1 due to a considerable reduction in NGT unit’s emission heat values from BIO unit lies between 25th percentile and median
as shown in Table 5. From Table 6, the total cost, emission, and energy and of the whiskers over 15 simulation runs. In case 3.3, the variation
loss are reduced by 2.06%, 74.42%, and 73.85% respectively compared in output power and heat is observed between 25th percentile and

12
H. Chaduvula and D. Das Energy 282 (2023) 128909

Fig. 8. The percentage of emission of each source in all cases.

Fig. 9. Boxplot of the per day power and heat output of each source obtained with Fuzzy-PSO over 15 runs for Case 3.1, Case 3.2, and Case 3.3.

75th percentile for both NGT and BIO units and also, a few values of individual objectives equivalent to a maximum mean value of com-
of data are present in outliers. The variation in output power from bined objective in fuzzy domain. The combined objective with a higher
renewables and heat output from the boiler are noticed in case 3.1 and mean value of 0.8503 per unit and lower standard deviation of 0.0085
case 3.2 respectively. The variation in reactive power output from shunt is obtained from the proposed fuzzy-PSO method taking 40.10 minutes
capacitors is observed in all the cases. of computation time per each simulation.
The results of case 3.1 obtained from proposed fuzzy-PSO method
is compared with fuzzy-GA and fuzzy-JAYA meta heuristic techniques. 8. Conclusion
The mean and standard deviation values of objectives over 15 simula-
tion runs for proposed fuzzy-PSO, fuzzy-GA, and fuzzy-JAYA algorithms The operation of a microgrid powered by several combinations of
are shown in Table 7. It is noticed from Table 7 that the proposed renewable and non-renewable energy sources has been optimized for
method yields minimum mean and standard deviation values than better performance in terms of the total cost, emission, and energy loss.
other methods with less computation time. The minimum mean values The optimal locations and ratings of sources present in each microgrid

13
H. Chaduvula and D. Das Energy 282 (2023) 128909

Table 7
Per day objective values and computation time obtained with Fuzzy-PSO, Fuzzy-GA, and Fuzzy-JAYA algorithms over 15 runs for Case 3.1.
Solution Total cost Emission Energy loss Combined Simulation
strategy ($/day) (kgCO2 /day) (kWh/day) objective time
(per unit) per one run
(minutes)
Mean Std Mean Std Mean Std Mean Std
Fuzzy-PSO 13518 57.66 9744.70 6.91 624.53 46.16 0.8503 0.0085 40.10
Fuzzy-GA 13534 72.11 9712.30 57.04 624.66 44.80 0.8498 0.0090 46.99
Fuzzy-JAYA 13572 173.92 9641.70 219.90 647.65 39.83 0.8436 0.0122 43.33

configuration which simultaneously minimize all the objectives have [2] Nazari-Heris M, Abapour S, Mohammadi-Ivatloo B. Optimal economic dispatch
been determined by using the fuzzy PSO algorithm. The point estimate of FC-CHP based heat and power micro-grids. Appl Therm Eng 2017;114:756–69.
http://dx.doi.org/10.1016/j.applthermaleng.2016.12.016.
method has been incorporated in the fuzzy-PSO approach for handling
[3] Arandian B, Ardehali M. Effects of environmental emissions on optimal com-
the uncertainty involved in PV and WT units. The base case, where bination and allocation of renewable and non-renewable CHP technologies in
the heat demand is solely fulfilled by the boiler, results in higher cost heat and electricity distribution networks based on improved particle swarm
and emission due to the characteristics of the boiler. In cases 1.1 and optimization algorithm. Energy 2017;140:466–80. http://dx.doi.org/10.1016/j.
1.2, the inclusion of NGT unit has succeeded in lowering the cost and energy.2017.08.101.
emission by reducing the electrical load on the grid and heat load on [4] Shao C, Ding Y, Wang J, Song Y. Modeling and integration of flexible demand
in heat and electricity integrated energy system. IEEE Trans Sustain Energy
the boiler. When the NGT unit acts as the zero bus DER in case 1.2, it 2018;9(1):361–70. http://dx.doi.org/10.1109/TSTE.2017.2731786.
further reduces the cost and emission as compared to case 1.1 where [5] Wang Y, Yu H, Yong M, Huang Y, Zhang F, Wang X. Optimal scheduling of inte-
the system power balance is maintained by the grid. On the other hand, grated energy systems with combined heat and power generation, photovoltaic
incorporating renewable energy sources like biomass, PV, and WT units and energy storage considering battery lifetime loss. Energies 2018;11(7). http://
in cases 2.1 and 2.2 results in a drastic reduction of emission and dx.doi.org/10.3390/en11071676, URL https://www.mdpi.com/1996-1073/11/
7/1676.
energy loss but increases the total cost compared with the base case.
[6] Barbieri ES, Melino F, Morini M. Influence of the thermal energy storage on
The combination of both NGT and renewables in cases 3.1, 3.2, and 3.3 the profitability of micro-CHP systems for residential building applications. Appl
achieve a reduction in cost, emission and energy loss simultaneously Energy 2012;97:714–22. http://dx.doi.org/10.1016/j.apenergy.2012.01.001, En-
compared with the base case. It is worth noting that using only NGT ergy Solutions for a Sustainable World - Proceedings of the Third International
which acts as zero bus DER (case 1.2) results in minimum total cost, Conference on Applied Energy, May 16-18, 2011 - Perugia, Italy, URL https:
//www.sciencedirect.com/science/article/pii/S0306261912000050.
using only renewables with biomass as zero bus DER (case 2.2) results
[7] Bianchi M, De Pascale A, Spina PR. Guidelines for residential micro-CHP systems
in minimum emission, and the combination of NGT and renewables design. Appl Energy 2012;97:673–85. http://dx.doi.org/10.1016/j.apenergy.
results in minimum energy loss. However, operating the microgrid 2011.11.023, Energy Solutions for a Sustainable World - Proceedings of the Third
using the combination of NGT and renewables as in cases 3.1, 3.2 and International Conference on Applied Energy, May 16-18, 2011 - Perugia, Italy,
3.3 results in a better performance of the same where all the objectives URL https://www.sciencedirect.com/science/article/pii/S0306261911007203.
[8] Mugnini A, Ferracuti F, Lorenzetti M, Comodi G, Arteconi A. Advanced control
are reduced to a satisfactory extent. This study can be taken forward
techniques for CHP-DH systems: A critical comparison of model predictive
by incorporating the thermal storage system to store the excess heat control and reinforcement learning. Energy Convers Manage: X 2022;15:100264.
produced in the microgrid and correspondingly the optimal sizing of http://dx.doi.org/10.1016/j.ecmx.2022.100264, URL https://www.sciencedirect.
the storage system will be determined. com/science/article/pii/S2590174522000873.
[9] Werner S. International review of district heating and cooling. Energy
2017;137:617–31. http://dx.doi.org/10.1016/j.energy.2017.04.045, URL https:
CRediT authorship contribution statement
//www.sciencedirect.com/science/article/pii/S036054421730614X.
[10] Buoro D, Pinamonti P, Reini M. Optimization of a distributed cogeneration
Hemanth Chaduvula: Conceptualization, Data curation, Methodol- system with solar district heating. Appl Energy 2014;124:298–308. http://dx.
ogy, Visualization, Writing – original draft. Debapriya Das: Investiga- doi.org/10.1016/j.apenergy.2014.02.062, URL https://www.sciencedirect.com/
tion, Supervision, Validation, Writing – review & editing. science/article/pii/S0306261914002074.
[11] Wanjiru EM, Sichilalu SM, Xia X. Optimal operation of integrated heat pump-
instant water heaters with renewable energy. Energy Procedia 2017;105:2151–6.
Declaration of competing interest
http://dx.doi.org/10.1016/j.egypro.2017.03.607, 8th International Conference
on Applied Energy, ICAE2016, 8-11 October 2016, Beijing, China, URL https:
The authors declare that they have no known competing finan- //www.sciencedirect.com/science/article/pii/S1876610217306604.
cial interests or personal relationships that could have appeared to [12] Tahir MF, Haoyong C, Mehmood K, Ali N, Bhutto JA. Integrated energy system
influence the work reported in this paper. modeling of China for 2020 by incorporating demand response, heat pump
and thermal storage. IEEE Access 2019;7:40095–108. http://dx.doi.org/10.1109/
ACCESS.2019.2905684.
Data availability
[13] Widl E, Jacobs T, Schwabeneder D, Nicolas S, Basciotti D, Henein S, et al. Study-
ing the potential of multi-carrier energy distribution grids: A holistic approach.
No data was used for the research described in the article. Energy 2018;153:519–29. http://dx.doi.org/10.1016/j.energy.2018.04.047, URL
https://www.sciencedirect.com/science/article/pii/S0360544218306510.
Appendix A. Supplementary data [14] Li G, Zheng X. Thermal energy storage system integration forms for a sus-
tainable future. Renew Sustain Energy Rev 2016;62:736–57. http://dx.doi.
org/10.1016/j.rser.2016.04.076, URL https://www.sciencedirect.com/science/
Supplementary material related to this article can be found online
article/pii/S1364032116301095.
at https://doi.org/10.1016/j.energy.2023.128909. [15] Carpaneto E, Lazzeroni P, Repetto M. Optimal integration of solar energy in
a district heating network. Renew Energy 2015;75:714–21. http://dx.doi.org/10.
References 1016/j.renene.2014.10.055, URL https://www.sciencedirect.com/science/article/
pii/S096014811400682X.
[1] Sadeghian H, Ardehali M. A novel approach for optimal economic dispatch [16] Lund H, Werner S, Wiltshire R, Svendsen S, Thorsen JE, Hvelplund F, et al.
scheduling of integrated combined heat and power systems for maximum 4Th generation district heating (4GDH): Integrating smart thermal grids into
economic profit and minimum environmental emissions based on benders de- future sustainable energy systems. Energy 2014;68:1–11. http://dx.doi.org/10.
composition. Energy 2016;102:10–23. http://dx.doi.org/10.1016/j.energy.2016. 1016/j.energy.2014.02.089, URL https://www.sciencedirect.com/science/article/
02.044. pii/S0360544214002369.

14
H. Chaduvula and D. Das Energy 282 (2023) 128909

[17] Rezaie B, Reddy BV, Rosen MA. Energy analysis of thermal energy storages with [35] Bornapour M, Hemmati R, Pourbehzadi M, Dastranj A, Niknam T. Probabilistic
grid configurations. Appl Energy 2014;117:54–61. http://dx.doi.org/10.1016/j. optimal coordinated planning of molten carbonate fuel cell-CHP and renewable
apenergy.2013.11.042, URL https://www.sciencedirect.com/science/article/pii/ energy sources in microgrids considering hydrogen storage with point estimate
S0306261913009483. method. Energy Convers Manage 2020;206:112495. http://dx.doi.org/10.1016/
[18] Wang H, Lahdelma R, Wang X, Jiao W, Zhu C, Zou P. Analysis of the location for j.enconman.2020.112495.
peak heating in CHP based combined district heating systems. Appl Therm Eng [36] Yan B, Luh PB, Warner G, Zhang P. Operation and design optimization of
2015;87:402–11. http://dx.doi.org/10.1016/j.applthermaleng.2015.05.017, URL microgrids with renewables. IEEE Trans Autom Sci Eng 2017;14(2):573–85.
https://www.sciencedirect.com/science/article/pii/S1359431115004597. http://dx.doi.org/10.1109/TASE.2016.2645761.
[19] Saloux E, Candanedo JA. Model-based predictive control to minimize pri- [37] Koltsaklis NE, Kopanos GM, Georgiadis MC. Design and operational planning of
mary energy use in a solar district heating system with seasonal thermal energy networks based on combined heat and power units. Ind Eng Chem Res
energy storage. Appl Energy 2021;291:116840. http://dx.doi.org/10.1016/j. 2014;53(44):16905–23. http://dx.doi.org/10.1021/ie404165c.
apenergy.2021.116840, URL https://www.sciencedirect.com/science/article/pii/ [38] Shams MH, Shahabi M, Kia M, Heidari A, Lotfi M, Shafie-khah M, et al. Optimal
S0306261921003366. operation of electrical and thermal resources in microgrids with energy hubs
[20] Khatod DK, Pant V, Sharma J. Evolutionary programming based opti- considering uncertainties. Energy 2019;187:115949. http://dx.doi.org/10.1016/
mal placement of renewable distributed generators. IEEE Trans Power Syst j.energy.2019.115949.
2013;28(2):683–95. http://dx.doi.org/10.1109/TPWRS.2012.2211044. [39] Zhang Y, Meng F, Wang R, Kazemtabrizi B, Shi J. Uncertainty-resistant
[21] Rezaee Jordehi A. Allocation of distributed generation units in electric power stochastic MPC approach for optimal operation of CHP microgrid. Energy
systems: A review. Renew Sustain Energy Rev 2016;56:893–905. http://dx.doi. 2019;179:1265–78. http://dx.doi.org/10.1016/j.energy.2019.04.151.
org/10.1016/j.rser.2015.11.086. [40] Bracco S, Delfino F, Rossi M, Robba M, Pagnini L. Optimal planning of the
[22] Pesaran H.A M, Huy PD, Ramachandaramurthy VK. A review of the optimal energy production mix in smart districts including renewable and cogeneration
allocation of distributed generation: Objectives, constraints, methods, and algo- power plants. In: 2016 IEEE international smart cities conference. 2016, p. 1–7.
rithms. Renew Sustain Energy Rev 2017;75:293–312. http://dx.doi.org/10.1016/ http://dx.doi.org/10.1109/ISC2.2016.7580795.
j.rser.2016.10.071. [41] Moradi MH, Eskandari M, Mahdi Hosseinian S. Operational strategy optimization
[23] Ishak R, Mohamed A, Abdalla AN, Che Wanik MZ. Optimal placement and sizing in an optimal sized smart microgrid. IEEE Trans Smart Grid 2015;6(3):1087–95.
of distributed generators based on a novel MPSI index. Int J Electr Power Energy http://dx.doi.org/10.1109/TSG.2014.2349795.
Syst 2014;60:389–98. http://dx.doi.org/10.1016/j.ijepes.2014.03.044. [42] Gabbar HA, Zidan A. Optimal scheduling of interconnected micro energy grids
[24] Vatani M, Alkaran DS, Sanjari MJ, Gharehpetian GB. Multiple distributed with multiple fuel options. Sustain Energy, Grids Netw 2016;7:80–9. http://dx.
generation units allocation in distribution network for loss reduction based on doi.org/10.1016/j.segan.2016.06.006.
a combination of analytical and genetic algorithm methods. IET Generation, [43] Maulik A, Das D. Optimal operation of droop-controlled islanded microgrids.
Transm Distribution 2016;10(1):66–72. http://dx.doi.org/10.1049/iet-gtd.2015. IEEE Trans Sustain Energy 2018;9(3):1337–48. http://dx.doi.org/10.1109/TSTE.
0041. 2017.2783356.
[25] Aman M, Jasmon G, Bakar A, Mokhlis H. A new approach for optimum [44] Roy NB, Das D. Optimal allocation of active and reactive power of dispatchable
simultaneous multi-DG distributed generation units placement and sizing based distributed generators in a droop controlled islanded microgrid considering
on maximization of system loadability using HPSO (hybrid particle swarm renewable generation and load demand uncertainties. Sustain Energy, Grids Netw
optimization) algorithm. Energy 2014;66:202–15. http://dx.doi.org/10.1016/j. 2021;27:100482. http://dx.doi.org/10.1016/j.segan.2021.100482.
energy.2013.12.037. [45] Maulik A, Das D. Determination of optimal reserve requirement for fuel cost
[26] Hadidian-Moghaddam MJ, Arabi-Nowdeh S, Bigdeli M, Azizian D. A multi- minimization of a microgrid under load and generation uncertainties. Arab J Sci
objective optimal sizing and siting of distributed generation using ant lion Eng 2019;44:2003–31. http://dx.doi.org/10.1007/s13369-018-3234-y.
optimization technique. Ain Shams Eng J 2018;9(4):2101–9. http://dx.doi.org/ [46] Das B, Mukherjee V, Das D. Optimum DG placement for known power injec-
10.1016/j.asej.2017.03.001. tion from utility/substation by a novel zero bus load flow approach. Energy
[27] Evangelopoulos VA, Georgilakis PS. Optimal distributed generation placement 2019;175:228–49. http://dx.doi.org/10.1016/j.energy.2019.03.034.
under uncertainties based on point estimate method embedded genetic algorithm. [47] Barik S, Das D. Impact of FFC distributed generations in a DNR in the presence
IET Generation, Transm Distribution 2014;8(3):389–400. http://dx.doi.org/10. of renewable and load uncertainties by mixed-discrete particle swarm-based
1049/iet-gtd.2013.0442. point estimation method. IET Renew Power Gener 2019;13(9):1431–45. http:
[28] Constante-Flores GE, Illindala MS. Data-driven probabilistic power flow analysis //dx.doi.org/10.1049/iet-rpg.2018.5834.
for a distribution system with renewable energy sources using Monte Carlo [48] Chaduvula H, Das D. Optimal energy dispatch based on zero bus load flow
simulation. IEEE Trans Ind Appl 2019;55(1):174–81. http://dx.doi.org/10.1109/ in microgrid having multiple sources using fuzzy-particle swarm optimization
TIA.2018.2867332. approach. In: 2021 IEEE Texas power and energy conference. 2021, p. 1–6.
[29] Mohammadi S, Soleymani S, Mozafari B. Scenario-based stochastic operation http://dx.doi.org/10.1109/TPEC51183.2021.9384931.
management of MicroGrid including wind, photovoltaic, micro-turbine, fuel cell [49] Baran M, Wu F. Network reconfiguration in distribution systems for loss
and energy storage devices. Int J Electr Power Energy Syst 2014;54:525–35. reduction and load balancing. IEEE Trans Power Deliv 1989;4(2):1401–7. http:
http://dx.doi.org/10.1016/j.ijepes.2013.08.004. //dx.doi.org/10.1109/61.25627.
[30] Bazmohammadi N, Tahsiri A, Anvari-Moghaddam A, Guerrero JM. A hierarchical [50] Grigg C, Wong P, Albrecht P, Allan R, Bhavaraju M, Billinton R, et al. The IEEE
energy management strategy for interconnected microgrids considering uncer- reliability test system-1996. a report prepared by the reliability test system task
tainty. Int J Electr Power Energy Syst 2019;109:597–608. http://dx.doi.org/10. force of the application of probability methods subcommittee. IEEE Trans Power
1016/j.ijepes.2019.02.033. Syst 1999;14(3):1010–20. http://dx.doi.org/10.1109/59.780914.
[31] Hung DQ, Mithulananthan N, Bansal R. Analytical strategies for renewable [51] Wu X, Wang X, Qu C. A hierarchical framework for generation scheduling of
distributed generation integration considering energy loss minimization. Appl microgrids. IEEE Trans Power Deliv 2014;29(6):2448–57. http://dx.doi.org/10.
Energy 2013;105:75–85. http://dx.doi.org/10.1016/j.apenergy.2012.12.023. 1109/TPWRD.2014.2360064.
[32] Kayalvizhi S, DM VK. Optimal planning of active distribution networks with [52] Kennedy J, Eberhart R. Particle swarm optimization. In: Proceedings of
hybrid distributed energy resources using grid-based multi-objective harmony ICNN’95-international conference on neural networks, Vol. 4. IEEE; 1995, p.
search algorithm. Appl Soft Comput 2018;67:387–98. http://dx.doi.org/10.1016/ 1942–8.
j.asoc.2018.03.009. [53] Ratnaweera A, Halgamuge S, Watson H. Self-organizing hierarchical particle
[33] Kayal P, Chanda C. Optimal mix of solar and wind distributed generations swarm optimizer with time-varying acceleration coefficients. IEEE Trans Evol
considering performance improvement of electrical distribution network. Renew Comput 2004;8(3):240–55. http://dx.doi.org/10.1109/TEVC.2004.826071.
Energy 2015;75:173–86. http://dx.doi.org/10.1016/j.renene.2014.10.003. [54] Rosenblueth E. Point estimates for probability moments. Proc Natl Acad Sci
[34] Abdelaziz MMA, Farag HE, El-Saadany EF. Optimum reconfiguration of droop- 1975;72(10):3812–4. http://dx.doi.org/10.1073/pnas.72.10.3812.
controlled islanded microgrids. IEEE Trans Power Syst 2016;31(3):2144–53. [55] Morales JM, Perez-Ruiz J. Point estimate schemes to solve the probabilistic power
http://dx.doi.org/10.1109/TPWRS.2015.2456154. flow. IEEE Trans Power Syst 2007;22(4):1594–601. http://dx.doi.org/10.1109/
TPWRS.2007.907515.

15

You might also like