INTERNATIONAL JOURNAL of RENEWABLE ENERGY RESEARCH

W. H. Tee et al., Vol.13, No.4, December, 2023

Sizing of Battery Energy Storage System: A Multi-

Objective Optimization Approach in DIgSILENT
PowerFactory

Wei Hown Tee* , Khaldon Ahmed Qaid* , Chin Kim Gan*‡ , Joe Siang Keek*,** , Junainah
Sardi*

* Faculty of Electrical Engineering, Universiti Teknikal Malaysia Melaka

** Texas Instruments Electronics Malaysia Sdn. Bhd.

([email protected], [email protected], [email protected], [email protected], [email protected])

‡Corresponding Author; Chin Kim Gan, Faculty of Electrical Engineering, Universiti Teknikal Malaysia Melaka, Jalan Hang
Tuah Jaya, 76100 Durian Tunggal, Melaka Tel: +60 62701310, [email protected]

Received: 19.05.2023 Accepted:27.06.2023

Abstract- In the paradigm of the increasing trend to prevent global warming, renewable energy sources applications integrated
with battery energy storage system (BESS) are gaining attention for reducing the usage of fossil fuels in electrical power
generation. In this regard, a multi-objective optimization script in DIgSILENT Programming Language (DPL) which links with
software modelling and scripting simulation is developed in this study. Formulation for multiple objectives for optimization of
BESS sizing with particle swarm optimization (MOPSO) and load flow simulation are applied in the DPL script. The considered
objective functions aim to improve the network performance by reducing power loss, voltage deviations and system costs. Pseudo
code of BESS optimal sizing with multi-objective algorithm is presented in this research. The BESS with optimal sizing was
discovered for improving the network performance in the tested reference network. The optimal BESS size obtained is 2.94 MW
with a system cost of MYR 2404.76. The total energy losses can be reduced by approximately 16% from the base case energy
losses with the optimal BESS size. The findings of the research reveal that the BESS sizing with MOPSO is applicable in DPL
operations alone to solve power system problems.
Keywords Particle swarm optimization; multi-objective optimization; battery energy storage system; optimal sizing;
DIgSILENT PowerFactory.

1. Introduction
Nomenclature
Countries around the globe are currently facing
Battery energy storage system BESS
challenges to reduce the electrical power generation
BloombergNEF BNEF
generated by the burning of fossil fuels which contributes to
DIgSILENT Programming Language DPL
global warming. As a result, the applications of renewable
Genetic algorithm GA
energy sources integrated with grid have received huge
Improve particle swarm optimization IPSO
attention by researchers nowadays. In the recent
Multi-objective optimization MOO BloombergNEF (BNEF) report, around half of the energy
Multi-objective particle swarm optimization MOPSO demand of the world is projected to be supplied by solar and
On-load tap charger OLTC wind energy by 2050 [1]. On top of that, renewable energy
Particle swarm optimization PSO
sources are expected to have a dominant power generation
Photovoltaic and Smart Grid PVSG
mix of almost 50% approaching 2050 [2][3]. However, the
Second order cone program SOCP unpredicted characteristics of the renewable energy sources
Universiti Teknikal Malaysia Melaka UTeM will cause instability of the power outputs [4][5]. Hence,
INTERNATIONAL JOURNAL of RENEWABLE ENERGY RESEARCH
W. H. Tee et al., Vol.13, No.4, December, 2023

energy storage technologies have emerged as an effective 2. Methodology

solution for integrating the grid system with renewable
energy sources because of their characteristics. Energy DIgSILENT PowerFactory is used to apply the
storage is widely applied to control the intermittency of methodology developed in this research. Consequently, a
renewable energy [6][7][8]. It has been demonstrated that series of operations to run the algorithm developed will be
energy storage improves the stability and reliability of carried out in the DPL script. This study is a continuous work
renewable energy generation of the network [9][10]. Besides, from [30], where the algorithm is developed to optimize two
the ideas have been also applied to both virtual and actual grid objective functions in the DPL script. The DPL script will be
system with the benefit of self-consumption of renewable used to execute both network model in the simulation
energy generation with storage systems [11][12]. interface and script operation in the software at the same time.
Furthermore, studies showed that energy storage is playing a As a result, the DPL script will interface with the database of
vital role to provide power quality improvement [13][14][15], the network model object to access, record and perform
frequency and voltage regulation [16][17], energy arbitrage operation based on the operation functions written in the
as well as ancillary services [18][19]. script.
At present, researchers show great interest in developing
1.1. Optimization Objectives and Constraints
battery energy storage system (BESS) optimization to serve
as an integrating option in power systems because of its
attributions with rapid reaction. Authors in [20] proposed a A multi-objective problem for the BESS sizing of the
second order cone program (SOCP) optimization method network is formulated in this study. The first objective
based on power flow equations to size the storage. In a study function aims to reduce the power loss and voltage deviations
[21], the application of DIgSILENT and MATLAB are used of the network with the integration of BESS which can be
in conjunction with improved particle swarm optimization expressed as
(IPSO) to solve the voltage fluctuation problem with optimal 𝑂𝑏𝑗𝑒𝑐𝑡𝑖𝑣𝑒 𝐹𝑢𝑛𝑐𝑡𝑖𝑜𝑛 1 = optimum (𝐵𝐸𝑆𝑆𝑠𝑖𝑧𝑒 ) (1)
BESS active power. However, the simulation requires
∑𝐻 𝑁
ℎ=1 ∑𝑖=1 𝑃𝑙𝑜𝑠𝑠𝐵𝐸𝑆𝑆
dynamic data exchange file between the software. An optimal optimum(𝐵𝐸𝑆𝑆𝑠𝑖𝑧𝑒 ) = [ ]+
∑𝐻 𝑁
ℎ=1 ∑𝑖=1 𝑃𝑙𝑜𝑠𝑠𝑏𝑎𝑠𝑒
BESS control is proposed to reduce excessive on-load tap
2
charger (OLTC) operation caused by renewable energy ∑𝐻 𝑁 𝐵𝐸𝑆𝑆 (𝑖))
ℎ=1 ∑𝑖=1(𝑉𝑝𝑢
[ 2 ] (2)
sources with PSO approach using MATLAB and ∑𝐻 𝑁 𝑏𝑎𝑠𝑒
ℎ=1 ∑𝑖=1(𝑉𝑝𝑢 (𝑖))
DIgSILENT software [22]. However, a switch CSV file is
needed for data communication between the software. The where 𝑃𝑙𝑜𝑠𝑠𝐵𝐸𝑆𝑆 is the power loss with BESS in
authors improved the network performance by combining the MW, 𝑃𝑙𝑜𝑠𝑠𝑏𝑎𝑠𝑒 is the initial power loss without BESS in
𝐵𝐸𝑆𝑆
linkage between MATLAB and DIgSILENT with genetic MW, 𝑉𝑝𝑢 is the voltage magnitude with BESS in per unit
𝑏𝑎𝑠𝑒
algorithm (GA) linear programming method to size BESS at bus i, 𝑉𝑝𝑢 is the base voltage magnitude in per unit, N
[23]. However, these studies did not apply both optimization denotes the total buses number, h denotes the study hours, and
algorithm and loadflow simulation in a single software, for H denotes the total number of study hours. This objective
example in DIgSILENT alone. function is a linear combination of two performance indices.
The denominators in the expressions (2) indicate the sum of
The studies showed that bio-inspired optimization
real power loss and the sum of the squares of the voltage
approaches are widely applied by the researchers due to their
deviations without the integration of BESS while the
flexibility, high accuracy rate and less computational time
numerators are presenting the variables with the integration
[24][25]. PSO is one of the established approaches due to its
of BESS [31].
simplicity with coding implementation, short computational
time as well as stable convergence properties [26]. It offers a Next, the algorithm aims to reduce the system costs. The
certain appealing feature of good memory where the particles equation can be expressed as
retain the knowledge of good solutions as compared to GA
𝑂𝑏𝑗𝑒𝑐𝑡𝑖𝑣𝑒 𝐹𝑢𝑛𝑐𝑡𝑖𝑜𝑛 2 = min (𝐶𝑜𝑠𝑡𝑠𝑦𝑠𝑡𝑒𝑚 ) (3)
approach [27][28]. PSO is also proven to converge faster and
able to escape from being trapped in local optima as min(𝐶𝑜𝑠𝑡𝑠𝑦𝑠𝑡𝑒𝑚 ) = 𝐶𝑣 + 𝐶𝑙 + 𝐶𝑝 (4)
compared to Bat algorithm and Tabu search. Moreover, PSO
is effective to determine the pareto front in optimizing multi- 𝐶𝑣 = [∑𝐻 𝑁
ℎ=1 ∑𝑖=1|𝑉𝑖 − 𝑉𝑟𝑒𝑓 |] × 𝜑𝑣 (5)
objective problems [29].
𝐶𝑙 = ∑𝐻 𝑁
ℎ=1 ∑𝑖=1|𝑃𝑙𝑜𝑠𝑠𝑖 | × 𝜑𝑙 (6)
In light of the above, the information of applying multi-
objective optimization (MOO) in DIgSILENT DPL script 𝐶𝑝 = Peakload × 𝐻 × 𝜑𝑝 (7)
alone is limited. To the best of authors’ knowledge, the where 𝐶𝑣 is voltage regulation charge, 𝐶𝑙 is power loss
structure of MOO scripting in DPL has not been reported so charge, 𝐶𝑝 is peak demand charge, 𝑉𝑖 is the magnitude of the
far. Hence, the major contributions of this work are voltage per unit of the bus, 𝑉𝑟𝑒𝑓 is set as 1 p.u. base voltage
1. To design MOO in programming script to apply both reference, 𝜑𝑣 is the cost rate of voltage regulation at 0.63
optimization algorithm and loadflow simulation in MYR/ p.u., 𝜑𝑙 is the cost rate of power loss at 1.26 MYR/
DIgSILENT without linkage between software. kWh and 𝜑𝑝 is the cost rate of peak demand at 2.43 MYR/
2. To improve the performance of the distribution network kW [32][33]. The objective functions are subjected to the
with optimal BESS sizing. following constraints

1605
INTERNATIONAL JOURNAL of RENEWABLE ENERGY RESEARCH
W. H. Tee et al., Vol.13, No.4, December, 2023

𝑃𝑔𝑟𝑖𝑑 = ∑𝑁 𝑁
𝑖=1 𝐿𝑜𝑎𝑑 + ∑𝑖=1 𝐿𝑜𝑠𝑠 𝑖 = 1,2, … , 𝑁 (8) 2. Based on the crowding degree in Eq. (15), a leader in the
non-dominated solutions is selected as global optimal.
0 ≤ 𝐿𝑜𝑠𝑠𝐵𝐸𝑆𝑆 ≤ 𝐿𝑜𝑠𝑠𝑏𝑎𝑠𝑒 (9)
|𝑓1(𝑥𝑖+1)−𝑓1(𝑥𝑖−1)| |𝑓2(𝑥𝑖+1)−𝑓2(𝑥𝑖−1)|
0 ≤ 𝑃𝐵𝐸𝑆𝑆 ≤ 𝑃𝑃𝑉 (10) 𝐶𝐷𝑖 = +
𝑓1𝑚𝑎𝑥 −𝑓1𝑚𝑖𝑛 𝑓2𝑚𝑎𝑥−𝑓2𝑚𝑖𝑛
(15)
0.95 ≤ 𝑉𝑖 ≤ 1.05 (11)
The maximum iteration (Ni) in this study is 30, with a
where 𝑃𝑔𝑟𝑖𝑑 is the total grid power in MW of the system, population (Np) of 100. The repository size (Nr) is 50. The
𝐿𝑜𝑠𝑠𝐵𝐸𝑆𝑆 is the power loss with BESS in MW, 𝐿𝑜𝑠𝑠𝑏𝑎𝑠𝑒 is
the base power loss of the system in MW, 𝑃𝐵𝐸𝑆𝑆 is the BESS learning factors of the algorithm, c1 and c2 are set as 0.5 and
power in MW, 𝑃𝑝𝑣 is the peak PV power in MW, 𝑉𝑖 is the bus 0.6. The operation of MOPSO is as follows:
voltage. 1. Define the objective function constraints and initialize
the optimization parameters.
1.2. Multi-Objective Algorithm
2. Define swarm iteration i = 1 with random particle
Particle swarm optimization (PSO) is a heuristic method velocity and position.
that was inspired by the birds’ behavior. In the algorithm, the 3. Create grid to store non-dominated solutions.
particle has its own velocity as well as position property
which is able to explore for a possible solution in search 4. Declare swarm particle j = 1.
space. The velocity of the particle represents the fast or slow 5. Objective function is executed for the respective particle
movement of the particle, whereas the position of the particle and iteration.
represents the particle’s direction. Each particle’s velocity
can be revised based on its personal and global optimal 6. Select leader according to crowding degree. Explore and
position. These two properties can be calculated as determine Pbest and Gbest of jth particle in ith iteration. If
the objective function of jth dominates in jth population,
𝑣𝑗𝑖+1 = 𝜔𝑣𝑗𝑖 + 𝑐1 𝑟1 (𝑃𝑏𝑒𝑠𝑡
𝑖
− 𝑥𝑗𝑖 ) + 𝑐2 𝑟2 (𝐺𝑏𝑒𝑠𝑡
𝑖
− 𝑥𝑗𝑖 ) (12) the program stores Pbest = f1(xj). f2(xj).
𝑥𝑗𝑖+1 = 𝑣𝑗𝑖+1 + 𝑥𝑗𝑖 (13) 7. Add Pbest to repository. Determine domination in jth
repository.
where 𝜔 denotes to the inertia weight of the algorithm, 𝑟1 and
𝑟2 are the values randomly ranged from 0 to 1, 𝑐1 and 𝑐2 are 8. Update grid.
𝑖 𝑖
cognitive and social parameters, 𝑃𝑏𝑒𝑠𝑡 and 𝐺𝑏𝑒𝑠𝑡 are personal 9. Update particle jth by 1. Next, the condition is validated:
and global best solution respectively. If particle jth +1 ≤ Np, go back to Step 5.
The inertia weight of the algorithm influences the 10. If Pbest dominates in jth population, adjust Gbest of ith = Pbest
velocity and direction of the particle, which brings a big effect at ith.
on the particle’s convergence. According to [34], PSO has a
higher convergence speed to achieve local optimum with 11. Plot pareto solutions.
smaller inertia weight while higher possibility to have global 12. Update iteration ith is by 1. Next, the condition is
search with greater inertia weight. However, the convergence validated: If iteration ith +1 ≤ Ni, go back to Step 4 and
speed is reduced and the number of iterations increases with the new particle velocity as well as position are updated
greater inertia weight. In this study, the particles initially aim for ith +1. Update new inertia weight.
to search globally. After obtaining the swarm particle, the
convergence speed increases to obtain the optimum value. 13. If iteration ith +1 ≥ Ni, the algorithm quits the operation
Hence, the 𝜔 of this study is to decrease between 0.9 to 0.4 and BESS size is acquired.
and be calculated as The flowchart of MOPSO BESS sizing operation
𝑖 between DPL script and DIgSILENT environment is
𝜔 = 𝜔ℎ𝑖𝑔ℎ − (𝜔ℎ𝑖𝑔ℎ − 𝜔𝑙𝑜𝑤 ) × 𝑖 (14)
𝑚𝑎𝑥 presented in Figure 1. The detailed process of the MOPSO is
MOPSO is popularly implemented in optimizing multi- presented in the Pseudo code, as illustrated in Figure 2.
objective problems because of its simple structure, parameter
settings as well as fast convergence characteristics [35][36].
MOPSO is relatively complex as compared to PSO in terms
of selecting personal and global optimal solutions. It obeys
the following principles
1. When the solutions are non-dominated, the best
individual position will be randomly selected from one
of the particle solutions.

1606
INTERNATIONAL JOURNAL of RENEWABLE ENERGY RESEARCH
W. H. Tee et al., Vol.13, No.4, December, 2023

Fig. 1. The flowchart of MOPSO in DIgSILENT.

1607
INTERNATIONAL JOURNAL of RENEWABLE ENERGY RESEARCH
W. H. Tee et al., Vol.13, No.4, December, 2023

Fig. 2. The Pseudo code of MOPSO in DIgSILENT.

3. Case Study 1.2

The normalized PV and load profiles of the study are 1

Power (p.u.)

presented in Figure 3. The modified urban reference network 0.8

with 33 & 11 kV feeders of Malaysia as shown in Figure 4
0.6
[37] is applied as the test system. The PV and BESS were
installed at the end of each feeder. The parameters of the 0.4
reference network are presented in Table 1. The PV data are 0.2
collected from the PV system installed at the Research
Laboratory of Photovoltaic System and Smart Grid, 0
Universiti Teknikal Malaysia Melaka (UTeM) [7]. 1 3 5 7 9 11 13 15 17 19 21 23
Time (hour)
Normalized Load Normalized PV

Fig. 3. PV power and load demand profile used in this study.

1608
INTERNATIONAL JOURNAL of RENEWABLE ENERGY RESEARCH
W. H. Tee et al., Vol.13, No.4, December, 2023

Fig. 4. Urban reference network with 33 & 11 kV feeders.

Table 1. Parameters of urban reference network with 33 & 11 feeders.

Parameters Average Value
F1 F2 F3 F4
PV capacity (MW) 4.00 3.50 3.00 2.50
Maximum load demand (MW) 3.59 3.01 2.77 2.26
Transformer
33/11 kV capacity (MVA) 30.00
132/33 kV capacity (MVA) 45.00
Number of 11/0.4 kV transformers 7.00 6.00 5.00 3.00
11/0.4 kV loading (MVA) 0.58 0.50 0.55 0.76
11/0.4 kV capacity (MVA) 1.00 1.00 1.00 1.00
11/0.4 kV transformer loading (MW) 0.52 0.45 0.50 0.76
Feeder
33 kV length (km) 4.43
11 kV length (km) 4.25 2.88 1.70 0.94
Distances between 11/0.4 kV transformers (km) 0.61 0.48 0.34 0.31

objective function 2 (OF2) cost of MYR 2404.76, with

4. Results and Discussions
optimal BESS solution of 2.94 MW. Besides, boundary
solutions can be observed also from the pareto fronts. Firstly,
MOPSO is implemented to take into account the BESS the maximum OF1 can be obtained at 3.12, with the increase
sizing with the beneficial objectives of reducing power loss, of 2.65% from the best compromise solution, followed by
voltage deviation and system costs. The simulation is MYR 2404.11. Next, the minimum OF1 is with MYR
conducted in DIgSILENT. The fitness function values and 2406.07, with a reduction percentage of 2.04% from OF1
optimal BESS size for F1 feeder are shown in Table 2 with value of the best compromise solution.
computation time of 4312.02s. Figure 5 presents the pareto
optimal front with a variety of leader selections for various The power losses for the solutions in comparison to the
objective function 1 (OF1) values correspond to the system base case are presented in Figure 6. The results reveal that the
costs obtained in the study. The ideal compromise position BESS sizing through MOPSO is able to reduce the power loss
obtained by the MOPSO is with OF1 value of 3.04 and of the system. For the base case, the losses are the highest

1609
INTERNATIONAL JOURNAL of RENEWABLE ENERGY RESEARCH
W. H. Tee et al., Vol.13, No.4, December, 2023

when there is no PV power (during midnight, and from late greatest PV capacity. Besides, F1 feeder has the longest 11
evening to 12.00 a.m.) with maximum value of 0.51 MW kV feeder that contributes to the highest losses among F1 to
losses, as well as when PV power is available (from morning F4 feeders. Hence, the implementation of optimal BESS size
until evening) with the maximum value of 0.45 MW losses. at F1 feeder will have the highest impact on the network
This can be explained that the demands are high when the performance. Initially, there will be losses in the feeder due
consumers are back from work in the late evening until early to the power transmitted from the main transformer to the end
morning. Low power usage during the afternoon when PV of the feeder. When the PV is connected, the voltage
power is highly available will also lead to power losses magnitude will increase at the end of the feeder due to the
because of the excess power generated by PV that flows back reverse power flow caused by the excess PV power. This
to the grid. The optimal solution with 2.94 MW BESS size issue can be solved by the installation of BESS at the location
can reduce the losses of the base case by storing excessive PV where BESS can reduce the reverse power flow by charging
power during afternoon hours with low demand, and the excess power generated by PV into the feeder and hence
discharging power to support the load when demand is high reduce the power losses across the feeder.
during nighttime. The base case has the highest energy losses
Table 2. Solutions of MOPSO obtained for F1.
per day, which is 7.30 MWh, followed by the maximum OF1
Item Minimum Best Maximum
(3.29 MW BESS) of 6.22 MWh, optimal solution (2.94 MW
OF1 Compromise OF1
BESS) of 6.13 MWh and lastly minimum OF1 (2.94 MW
BESS) with 6.04 MWh. PV size 4.00 4.00 4.00
(MW)
The voltage profile for the base case and optimal solution BESS size 2.52 2.94 3.29
with respect to BESS power is shown in Figure 7. The BESS (MW)
improves the voltage profile during high demand hours by OF1 2.97 3.04 3.12
discharging power into the network. Moreover, the voltage magnitude
deviations are reduced with the charging of excess PV power System cost 2406.07 2404.76 2404.11
into the BESS during low load demand hours in the afternoon (OF2)
period. Table 3 shows the summary of optimal BESS sizing (MYR)
of MOPSO with F1, F2, F3 and F4. It can be observed that
the deployment of BESS at F1 has the highest energy losses
reduction as compared to the remaining feeders. This is
because F1 feeder has the largest load demand with the

Fig.5. Pareto optimal solutions produced by MOPSO.

1610
INTERNATIONAL JOURNAL of RENEWABLE ENERGY RESEARCH
W. H. Tee et al., Vol.13, No.4, December, 2023

Table 3. Summary of optimal BESS sizing in the feeders. 5. Conclusion

F1 F2 F3 F4
PV size 4.00 3.50 3.00 2.50 This study presented optimal BESS sizing with MOPSO
(MW) in the DPL script of DIgSILENT PowerFactory. The
BESS size 2.94 2.67 2.05 1.93 developed MOPSO reduced the total power losses, improved
(MW) the voltage profile as well as the costs associated with the
OF1 3.04 3.01 3.07 2.99 distribution network. In the MOPSO, various candidate
magnitude solutions of OF1 and OF2 are acquired with optimal BESS
System cost 2404.76 2343.3 2286.24 2250.33 size through the pareto optimal front. The best solution
(OF2) identified is OF1 value of 3.04 and OF2 of MYR 2404.76
(MYR) with optimal BESS size of 2.94 MW. Besides, the optimal
Energy 6.13 5.98 5.47 5.21 BESS size can reduce the total energy losses by 6.13 MWh,
losses which is approximately 16% reduction from 7.30 MWh. The
reduction findings reveal that the network has enhanced voltage profile
(MWh) with BESS deployment, resulting in more system costs that
can be saved. However, it is noted that this work does not
By taking into account the network performance, it is consider the smoothing of solar PV output with optimal BESS
also demonstrated that the overall improved voltage profile in sizing. Hence, the future potential of this study consists of the
one day is sustained within the operating voltage, with development of objective functions in MOPSO that focus on
reduced power losses by implementing the optimal sizing of the limitation of renewable energy intermittency, the impact
BESS through MOPSO. This enhances the power quality and of BESS applications and techno-economic areas.
improves the reliability of the network. The MOPSO shown
in this study can offer optimal BESS sizing integrated with Acknowledgements
PV in the distribution network as well as benefit in terms of
cost. Moreover, it can also be effectively deployed to other The authors gratefully acknowledge the support given by
real networks connected to grid systems with more updated Universiti Teknikal Malaysia Melaka (UTeM) for this work.
problem formulation or objectives to optimize the BESS size
based on the idea and Pseudo code provided in this study. References
0.55 [1] “Bloomberg New Energy Outlook 2019: The future
0.5 of the energy sector - Power Technology.” [Online].
Power Losses (MW)

0.45
Available:https://www.power.technology.com/news/
bloomberg-new-energyoutlook-2019-2/ (accessed
0.4 Jul. 22, 2022).
0.35
[2] A. A. Kebede, T. Kalogiannis, J. Van Mierlo, and M.
0.3 Berecibar, “A comprehensive review of stationary
0.25 energy storage devices for large scale renewable
energy sources grid integration,” Renew. Sustain.
0.2
Energy Rev., vol. 159, p. 112213, 2022.
0.15
1 2 3 4 5 6 7 8 9 101112131415161718192021222324 [3] V. Henze, “Energy Storage Is a $620 Billion
Time (hour) Investment Opportunity to 2040 | BloombergNEF,”
Base case Optimal solution BloombergNEF, 6 Nov. 2018,
Minimum OF1 Maximum OF1 about.bnef.com/blog/energy-storage-620-billion-
Fig. 6. Power losses comparison of the solutions. investment-opportunity-2040., 2018.
https://about.bnef.com/blog/energy-storage-620-
1.03 3
billion-investment-opportunity-2040/ (accessed Nov.
26, 2021).
Active Power (MW)

1.02 2
Voltage (p.u.)

[4] W. Wang, B. Yuan, Q. Sun, and R. Wennersten,

1.01 1
“Application of energy storage in integrated energy
1 0 systems—A solution to fluctuation and uncertainty of
0.99 -1 renewable energy,” J. Energy Storage, vol. 52, p.
104812, 2022.
0.98 -2
[5] H. Shahinzadeh, J. Moradi, W. Yaïci, M. Longo, and
0.97 -3
1 3 5 7 9 11 13 15 17 19 21 23 25 Z. Azani, “Impacts of Energy Storage Facilities on
Resilient Operation of Multi-Carrier Energy Hub
Time (hour)
Systems,” in 2022 10th International Conference on
BESS Base case Optimal solution
Smart Grid (icSmartGrid), pp. 339–344, 2022.
Fig. 7. Voltage profiles that correspond to BESS power.
[6] J. Dey, N. Mohammad, and M. T. Islam, “Analysis of

1611
INTERNATIONAL JOURNAL of RENEWABLE ENERGY RESEARCH
W. H. Tee et al., Vol.13, No.4, December, 2023

a Microgrid having Solar System with Maximum [17] N. McIlwaine, A. M. Foley, D. J. Morrow, D. Al Kez,
Power Point Tracking and Battery Energy System,” C. Zhang, X. Lu, and R. J. Best, “A state-of-the-art
in 2022 10th International Conference on Smart Grid techno-economic review of distributed and
(icSmartGrid), pp. 179–184, 2022. embedded energy storage for energy systems,”
Energy, vol. 229, p. 120461, 2021.
[7] W. H. Tee, C. Kim Gan, J. B. Sardi, K. Azmi Baharin,
and K. K. Kong, “Probabilistic sizing of battery [18] E. Pusceddu, B. Zakeri, and G. C. Gissey, “Synergies
energy storage system for solar photovoltaic output between energy arbitrage and fast frequency response
smoothing,” PECon 2020 IEEE Int. Conf. Power for battery energy storage systems,” Appl. Energy,
Energy, pp. 350-355, 2020. vol. 283, p. 116274, 2021.
[8] M. A. Pinazo and J. L. R. Martinez, “Intermittent [19] K.Oureilidis et al., “Ancillary services market design
power control in wind turbines integrated into a in distribution networks: Review and identification of
hybrid energy storage system based on a new state- barriers,” Energies, vol. 13, no. 4, p. 917, 2020.
of-charge management algorithm,” J. Energy
[20] E. Grover-Silva, R. Girard, and G. Kariniotakis,
Storage, vol. 54, p. 105223, 2022.
“Optimal sizing and placement of distribution grid
[9] L. A. Wong, V. K. Ramachandaramurthy, P. Taylor, connected battery systems through an SOCP optimal
J. B. Ekanayake, S. L. Walker, and S. Padmanaban, power flow algorithm,” Appl. Energy, vol. 219, pp.
“Review on the optimal placement, sizing and control 385–393, 2018.
of an energy storage system in the distribution
[21] K. Tantrapon, P. Jirapong, and P. Thararak,
network,” J. Energy Storage, vol. 21, pp. 489–504,
“Mitigating microgrid voltage fluctuation using
2019.
battery energy storage system with improved particle
[10] D. Al kez, A. M. Foley, N. Mcllwaine, D. J. Morrow, swarm optimization,” Energy Reports, vol. 6, pp.
B. P. Hayes, M. A. Zehir, L. Mehigan, B. Papari, C. 724–730, 2020.
S. Edrington, and M. Baran, “A critical evaluation of
[22] T. Akapan, P. Thararak, and P. Jirapong, “Optimal
grid stability and codes, energy storage and smart
real-time operation of battery energy storage systems
loads in power systems with wind generation,”
for reducing on-load tap changer operation of
Energy, vol. 205, p. 117671, 2020.
substation transformers,” in AIP Conference
[11] M. Shamshiri, C. K. Gan, J. Sardi, M. T. Au, and W. Proceedings, vol. 2681, no. 1, p. 20033, 2022.
H. Tee, “Design of Battery Storage System for
[23] M. R. Jannesar, A. Sedighi, M. Savaghebi, and J. M.
Malaysia Low Voltage Distribution Network with the
Guerrero, “Optimal placement, sizing, and daily
Presence of Residential Solar Photovoltaic System,”
charge/discharge of battery energy storage in low
Energies , vol. 13, p. 4887, 2020.
voltage distribution network with high photovoltaic
[12] J. Sardi, N. Mithulananthan, M. M. Islam, and C. K. penetration,” Appl. Energy, vol. 226, pp. 957–966,
Gan, “Framework of virtual microgrids formation 2018.
using community energy storage in residential
[24] M. A. Hannan, S. B. Wali, P. J. Ker, M. S. Abd
networks with rooftop photovoltaic units,” J. Energy
Rahman, M. Mansor, V. K. Ramachandaramurthy, K.
Storage, vol. 35, p. 102250, 2021.
M. Muttaqi, T. M. I. Mahlia, and Z. Y. Dong,
[13] Y. Iwasaki, “Dynamic Switching Method with “Battery energy-storage system: A review of
Energy Storage Devices in Wind Power Generation,” technologies, optimization objectives, constraints,
in 2022 10th International Conference on Smart Grid approaches, and outstanding issues,” J. Energy
(icSmartGrid), pp. 165–168, 2022. Storage, vol. 42, p. 103023, 2021.
[14] M. Sarr, M. Thiam, B. Niang, O. Ba, and L. Thiaw, [25] R. Raghutu, M. Sankaraiah, R. S. S. Nuvvula, and M.
“BESS deployment and Virtual Power Plant: Venkatesh, “Dispatchable and Non-dispatchable
Technical and financial analysis of the Senelec Distributed Generation Reactive Power Coordination
network to assess the relevance,” in 2022 10th with Reactive Power-controlled Devices using Grey
International Conference on Smart Grid Wolf Optimizer,” in 11th International Conference
(icSmartGrid), pp. 118–123, 2022. on Renewable Energy Research and Application
(ICRERA), 2022, pp. 33–41, 2022.
[15] S. Hajiaghasi, A. Salemnia, and M. Hamzeh, “Hybrid
energy storage system for microgrids applications: A [26] M. H. K. Roni, M. S. Rana, H. R. Pota, M. M. Hasan,
review,” J. Energy Storage, vol. 21, pp. 543–570, and M. S. Hussain, “Recent trends in bio-inspired
2019. meta-heuristic optimization techniques in control
applications for electrical systems: A review,” Int. J.
[16] U. Akram, M. Nadarajah, R. Shah, and F. Milano, “A
Dyn. Control, pp. 1–13, 2022.
review on rapid responsive energy storage
technologies for frequency regulation in modern [27] T. A. Jumani, M. W. Mustafa, A. S. Alghamdi, M. M.
power systems,” Renew. Sustain. Energy Rev., vol. Rasid, A. Alamgir, and A. B. Awan, “Swarm
120, p. 109626, 2020. intelligence-based optimization techniques for
dynamic response and power quality enhancement of

1612
INTERNATIONAL JOURNAL of RENEWABLE ENERGY RESEARCH
W. H. Tee et al., Vol.13, No.4, December, 2023

AC microgrids: A comprehensive review,” IEEE

Access, vol. 8, pp. 75986–76001, 2020.
[28] T. O. Ajewole, O. Oladepo, K. A. Hassan, A. A.
Olawuyi, and O. Onarinde, “Comparative Study of
the Performances of Three Metaheuristic Algorithms
in Sizing Hybrid-Source Power System,” Turkish
Journal of Electrical Power and Energy Systems, vol.
2, no. 2, pp. 134-146, 2022.
[29] Q. Gu, Q. Wang, X. Li, and X. Li, “A surrogate-
assisted multi-objective particle swarm optimization
of expensive constrained combinatorial optimization
problems,” Knowledge-Based Syst., vol. 223, p.
107049, 2021.
[30] W. H. Tee, K. A. Qaid, C. K. Gan, and P. H. Tan,
“Battery Energy Storage System Sizing Using PSO
Algorithm in DIgSILENT PowerFactory,” Int. J.
Renew. Energy Res., vol. 12, no. 4, pp. 2143–2151,
2022.
[31] B. Mukhopadhyay and D. Das, “Multi-objective
dynamic and static reconfiguration with optimized
allocation of PV-DG and battery energy storage
system,” Renew. Sustain. energy Rev., vol. 124, p.
109777, 2020.
[32] P. Boonluk, S. Khunkitti, P. Fuangfoo, and A.
Siritaratiwat, “Optimal siting and sizing of battery
energy storage: Case study seventh feeder at Nakhon
Phanom substation in Thailand,” Energies, vol. 14,
no. 5, p. 1458, 2021.
[33] N. Jayasekara, M. A. S. Masoum, and P. J. Wolfs,
“Optimal operation of distributed energy storage
systems to improve distribution network load and
generation hosting capability,” IEEE Trans. Sustain.
Energy, vol. 7, no. 1, pp. 250–261, 2015.
[34] B. Sun, “A multi-objective optimization model for
fast electric vehicle charging stations with wind, PV
power and energy storage,” J. Clean. Prod., vol. 288,
p. 125564, 2021.
[35] M. Kharrich, O. H. Mohammed, N. Alshammari, and
M. Akherraz, “Multi-objective optimization and the
effect of the economic factors on the design of the
microgrid hybrid system,” Sustain. Cities Soc., vol.
65, p. 102646, 2021.
[36] C. A. C. Coello, G. T. Pulido, and M. S. Lechuga,
“Handling multiple objectives with particle swarm
optimization,” IEEE Trans. Evol. Comput., vol. 8, no.
3, pp. 256–279, 2004.
[37] H. Salah Mohammed, C. Kim Gan, and M. R. Ab
Ghani, “Technical impacts of solar photovoltaic
systems integration into malaysian medium voltage
reference networks,” Int. J. Nonlinear Anal. Appl.,
vol. 11, pp. 265–276, 2020.

1613