Download as pdf or txt
Download as pdf or txt
You are on page 1of 4

2022 19th International Conference on Electrical Engineering/Electronics, Computer, Telecommunications and Information Technology (ECTI-CON) | 978-1-6654-8584-5/22/$31.

00 ©2022 IEEE | DOI: 10.1109/ECTI-CON54298.2022.9795484 2022 19th International Conference on Electrical Engineering/Electronics, Computer, Telecommunications and Information Technology (ECTI-CON)

Distribution Systems Reconfiguration Using PSO


based on Co-Simulation of PowerFactory and
Python
Naowarat Tephiruk1 , Budsarakum Ruenroeng2, Sirasit Fongjangvang3, Thunwa Boonlert4, Komsan Hongesombut5
Department of Electrical Engineering, Faculty of Engineering, Kasetsart University, Bangkok, Thailand
Email: 1*Naowarat.teph@ku.th, 2Budsarakum.r@ku.th, 3Sirasit.f@ku.th , 4Thunwa.boo@ku.th, 5Komsan.h@ku.th

Abstract—This paper proposes power loss reduction and the best outcome becomes challenging. Many research
voltage profile improvement in the distribution system using papers have applied Meta-heuristic algorithms to obtain an
feeder reconfiguration. There are many methods to reduce optimal configuration that minimizes power loss. The ant
power loss and improve voltage profile in the distribution colony search algorithm, Fuzzy logic, Genetic algorithm
system. Feeder reconfiguration is one of the most efficient ways (GA), and Particle swarm optimization (PSO) have been
that consume less investment cost. However, obtaining the proposed in [2]-[4], respectively.
optimal solution is somewhat complex and challenging since
many related variables exist. A modified particle swarm The PSO algorithm has been applied to find the optimal
optimization (MPSO) is applied to solve this complicated switching state of the IEEE 33-bus system to achieve
problem. This method has an operating pattern that mimics minimum power loss and maintain bus voltage within
searching for the food of the swarm by exchanging experiences requirements [4]. In [5], The comparison of PSO and GA in
and communicating data to observe the excellent result of the IEEE 33-Bus reconfiguration and DG sizing problems shows
problem. Therefore, this paper proposes feeder that PSO results surpass GA results. In [6], The selective
reconfiguration in the distribution system by using a modified random search space and position control improvement of
particle swarm optimization to find the appropriate network PSO have been proposed, including comparison results from
topology of the distribution system that causes minimum loss each PSO algorithm.
and improves voltage profile to acceptable criteria. The
proposed method was tested on a radial IEEE69-Bus This paper presents an MPSO algorithm that can narrow
distribution system in the case of balanced loads and an initial search area to help determine the new distribution
unbalanced loads by simulating on DIgSILENT Simulation system structure. To achieve minimum power loss and
Language (DSL) of the DIgSILENT PowerFactory software improve the voltage profile within grid code requirements.
using Python programming language to write MPSO scripts to The algorithm has a working pattern that mimics the flight of
analyze the system. The result shows that the power loss of the birds by exchanging experiences and communication
distribution system decreased to 76 % after feeder information exchanged to find the most suitable answer,
reconfiguration, which is satisfactory and indicates that the global optimization. The proposed method was tested on the
distribution system has high efficiency compared with the
IEEE 69-Bus, a radial power distribution system. The system
conventional one.
is simulated on DIgSILENT PowerFactory using DSL that
Keywords—Distribution system, Power loss, Feeder works with MPSO scripting in Python for system analysis.
reconfiguration, Distribution system efficiency, Python The simulation results show that system power loss is
programming language, Particle Swarm Optimization method reduced and voltage profile is improved.

I. INTRODUCTION II. RESEARCH METHODOLOGY


Power systems have expanded continuously due to A. Particle Swarm Optimization
economic activities and population growth, especially The Particle Swarm Optimization (PSO) was proposed in
distribution networks. The radial system configuration is 1995 by Eberhart and Kennedy [7]. The original idea came
selected mainly by many utilities for distribution networks from studying the social behavior of a bird flock. A bird
because of its simplicity in operating. While distribution flock split itself. Each tries to search for food and transmits
systems keep expanding, Increasing connected loads to grids signals between them to communicate and exchange
cannot be avoidable. High load current can cause increasing information about where food is available and move to the
system power loss and voltage drop below grid code location where the best food source is informed.
requirements. Thailand's grid code requirement has specified
voltage requirements in normal conditions must maintain In PSO, To find the optimized solution, Particles (or
within a range of 0.95 to 1.05 pu [1]. birds) are defined as a data struct that contains position and
velocity. First, The population size of particles must be
Those problems can be solved in many ways, such as by determined, and each particle position is assigned randomly.
installing DGs (Distributed Generations), ESSs (Energy The population then moves to the search area to find its best
Storage Systems), or FACTS (Flexible AC Transmission position relative to each particle and compare it to others to
System) devices. However, the methods mentioned above find the best place in the swarm. The best value from each
have limitations, requiring the appropriate installation particle is called “Pbest,” and the best in the swarm is called
location and expensive cost. Alternatively, Network “Gbest.” And then, each Gbest value from the PSO method
reconfiguration is implemented to reduce power loss and will continue to improve and search for the answer until the
improve the voltage profile in distribution systems because most suitable solution is found. Finally, The best position
of the most cost-effective and efficient methods. Even some value obtained in the last generation is used to answer the
configurations can achieve the goal, but which one provides

978-1-6654-8584-5/22/$31.00 ©2022 IEEE

Authorized licensed use limited to: Universidad de Talca. Downloaded on April 22,2023 at 01:46:39 UTC from IEEE Xplore. Restrictions apply.
2022 19th International Conference on Electrical Engineering/Electronics, Computer, Telecommunications and Information Technology (ECTI-CON)

PSO method. The velocity, Inertia weight, and position of reconfiguration scheme must be found. Then, check the
PSO can be described in Equations 1 to 3. radial power supply scheme so that all loads are continuously
energized in (7)
vt + 1 vt vt vt
Vk = ω V k + c1r1 ( pbest − X k ) (
+ c 2 r2 gbest − X k ) (1)
det( A) = 1 or − 1 (7)
 ωmax − ω min 
ω = ωmax −  Iter (2) Where, det( A) is a determinant of Bus Incidence Matrix.
 Itermax 
C. The Network Reconfiguration using MPSO
v t +1 vt v t +1
Xk = X k +Vk (3) This section describes the MPSO algorithm, which is
vt + 1 based on the interaction of DIgSILENT Powerfactory with
Where, Vk is the velocity of particle kth at t+1 iterations. Python, where data is exchanged through an Application
Programming Interface (API), as shown in Fig. 1, Fig. 2 is an
vt example of a Python script to retrieve various distribution
Vk is the velocity of particle kth at t iterations.
system parameters for analysis in MPSO's algorithm.
pbest is the best value of particle kth (local).
Fig. 3 illustrates the MPSO algorithm written in Python
gbest is the global best value.
script and can be explained as follows:
c1 , c2 is an acceleration coefficient. Step 1: Create a search area by defining a possible power
ω is an Inertia weight. supply structure using the tree diagram according to
r1 , r2 is a uniform random value in a range of [0,1].
Table I and a Python script example of this step is shown in
Fig. 4.
uv t +1
Xk is a position of particle kth at t iterations. Step 2: Start MPSO by randomly placing every transfer
switch in the system and checking the power supply structure
B. The MPSO Application in the Network Reconfiguration condition. If the power supply scheme is not the radial
system, repeat step 2 and an example Python script is shown
in Fig. 5.
Network reconfiguration is inexpensive to minimize
power loss while maintaining each bus voltage within grid
code requirements. Still, the question of “How to achieve this
goal” becomes more challenging. The PSO method is an
efficient tool that can help find the optimal solution. To
achieve those goals, In This Paper, MPSO is proposed. This
MPSO can narrow an initial search area by defining a
possible power supply scheme using the tree diagram
principle, as shown in Table I. The objective function is
defined in (4), which power loss can be calculated using (5).
The voltage level requirement is a constraint, as shown (6). Fig. 1. The data exchange between Python programs and DigSILENT
Powerfactory programs
total
Minimize Objective = Ploss (4)
n
total
Ploss = 3 x  (I i2 ⋅ Si ⋅ Ri ) ; i ∈ n (5)
i =1

Vmin ≤ Vbus ≤ Vmax (6)

total
Where, Ploss is a total power loss (kW). Fig. 2. An example of a Python script to retrieve various distribution
system parameters
Ii is a current in a feeder section (kA).
TABLE I. SEARCH AREA OF MPSO
Si is a switching state for each feeder section i
Parameters Values
Ri is a feeder section resistance (Ω). Sw11, Sw12, Sw13, Sw14, Sw43, Sw44, Sw45,
1
n is a number of wires in a feeder section Sw70
Sw4, Sw5, Sw6, Sw7, Sw8, Sw46, Sw47, Sw48,
Vmin is the lowest voltage limit (pu). 2 Sw49, Sw52, Sw53, Sw54, Sw55, Sw56, Sw57,
Sw58, Sw73
Vbus is a bus voltage (pu). Sw3, Sw9, Sw10, Sw35, Sw36, Sw37, Sw38, Sw39,
3
Sw40, Sw41, Sw42, Sw69
Vmax is the maximum voltage (pu). Sw21, Sw22, Sw23, Sw24, Sw25, Sw26, Sw59,
4
Sw60, Sw61, Sw62, Sw63, Sw64, Sw72
The objective function is considered by combining the 5 Sw15, Sw16, Sw17, Sw18, Sw19, Sw20, Sw71
power loss, which can be calculated by (5), and the effect of
the voltage levels at the buses in (6). If the voltage at the bus
is without the specified benchmark range, a feeder

Authorized licensed use limited to: Universidad de Talca. Downloaded on April 22,2023 at 01:46:39 UTC from IEEE Xplore. Restrictions apply.
2022 19th International Conference on Electrical Engineering/Electronics, Computer, Telecommunications and Information Technology (ECTI-CON)

Step 7: If the calculation has not been yet completed, go


back and repeat from step 4 until the end of the MPSO
calculation.
D. The IEEE 69-Bus System Configuration
To investigate the effectiveness of the proposed method,
this paper applies the above approach to the IEEE 69-Bus
system before reconfiguration presented in [2]. The IEEE
69-Bus system is a 12.66 kV -distribution system consisting
of 68 feeder sections, 48 load groups with the total loads
equal to 3.8 MW, and 73 switches which 68 of 73 are
sectionalized switches, and 5 are transfer switches. A
distribution system is studied and analyzed on DIgSILENT
PowerFactory with Python programming language. The
initial parameters of MPSO are demonstrated in Table II.
Before optimizing network configuration, the power loss of
this system is 224.96 kW, and the minimum voltage is 0.90
p.u. at bus number 65 (Bus65). Python script can help control
switch-state and coding MPSO easier. The IEEE 69-Bus
system after reconfiguration is illustrated in Fig. 6.
TABLE II. INITIAL CONDITION OF MPSO

Parameters Simulation values


Particles 5

Iterations 300

c1 , c2 1,2

Fig. 3. The MPSO Flowchart


ωmax , ωmin 0.9,0.4

A number of variables 5

r1 , r2 Random, Random

Fig. 4. An example Python script of Step 1

Fig. 5. An example Python script of Step 2

Step 3: Execute load flow calculation and check the


voltage conditions of all buses.
Step 4: The velocity of each particle is randomized. Next,
update the position of each particle. Then, check the power
supply structure condition if it is the radial system or not. If
the power supply structure condition is not radial, repeat step
4.
Step 5: Execute load flow calculation and check the
voltage conditions of all buses again.
Step 6: Calculate the power loss (Fitness Function) and
keep the power supply structure that provides the minimum
power loss.
Fig. 6. The IEEE 69-Bus system after reconfiguration

Authorized licensed use limited to: Universidad de Talca. Downloaded on April 22,2023 at 01:46:39 UTC from IEEE Xplore. Restrictions apply.
2022 19th International Conference on Electrical Engineering/Electronics, Computer, Telecommunications and Information Technology (ECTI-CON)

III. SIMULATION RESULTS AND DISCUSSION

TABLE III. SIMULATION RESULTS

Parameters Before After


Sw69, Sw70, Sw71, Sw13, Sw57, Sw62,
Open-Switch
Sw72, Sw73 Sw69, Sw71
Power Loss 224.96 kW 53.66 kW

Vmin 0.90919 pu 0.97340 pu

Energy Loss $147,903.44 $38,623.58

Fig. 9. The comparison of voltage of all buses between before and after
network reconfiguration

IV. CONCLUSION
This paper presents the optimal network reconfiguration
of the IEEE 69-Bus system using the MPSO method. The
distribution system reconfiguration using MPSO, simulated
Fig. 7. The number of convergence cycles in the MPSO on DSL with Python scripts, showed that the power loss was
reduced to 76 % from the base case (53.66 from 224.96 kW).
After finishing the MPSO process, The optimal network In other words, this method can save energy loss costs.
configuration is obtained that power loss is reduced to 53.66 Moreover, The voltage profile impact can also improve,
kW. The switching operation pattern is shown in Table III. which helps operation voltage at each bus stay within the
The voltage level of all buses is operated within Thailand's grid code limit. Therefore, the proposed method is an
grid code requirement. The MPSO can convert to the optimal efficient alternative way to find the optimum network
solution within 78 iterations from predefining 300 iterations configuration in the distribution system. In addition, the
spending approximately 27 minutes shown in Fig. 7. The proposed method can be further developed for other power
power loss and voltage profile comparison between before distribution purposes, such as finding the network
and after network configuration is shown in Fig. 8 and 9, configuration that provides the minimized reliability indexes.
respectively.
REFERENCES
Fig. 8 represents the comparison of power loss of all
[1] Regulation on the Power Network System Interconnection Code B.E.
buses between before and after network reconfiguration. A 2016, [online] Available:
red line represents before network reconfiguration. Line https://www.pea.co.th/Portals/0/Document/vspp/PEA%20Interconnec
numbers 56 and 6 have maximum power losses in this case. tion%20Code%202016.pdf [Accessed April 10, 2022].
However, both transmission lines have reduced power loss [2] J. S. Savier and D. Das, "Impact of Network Reconfiguration on Loss
after network reconfiguration, represented by a black line. Allocation of Radial Distribution Systems," in IEEE Transactions on
Power Delivery, vol. 22, no. 4, pp. 2473-2480, Oct. 2007, doi:
Fig. 9 represents the comparison of the voltage of all 10.1109/TPWRD.2007.905370.
buses between before and after network reconfiguration. [3] K. Nara, A. Shiose, M. Kitagawa and T. Ishihara, "Implementation of
Before network reconfiguration, at bus65, the voltage genetic algorithm for distribution systems loss minimum re-
decreased to 0.91 p.u., lower than 0.95 p.u., which was configuration," in IEEE Transactions on Power Systems, vol. 7, no. 3,
pp. 1044-1051, Aug. 1992, doi: 10.1109/59.207317.
below the grid code requirement. However, after network
[4] V. J. Shetty and S. G. Ankaliki, "Electrical Distribution System
reconfiguration, the voltage at Bus65 increased to 0.97 p.u., Power Loss Reduction and Voltage Profile Enhancement by Network
upper than 0.95 p.u., which passes the grid code requirement. Reconfiguration Using PSO," 2019 Fifth International Conference on
Electrical Energy Systems (ICEES), 2019, pp. 1-4, doi:
10.1109/ICEES.2019.8719292.
[5] W. M. Dahalan and H. Mokhlis, "Network reconfiguration for loss
reduction with distributed generations using PSO," 2012 IEEE
International Conference on Power and Energy (PECon), 2012, pp.
823-828, doi: 10.1109/PECon.2012.6450331.
[6] I. I. Atteya, H. A. Ashour, N. Fahmi and D. Strickland, "Radial
distribution network reconfiguration for power losses reduction using
a modified particle swarm optimisation," in CIRED - Open Access
Proceedings Journal, vol. 2017, no. 1, pp. 2505-2508, 2017, doi:
10.1049/oap-cired.2017.1286.
[7] J. Kennedy and R. Eberhart, "Particle swarm optimization,"
Proceedings of ICNN'95 - International Conference on Neural
Networks, 1995, pp. 1942-1948 vol.4, doi:
10.1109/ICNN.1995.488968.
Fig. 8. The comparison of power loss of all buses between before and after
network reconfiguration

Authorized licensed use limited to: Universidad de Talca. Downloaded on April 22,2023 at 01:46:39 UTC from IEEE Xplore. Restrictions apply.

You might also like