Comparison of optimal reactive power dispatch methods

in IEEE 30 bus system

Sulabh Sachan1[0000-0003-0309-5001], Sambeet Mishra1[0000-0003-2970-7598],
Thomas Øyvang1[0000-0002-0529-0620], Chiara Bordin2 [0000-0002-9869-3276]
1 Department of Electrical, IT and Cybernetics, University of South-Eastern Norway
2 Department of Computer Science, University of Tromsø, Norway

[email protected] , [email protected] ,
[email protected] , [email protected]

Abstract. It has been noted that a number of metaheuristics are applied with suc-
cess to power system optimal reactive power dispatch (ORPD) challenges. These
algorithms' convergence rates are also determined to be low, and the results they
provide are deemed to be inadequate. It suggests that there is insufficient inves-
tigation and exploitation in the algorithm. Therefore, an appropriate approach is
needed to improve the algorithm's search performance. The enhanced butterfly
optimization algorithm (EBOA) is used in this work to address the power sys-
tem's ORPD problems. IEEE 30 bus network is chosen as a base network to test
and validate the performance of the methods. On this systems, all three goals—
Power Loss Minimization, TVD Minimization, and L-Index Minimization—are
taken into account. The outcomes of EBOA have been equated with those of the
novel BOA and additional modern heuristics. Additionally, statistical analysis is
performed to evaluate the algorithm's resilience.

Keywords: L-Index, EBOA, Power Loss, and ORPD.

1 Introduction

The power system is operating under heavy load as more distributed resources and de-
mand are being integrated with the grid. Its operation, control, and planning cannot be
done by hand. Intelligent computing-based support systems are necessary for power
systems [1]. Currently, one of the main sources of electricity worldwide is thermal
power generating units. To meet the energy demand, resources must be used as effi-
ciently as possible. Environmental constraints are starting to have a significant impact
on how the electricity system operates.
These generators can operate more economically and reduce their emissions into the
environment if they do so. A secure and stable operating state may not always be guar-
anteed by power system operation based solely on economic or environmental consid-
erations. Thus, further research on generator scheduling may be done to keep power
systems operating in an economical and reliable manner. In the current power system
context, power system operators are primarily concerned with power system security
challenges, aside from the economy.
2

Voltage security is greatly impacted by reactive power, which is intimately tied to bus
voltage. Voltage instability or, in severe situations, voltage collapses can result from an
inadequate reactive power supply [2–3]. Thus, careful consideration must be given to
the distribution and management of reactive power. Additionally, many power appa-
ratus in contemporary power networks are currently operating at full capacity. As a
result, there is a chance of component outages when the system operates in an unstable
area.
Tools like voltage control, reserve management, and preventative dispatch are found
to be appropriate in preventing this. Reactive power rescheduling is another way to
provide a steady condition for power system functioning. Reactive power sources can
be optimally adjusted to achieve this. In order to meet the aforementioned objectives,
OPF-based restricted ORPD (Optimal reactive power dispatch) challenges must be re-
solved. An instrument for controlling reactive power needs in power systems is ORPD.
These kinds of generation scheduling are quite difficult and complex as optimization
problems. The use of traditional optimization approaches to solve these kinds of issues
has reached its limit [4]. The global power system community is open to non-traditional
optimization methods like EBOA, TLBO, CSO, EPSO etc. Furthermore, little research
has been done on the application of modern optimization techniques in these fields;
more investigation is needed. A review of the literature is done in order to learn more
about the operational issues with the power system and how optimization approaches
are applied to address those issues. The most widely used OPF solution for a network
was created in [5]. Different facets of the OPF issue are covered. To create an OPF,
Newton's power flow solution method is expanded upon.
Minimal cost and loss have been achieved by solving a 500-bus test problem [6]. Real
and reactive power of the generator along with the tap positions of the transformers
have been employed as variables. In [7], an expanded OPF problem based on the suc-
cessive linear programming approach is put out. [8] Talks on the fundamental ORPD
issue. Real power loss and the cost of injecting reactive power are reduced using an
interior point-based technique [9]. The majority of clever ways to OPF solutions with
different objective functions. In [10], several computational strategies have been uti-
lized to tackle power system problem. An effective OPF based on evolutionary pro-
gramming is developed in [11]. The convergence is improved by using the gradient
information. The new hybrid PSO method solves a difficult nonlinear non convex op-
timization issue.
A harmony search algorithm for ORPD power system problems is presented by the
authors in [12–13]. The approach is contrasted with GA and PSO-based ORPD problem
solutions. The multi-objective optimization problem in [14–16] is handled with consid-
eration for environmental emissions and economic load dispatch as objective functions.

1.1 Objectives and Contribution of the paper

The aim of the proposed work is to properly allocate reactive power derived from
the numerous sources that are found in power systems in order to achieve the power
 3

system's security- related objectives. In addition, the following objectives have been
taken into account in this paper.
• The ORPD issue has been resolved.
• Control variables, both discrete and continuous, are taken into account.
• The goal of the task, which is presented as an OPF, is to:
(a) Curtail the system's active power loss (b) Enhancing the voltage stability index
and (c) reduction of the load voltage variance.

2 Optimal reactive power dispatch problem

An instrument for controlling reactive power needs in power systems is ORPD [17].
The goal of the ORPD issue is to reduce Voltage Stability Index, Load Voltage Devia-
tions, and Real Power Losses. By controlling the generator bus voltage (a continuous
variable), turning on and off discrete variable controllers (SVCs), modifying the trans-
former's tap-settings (Discrete Variable) and controlling more reactive control equip-
ment, this can be guaranteed.

2.1 Objectives Function

Objective functions considered in this work are depicted in eqn. (1-3).

Minimization of Active Power Loss (PLoss)
NL
PLoss =  g k (Vi 2 + V j 2 − 2ViV j cos ij )
k =1 (1)

where
ij =  −
i j

gk = conductance of kth line

Minimization Total Load Voltage Deviation (VD)
NL
VD =  Vi − Vrefi
i =1 (2)
Voltage Stability Enhancement (L-Index)
NPV
Vi
L − Index = 1 −  Fi j where, j = 1, 2,3...NPQ
i =1 Vj
(3)
 I PQ  Y1 Y2  VPQ 
 I  = Y Y4  VPV 
 PV   3 (4)
L-Index value vary from 0 to 1.
4

2.2 Control and Dependent Variables

In this work, terminal voltage at voltage control bus is shown as (VG1, VG2. . .VGn),
output of shunt VAR compensator as (QC1, QC2 . . . QCn) and tap ratio of tap changing
transformers as (T1, T2 . . .Tn).
Voltage of PQ bus is depicted as (VL1, VL2 . . .VLm) and generator reactive power
output as (QG1, QG2 . . .QGm) along with slack bus power (P1).

3 Enhanced Butterfly Optimization Algorithm Outlines

The EBOA is an algorithm inspired by nature and derived from butterfly behavior. for
both local and worldwide searches, we can refer eqn. (5-6)
xi t +1 = xi t + (r 2  g * − xi t )  f i
(5)
t +1
xi = xi + (r  x j − xk )  fi
t 2 t t
(6)
where;
t: current iteration number
xkt : solution of x for kth butterfly with iteration number t
Fragrance value, f is calculated by eqn. 7.
f= (SM)Ia (7)
Here, SM is sensory modality, I is the stimulus intensity, and a is the power exponent.
The pseudo code for EBOA heuristics is shown in Fig.1.

Fig. 1. The EBOA Pseudocode

 5

4 Analysis and Result Discussion

In order to address the power system's ORPD issues, the IEEE-30 bus system has
been considered. Nineteen control variables total—six general voltages, four on-line
taps, and nine VAR outputs from shunt capacitors—are present in case study. The algo-
rithm's performance is primarily dependent on choosing the right value for one of its
parameters. The three control parameters of EBOA are Population Size (M), Switch
Probability (P), and Sensory Modality (SM). In this work, two parameters of the EBOA
are fixed while one is tuned. Keeping P and M constant, SM is first adjusted between
0.01 and 0.09, and so forth. Before the ideal solution is declared, 50 trials are conducted.
Fig. 2 depicts the convergence plot for IEEE-30 bus system. This illustrates that Switch
Probability (P) converges after 20 iterations, Sensory Modality (SM) converges after
60 iterations and Population Size (M) converges after 192 iterations.

Fig.2 Convergence plot of Active Power Loss

Fig. 3, Fig. 4 and Fig. 5 depict the comparison of power losses, TVD and L-index with
different heuristics respectively. Fig 6 depicts the convergence plot for IEEE-30 net-
work corresponding to P-Loss, TVD and L-Index.
Table 1 shows the statistical analysis for IEEE-30 bus system. From Fig 3,4,5 and Table
1, it can be depicted that power losses, TVD and L-index reduces significantly as com-
pared to other heuristics. It's evident from the findings that EBOA has superior conver-
gence properties.

Fig.3 Comparison of P-Loss with different heuristics

6

Fig.4 Comparison of TVD with different heuristics

Fig.5 Comparison of L-Index with different heuristics

Table 1: Statistical Analysis (Case-1, 2 & 3) IEEE-30 Bus

Case1, IEEE 30 Bus, P-Loss

50 Trial Best Value Worst

Mean Value SD Value
Runs (MW) Value

BOA 4.646 5.5608 5.1468 0.1669

EBOA 4.487 4.8092 4.7603 0.0176
Case2, IEEE 30 Bus, TVD Minimization
50 Trial Worst
Best Value Mean Value SD Value
Runs Value
BOA 0.1425 0.2312 0.1739 0.0186
EBOA 0.0886 0.1175 0.0974 0.0061

Case3, IEEE 30 Bus, L-Index Minimization

50 Trial Worst
Best Value Mean Value SD Value
Runs Value
BOA 0.1167 0.1212 0.1189 0.0021
EBOA 0.0977 0.098 0.0979 0.0000677
 7

Fig 6 Convergence Plot for IEEE-30 bus: P-Loss, TVD and L-Index

Fig 7 Load voltage profile for 24 hours

8

Exploitation and Exploration are balanced as a result of intelligent EBOA implementa-

tion. EBOA performs better than BOA and a number of other modern algorithms. Ac-
cording to statistical study, EBOA is a reliable and consistent algorithm. The outcome
of benchmark problems shows that EBOA is qualified to be an effective member of the
Meta-Heuristic Family.

5 Conclusion

Under varying degrees of operational and physical limitations, OPF-based ORPD is-
sues are solved using a new meta-heuristic method named EBOA. The outcomes of the
simulations and the comparisons with other modern algorithms verify that EBOA per-
forms better in nearly every scenario. The power losses are reduced to 1.76%, bus volt-
age variation is reduced to .61%, and L-index factor is reduced to .006%. The effec-
tiveness of evolving machine algorithms (EMA) in addressing single-objective, multi-
objective, and optimization under intricate restrictions and variable problems suggests
that EMA can be considered among the most efficient members of the evolutionary
algorithm community. It can also be used to address optimization issues from various
scientific and engineering fields, both confined and unconstrained. This study guaran-
tees optimal operating cost while maintaining adequate voltage stability under increas-
ing loading situations. Further the robustness of this methodology can be checked on
large system. Also, reliability-oriented network restructuring can be investigated along
with ORPD in future scope of work.
Acknowledgement
The Research Council of Norway (RCN) provided funding for this work under Grant
number. 326673 (SysOpt project).
Nomenclature
Indices
i : generating unit
t : time duration
Parameters & Abbreviations
ORPD : Optimal reactive power dispatch
EBOA : Enhanced butterfly optimization algorithm
TLBO : Teaching learning based optimization
CSO : Chicken swarm optimization
EPSO : Evaluated particle swarm optimization
PLoss : Active Power Loss
TVD : Total voltage deviation
gk : Conductance of kth line
L-Index : Voltage Stability Enhancement
VPQ : Voltage of load bus
VPV : Voltage of generating bus
 9

References
[1]. Wood and Wollenberg, “Power Generation Operation and Control” John Wiley and sons, 1996.
[2]. S. Arora and S. Singh, “Butterfly optimization algorithm: a novel approach for global optimi-
zation”, Soft Computing, vol. 23, pp. 715-734, 2019.
[3]. Bordin C, Mishra S, Palu I., “A multihorizon approach for the reliability-oriented network
restructuring problem, considering learning effects, construction time, and cables mainte-
nance costs”, Renewable Energy, vol. 168, pp. 878-895, 2021.
[4]. Cheng Yang, Yupeng Sun, Yujie Zou, Fei Zheng, Shuangyu Liu, Bochao Zhao, Ming Wu,
Haoyang Cui, "Optimal Power Flow in Distribution Network: A Review on Problem Formu-
lation and Optimization Methods", Energies, vol.16, no.16, pp.5974, 2023.
[5]. J. Amomoh, R. Adapa and M.E. El-Hawary, “A review of selected optimal power flow liter-
ature to 1993-I: Nonlinear and quadratic programing approach” IEEE Transaction
Power System, Vol. 14, pp 96-104, 1999.
[6]. L.S. Vargas, V.H. Quintana and A. Vannelli, “A tutorial description of an interior point
method and its application to security constrained economic dispatch” IEEE Transactions on
Power Systems, Vol. 8, pp 1315-1324, 1981.
[7]. K. Iba, “Reactive power optimization by genetic algorithm” IEEE Transactions on Power
Systems, Vol. 9, pp 685-692, 1994.
[8]. Mishra S, Bordin C, Palu I., “RNR: Reliability oriented Network Restructuring," 2018 IEEE
59th International Scientific Conference on Power and Electrical Engineering of Riga Tech-
nical University (RTUCON), Riga, Latvia, 2018.
[9]. A. Neubauer, “Adaptive Non-uniform Mutation for Genetic Algorithms” International Con-
ference on Computational Intelligence, pp 24–34, 2005.
[10]. M. Dorigo, M. Birattari and T. Stutzle “Ant colony optimization” IEEE Computational Intel-
ligence Magazine, Vol. 1, pp 28-39, 2006.
[11]. Abhishek Rajan, T Malakar, “Optimum economic and emission dispatch using exchange mar-
ket algorithm”, Int. J. of Electrical Power and Energy Systems, Vol. 82, pp. 545–560, Apr
2016.
[12]. Abhishek Rajan, Tanmoy Malakar, “Exchange market algorithm based optimum reactive
power dispatch”, Applied Soft Computing, Vol. 43, pp. 320–336, Feb 2016.
[13]. Chong Yao, Youjun Zhang, "Direct Power Flow Controller with Continuous Full Regulation
Range", IEEE Transactions on Power Electronics, vol.39, no.5, pp.5449-5461, 2024.
[14]. Mahmuda Akter, Hamidreza Nazaripouya, "A Review of Data-Driven Methods for Power
Flow Analysis", 2023 North American Power Symposium (NAPS), pp.1-6, 2023.
[15]. Y. Chen, C. Wu and J. Qi, "Data-driven Power Flow Method Based on Exact Linear Regres-
sion Equations," Journal of Modern Power Systems and Clean Energy, vol. 10, no. 3, pp.
800-804, May 2022.
[16]. Y. Liu, N. Zhang, Y. Wang, J. Yang and C. Kang, "Data-Driven Power Flow Linearization:
A Regression Approach," IEEE Transactions on Smart Grid, vol. 10, no. 3, pp. 2569-2580,
May 2019.
[17]. P. Li, W. Wu, X. Wang and B. Xu, "A Data-Driven Linear Optimal Power Flow Model for
Distribution Networks," IEEE Transactions on Power Systems, vol. 38, no. 1, pp. 956-959,
Jan. 2023.