You are on page 1of 6

Minimization of Ohmic Losses in Power Networks by

Utilization of Interphase Power Controllers

Behrouz Azimian Amin Helmzadeh Amir Shahirinia

Inamori School of Engineering Electrical Engineering Department Center of Excellence for Renewable Energy
Alfred University Amirkabir University of Technology University of the District of Columbia
Alfred, USA Tehran, Iran Washington, D.C., USA

Abstract—Installation of flexible alternating current A genetic algorithm (GA) is one of the intelligent
transmission system (FACTS) devices is a widespread approach optimization algorithms utilized for the efficient placement of
for reducing ohmic losses in power networks. However, the these devices. It performs a search process imitated from the
location and control parameters of these devices should be mechanism of biological selection and biological genetics [1].
determined through an optimization procedure. An interphase Allocation of FACTS devices in power systems is a relatively
power controller (IPC) is one of the FACTS devices that can new research area. These devices can contribute in the
change the active power flow in network branches. However, reduction of power flow on overloaded lines, which in turn
the placement of this device has not been discussed yet. increases the loadability of the power system. In addition,
Therefore, an efficient optimization method will be introduced
FACTS devices can improve the stability, security and
in this paper based on a genetic algorithm (GA) combined with
optimal power flow (OPF) for IPC placement. Firstly, a simple
finally, provide a more energy-efficient transmission system.
novel method for entering the steady state model of IPC into the In [2], a GA is used in a graphical user interface (GUI)
power flow equations is proposed. Secondly, the optimal values application to determine the optimal multi-type FACTS
of decision variables such as IPC location and its control location in a more convenient and flexible way for power
parameters will be determined by the proposed optimization grids. Based on the GA, authors in [3] have considered
method. The simulation results on IEEE 30-bus and 118-bus contractual requirements to identify optimal thyristor
test systems show that the GA-based optimization process is controlled phase shifting (TCPS) location in a power system
able to obtain optimal solutions for the mixed integer placement network for maximization of system capabilities in an open
problem, which would result in a more energy-efficient power market. In [4], the extended voltage phasors approach
transmission system. (EVPA) and the well-established line flow index (LFI)
method are compared and it is shown that the former provides
Index Terms—Genetic algorithm; IPC placement; ohmic loss a more accurate indication for the placement of FACTS
reduction; optimal power flow; optimization controllers. In the most recent literature about the FACTS
device allocation problem, Duan et al, [5] has proposed a
I. INTRODUCTION sparse optimization method based on the sparse
Ohmic loss occurs due to energy dissipation in conductors characteristics of device placement in power system and the
used for transmission lines, transformers and etc. These losses investment cost of different types of FACTS devices (i.e.,
are intrinsic in real-world power systems. However, ohmic static VAR compensator (SVC), thyristor-controlled series
losses can be reduced to an optimum level by utilizing proper compensation (TCSC) and thyristor controlled phase shifting
strategies and installing compensation devices. Due to the (TCPS)). This method is combined with optimal power flow
global energy crisis, power ohmic loss minimization has (OPF) to investigate the numbers, locations, initial settings
attracted more attention in recent years. Many optimization and even types of FACTS devices simultaneously.
methods have been proposed for installing compensation Further studies about the FACTS device allocation and
devices to minimize these losses in power networks. Optimal the corresponding intelligent optimization approaches are
placement of the power system equipment such as shunt surveyed as follows. The optimal allocation of a unified
capacitors, distributed generations (DG) and FACTS devices power flow controller (UPFC) using a self-adaptive
have been widely investigated with the aim of power loss
evolutionary programming (SAEP) is performed in [6].
Optimal location of FACTS shunt-series controllers (i.e.,
Various algorithms have been developed to determine the phase-shifting transformer (PST), hybrid flow controller
most efficient place for compensation devices in power grids. (HFC), and UPFC) by means of nonlinear programming

978-1-5386-7138-2/18/$31.00 ©2018 IEEE

(NLP) and mixed-integer nonlinear programming (MINLP)) enormous design flexibility and makes a wide variety of
is investigated in [7]. Sensitivity analysis is used for applications possible. The capability of current control
determination of optimal location of the interline power flow enables the operator to control the power carried by the IPC
controller (IPFC) and optimal unified power flow controller and this could reduce the ohmic loss in network branches
(OUPFC) in [8]. Improved particle swarm optimization [12]. Indeed, IPC can be adapted to specific operating
(IPSO) is proposed to determine the optimal location of static conditions. In general, this adaptation can result in an
compensator (STATCOM) units in [9], and the brainstorm optimized operation of power systems.
optimization algorithm (BSOA), which is a novel promising
heuristic optimization algorithm inspired by brainstorming The main capabilities of IPC are robust control of active
process in human beings, is applied to SVC and TCSC and reactive power, fault current limitation and
allocation problems [10]. interconnection of non-synchronous networks [13-15]. The
authors in [16] have shown that power quality can be
Although many of the previous studies have focused on enhanced with the use of IPC in power networks. Due to the
the allocation problem for different kinds of FACTS devices, capability of controlling active power flow in network
allocation of the interphase power controller (IPC) has not branches, IPC can be installed in power grids with the aim of
been investigated. The purpose of this paper is to present a power loss minimization.
new model of IPC that can be used in load flow studies as
well as finding the most efficient place for one IPC device so III. MODELLING IPC FOR LOAD FLOW STUDIES
as to minimize the power grid loss. The presence of an IPC in Minimization of the active power losses in network
a power grid and its impact on the load flow can be simply branches is the main objective of the presented IPC
investigated by the proposed model. placement algorithm. Therefore, a simulation package is
required to calculate the power losses for the both networks
Authors in [11] have proposed a method that maximizes the with and without IPC. MATPOWER, which is a simulation
total transfer capability (TTC) while minimizing the power package based on MATLAB, is selected in this paper for
loss. Comparative results are presented to show that the performing load flow calculations. MATPOWER can provide
optimum place and settings of UPFC can improve TTC a high-level set of tools for steady-state power system studies
without compromising the minimization of power loss which such as power flow and optimal power flow [17]. This
is a challenging issue in UPFC placement. simulation package contains a set of MATLAB M-files that
can be easily understood and customized by the users.
In the following, we first establish a mathematical model of
Fortunately, the steady-state model of all FACTS devices
the optimal IPC planning problem and present a new model
including IPC can be combined with MATPOWER. For
of IPC, then we solve the optimization problem by a genetic example, the steady-state model for multi-terminal voltage
algorithm by considering the loss minimization and the TTC source converter high voltage direct current (VSC MTDC) is
maximization. introduced in [18] for carrying out sequential AC/DC power
The remainder of this paper is organized as follows. flow. Simulation results of this paper are obtained by
Section II describes the IPC technology. In Section III IPC combining the proposed methodology with the
will be modeled for load flow studies. Section IV formulates MATPOWER.
the proposed objective function optimization approach for In MATPOWER, all branches are modeled with a
IPC placement with the aim of ohmic loss reduction. Section standard transmission line π model in series with an ideal
IV provides the simulation results. Section V summarizes the phase shifting transformer which is located at the sending end
main contributions and conclusions of this paper. of the branch. Therefore, phase shifting can be easily applied
II. IPC TECHNOLOGY for desired branches in MATPOWER.
IPC is a power electronic device which is equipped with The capability of having phase shifting for network
thyristor switches. A general electrical model of IPC is shown branches is very helpful for combining the steady state model
in Fig. 1. As shown in the figure, each phase consists of two of IPC with the MATPOWER. Suppose that an IPC should be
parallel branches and each branch contains a reactance installed on the k-th branch of the network which is located
(inductor or capacitor) connected in series with the phase- between i-th and j-th buses. As it is shown in Figure 2 the
shifting device. model the network can be modified by adding one bus and
two branches to the existing branch model.

Figure. 1 IPC model.

What distinguishes IPC from other FACTS devices is the

way in which the series elements are connected between two
buses. Adjusting the values of the reactances and internal Figure 2. Modifying existing network model due to installing an IPC on the
phase angles, which are the control variables, allows k-th branch.
A. Objective Function The mathematical equations of the cost function and
The main goal of the proposed IPC placement algorithm optimization constraints are not clear in terms of decision
is minimizing the sum of ohmic losses in network branches. variables. In addition, the optimization problem is a mixed
Therefore, the objective function can be defined as follows: integer one. In such situations, heuristic methods such as GA
can be utilized for minimizing the cost function. As
mentioned before, determination of the cost function value
Ploss for a special set of optimization variables is based on OPF
for the network with IPC
f= (1) results in MATPOWER. As a result, the proposed objective
Ploss function will be formed in MATLAB. A set of M-files should
for the network without IPC be prepared in this software to determine the optimal values
of decision variables. However, the GA toolbox in MATLAB
can be utilized for the objective function minimization. It is
If the active power flows at the sending and receiving noteworthy that this toolbox can solve the mixed integer
ends of all branches are known, the value of Ploss can be optimization problems. Another advantage of the
calculated as below: combination of OPF with GA is the limitation of the
optimization search space. When the cost function for any
N br
iteration is obtained based on OPF calculation, search space
Ploss =  (Psendi + Preceivei ) (2) for GA can be limited significantly. Therefore, GA can
i =1
achieve the optimal solution much faster and with high
probability the best global solution [19].
Therefore, the value of Ploss can be determined by
performing power flow analysis on the modified network Genetic algorithm begins with a set of solutions
model (with IPC). This value will change if the location (represented by chromosomes) called population. A
and/or control parameters of the IPC are changed. Hence, the chromosome consists of genes that are optimization variables.
value of proposed cost function can be minimized by Based on fitness function values to the current population, a
determining optimal location for the IPC and optimal values new population will be formed by reproduction and crossover
of control parameters. actions. This process will be repeated until some conditions
are satisfied. Figure 3 shows the flowchart of GA
B. Optimization Variables optimization algorithm.
IPC location is the most important variable in the
optimization procedure. In MATPOWER case studies, each
branch has a number that indicates its location. Hence, the
number of selected branches is an integer decision variable in
the optimization process. On the other side, the parameters
Φ1 , Φ 2 , XL , and X C are continuous decision variables that
should be calculated with the aim of ohmic losses
minimization. Hence, the proposed IPC placement is a mixed
integer optimization problem with one integer and four
continuous decision variables.
C. Optimization Constraints
It is noteworthy that there are some equality and
inequality constraints in IPC placement optimization
procedure. For example, voltage magnitudes of all buses and
loading factors of branches should be maintained at an
acceptable range for the network with IPC (inequality
constraints). In addition, the total power demand of the
Figure 3. Flowchart of GA optimization algorithm.
system must be provided by the generating units with the
minimum cost of generation (equality constraint).
A sub-process should be done in all iterations of GA for
All the above equality and inequality constraints must be evaluating the fitness function values for the population.
satisfied in the OPF process. Hence, we preferred to calculate Fitness function evaluation is based on OPF results.
the value of ohmic losses and consequently the value of the Therefore, several OPF should be done in all GA iterations.
cost function by performing OPF calculations. In such OPF calculations may diverge for some chromosomes (each
manner, all equality and inequality constraints will be chromosome is a set of decision variables). In such situations,
satisfied during the optimization procedure. MATPOWER the value of cost function will be set equal to an extremely
includes codes for solving the OPF problem. Therefore, the large value. Therefore, these chromosomes have little chances
value of proposed cost function for the networks with and to be selected in the next population. The flowchart of fitness
without IPC can be calculated by using this package. function evaluation is shown in Figure 4.
Check Stopping Criteria
Satisfied k=1

Modify simulation model for Extract the values of

the current set of decision decision variables Figure 6. Chromosome structure
variables from k-th chromosome
Simulation results are depicted in Figure 7. As it is shown,
Reproduction the best value of the cost function is equal to 0.7532. It
Run OPF for the modified means that the ohmic losses can be reduced by 24.68% via
simulation model NO optimal installation of an IPC in the IEEE 30-bus test system.

Set the cost function value

equal to an extremely large NO Successful OPF k > population size


Calculate the value of cost Calculate the sum of losses k=k+1

function for the chromosome of network branches

Figure 4. Flowchart of sub-process for fitness function evaluation in GA


A. IEEE 30-Bus Test System
The standard IEEE 30-bus test system is selected for Figure 7. GA toolbox outputs for IPC placement in IEEE 30-bus test system
examination of the proposed IPC placement algorithm. The
location of IPC and the values of control parameters should Optimal values of decision variables are shown in the
be determined for the selected test system. The single line TABLE I.
diagram of the IEEE 30-bus test system is shown in Figure 5.
Number of
Phi1 Phi2 XL XC
The selected
(degree) (degree) (ohm) (ohm)
36 -20.6995 -35.1144 187.2990

As it can be seen, the branch with number 36 (that is

located between 27-th and 28-th buses) has been selected for
IPC installation. The voltage magnitudes and phase angles for
both networks with and without IPC, which are obtained from
OPF results, are depicted in TABLE II.
TABLE II shows that voltage magnitudes for both
networks with and without IPC are maintained between
acceptable ranges. Loading factors of the network branches
are shown in TABLE III.
TABLE III shows that by optimal placement of IPC,
Figure 5. Single line diagram of the IEEE 30-bus test system. loading factors have been changed significantly in branches
with high resistances values. For example in branches with
The GA optimization method is used for minimizing the
numbers 33, 35 and 20 that have high resistances, the values
proposed cost function. It must be noted that there are no
of loading factors have been decreased by 88.05%,74.86%
equality and inequality constraints in the GA optimization
and 48.3% respectively.
process. However, all constraints for electrical quantities
(such as voltage magnitudes and loading factors of branches)
will be satisfied via OPF procedure at the stage of fitness
function evaluation. Chromosome structure which consists of
decision variables is shown in Figure 6.
TABLE II. PHASE ANGLES AND MAGNITUDES OF VOLTAGES FOR optimization (PSO) and interior point methods can be
implemented by MATLAB for minimizing desired cost
Without IPC With IPC functions. However, the mentioned methods cannot solve the
Voltage Voltage Vmax Vmin Voltage Voltage MIP optimization problems. Therefore, the optimum location
mag angle mag angle of the IPC (as an integer decision variable) is assumed to be
1 0.982 0.000 1.05 0.95 1.050 0.000
known. It means that four continuous decision variables
2 0.979 -0.763 1.1 0.95 1.048 -0.710
3 0.977 -2.390 1.05 0.95 1.035 -2.040
should be determined by the mentioned optimization
4 0.976 -2.839 1.05 0.95 1.033 -2.427 methods. Simulation results are shown in TABLE IV.
5 0.971 -2.486 1.05 0.95 1.034 -2.170
6 0.972 -3.229 1.05 0.95 1.028 -2.706 TABLE IV. OPTIMAL VALUES OF DECISION VARIABLES
7 0.962 -3.491 1.05 0.95 1.021 -2.985
8 0.961 -3.682 1.05 0.95 1.017 -3.062 Cost
9 0.990 -4.137 1.05 0.95 1.034 -4.121 Opt. Phi1 Phi2 XL XC
Method (degree) (degree) (ohm) (ohm) Fun.
10 1.000 -4.600 1.05 0.95 1.037 -4.857 Value
11 0.990 -4.137 1.05 0.95 1.034 -4.121
12 1.017 -4.498 1.05 0.95 1.050 -4.524 Genetic
13 1.064 -3.298 1.1 0.95 1.081 -3.379 -20.699 -35.114 178.255 187.299 0.7532
14 1.007 -5.040 1.05 0.95 1.039 -5.148 Particle
-2.580 -16.765 113.944 166.041 0.7537
15 1.009 -4.814 1.05 0.95 1.041 -5.039 Swarm
16 1.003 -4.839 1.05 0.95 1.038 -4.946 Simulated
-14.012 -38.129 125.085 195.144 0.7603
17 0.995 -4.887 1.05 0.95 1.032 -5.086 annealing
18 0.993 -5.484 1.05 0.95 1.027 -5.675 Interior
-17.055 -28.458 144.577 156.997 0.7560
19 0.987 -5.688 1.05 0.95 1.023 -5.868 point
20 0.990 -5.472 1.05 0.95 1.025 -5.667
21 1.009 -4.621 1.05 0.95 1.042 -5.002 As it can be seen in TABLE IV many combinations of
22 1.016 -4.503 1.1 0.95 1.047 -4.924
23 1.026 -3.756 1.1 0.95 1.052 -4.453 other decision variables ( Φ1 , Φ 2 , X L , and X C ) can lead to
24 1.017 -3.885 1.05 0.95 1.037 -5.100 approximately optimal value for the proposed objective
25 1.044 -2.072 1.05 0.95 1.037 -5.374
26 1.027 -2.476 1.05 0.95 1.020 -5.783
27 1.069 -0.715 1.1 0.95 1.046 -5.287 C. IEEE 118-Bus Test System
28 0.982 -3.215 1.05 0.95 1.035 -2.390
29 1.050 -1.849 1.05 0.95 1.027 -6.473 The standard IEEE 118-bus test system is selected for
30 1.039 -2.643 1.05 0.95 1.016 -7.303 examination of the proposed IPC placement algorithm.
31 - - 1.1 0.9 1.100 -9.462 Simulation results are depicted in Fig.8.



Line R LF Line R LF
Num. (pu) Change Num. (pu) change
1 0.02 5.00% 22 0.11 -7.96%
2 0.05 2.21% 23 0.06 -14.17%
3 0.06 1.95% 24 0.03 8.08%
4 0.01 2.46% 25 0.09 6.18%
5 0.05 3.64% 26 0.03 0.32%
6 0.06 2.19% 27 0.03 -40.79%
7 0.01 -2.92% 28 0.07 -36.53%
8 0.05 6.73% 29 0.01 -21.34%
9 0.03 -21.07% 30 0.1 -35.01%
10 0.01 -1.87% 31 0.12 -3.01%
11 0 16.16% 32 0.13 113.22%
12 0 16.16% 33 0.19 -88.05%
13 0 - 34 0.25 0.02%
14 0 14.69% 35 0.11 -74.86%
15 0 -11.91% 36 0 8.64%
16 0 -24.67% 37 0.22 0.12% Figure 8. GA toolbox outputs for IPC placement in IEEE 118-bus test
17 0.12 12.21% 38 0.32 0.13% system.
18 0.07 38.58% 39 0.24 0.06%
19 0.09 -4.59% 40 0.06 2.98% As it is shown, the best value of the cost function is equal
20 0.22 -48.30% 41 0.02 1.12% to 0.8732. It means that the ohmic losses can be reduced by
21 0.08 -15.89%
12.38% via optimal installation of an IPC in the IEEE 118-
B. Comparision With Some Optimization Methods bus test system. Optimal values of decision variables are
In addition to the GA solver, many other optimization shown in TABLE V.
methods such as simulated annealing (SA), particle swarm
TABLE V. OPTIMAL VALUES OF DECISION VARIABLES IN 118-BUS IEE Proceedings-Generation, Transmission and Distribution, vol.
TEST SYSTEM 151, no. 6, pp. 705-712, 2004.
[7] A. L. Ara, A. Kazemi, and S. N. Niaki, "Multiobjective optimal
Number of Phi1 Phi2 XL XC
location of FACTS shunt-series controllers for power system
The selected Line (degree) (degree) (ohm) (ohm)
operation planning," IEEE Transactions on Power Delivery, vol. 27,
9 74.93 74.94 117.28 120.54 no. 2, pp. 481-490, 2012.
[8] V. S. Rao and R. S. Rao, "A generalized approach for determination
VI. CONCLUSION of optimal location of OUPFC," in 2015 International Conference
on Electrical, Electronics, Signals, Communication and
This paper proposed a novel model for optimal placement Optimization (EESCO), 2015.
of an IPC in power systems. IPC is a FACTS device which [9] K. Ravi and M. Rajaram, "Optimal location of FACTS devices using
can potentially be used for network applications where a improved particle swarm optimization," International Journal of
flexible adjustment of the power flow is required. Therefore, Electrical Power & Energy Systems, vol. 49, pp. 333-338, 2013.
[10] A. R. Jordehi, "Brainstorm optimisation algorithm (BSOA): An
IPC can be installed in power networks in order to minimize efficient algorithm for finding optimal location and setting of
power losses in transmission lines. Along with the optimum FACTS devices in electric power systems," International Journal of
placement of IPC, its control variables should be set as well. Electrical Power & Energy Systems, vol. 69, pp. 48-57, 2015.
GA has been purposed to simultaneously determine the [11] H. I. Shaheen, G. I. Rashed, and S. Cheng, "Optimal location and
location and setting values of IPC. As there are several parameter setting of UPFC for enhancing power system security
based on differential evolution algorithm," International Journal of
constraints in an optimization problem in a power network, a Electrical Power & Energy Systems, vol. 33, no. 1, pp. 94-105,
GA is combined with OPF to meet all the required constraints 2011.
such as voltage magnitude at each bus, loading factor and [12] J. Pourhossein, G. Gharehpetian, and S. Fathi, "Unified interphase
minimum cost of power generations. The proposed method power controller (UIPC) modeling and its comparison with IPC and
has been tested on IEEE 30-bus and 118-bus test systems. UPFC," IEEE transactions on power delivery, vol. 27, no. 4, pp.
1956-1963, 2012.
The results show that if optimum place of IPC is found and [13] F. Beauregard, J. Brochu, G. Morin, and P. Pelletier, "Interphase
corresponding variables are set, power losses could be power controller with voltage injection," IEEE transactions on
reduced by almost 24.68% and 12.38% in IEEE 30-bus and power delivery, vol. 9, no. 4, pp. 1956-1962, 1994.
118-bus test systems, respectively. [14] J. Brochu, P. Pelletier, F. Beauregard, and G. Morin, "The
interphase power controller: a new concept for managing power
REFERENCES flow within AC networks," IEEE transactions on power delivery,
vol. 9, no. 2, pp. 833-841, 1994.
[1] X. Yin and N. Germay, "Investigations on solving the load flow [15] K. Habashi et al., "The design of a 200 MW interphase power
problem by genetic algorithms," Electric Power Systems Research, controller prototype," IEEE transactions on power delivery, vol. 9,
vol. 22, no. 3, pp. 151-163, 1991. no. 2, pp. 1041-1048, 1994.
[2] E. Ghahremani and I. Kamwa, "Optimal placement of multiple-type [16] V. S. Rao, "Enhancement of power quality in IEEE-14 bus systems
FACTS devices to maximize power system loadability using a using Interphase power flow controller," Majlesi Journal of Energy
generic graphical user interface," IEEE Transactions on Power Management, vol. 2, no. 3, 2013.
Systems, vol. 28, no. 2, pp. 764-778, 2013. [17] R. D. Zimmerman, C. E. Murillo-Sánchez, and R. J. Thomas,
[3] L. Ippolito and P. Siano, "Selection of optimal number and location "MATPOWER: Steady-state operations, planning, and analysis tools
of thyristor-controlled phase shifters using genetic based for power systems research and education," IEEE Transactions on
algorithms," IEE Proceedings-Generation, Transmission and power systems, vol. 26, no. 1, pp. 12-19, 2011.
Distribution, vol. 151, no. 5, pp. 630-637, 2004. [18] J. Beerten, S. Cole, and R. Belmans, "Generalized steady-state VSC
[4] N. K. Sharma, A. Ghosh, and R. K. Varma, "A novel placement MTDC model for sequential AC/DC power flow algorithms," IEEE
strategy for FACTS controllers," IEEE Transactions on Power Transactions on Power Systems, vol. 27, no. 2, pp. 821-829, 2012.
Delivery, vol. 18, no. 3, pp. 982-987, 2003. [19] M. H. Aliabadi, B. Behbahani, and A. Jalilvand, "Combination of
[5] C. Duan, W. Fang, L. Jiang, and S. Niu, "FACTS devices allocation GA and OPF for allocation and active and reactive power
via sparse optimization," IEEE Transactions on Power Systems, vol. optimization in Distributed Generation units," in Power and Energy
31, no. 2, pp. 1308-1319, 2016. Conference, 2008. PECon 2008. IEEE 2nd International, 2008,
[6] J. Hao, L. Shi, and C. Chen, "Optimising location of unified power pp.1541-1544: IEEE.
flow controllers by means of improved evolutionary programming,"