2015 International Conference on Nascent Technologies in the Engineering Field (ICNTE-2015)

Optimal Reactive Power Dispatch by Furnishing

UPFC Using Multi-Objective Hybrid GAPSO
Approach For Transmission Loss Minimisation and
Voltage Stability
Shilpa S. Shrawane Dr. Manoj Diagavane Dr. Narendra Bawane
Department of Electical Engg. Principal Principal
GHRCE, VIT, SBJIT,
Nagpur,India Nagpur,India Nagpur,India
shilpashrawane@gmail.com mdai@rediffmail.com narendra.bawane@yahoo.com

Abstract: The goal of optimal reactive power dispatch is to make production cost while satisfying constraints. Some
the transmission loss minimal in addition to control of voltage uncertainties have developed due to tremendous growth of
profile such that the voltage deviations at the load buses for power systems .At the same time, electrical energy is in great
various loading conditions. By furnishing Flexible AC demand due to urbanization of life style and industrialisation.
Transmission System devices in the transmission system , the
With the tremendous advancement of technology, to increase
power control can be finely achieved. The paper describes
Optimal Reactive Power Dispatch for the power loss the power delivery, transmission systems are being pushed to
minimisation using optimal location of Unified Power Flow operate closer to their stability limits and at the same time
Controller for loss minimisation. The power flow is first solved reaching their thermal limits [9].
by Newton Raphson method without implementing or furnishing While matching supply and demand, main problems faced by
Unified Power Flow Controller and then by furnishing them power industry are stability problem and constraint associated
randomly in the IEEE -30 bus system. Unified Power Flow with the thermal limit which affects the quality of power
Controller allows control of real and reactive power both in delivered [2]. One of the best methods for reducing these
addition to voltage magnitude control at various buses. In this constraints and better utilisation of available capacity is
paper, optimal reactive power dispatch is applied using the
FACTS devices which have made the present transmission
hybrid GA -PSO. Simulations are performed on IEEE-30 bus
test system using MATLAB software package to ensure efficacy and distribution of electricity more controllable and more
of the proposed algorithm. This paper aims at to find the efficient which help in controlling the flow in heavily loaded
optimum usage of Unified Power Flow Controller which means lines and enhance system stability.
the finding of the optimal location where their influence would In this paper, the ORPD problem is solved by furnishing the
be more useful. FACTS devices as an additional control parameter in. Power
flow model of UPFC as an additional control parameter has
Keywords: Unified Power flow controller UPFC, Genetic been used in this study which finds first the power loss and
algorithm GA, Particle swarm optimisation PSO , Flexible AC then optimal location of UPFC and its cost. Earlier, the
transmission devices FACTs, Optimal Reactive Power Dispatch
gradientbased optimization algorithm was used to solve
ORPD
ORPD problem [1,4]. But ,ORPD being a global optimization
I. INTRODUCTION problem with several local minima that can lead the
conventional optimization solutions to a local minimum [3].
For security to be the best policy, the secured economic Even, analytic and differential properties of the objective
performance of power systems, ORPD plays a vital role. The function have to be known in conventional methods to
reactive power is automatically generated with no cost .But ,it simplify the problem. Earlier , many stochastic algorithms
affects the total generating cost by increasing the transmission such as simulated annealing, GA and EP(Evolutionary
loss. Concept that dispatches the reactive power to minimize programming) were developed to solve the global
the transmission loss will be consequent to the lowest optimization problem [5].

978-1-4799-7263-0/15/$31.00 2015 IEEE

 2015 International Conferencce on Nascent Technologies in the Engineering Field (ICN
NTE-2015)

Recently, few algorithms including FACTS devices are , , ,

proposed to solve the ORPD problem. However, these
researches consider the problem as single-objjective and it was
solved using several methods, such as, iterrative techniques
[1], particle swarm optimization (PSO) [44-5], differential
(4)
evolution [6] and (GA) [7].
This paper aims at implementation of hhybrid GA-PSO
approach for multi objective ORPD, to find UPFC optimum where , ra, rb, rc, rd and re are the positive coefficients to be
location in a power system to minimum traansmission losses varied to minimise the objective fu unction.Pgref and Pgreflim are
reference value and limiting value of o real power generation at
and voltage deviation at load buses. An existing NRLF
the PV busses. NPQ, NPV , Nl and Ng represent number of load
algorithm is used to include FACTS deviices [10] and is
busses, generator busses, number off transmission lines
presented here. The proposed algorithm is tessted on the IEEE-
30 bus test system and using MATLAB softw ware package.
III. UPFC MODEL FOR ORPD :
II. PROBLEM FORMULATION
N OF ORPD
The UPFC consists of the serries inverter coupled to
transmission line via a series transformer and the shunt
ORPD mainly related with minimization actiive power loss of
inverter coupled to a local buss via a shunt connected
power system, while satisfying the unit andd system equality
transformer. The shunt inverterr absorbs or generates
and inequality constraints such as geenerator voltage
controllable reactive power and d provides active power
magnitude (VGi), reactive power generation oof capacitor bank
exchange to the series inverter to
o satisfy operating control
(QCi) and transformer tap settings (tK) [8].
requirements. Fig. 1 shows the twoo voltage source model of
Mathematically ORPD problem may be reepresented as an
UPFC.
optimization problem as
, 1
, 0
, 0

where, J is the objective function to be minim

mized, g(x,u) and
h(x,u) represent equality and inequalityy constraints, x
represents state variables vector and u is a vvector of control
variables . Here , in this research, J is multi objective
functions considering transmission loss, parrameter selection
for location of UPFC device, then real andd reactive power
transferred through UPFC leaving bus i to bbus j through the
dc link, voltage deviation .The equality and inequality
Fig. 1 Two voltage source model of UPFC
constraints are selected as the transmissionn line constraints
and security constraints[6]. Using all thesee becomes the
Vse and Vsh represent injected seriess and shunt voltage sources
vector of dependent variables and can be giveen as follows:
which are controllable in magnitud de and phase. Xse and Xsh
are the UPFC series and shunt coup pling transformer reactance
, , , , 2 resp. Vi and Vi are voltages at busees i, iand Vj is the voltage
of bus j of the receiving end of traansmission line. Bus i is a
Pgref is real power of generator bus, VPQ is vvoltage at the PQ fictitious bus to be added betweeen bus I and bus j. Ish is
busses and Lpij and Lqij represent real and reaactive power loss current through the UPFC shunt con nverter [11].
through transmission line connected between bus i and bus j.
The control vector is given by is given by: 0

(5)
, , , 3
(6)
PPV, VPV are the real power and voltage at thhe PV busses and (7)
PijUPFC and QijUPFC the real and reactive pow
wer to be injected
by UPFC to be connected in between bus i annd bus j.
sin
Therefore , the objective function can be writtten as follows:
cos (8)
sin
 2015 International Conference on Nascent Technologies in the Engineering Field (ICNTE-2015)

cos (9) 1.02 sin 0.02 sin

for the lossless operation of UPFC sin (16)
1.02 if losses have to be included. cos
cos (17)
The total switching losses of the two converters is estimated
to be about 2 % of the power transferred, for thyristor based Where PiUPFC, PjUPFC, QiUPFC and QjUPFC are the real and
PWM converters. reactive power injected by UPFC at bus i and bus j
Psh and Pseries are the shunt and series active power exchange respectively.
of series and shunt voltage sources. Pij,Qij and Pji,Qji are the
real and reactive power flows leaving bus i and bus j
respectively.bse , bsh, gse , gsh,are the series and shunt
suseptances and conductance ces of the transmission line
resp. i and j are the voltage angles at bus i and j resp. And r
and are the coefficients for the calculation of Vse from
Vi.Therefore , for more simplicity the expressions for Psh ,
Qsh and Pseries , Qseries are given by following expressions:

cos
sin (10)
sin
cos (11)
Fig. 2 Steady state model of UPFC
The active and the reactive power supplied by the series
converter of UPFC is given by
IV. HYBRID PARTICLE SWARM OPTIMIZATION
(PSO)
sin
sin 12 The hybrid approach consists of two stages.

cos 1. Implementation of GA
sin 13 2. Implementation of PSO

The operating constraints of the UPFC (active power IV.1 IMPLEMENTATION OF GA

exchange between two inveters via the DC link) are:
The common terms used while solving problem using GA are:
0 14 Search space: The space of all feasible solutions which

all desired solutions resides is called search space. Each
point in the search space represents one possible solution
Where,
which can be marked by its fitness value for the problem.
Population: it is the number of individuals present with
There are in all 13 modes of operation of UPFC in power flow the same length of solutions.
control [11]. The control equation of any mode can be Fitness: it is the value assigned to an individual present
generally written as: with the same length of solution it contains.
Fitness function: it is a function that assigns fitness value
, 0 to the individual . It is problem specific.
, 0 (15) Selection : Selecting individual for creating the next
generation .
where , , , , , ; Crossover: The main objective of crossover is to
fspec and gspec are the control references. recognize the information of two different individuals and
produce a new one.
For the analysis purpose, the steady state model of UPFC is Mutation: It is nothing but randomly changing the values
described by combining both series and shunt power in a solution. It is used to introduce some part of artificial
injections at bus i and bus j and is described by following diversification in the population to avoid premature
equations and is as shown in Fig. 2 convergence to local optimum.
 2015 International Conference on Nascent Technologies in the Engineering Field (ICNTE-2015)

In stage-1, GA has been applied. Firstly to generate the where, Xi ,Vim+1and w represent the particle, velocity of the
chromosome generation based on random voltages values on particle and inertia weight respectively. a1 and a2 are the
the load busses. Using this, the values of Pi and Qi at various positive coefficients , iter stands for no. of iterations.
busses have been found out for the system. The Jacobean
In stage 2, PSO has been applied. Firstly the lines are
matrix elements of NRLF have been calculated .Then
crossover has been applied using these load values. All these arranged in the descending order of the power loss and using
give the children solutions. Then using these random load this data, the velocity function has been written to give the
values , mutation has been done and fitness function selects based on the voltages at various busses .Here, GA has been
the best line with power output has been calculated to give employed to obtain the optimal region quickly and in stage-2,
maximum loss in the transmission lines for the proper the PSO with systematic reduction of the size of the search
placement of UPFC. region [16] is used to find the local optimum. Here, (22) gives
For optimum location of UPFC following function has been
function used for optimum capacity of UPFC calculation is
used
N
(18)
= Max Vi V j Yij cos( i j ) N
(22)
(V )
B
ij
n =1 = Min Normal Vi
i =1

IV.2 IMPLEMENTATION OF PSO where, Vnormal is the normal voltage and Vi is the voltage at ith
bus .
PSO is based on evolutionary algorithm of swarm
intelligence population for global optimization and it is
V. THE OPERATIONAL INDICES
developed using bi direction space simulation of bird
flocking. Every agents position is represented along XY axis Various operational indices considered in the algorithm are as
and the velocity is expressed by Vx and Vy (the velocity along follows:
X-Y axes). Each agent knows its best value so far (pbest) and i)Voltage Profile Index(VPI): Basically this index will
its XY position and in the group (gbest). This information is measure the variations in voltages at all the busses with
analogy of personal experiences of each agent. Namely, each respect to Vref .For accuracy flat voltage profile is always
agent tries to modify its position using the following preferred. VPI is calculated as in (23)
information: - the current positions (x, y); the current
(23)
velocities (Vx, Vy); the distance between the current position
and pbest; the distance between the current position and gbest. The per unit value of this VPI is given by (24)
Various terminologies used in PSO are
Particle definition (24)
PSO Parameters /

Number of particles ii)Voltage Stability Index(VSI): Basically this index is

Inertia weight (w) calculated for all busses except slack bus from bus no.2 to all
Acceleration constants other busses as follows:
Number of iterations
,
Values of velocity which is given as follows 1 (25)

These terms are given as follows: iii) Power loss sensitivity Index(PLSI):The variation in power
loss has inverse relation with the voltage profile at various
V wV a rand1 P X busses. Power loss sensitivity index at each bus is given by
a rand2 g X (19)
Where (26)
w w iter (20)
The transmission loss minimises as the PLSIi becomes more
X X V (21) and more negative. The variations To ensure validity of the
proposed hybrid method, it is tested on standard IEEE- 30 bus
test system having non-linear characteristics. The results of
 2015 International Conference on Nascent Technologies in the Engineering Field (ICNTE-2015)

the proposed hybrid approach prove the effectiveness of the performence of bus no vs bus voltage
1.1
proposed approach in terms of solution quality and NRLF
computation time. 1.08
increased load
proposed

VI. SIMULATION RESULTS AND DISCUSSIONS 1.06

BUS voltage
1.04
The IEEE -30 BUS system is used as the test system to prove
the effectiveness of the proposed algorithm for ORPD. Firstly, 1.02
it is simulated for normal load flow using NRLF analysis .The
total power losses have been found to be 10.8095pu. Then the 1

load on the load bus is randomly increased. Fig.3 shows the

0.98
plot for variations in the voltages at various busses without
increasing the load, with increasing the load and voltage 0.96
variations with the proposed algorithm. It can be concluded 0 5 10 15
BUS NO
20 25 30

from the Fig.3 that the voltage variations at all the busses
have been reduced by proposed algorithm though there is Fig.3 Plot of bus voltages at all the busses of IEEE-30 bus system
sudden increase in load at the load busses. performence of Lineno vs Line loss
3.5
It has been observed that the losses have increased to NRLF
11.461pu. Here, now programming is done such that the 3 increased load
proposed
proposed algorithm specifies the five transmission lines in the
2.5
descending order of their of their power losses .So, we specify
the transmission line where power loss is having highest 2

magnitude for the placement of UPFC. IEEE 30 bus system is

Loss
1.5
simulated for the load variations and again the power losses
have been found to be10.9983pu. The proposed approach is 1

proved to be efficient from the results of hybrid GAPSO to be

0.5
9.4652pu.
Fig.4 depicts the performance of power losses in all lines of 0

IEEE 30 bus system for the normal load flow, with increase in
load and with proposed approach. Fig. 5 gives the -0.5
0 5 10 15 20 25 30 35 40 45

Comparison of Power loss in a line randomly chosen at Line no

normal load, increased load and after applying proposed

approach .Now for analysis purpose, three transmission lines Fig. 4 Plot of power loss in all the transmission lines of IEEE -30 bus system
are randomly chosen and the comparative results of power 12
performence of Loss

loss for normal load condition, for increased load and NRLF

optimised values using the proposed algorithm have been 10

increased load
proposed
drawn in Fig. 6. It has been observed from the figure that the
power loss has reduced to a great extent after applying the
8
proposed approach.
Fig. 7 shows the error function calculation for the placement
Power Loss

of UPFC in the transmission line between bus no. 6 and bus 6

no.7.
Fig. 8 shows the variation of power loss sensitivity index with 4

the variation of voltage profile for the placement of UPFC in

transmission line between bus no. 6 and bus no.7. 2

0
1
loss

Fig. 5 Comparison of Power loss in a line at normal load, increased load and
after applying proposed approach
 2015 International Conference on Nascent Technologies in the Engineering Field (ICNTE-2015)

comparison of line no 10-21 6-7 and 22-24 given number of Unified Power Flow Controllers by satisfying
12
NRLF
the voltage constraints at various busses during optimisation.
increased load A multi objective hybrid computational intelligence approach
10 proposed
i.e. GA-PSO is used to solve this multi objective nonlinear
programming problem. The computation schemes are
8
discussed in detail, such as consideration of network losses,
voltages at various buses, the location of Unified Power Flow
Power Loss

6 Controller to be implemented and the power flow computation.

Using the proposed algorithm, the losses have been reduced
4 by optimally placing the Unified Power Flow Controller in the
transmission lines . It also gives the value of the real and
2 reactive power to be injected by Unified Power Flow Controller.
The case studies of the IEEE- 30 system has confirmed that
0 this Hybrid intelligence approach is very effective and exact
1 2 3
Line NO for solving the optimal reactive power dispatch.
Fig. 6 Comparison of Power Loss of three lines

VIII. REFERENCES
4
[1] Sen K.K.,Stacey, E.J,UPFC: unified power flow controller:theory,
3 modeling, and applications, IEEE Trans.Power
Deliv.,1998,13,(4),pp.145360
2 [2] Tambey, N., and Kothari, M.L.: Damping of power system
oscillations with unified power flow controller (UPFC), IEE Proc.,
1 Gener. Transm. Distrib., 2003, 150, (2), pp. 129140
[3] Gyugyi L.,Schauder,C.D.,Williams, S.L., Rietman, T.R., and
Torgerson, D.R.: The unified power flow controllers: a new approach
0 to power transmission control,IEEE Trans. Power Deliv.,1995,
0 20 40 60 10,pp.1085-1097
[4] Nabavi-Niaki, A., and Iravani, M.R.: Steady -state and dynamic
models of unified power flow controller (UPFC) for power system
studies, IEEE Trans. Power Syst., 1996, 11, (4), pp.19371943
[5] Fang, W.L., and Ngan, H.W.: Optimising location of unified power
Fig.7 Error function for UPFC placement in line6-7 flow controllers using the method of augmented Lagrange multipliers,
IEE Proc., Gener, Transm, Distrib., 1999, 146, (5),pp. 428434
[6] Kursat Ayan, Ulas Kilic, Artificial bee colony algorithm solution for
0.2 optimal reactive power flow , Applied soft computing 12(2012), pp
1477-1482
0 [7] S.M.Sadeghzadeh,A.H.Khazali,S.Zare; Optimal reactive power
Dispatch considering TCPAR and UPFC, IEEE 2009,pp577-583
0 20 40 60 [8] Dommel, H. W. and Tinney, W. F., Optimal power flow solutions.
-0.2
IEEE Transactions on Power Apparatus and Systems, 1968, 87, 1866
1876
-0.4 [9] N.G.Hingorani, L. Gyugi, Understanding FACTS Concepts and
Technology of Flexible AC Tansmission Systems, New York: IEEE
-0.6 Press ,2000.432p. ISBN 0-7803-3455-8
[10] P. Preedavichit and S.C. Srivastava, Optimal reactive power dispatch
-0.8 considering FACTS devices, ElectricPower Systems Research. 46
(3),1998 , pp. 251257.
[11] Xiao-Ping Zhang, Christian Rehtanz, Bikash Pal, Flexible AC
Transmission Systems: Modelling and Control, Springer Publishers
Fig.8 Power loss sensitivity for UPFC placement in line 6-7
(2006),
[12] P.Subbraj,P.N. RajarayananOptimal Reactive Power Dispatch by
VII. CONCLUSION Particle Swarm Optimization with Cauchy and adaptive Mutations,
2010 International Conference on Recent Trends in Information,
A mathematical model for Optimal Reactive Power Dispatch Telecommunication and Computing , pp 110-116
[13] Marco R. Silva, Zita Vale, H.M. Khodr, Carlos Ramos, J. M. Yusta,
simultaneously optimally furnishing placement and size of Member IEEE, Optimal Dispatch with Reactive Power Compensation
UPFCs is presented in this paper. It includes two important by Genetic Algorithm 978-1-4244-6547-7/10,2010 IEEE
aspects: to minimise power loss and maximise the system
load ability via optimising location and parameters of the