Volume 53, Number 1, 2012
69
Allocation of Static VAR Compensator in
western Algerian Network
F. BENZERGUA, N. KHALFALAH, A. CHAKER,
J.L. MARTINEZ RAMOS, J. RIQUELME and A. MARANO
Abstract: Due to the ever increasing load demand and reduced transmission rights, modern power transmission systems are
forced to carry increasingly more power over long distances. Consequently, the transmission system becomes more stressed.
Advances in power electronics and control technology, concretely the thyristor-controlled Static VAR Compensator (SVC), have
introduced powerful tools to power utilities and System Operators (SO) in order to guarantee an adequate power transfer
capability. This paper presents a based approach to determine optimal placement of Static VAR compensator (SVC) for voltage
security enhancement, the proposed method has been applied to the Algerian distribution system.
The SVC placement problem considers both, practical operation constraints of SVC, and the bound constraints on voltages which
minimizes the system losses. A sensitivity-based analysis method is used to select the candidate buses for the installation of the
SVC’s.
SVC is integrated into FDLF program. More specifically the problem of low voltage profile in Western Algerian system is
investigated with respect of the dynamic voltage margins.
Keywords: Optimal power flow, Sensitivity analysis, Voltage control, Optimal SVC allocation.
1.
INTRODUCTION
The trend, which appears in the development of
modern transmission systems, is the more intensive
utilization of existing networks. Moreover, a new
situation appears in distribution systems as is the
spreading of distributed generation due to renewable
energies employment. That provides a new aspect that
the conventional analytical or control methods do not
necessarily face conveniently. As a result, it is
important to reconsider voltage and reactive power
control in distribution systems [3].
In addition to this, due to deregulation and
restructuring of the electric power industry arises the
need to transport large blocks of energy among the
different system areas through some defined corridors.
Furthermore, the voltage profile of some remote buses
needs to be kept within a pre-specified range. These
reasons justify the installation of modern devices, such
as SVC, into power systems.
Static VAR compensators (SVC) are devices that
control the reactive power injection at a bus using
power electronics switching components [4].
Those SVC devices can be used to improve the
performance of the electric power system providing
reactive support, helping to control the voltage profile
of the network, getting as a result a better power factor
[9] and power system damping.
The problem of VAR sources planning has
received considerable attention in recent years, where

different optimization methods have been used to solve
the problem.
Conventionally, the optimal operation of the
power system networks has been based on economic
criterion. However, recent concerns about power
quality have forced power engineers to incorporate
other criteria such as improving the system voltage
profile, the transmission loss minimization etc.
Generally, the transmission power losses cause a
diminution of revenue due to increased generation
capacity requirement. Optimal reactive power dispatch
(ORPD) and incorporation of FACTS devices are some
of the applications used in modern energy management.
They are used to minimize total system transmission
loss and improve voltage profile.
Although earlier attempts to address this problem
were made in the sixties, it is not until the eighties
when more rigorous approaches can be found [5]. The
work reported in [1] constitutes the reference frame
from the point of view of classical optimization
methods, like mixed-integer linear programming. More
recently, several AI-related techniques, such as genetic
algorithms [3], simulated annealing [6], Fuzzy tabu
search [7], differential evolution [2], and hand hybrid
method, have been explored.
In this paper a heuristic technique based on
sensitivities is used to determine the location and size of
the VAR sources to be installed.
The article consists of the following parts:
Initially, an introduction to deregulation and
restructuring of the electric power industry was made.
Following, the problem is formulated showing how the
Manuscript received February 5, 2012.
© 2012 – Mediamira Science Publisher. All rights reserved.
70
ACTA ELECTROTEHNICA
SVCs are used as active elements to retain the voltages
magnitudes in allowable values for different loads
levels. In Section 3 it is proposed the calculation of
sensitivity factors and how to use them to select
adequate locations for SVCs in order to regulate
voltages. Later the needed steps to incorporate these
factors into the electrical system are detailed. Finally,
the method is applied to the problem of low voltage
profile in the western Algerian transmission and
subtransmission system (220/60 kV). The results are
discussed and compared with the ones obtained using
an OPF method.
2.
PROBLEM FORMULATION
The SVC placement problem considered in this
paper is to determine the locations, number and sizes of
SVC to be installed in the Algerian electrical
distribution system.
The objective is aimed to retain the voltage
magnitudes of the system within their prescribed
maximum and minimum allowable values for different
loads levels.
The Static Var Compensator (SVC) equipment is
composed of capacitors, thyristors and inductances.
There are two ways for modelling these devices. The
first model considers SVC as variable impedance,
which is adapted automatically to achieve the voltage
control. This is called the passive model and its main
disadvantage is the changing of nodal admittance
matrix whenever there is a variation in the operation
conditions of the power grid.
The second model, called active model, represents
SVC as a nodal reactive power injection. This is the
model used in this work. The injected reactive power at
bus i is Qsvc .
Sensitivity factors are used to determine at which
nodes additional SVC will have greatest effect in
controlling the voltage limits and the size of those
necessary SVC.
3.
SOLUTION METHODOLOGY AND
ALGORITHM OF THE PROPOSED
METHOD
The proposed optimal SVC placement method
uses sensitivity factors to rank the feeder nodes in order
to control the system voltages and reduce the power
losses.
The voltages along the feeder buses are required
to remain within upper and lower limits (typically,
within  5% of the rated feeder voltage) prior to, or
after, the addition of the SVC on the feeder.
Sensitivity factors are evaluated at each node
every time a change occurs in the feeder, such as the
addition of the SVC. The nodes are ranked in
descending order of the values of their sensitivity
factors. The top ranked nodes in the priority list are the
first to be considered for adding new SVC.
For the next iteration, the sensitivity factors are
recalculated and the nodes are again ranked with the
modification described in the previous paragraph. The
nodes that have been selected for SVC addition during
previous iterations occupy the highest positions in the
priority list.
However, in these algorithms, once a node has
been selected and a SVC device has been placed, a load
flow is performed to ensure that no voltage constraints
have been violated. In the proposed method, an ideal
candidate node for SVC placement will be selected
based on both, the reactive power losses and voltage
sensitivity. Thus, the likelihood of voltage violations
accruing after the installation of the SVC is reduced [8].
Each node voltage magnitude is checked against
its upper and lower limits. If a node voltage is not
within limits, the particular alternative is rejected. If no
constraint violation occurs the peak power loss, total
energy loss, and released SVC are computed [9].
The proposed algorithm is summarized by the
following steps:
1. Perform the load flow program to calculate bus
voltages using the fast decoupled method.
2. Compute the sensitivity factors.
3. On the first iteration, arrange the nodes in a priority
list in descending order of the value of their
sensitivity factors.
4. Identify the candidate location as the bus with
largest sensitivity factor.
5. In the next iterations, place the nodes with
permanent SVC additions at the top of the list, and
the remainder ones in descending order of their
sensitivity factors.
6. Add Qsvc at bus k and perform a load flow to find
new bus voltages. Exit if there is no voltage
violation, otherwise go to step (1).
Terminate the procedure when the maximum
number of SVC additions has been allocated, or if no
better solution can be found.
4.
NUMERICAL RESULTS
The assumption being tested is that the largest
sensitivity coefficients indicate the buses at which
reactive power injection is more effective in order to
control the voltage limits. The purpose of the tests is
not to study the economy of the reactive power
injection. But, it may happen that the required Qsvc
injection at a bus to take the bus voltage inside the
acceptable range is also useful to minimize the system
losses.
The proposed method has been applied to the western
Algerian transmission/ subtransmission system
220/60 kV (fig.1). Its main data and operational limits
are summarized in table 1 and 2.
This system has 64 load and 4 generator buses.
Table 1. Main data of the Western Algerian system
Load buses
Generator buses
Lines
Transformers
Shunt capacitors
64
4
78
12
5
71
Volume 53, Number 1, 2012
2
36
15
42
39
38
20
1
3
16
17
28
37
40
41
26
27
29
31
32
47
55
14
30
35
6
56
7
45
46
34
33
22
44
13
Morocco interconnection
52
53
11
25
51
48
67
49
54
50
12
61
63
68
66
64
43
21
62
60
65
59
18
19
24
23
4
5
10
9
8
57
58
Fig. 1. Algerian 220/60 KV transmission/ sub-transmission system.
Table 2. Limits of control variables and bus voltages
Magnitude
220 KV level
60KV level
Taps (20steps)
QShunt (2 steps)
Q1g
Q9g
Q23g
Q36g
Lower
0.99
0.95
0.9
0
-250 Mvar
-90 Mvar
-15 Mvar
-20 Mvar
Upper
1.11
1.10
1.1
10 Mvar
500 Mvar
180 Mvar
35 Mvar
36 Mvar
The inequality constraints for optimal Static VAR
Compensator placement are the quality of service
considerations and equipment limitations based on
company’s experiences.
Voltage magnitude constraints are expressed as:
Vi,min  Vi  Vi,max
Where | Vi ,min | and | Vi ,max | are the acceptable
voltage magnitude limits at bus i.
The adopted maximum and minimum tolerance
have been defined in such a way that the allowed
voltage interval is reduced to 0.95 <Vi <1.1 pu for all
buses I.
A large value of | Vmin | and | Vmax | indicates that
one or more of the system buses has a voltage
magnitude outside the desired limits, table 3.
FDLF method is employed both, during the line
search process and is updating the state before the next
iteration.
Table 3 shows the buses whose voltage is initially
bellow the lower limit.
Table 3. Buses with voltage violation
Bus
Vi
46
68
60
67
53
45
35
0.81281
0.82320
0.83915
0.87646
0.93768
0.94168
0.94690
Vi
0.137
0.127
0.111
0.074
0.012
0.008
0.003
After the calculation of the sensitivities, the
recommendations at this step consist of introducing a
SVC of 20 Mvar at bus 46.
This first action could correct the voltages of
buses 45 and 46.On the other hand, the voltage at bus
60 has not changed and buses 53, 60, 67 and 68 still
remain out of limits but with a lower voltage violation,
as shown in table 4.
Table 4. Buses with voltage violation after the first iteration
Bus
Vi
68
60
67
53
35
0.82438
0.83915
0.87754
0.94088
0.94702
Vi
0.12562
0.11085
0.07246
0.00912
0.00298
New sensitivities are calculated for each bus and
tested according the proposed algorithm. As a result, a
new SVC located at bus 68 is needed, being the
required reactive injection of 10 MVAR.
This correction eliminated the bound violation
from bus 68 and improved the voltage of bus 67, but it
72
ACTA ELECTROTEHNICA
was not enough to eliminate the bound violation in this
bus, as shown in table 5.
Table 5. Buses with voltage violation after the second iteration
Vi
0.1105
0.0463
0.0098
0.0031
Description
The third correction obtained after running the
algorithm is to place an SVC of 20 Mvar at bus 60. This
injection eliminates all violations, remaining now all
the bus voltages inside their feasible bounds. The active
power losses in this state are 23.62 MW.
The evolution of the active power losses is shown
in table 6.
27.16
23.10
24.00
24.10
25.10
23.9
--14.94
11.63
11.26
7.58
12.0
Table 6. Evolution of total active losses
Initial
It 1
It 2
It 3
27.16
24.72
24.72
23.9
Reduction (MW)
2.17
2.44
3.26
Reduction (%)
7.98
8.98
12
Losses (MW)
It should be noticed that after each voltage
correction the active power generated by the slack
generator (bus 1) decreased, which justifies that the
voltage correction in the network decreases the active
power losses as well.
To compare the results obtained with the proposed
algorithm two conventional OPF have been used. The
first one (denoted by the subindex 1) has as objective
function the minimization of the system losses, while
the other one (denoted by the subindex 2) has as
objective function the minimization of the reactive
power injection. In both cases, two possibilities were
taken into account, let the PV bus voltages free ( OPF1
and OPF2 ), or fix their values to a scheduled set-point
obtained from a previous optimization ( OPF1
and
OPF2 ), table 7.
Table 7. Optimization results
BUS
OPF1 (Mvar)
OPF1’(Mvar)
OPF2 (Mvar)
OPF2’ (Mvar)
Sensitivity based
solution (Mvar)
46
12.2
17.5
8.6
9.0
20
53
9.9
10.1
1.8
2.3
-
60
23.0
22.2
14.7
19.3
20
67
4.9
6.6
1.7
1.8
-
68
7.4
9.2
3.7
3.8
10
In all cases it can be observed that the use of SVC
increases the system loadability, table 8.
The results of OPF2 and OPF2 establish to add
5 SVC with a total reactive power of 30.5 Mvar and
36.2 Mvar respectively, located in the same buses than
the previous OPF. In this case there is a fewer active
power losses reduction, being of 11.26% and 7.58%,
respectively.
Base case
OPF1
OPF1’
OPF2
OPF2’
Proposed
method
5 SVCs
5 SVCs
5 SVCs
5 SVCs
3 SVCs
The total reactive power
injected (Mvar)
Table 8. Comparison of system’s active power losses (Propsed
method vs OPF)
Number of VAR sources Added
0.8395
0.9037
0.9402
0.9469
% Loss reduction
Vi
60
67
53
35
PL (MW)
Bus
The results of OPF1 and OPF1 show that the
placement of 5 SVC with a total reactive power of
57.4 Mvar and 65.6 Mvar respectively, located in the
buses 46,53,60,67 and 68, are sufficient to avoid the
bus voltage violations and reduce the losses at about
14.94% and 11.63, respectively.
57.4
65.6
30.5
36.2
50
The proposed sensitivity algorithm determines the
placement of 3 SVC, with 50 Mvar of total reactive
power, located in the buses 46, 60 and 68. In this case,
the active power losses decrease about the 12%.
The results show that both, the proposed method
and an OPF, give good results to avoid voltage
violation, but the proposed sensitivity-based algorithm
minimizes at the same time the number of SVC to be
installed, this becomes an important aspect if the
installation costs are taken into account.
5.
CONCLUSION
In this paper, we have proposed a sensitivity based
solution methodology to determine the optimal
locations of SVC devices, their number and size.
The SVC devices should be placed on the most
sensitive buses. With the loss sensitivity indices
computed for each bus.
The effectiveness of the sensitivity method to
solve the combinatorial optimization problem of SVC
placement has been demonstrated through the
numerical examples.
Results show the impact that the use of SVC
devices has in transmission losses of real networks.
Taking into account the obtained results, the
following can be stated:
 The use of SVC devices helps to improve the
voltage profile of the system.
 Including SVC devices reduce the system’s active
power losses.
 An important advantage of the proposed method
compared with the OPF is that it allows checking
the changes caused in the system by each
modification.
Volume 53, Number 1, 2012

The proposed methodology is easy to implement
and the control of the magnitudes can be
determined after each action decision.
REFERENCES
1.
2.
3.
4.
5.
6.
7.
8.
9.
M.E. Baran and F.F. Wu, Optimal capacitor placement on radial
distribution Systems, IEEE Trans. on Power Delivery. 4 (1989)
725-734.
Ji-Pyng Chiou, Chung-Fu Chang, Ching-Tzong Su, Ant direction
hybrid differential evolution for solving large capacitor
placement problems, IEEE Transactions on Power Systems. 19
(2004) 1794 – 1800.
D. Das, Reactive power compensation for radial distribution
networks using genetic algorithm, Electrical Power and Energy
Systems. 24 (2002) 573-581.
Delfanti. M, Granelli. G.P, Marannino. P, Montagna. M, Optimal
capacitor placement using deterministic and genetic algorithms
Power Systems, IEEE Transactions Power Systems. 15 (2000)
1041 -1046.
J.J. Grainger, S. Civanlar, S.H. Lee, Optimum size and location
of shunt capacitors for reduction of losses on distribution
feeders, IEEE Trans. Power Apparatus and Systems. 100 (1981)
1105-1116.
Y.C. Huang, H.T. Yang, and C.L. Huang, Solving the capacitor
placement problem in a radial distribution system using tabu
search approach, IEEE Trans. Power Systems. 11 (1996) 18681873.
G. Levitin, A. Kalyuzhny, A. Shenkman and M. Chertkov,
Optimal capacitor allocation in distribution systems using a
Genetic Algorithm and a fast energy loss computation
technique, IEEE Trans Power Delivery. 15 (2000) 623-628.
Rodrigo Palma-Behnke, Luis S.vargas, Juan Pérez .R and alls,
OPF with SVC and UPFC modelling for longitudinal systems,
IEEE Transactions on Power Systems. 19 (2004) 1742-1753.
Walid Hubbi, Takashi Hiyama, Placement of Static VAR
Compensators to minimize power system losses, Electrical
power systems research. 47 (1998) 95-99.
73
Prof. Fadela BENZERGUA
Prof. Naima KHALFALAH
Prof. Abdelkader CHAKER
Laboratory SCAMRE
Department of Electrical Engineering
ENSET Oran, Algeria
E-mail: benz_fad@yahoo.fr
Prof. José Luis MARTINEZ RAMOS
Prof. Jésus Manuel RIQUELME SANTOS
Prof. Alejandro MARANO MARCOLINI
Departamento de Ingeniería Eléctrica
Escuela Superior de Ingenieros
Universidad de Sevilla, Spain
Fadela BENZERGUA is a professor in the University Sience
and technologies USTO, Oran, Algeria. She received a doctorate in
science degree in electrical networks then university habilitation from
the University of USTO, Oran. Member of the “SCAMRE”
laboratory. Her search activities include the control of large power
systems, FACTS devices and wind farm energy.
Naima KHALFALLAH is a professor in the department of
electrical engineering at the ENSET, Oran, Algeria. She received a
Master degree in Electrotechnics from the ENSET. She prepares a
doctoral thesis in metaheuristics methods applied to the electrical
networks. Member of the “SCAMRE” laboratory.
Abdelkader CHAKER is a professor in the department of
electrical engineering of the ENSET, in Oran, Algeria. He received a
Ph.D degree in engineering systems from the university of SaintPetersbug. Director of “SCAMRE” laboratory. His search activities
include the control of large power systems, multiconverter systems
and unified power flow controller. His teaching includes neural
process control and real time simulation of power systems.