International Journal of Engineering and Technical Research (IJETR)

ISSN: 2321-0869, Volume-2, Issue-12, December 2014

Improvement of Voltage Stability by Optimal

Capacitor placement using Sensitivity Matrix
Gogu Rajendar, Basavaraja Banakara
It has been observed that the distribution system experiences
Abstract This paper presents a method to improve the voltage collapse under certain critical loading conditions,
Voltage Stability of the radial distribution system by Optimal Brownell and Clarke [1] have reported the actual system
Capacitor placement using L-Index Sensitivity Matrix. The phenomenon in which system voltage collapses periodically
proposed methodology calculates a sensitivity matrix, which and reactive compensation is to be supplied to avoid repeated
gives the capacitor location; it is effective than the conventional
voltage collapse. In distribution networks, voltage instability
optimization or meta-heuristic optimization usage. The objective
of the capacitor placement is to enhance the voltage stability.
is a local phenomenon and occurs near the buses which have
The proposed sensitivity matrix approach requires calculation high variation in loading and low-voltage profiles. So voltage
of the matrix once and using it for multiple load conditions, stability analysis is very much important in distribution
against the iterative solution approach of the optimization system as in transmission system for reliable and secured
methods. The characteristics of the radial distribution system is operation of the system.
considered and have been used for modifications in the present
approach. In Literature most of the work has been done on transmission
system, voltage stability analysis [2-10], but very less work
Index Terms Voltage stability, Sensitivity matrix, has been done on the voltage stability analysis in the radial
Capacitor Placement.
distribution system. G.B.Jasmon and L.H.C.C Lee, [11-13]
discussed about voltage stability analysis using various load
flow analysis and network reduction technique and
I. INTRODUCTION
maximizing voltage stability by loss reduction. In [14]
M.Moghavvemi, M.O.Faruque proposed a technique to study
According to voltage and power levels, an electric power the voltage collapse situation by reducing the network to a
system as classified into a generating system, a transmission single line equivalent circuit and then represented through a
system and a distribution system. In many utilities, even -network. A stability index is developed to indicate the
though new generators have been added to augment the severity of the loading situation of the system through which
existing capacity for meeting the increase in demand, but the voltage collapse is determined. An artificial Neural Network
transmission capacity has not changed much due to based technique [15] is used to predict the voltage stability
constraints like rights of way and cost of tower etc. Hence, and the learning is done is based on a new voltage stability
transmission systems are forced to carry increasingly more index for assessment of radial distribution systems taking
and more power over long distances using the FACTS loading values P and Q as the inputs. Catastrophic theory [16]
devices. Voltage stability analysis is important to assess the is proposed to identify how the equilibrium voltage point of
criticality of the existing networks and voltage Instability is the system is changed as the system parameters change during
a very challenging problem being faced by power systems sudden variations in system operation. This paper explores
all over the world. the applicability of the catastrophe theory to identify a new
The transmission system which is quite different, in both VSI (voltage stability index) and finding the critical voltage
its characteristics and operation, from the distribution system. and maximum loading for each bus in the system in terms of
We cannot use same analysis approach that used in one of the elementary catastrophe. In this paper different
transmission system to distributions system. The analysis of a indices which are used in voltage stability assessment of the
distribution system is an important area of activity, as radial distribution system is discussed.
distribution systems provide the vital link between the bulk
power system and the consumers. A distribution circuit II. VOLTAGE STABILITY INDICES
normally has a primary or main feeders which further rooted
into lateral distributors and sub lateral distributors. Radial
Voltage Instability is a very challenging problem being
systems are popular because of their simple design and
faced by power systems all over the world. In many utilities,
generally low cost.
even though new generators have been added to augment the
existing capacity for meeting the increasing load, the
transmission capacity has not increased correspondingly due
to problems like rights of way, environmental concerns etc.
Manuscript received December 02, 2014. Hence, transmission systems are forced to carry increasingly
Gogu Rajendar, Research Scholar,JNTUH, Electrical & Electronics
more and more power over long distances.
Engineering Department.
Basavaraja Banakara, Professor & Chairman (HOD), Electrical & An Index has to be calculated for assessing the voltage
Electronics Engineering Department, University of BDT Engineering stability and by ranking the buses, the most prone bus for
college Davanagere. voltage instability is chosen. This bus becomes the location

32 www.erpublication.org
 Improvement of Voltage Stability by Optimal Capacitor placement using Sensitivity Matrix

for the Capacitor to be placed in the system and the TABLE I. LINE FLOW INDICES OF 15- BUS SYSTEM
sensitivity indices for the buses remain same as long as the
system configuration remains the same. From To LFISP LFISQ LFIRP LFIRQ

1 2 0.1075 -0.1111 0.1048 -0.1082

A. Line Flow Indices:
2 3 0.0495 -0.0505 0.0483 -0.0493
The four line flow indices are calculated as follows.
Line flow index for the sending end real power: 3 4 0.0185 -0.0186 0.0180 -0.0182

4 5 0.0025 -0.0025 0.0017 -0.0017

r r
LFISP : 4 i2 Pr i2 Qi2
(1)
2 8 0.0081 -0.0081 0.0054 -0.0055
Vi Vi
Line flow index for receiving end real power: 8 9 0.0026 -0.0026 0.0018 -0.0018

ri r 2 6 0.0445 -0.0454 0.0299 -0.0305

LFIRP : 4 2
Pi i2 Qr2 (2)
Vi 1 Vi1 6 15 0.0055 -0.0055 0.0037 -0.0037
Line flow index for sending end reactive power:
6 7 0.0032 -0.0032 0.0021 -0.0021
Xi X
LFISQ : 4 Q 2i Pi 2
2 r
(3) 3 10 0.0166 -0.0168 0.0112 -0.0113
Vi Vi
Line flow index for receiving end reactive power: 10 11 0.0103 -0.0103 0.0069 -0.0070

Xi X
LFIRQ : 4 2
Qi 2i Pr2 (4)
11 12 0.0033 -0.0033 0.0022 -0.0022
Vi 1 Vi 2 4 13 0.0057 -0.0057 0.0039 -0.0039
where
Vi :- voltages at the sending end node i 4 14 0.0061 -0.0062 0.0041 -0.0042
Vi+1:-voltages at the receiving end node i +1
ri:-resistance of the line joining the nodes i and i +1
Xi:-reactance of the line joining the nodes
Pi :- real power flowing from node i B. L index:
Qi :- reactive power flowing from node i
Pr :- real power entering from node i +1 In this method the Y-bus matrix of the system is split
Qr :- real power entering from node i +1 into rows and columns of generators and load buses.
These line flow indices are calculated for all the lines in I G YGG YGL VG
the system and the lines with a high value (close to 1) are I Y (6)
considered as critical lines. The receiving end bus of the L LG YLL VL
critical line/lines is identified as the weakest bus from the
VL Z LL FLG I L
I K
voltage stability perspective.
YGG VG
(7)
Issues with Line flow indices: G GL
Consider the expressions for LFISP and LFIRP. If i, = 0 then FLG = [YLL]-1 [YLG] (8)
LFISP = 0 and LFIRP = 0 indicating that the lines are stable ig
Vi
in the case of real power flow which could be misleading. L j 1 Fji (9)
This is because, the power transferred over a line is given by i 1 Vj

ViVi 1 where the subscript,

Pr sin (5)
G:- refers to the generator buses in the system.
Xi
where :- difference in the angles of the sending and L:- refers to the load buses in the system.
receiving end voltages.
Say the resistance of a line feeding a PQ bus B of a An L-index value away from 1 and close to 0
system is zero. Let the maximum Pload at the bus B at a indicates a large voltage stability margin. The maximum
fixed value of Qload be Pmax. If the indices LFISP and of L-indices (Lmax) of the buses to which it corresponds is
LFIRP are calculated for the line 'L' under these the most critical bus. Also the summation of the squares of
conditions, the values turn out to be zeros, since the the L indices of individual buses (L2) is used as a relative
resistance of the line is zero. Hence the line L will be indicator of the overall voltage stability of the system at
classified as stable as per the real power loading based on different operating conditions.
this index whereas the system is in the verge of collapse.

33 www.erpublication.org
 International Journal of Engineering and Technical Research (IJETR)
ISSN: 2321-0869, Volume-2, Issue-12, December 2014
mode i are defined as kiik

TABLE II. L- INDICES OF 15- BUS SYSTEM

Bus no L-Index
2 0.0292
3 0.0448
4 0.0511
5 0.0521
6 0.0431
7 0.0445
8 0.0327
9 0.0339 Fig. 2. Participation factor of 15- bus system

10 0.0521
11 0.0567 III. SENSITIVITY MATRIX

12 0.0581
L- Index as introduced in section. 2. would predict voltage
13 0.0536 stability.
14 0.0538 To derive L-Index sensitivities Let {1,2,3.g} be no of
15 0.0455 generator nodes{g+1, g+2, g+3n} be no of load nodes
The L-index is,
ig
Vi
L j 1 Fji (13)
i 1 Vj
where the subscript
G :- refers to the generator buses in the system
L :- refers to the load buses in the system
Squaring the equation then,

2

2
ig
V ig
2
1 Fji i 1 F V i (14)
L j Vj ji
Vj
i 1 i 1
Put,

2
K L (15)
j j
2
ig
Vi
Fig. 1. L- index of 15- bus system.
K 1 F ji (16)
j Vj
i 1
C. Participation factor: Differentiating the above equation, the new equation is:
K j ig V ig V i
21 Fji i Fji 2 (17)
It is developed as follows. Consider the load flow
V j i 1 V i 1 V
Jacobian. j j

P H N
Q M
L V
(10)
P H N
Q M V
L
(18)
Although both P and Q changes affect system conditions,
it is possible to study the effect of reactive power
injections on the voltage stability by setting P (P As the inclusion of an extra capacitor would change only
constant) and deriving the Q V sensitivities at different the reactive power, we would make P in to be zero. From
loads. Thus this,
Q [ L MH 1 N ]V J R V (11)
J R (12)
Q L MH 1N V J R V V J R 1 Q (19
where )
: Left Eigen matrix of JR where (JR) is the reduced Jacobian.
: Right Eigen matrix of JR where Q is a vector of change in reactive power
injections
- Eigenvalues JR
The participation factors for the bus k and the critical
Q Qg 1Qg 2Qg 3......... '
(20)

34 www.erpublication.org
 Improvement of Voltage Stability by Optimal Capacitor placement using Sensitivity Matrix

IV. ALGORITHM FOR CAPACITOR PLACEMENT

where V is a vector of change in voltages
Calculate L- Index of all the load buses as given in

V Vg 1Vg 2Vg 3......... '
(21) section 2.
Calculate the sensitivity matrix as given in section 3.
ag 1, g 1 ag 1, g 2 ag 1, n Choose a limit for the L- Index, let it be L-limit

ag 2, g 1 ag 2, g 2 ag 2, n Find the buses which is having L- Index more than
J R1 (22)
L-limit, let the buses be [B2 B5 B7].
Find Lred, Lred is taken for the buses which are
a an , g 2 an , n
n , g 1 exceeding L-limit.
Lred = [L2 L5 L7].
i n
V j a ji Qi (23) Find the reduced sensitivity matrix (Lrq), reduced
i g 1 sensitivity matrix would relate Lrq and Qred where
Qred = [Q2 Q5 Q7].
Kj (24) Lred = LrdQred.
V j
ig
V ig V Now perform optimization minimizing objective
21 F ji i F ji i2

V j i 1 Vj
i 1 function satisfying the constraints
ig V ig V ig Lmin L Lmax
K j 21 Fji i 1 Fji i2 * a ji Qi (25)
i 1 V i 1 V i 1 where
j j

Lmin Lactual Llim it

As K j = 2L j L j

L
max [1] L
actual
ig ig ig
L j
1 1 F ji Vi 1 Fji Vi2 a ji * Qi (26) The optimal location of capacitor along with the Q value
Lj Vj Vj i 1
i 1 i 1 is found out.

1 V i g V i g
Lq 1 i 1 Fji i i 1 Fji i2 i 1 a ji
ig
(27) V. CASE STUDY AND RESULTS
L j V j V j
The method is carried out for finding the capacitor
This is called L- Index sensitivity matrix. location, in 15 bus system with four times the normal load
and the results are presented below.
TABLE III. COMPARISON OF L OBTAINED BOTH IN CONVENTIONAL
AND L-INDEX SENSITIVITY
Buses exceeding Llimi are [B4 B5 B10 B11 B12 B13 B14].

Bus L from L-Index L from conventional Lred = [L4 L5 L10 L11 L12 L13 L14].
no sensitivities method
TABLE IV. REDUCED SENSITIVITY MATRIX OF 15 BUS EXPERIMENT
2 0.001561375 0.00153235 SYSTEM

3 0.002982529 0.002926712 4 5 10 11 12 13 14
0.18 0.186 0.160 0.178 0.184 0.192 0.193
4 0.003036517 0.002977799 4
241 57 42 13 33 5 3
5 0.003045882 0.002986667 0.18 0.224 0.165 0.184 0.190 0.199 0.199
5
861 97 87 18 59 03 86
6 0.001625685 0.00159306 0.16 0.166 0.209 0.233 0.241 0.172
10 0.172
7 0.001632134 0.001599146 299 71 07 26 74 72
0.19 0.194 0.244 0.335 0.348 0.200 0.201
11
8 0.001577317 0.001547467 035 69 16 98 64 87 71
0.20 0.204 0.256 0.353 0.419 0.211 0.212
9 0.001582524 0.001552402 12
023 79 83 42 22 3 18
10 0.004477482 0.004394593 0.19 0.202 0.173 0.192 0.199 0.258 0.209
13
758 08 77 95 66 67 37
11 0.006484893 0.006368077
0.19 0.203 0.174 0.194 0.200 0.209 0.237
14
12 0.008109174 0.007966572 88 33 84 14 9 8 82

13 0.003058382 0.002998492
14 0.003059996 0.003000017 Lmax = [0.53262 0.54971 0.55208 0.63254 0.66014
15 0.001636937 0.001603675 0.57359 0.57676]

35 www.erpublication.org
 International Journal of Engineering and Technical Research (IJETR)
ISSN: 2321-0869, Volume-2, Issue-12, December 2014
Lmin = [0.032616 0.049709 0.052083 0.13254 0.16014 [7]. W. Xu and Y. Mansour, Voltage stability analysis using generic
dynamic load models, IEEE Trans. on Power Systems, vol. 9, pp.
0.073588 0.076765] 479-493, Feb. 1994.
[8]. O. Crisan and M. Liu, Voltage collapse prediction using an improved
Imposing the condition Lmin L Lmax and
sensitivity approach, Electric Power Systems Research, vol. 28, no.
running optimization program we get values of capacitors 3, pp. 181-190, Jan. 1994.
as given below: [9]. J. Lu, C. W. Liu, and J. S. Thorp, New methods for computing a
saddle-node bifurcation point for voltage stability analysis, in IEEE
TABLE V. CAPACITOR PLACEMENT FOR VOLTAGE STABILITY Trans. on Power Systems, vol. 10, pp. 978-89, May 1995
IMPROVEMENT
[10] T. Van Gutsem and R. Mailhot, Validation of a fast voltage stability
Bus no Value of Capacitor analysis method on the Hydro Quebec system, IEEE PES Winter Power
Meeting, 96 WM 280-8-P W RS
4 2.3036e-011
5 1.8078e-011
10 1.6882e-011
11 1.5086e-010
12 0.38191
13 1.2864e-011
14 0.00016339

The values of capacitor to be placed at bus 12 is

0.38191MVAR and at bus 14 is 0.00016339MVAR. This
would reduce the values of L- Index to maximum of 0.5.
This would make the system stable.

VI. CONCLUSION
This paper emphasis on the analyzing and improvement of
voltage stability of radial power systems. The various indices
which have been proposed for assessing voltage stability are
studied. Patterns of voltage stability indices are been
observed and their dependence on the electrical distance has
been studied. An algorithm is proposed for optimal placement
of the capacitors to improve the voltage stability to the
required level based on the new sensitive index proposed. The
effect of reactive power injections at a bus on the entire power
system in consideration is studied. The algorithms which have
been proposed for radial power systems can also be suitably
modified and can be applied to meshed systems.

REFERENCES

[1]. Brownell G, Clarke H. Analysis and solutions for bulk system

voltage instability IEEE Computer Applications in Power vol. 2(3)
pp. 31-35. 1989
[2]. Drexel University, Performance Indexes for Predicting Voltage
Collapse, Final Report EPRI EL-6461, Research Project RP1999-10,
July 1989
[3]. P-A. Lof, T. Smecl, G. Anderson, and D. J. Hill, Fast calculation of a
voltage stability index, IEEE Trans. on Power Systems, vol. 7, pp.
54-64, Feb. 1992.
[4]. M. M. Begovic and A. G. Phadke, Control of voltage stability using
sensitivity analysis, IEEE Trans. on Power Systems, vol. 7, pp.
114-123, Feb. 1992.
[5]. G. K. Morison, B. Gao and P. Kundur, Voltage stability analysis
using static and dynamic approaches, IEEE Trans. on Power
Systems, vol. PWRS-8, no. 3, pp. 1159-1171, Aug. 1993.
[6]. P.A. Lof, G. Anderson, D.J. Hill: Voltage Stability Indices for
Stressed Power Systems, IEEE Trans. on Power Systems No. 1,
1993;

36 www.erpublication.org