You are on page 1of 6

2004 InternationalConference on

Power System Technology POWERCON 2004
Singapore, 21-24 November2004

Monitoring the effects of On-Load

Tap Changing Transformers on Voltage stability
D.Thukaram, Senior Member, IEEE, Lawrence Jenkins, Senior Member IEEE,

H. f.Khincha, Senior Member, IEEE, G . Yesuratnam, B. Ravi Kumar

power capability limits. The process will eventually may lead
Abstract-- It is becoming increasingly important for power to voltage collapse. Thus the operation of on-load tap changers
system planning and operating engineers to be capable of (OLTCs) has a significance influence on voltage instability.
performing comprehensive voltage stability analyses of the Zhu et al. [l] addresses the effects of OLTC operation on
systems. This need is largely due to recent trends towards
operating systems under stressed conditions as a result of
voltage collapse from the point of view of how the limit of
increasing system loads without sufficient transmission and/or power transfer from the generation to the load center can be
generation enhancements. This paper discusses effect of the On- affected by OLTC operation. Isaias et al. [2] shown that
load tap changing transformem (OLTCs) on voltage stability and blocking the tap changes in the OLTCs may prevent voltage
identifying critical OLTCs to avoid possibilities of voltage collapse, while maximizing the load recovered. For this sake,
instability conditions. The developed approach bas been tested on the blocking time is determined by a tangent vector - based
sample systems. Results obtained on an equivalent system of 24-
index. The index proposed is based on voltage level variation
node practical power system are presented.
at the bus controlled as a function of tap changes. Costas
Zndex Terms-- Voltage stability, OLTC transformers, Vournas et al. [3] presents emergency control of bulk power
delivery load tap changers focusing on tap reversing, when
I. INTRODUCTION transmission voltage becomes low. In this an algorithm is
HE voltage stability problem has become a major concern suggested for the determination of optimal LTC settings that
T in operating and planning today's power system as a result
of heavier loading conditions, without sufficient transmission
maximize the loadability margin for a specific active power
pattern. Mamandur et al. [4] addresses the effect of the control
and/or generation enhancements. Voltage stability problems variables (transformer taps) on the system losses. Thukaram et
normally occur in heavily stressed systems. While the al. [ 5 ] is mainly concerned with analysis and enhancement of
disturbance leading to Voltage collapse may be initiated by a steady state voltage stability based on L-index an algorithm is
variety of causes, the underlying problem is an inherent proposed for optimization of reactive power control variables
weakness in the power system. The factors contributing to using Linear-programming technique. This paper presents an
Voltage collapse are the generator reactive power Noltage approach for monitoring the on - load tap changing
control limits, load characteristics, characteristics of reactive transformers effect on voltage stability and identifying the
compensation devices, and the action of the voltage control critical transformers which may lead to voltage instability
devices such as transformer on - load tap changers (OLTCs). under peak load conditions. The approach is based on voltage
Power system experiencing an abnormal operating conditions stability L-index. The proposed approach is tested on few
following a disturbance, the EHV level voltage reduction at sample systems and results obtained on a 24-node EHV 400kV
load centers would be reflected into the distribution system. Indian power system network.
The OLTCs of distribution transformers would restore
distribution voltages, with each tap change operation, the n. APPROACH
resulting increment in load on EHV lines would increase the Network security analysis and optimization form the core
MWs, MVARs, losses, which intern would cause a great drops functions in a modem energy control center (ECC). The
in EHV levels. As a result, with each tap - changing operation, system statu$ is obtained from the output of an on-line state
the reactive output of generators throughout the system would estimator. The output of the state estimator is checked for limit
increase gradually, the generators would hit their reactive violations and if the system is insecure then corrective control
action has to be taken. If the system is operating in a secure
D. Thukaram is with Department of Electrical Engineering, Indian state, then the system is subjected to a subset of credible
Instirule of Science. Bangalore-560012, INDM ( contingencies and the system status is evaluated. If after a
Lawrence Jenkins is with Department of Electrical Engineering, Indian
Institute of Science, Bangalore-560012, INDIA ( contingency study the system status indicates insecure
H. P. Khincha is wirh Department of Electrical Engineering. hdim operation, then suitable preventive control action is
Institute of Science, Bangalore-560012, NDIA ( formulated. If voltage violations arc present, then an optimal
G. Yesuratnam is with Department of Electrical Engineennp, Indian
Institute of Science, Bangalore-560012, INDIA (mtnanipy2003 reactive power dispatch is obtained. The objectives generally
B. Ravi Kumar is with Department of Electrical Engineering, Indian used are:
Institute of Science. BangaIore-56W12. INDIA ( r ~ v i ~ e e . i i s c . e ~ e t , j n )

0~7803-8610-81041$20.00 0 2004 IEEE 419

Minimization of real power losses ( P d in the system or
Minimization of voltage deviations from desired voltage (3)
values (Vdcrid).
where LG = -[ LL]-' [y ]
LG are the required values. The
Both the objectives achieve their desired minimum by
optimal control of the reactive power control variables .The L-indices for a given load condition are computed for all load
controls recommended while achieving this objective may in busses.
certain cases actually deteriorate the system voltage stability For stability, the bound on the index Lj must not be violated
condition under stressed conditions, especially when using (maximum l i m k l ) for any of the nodes j . Hence, the global
OLTC transformers. While OLTC transformers improve the indicator L describing the stability of the complete subsystem
voltage profile at the load end (especially with objective is given by kmaximum of Lj for all j (load buses). An L-index
Vdesired) by change of the tap position (boost to low voltage value away from 1 and close to zero indicates an improved
side), this may lead to voltage instability conditions due to system security. For a given network, as the load/generation
inadequate reactive power support. Hence, from the system increases, the voltage magnitude and angles change, and for
security point of view, an objective function, which near maximum power transfer condition, the voltage stability
incorporates improvement of the voltage stability margin, is index Lj values for load buses tend to close to I, indicating that
found to be necessary. Since the L-index value indicates the the system is close to voltage collapse. The stability margin is
proximity of the system to voltage collapse; we have selected obtained as the distance of L from a unit value i.e. ( 1 4 ) .
minimization of the sum square of L - indices as the objective
(Valabi~Ly).T h i s objective recommends optimal control of the N.DESCFUPTION OF THE REACTIVE POWER OFITMEATION
reactive power control variables such that the overall system PROBLEM
stability is improved. The model selected for the reactive power optimization
In case of objective V k s a or P r
, the controller action, uses linearized sensitivity relationships to define the
especially the direction of tap change recommended may be optimization problem. The objective is to minimize the voltage
opposite to the tap change direction recommended by the
stability objective function v, = CL' in the system. The
proposed objective. This is because objective Vdesired or
Ploss tries to improve the secondary side voltage by constraints are the linearized network performance equations
recommending tap variation towards the 0.9 setting , but relating the control and dependent variables and the limits on
under heavy load conditions this may actually deteriorate the the control and dependent variables. The control variables are:
system voltage st3bility margin. Therefore, the proposed The transformer tap settings (T),
objective also helps in identifying critical OLTCs, which The generator excitation settings (V),
should be made manual to avoid possibilities of voltage The switchable VAR compensator (SVC) settings (Q).
instability conditions due to the operation of OLTC These variables have their upper and Iower limits. Changes
transformers based exclusively on voltage stability criteria. in these variables affect the distribution of the reactive power
and therefore change the reactive power at generators, the
m. VOLTAGE STAB- ANALYSIS USING L INDEX voltage profile and thus the voltage stability of the system. The
Consider a system where, n=total number of busses, with 1, dependent variables are:
2... g generator busses (g), g+l, g+2 ... g+s SVC busses ($1, The reactive power outputs of the generators (Q),
gts+l ... n the remaining busses ( S n - g - s ) and t =number of The voltage magnitude of the buses other than the generator
OLTC transformers. buses (V).
A load flow result is obtained for a given system operating These variables also have their upper and lower limits. In
condition, which is otherwise available from the output of an mathematical form, the problem is expressed as:
on-line state estimator. Using the load flow results, the L-index
[6] is computed as Minimize,
= cx
Lj = i=l
wherej=g+l ... n and all the terms within the sigma on the RHS
Subject to,
Ib = sx Ib-
of (1) are complex quantities. The values Fji are obtained from and

the Y bus matrix as follows

]: E] (2)
5x2 X"

Where C is the row matrix of the linearized loss sensitivity

coefficients, S the linearized sensitivity matrix relating the
where IC, I L f and V G, V represent currents and voltages dependent and control variables, b the column matrix of
at the generator nodes and load nodes. Rearranging (2) we get linearized dependent variables, x the column matrix of the
linearized control variables, bmaaand b"'" are the column
matrices of the linearized upper and lower limits on the

dependent variables and 1"" and 1"'" are the column matrices B. Computarionof voltage stability objective frmction
of linearized upper and lower limits on the control variables. ( vL = L2 ) sensitivities (C) with respect to control
The linear programming technique is now applied to the
above problems to determine the optimal settings of the
control variables. The sensitivities of the voltage stability objective function
The control vector in incremental variables is defined as ( v r ) with respect to the real and reactive power injections at all
the buses except the swing bus (angle reference bus) are first
x =[AT, ?. .+,AT,+,
,..,,AT ,Ay ..?A& ,Mg+l,. 7
computed and these values are used to compute the objective
function sensitivities with respect to the control variables.
(7) Considering the fact the real power injection does not change
and the dependent vector in incremental variables as
for a small change in voltage magnitude of the bus and the
]I , A ~ power injection at a bus does not change for a small
b=[AQl y..., A ~ , A ~ + ~ y , . . ? A ~ + ~ + ~(8), . .- .reactive
change in the phase angle of the bus voltage, the relation
The upper and lower limits on both the control and between the sensitivities of the objective function with respect
dependent variables in linearized form are expressed as to the real and reactive power injections at all the buses except
b" =lAQ;D" ,... y A ~ ~ " , A Vr*. ~. . , A V ~ ~ 7 A V ~ ~ + i ,
aVL- -a4
..., A V , i i , l t -- - ap, --avL
... - -
bm =[AQlm" ,...,AQgmi"yAVgT,..., AV$,AV8~l;ntl, as, as, as, ap,
.. -- .. ... *
xmax= [AT,- .. .,AT,"", AV," ,...,AVx-, A Q g y , avL -
- apz ... ap. -
- avL
...,AQg% ,I'
- as, - - aa,, as, - - ap. -
xmin= [AT,"'",. .., A T ~ d n ,...,AV,.On,AQg$ ,
. . . , AQg,l'
(9) ...
ATmx = Tm" - T a c r , ATmin = Tmin -Tuct ...
A Q w = Q m x Qocf, I
AQmin = Qmin - Qacr
AV" = Vim - V a c r , AV& = V d U -V Q C t Knowing the terms av@, dvL/av, d P p 8 , and
aQ/aV in the above relation, the sensitivities of the objective
A. Computation of sensitivity mtrix (S) function with respect to the real and reactive power injections
The sensitivity matrix S relating the dependent and control at all the buses except the swing bus,
variable is evaluated [7] in the following manner. Considering &,/ap, ,av, fdQ,, k=2 ...n @us 1 is considered as a
the fact that the reactive power injections at a bus does not reference bus) can be computed. From the solution of the
change for a small change in the phase angle of the bus above the objective function sensitivities with respect to
voltage, the relation between the net reactive power change at transformer taps and generator excitation voltages can be
arty node due to change in the transformer tap settings and the calculated using the equations given below.
voltage magnitudes can be written as

A9 A10 A l l A12

Then, transferring all the control variables to the right hand

side and the dependent variables to the left hand side and The values for L a r e obtained from the solution of (1 3)
rearranging: aQ&
C. Computational Procedure
In the day-to-day operation of the power systems, the
following are the steps used to obtain the optimal reactive
power allocation in the system for improvement of voltage

Step I: Perform the initial operationai load flow (or output of Case I -Effect of tup - controllersettings on the system:
state estimation) to obtain the values of L-index at
each load bus and check for voltage violations in the In this case, tap setting of each transformer is varied in its
system. full range from lower limit (0.9 p a ) to the upper limit
Step 2: Advance the VAR control iteration count. (1.1p.u.) in steps of 0.0125 by keeping the tap settings of the
Step 3: Compute the column matrices (b-, bdn) of the remaining transformers at nominal value (1.0 pa.), under peak
linearized upper and lower limits on the dependent load conditions. For each tap setting of controlling transfomer
variables. the power flow solution and voltage stability indices of all the
Srep 4: Compute h e column matrices (x-, x ~ , Jof the load buses for the system are obtained. The corresponding
linearized upper and lower limits on the control results of voltage magnitudes and voltage stability indices (L
variables. values) are shown in Fig.2 and Fig.3 respectively.
Step 5: Modify the matrices x, and x ~ , ,to reasonably small
Step 6: Compute the sensitivity matrix (S), relating the
dependent variables and the control variables.
Step 7: Compute the row matrix (C) of the objective function
sensitivities with respect to the control variables.
Step 8: Solve the optimization problem using the linear
programming technique.
Step 9: Obtain the optimum settings of control variables.
Step 10: Perform the operational load-flow with the optimum
settings of the control variables. Find the L-index for
all the load buses in the system.
S f q 11: Check for satisfactory limits on the dependent
variables. If no, go to step 2.

v. RESULTS AND DISCUSSIONS 0.95 1 1M 1.1 1.15

Tap settings
The proposed approach has been tested on a 24-node EHV
400kV Indian power system network shown in Fig.1. The Rg. 2. VoQagesat the controlled bus= (in p.u,) wirh respect to tap semings
system has seven 400t220 kV tap changing transfor", 4 .
generator buses and 20 other buses. To identify the critical n.7
OLTC transformers in the system under peak load conditions
only OLTC transformers were used as the control variables.

Tap senings

Fig. 3. L index values at the controlled buses with respecc to tap settings

From these results it is clear that the effect of tap changing

transformer connecting between the buses 14 and 8 (near the
load bus 8 ) is different from other transformers. The voltage
at the bus 8 is very low (0.808 P.u.) and the voltage stability
index (L index) value is very high (0.639) even at the limit 0.9
tap setting of controlling transformer T14.8.
The effect of tap setting of each transformer on the all
generator reactive power outputs is obtained from power flow
Fig. 1. A Practical system.

solution. The results are shown from Fig.4 and Fig. 5
corresponding to Generator GI and Generator G3 respectively.
From these results it is clear that the transformer connecting
between the buses 14 and 8 (T,&S) drawing more reactive
=--I P
power from the generators in comparison to other
transformers, when the tap is moved from nominal 1.0 towards
0.9 setting. Hence this transformer i s critical for the system.

Tap settings

Fig. 5. Reactive Power output of Generator 0 3 with respect to various

controlling transfonners.

0.95 1 1.E 1.1 1.15

Tap settings

Fig. 4. Reactive output of Generator G1 with respect to various contolling



0 1 .o 1 .o 1.0 1.os 1.o. 1.o - 2.4217

1.0 2.42
1 0.9875 0.9875 0.9875 1,0125 0.9875 0.9875> 0.9875 2.3577 2.38
2 0.975 0.975 0.975 1.025 0.975 0.975 0.975, 2.2963 2.35
3 0.9625 0.9625 0.9625 1.0375 0.9625 0.9625 0.9625 2.240 2.33
4 0.950 70.950 0.950 1.050 0.950 0.950 0.950 2.1863 2.39
5 0.9375 0.9375 0.9375 1.0625 0.9625 0.9625 0.9375 2.1405 2.29
6 0.950 0.925 0.925 1.050 0.935 0T975 .0.925 2.1089 2.27
7 0.9625 0.9375 0.9125 1.0375 0.9625 0.9625 0.9125 2.0798 2.26
8 I 0.950 I 0.925 I 0.925 1 -1.025 I 0.975 1 0.975 I 0.9000 I 2.0512 2.24
Table I. All the transformer taps are at nominal value (1.0)
Case 2. Critical OLTC idenrijkdon using optimization initially. It is observed from the VAR control iterations, that
the objective VsmbiIily recommends tap variation from nominal
To identify critical OLTC transformers in the system under 1.0 towards 0.9 at all OLTC transformers except for
peak load condiiions only OLTC transformers were used as Transformer between the buses 14 and 8 The optimum
control variables. Results were obtained using two different movement for the transformer Tlbe tap is from nominal 1.0
objective functions: towards 1.1. Hence OLTC T14.8 is critical for the system and
Minimization of voltage deviations from pre-desired values may be made manual to avoid possible voltage instability due
(Objective Vdesired); to the operation of the OLTC transformer to improve the
Minimization of the sum square of L-indices for the system voltage at the secondary side of the transformer.
(Objective V stability).
Results for the VAR control optimization are presented in

The problem of voltage stability in power system focusing on D. Thukaram (SM, 90) received the B.E. degree in
the effect of On Load Tap Changing transformers i s Electrical Engineering from Osmania University,
Hydembad in 1974, M.Tech degree in Intepted
discussed. An approach is presented for monitoring the on - Power Systems from Nagpur University in 1976 and
load tap changing transformers effect on voltage stability and Ph.D. degree from Indian Institute of Science,
identifying the critical transformers which may lead to voltage Bangalore in 1986. Since 1976 he has been with
instability under peak Ioad conditions. The approach is based i
Indian Institute of Science as a research fellow and
faculty in various positions and currently he i s
on voltage stability L- index. The effect of OLTCs on voltage Professor. His research interests include computer
stability are simulated and explained for a 24 - node practical aided power system Analysis, reactive power
power system.. The studies show that under varying load optimization, voltage stability, distribution
conditions the OLTC directions obtained with different automation and AI applications in power systems.
objectives are similar, if the system conditions are not close to Lawrence Jenkins received the B.Tech degree in
critical voltage instability situation. However, as the system Electronics from IlT Madras in 1972. In 1974 and
conditions are approaching closure to voltage instability 1976 he received MS and Ph.D. d e g m both in
conditions, sensitivities of some critical transformers may be Electricat Engneering from the University of Nortre
Dame, USA. Since 1979 he bas been with Indian
reversed. Institute of Science, Bangalore where currently he is
Professor. His research interests include power system
VrI. REFERENCES analysis, fault tolerant and real time systems.
T. X. nu,S . K. lso, and K L Lo, A0lnvesdgadon into the OLTC
effects on Voltage Collapse, IEEE Trans. on Power Systems, Vol. 15,
H. P. Khincha (SM. 82) received the B.E degree in
no. 2, May 2000.
Electncal Engineenng from Bangalore Umversity in
B. Isaias Lima Lopes, and A.C. Zambroni de Souza, On multiple tap
1966 He received M E . d e g m in 1968 and Ph.D.
blocking to avoid voltage collapse, Electric Power Systems Research,
degree in 1973 both m Electncal hgneenng fmm
Vol. 67,2003, pp. 225-231.
the Inman Inshtute of Science, Bangalore Since 1973
Costas Vownas. and Michael Karystianos. Load Tap Changers in

he has been with h&an Institute of Science,
Emergency and Preventive Voltage Stability Control, IEEE
Bangdore as faculty where currently he IS Professor.
Transacfionson Power Sysfems, Vol. 19, no. I , Feb 2004.
HISresearch interests include computer nded power
K R. C. Mamandur, and R. D. Chenoweth, Optimal control of
system analysis, power system protechon, mstnbution
Reactive power flow for improvements in voltage profiles and for real
automation and AI apphcations in power systems
power loss minimization IEEE Trunsuctiom on Power Appararus and
Syslemr. Vol. PAS-lOO,no.7. July 1981.
G. Yewrainam received the BTech degree in
D. Thu!amq K. Parthasarathy, H. P. Khincha, Nmndranath Udupa
EIectncaI and EIectromcs Engneenng from
and A Bansilal. Voltage Stability improvement: case studies of lndian
power networks kcm*cpower sysrems reseurch Vo1.44, 1998, pp. 35-
lawaharlal Nehru Technologcal Umversiry,
44. Hyderabad, In&a in 1995. He recaved M.Tech degree
in 1998 form National Institute of Technology.
P. Kessel and H. Olavitsch, Esrimating the voktage slabitity of a power
Warangal in the filed of Power Systems. Currently he
system, IEEE Trarrs. Power Delivety. Vol. PWRD-1(3) July 1986, pp.
is worhng towards his PhD III the Lkpamnent of
Elecmcal Engneenng, h&an Institute of Science,
D. Ihukaram and K. Paahasarathy. D. L,. Prior. Improved algorithm
for optimum reactive power allocation, Int. l o u m l ofElecrnc. Power.
Bangalore. HIS research mterests include computer
aided power system Analysis, reactive power
Energy SysSystem., Vol. 6, no.2. 1984, pp. 72-74.
optImzahon, voltage stabhty and AI apphcations m
power systems.

B. Rnvi Kumar received &e B.Tech degree in

Electncal B Elecnomcs E n g m m n g from Nagaguna
Utuversity (AI), In&a m 2002. Currently he is
pursuing his M.Sc (Engg) degree III Elecmcal
En@ne&ng at the Ind~an Insntute of Science,
Bangalore. Kls research mterests incfude computer
aided power system analysis, dmnbution automanon,
Al techques and apphcauons.