International J ournal of Energy Engineering (IJ EE) Nov. 2012, Vol. 2 Iss. 4, PP.

126-136
- 126 -
A Fast Continuation Load Flow Analysis for an
Interconnected Power System
D. Hazarika
Assam Engineering College, Gauhati-13, Assam, India
dlhazarika@sify.com
Abstract- The paper provides an algorithm for a fast continuation
load flow to determining critical load for a bus with respect to its
voltage collapse limit of an inter connected multi-bus power
system using the criteria of singularity of load flow Jacobian
matrix. For this purpose, load flow Jacobian matrix of an
interconnected multi-bus power system is transformed into a two
by two elements Jacobian matrix with respect to a target/selected
bus by incorporating the effect of all the other buses of the system
on the target bus. The validity of the proposed method has been
investigated for the IEEE 30 and IEEE 118 bus system.
Keywords- Voltage Stability; Load Margin; Voltage Collapse;
Power System
List of symbols
N =Total number of buses in the system.
NG =Total number of generation buses in the system.
npv =Total number of PV buses (including slack bus) in
the system.
Pi =Injected active power at ith bus.
Qi=Injected reactive power at ith bus.
PDi =Active power demand at ith bus.
QDi=Reactive power demand at ith bus.
PGi =Active power generated by ith generator.
QGi=Reactive power generated by ith generator.
Vi =Magnitude of voltage at ith bus.
i = Angle of the bus voltage at ith bus.
Cos i =Load power factor of the ith load bus..
Gij +jBij =Element of Y-BUS matrix at ith row and jth
column.
I. INTRODUCTION
Information/knowledge of load margin of a bus with
respect to its voltage collapse limit constitutes important
criteria for load pick-up step during power system restoration
planning and operation. In a situation, when a load bus has to
be energized through network building process (i.e.,
connecting the load to energized bus through transmission
line(s)) and several energized buses are available for the
connection, it becomes essential for power system
planner/operator to select a proper energized bus for this
purpose. Under such circumstances, the bus having highest
load margin appears to be the best choice
[1]
and magnitude of
load to be picked-up must be less than the load margin for the
connected bus. Voltage collapse of a bus is characterized by a
slow variation in system operating point, due to increase in the
loads, in such a way that the voltage magnitude gradually
decreases until a sharp accelerated change occurs. It has been
observed that voltage magnitudes in general, do not give a
good indication of proximity to voltage stability limit
[2]
. In
recent literature many voltage stability and voltage collapse
prediction methods have been presented. Some of the
important ones are: voltage collapse index based on a normal
load flow solution
[3, 4, 5]
, voltage collapse index based on
closely located power-flow solution pairs
[6]
, voltage collapse
index based on sensitivity analysis
[7]
, and minimum singular
value of Newton - Raphson power flow J acobian matrix
[8, 9]
.
Continuation power flow analysis is based on locally
parametrized continuation technique
[10]
. The key idea here is
to avoid the singularity of the J acobian by slightly
reformulating the power flow equations, and applying the
locally parametrized continuation technique
[11, 12]
. The method
uses predictor-corrector approach and some parameter
(continuation parameter) as additional condition to overcome
ill conditioning of J acobian matrix. Network partitioning
technique has been employed in determination of load margin
at a load bus
[13]
. A two parameter continuation technique has
been proposed for evaluating branch outage contingencies
introducing new branch parameter
[14]
.
A voltage stability index VMPI (voltage margin proximity
index) is proposed considering voltage limits, especially lower
voltage limits
[15]
. A simple, computationally very fast local
voltage-stability index has been proposed using Tellegens
theorem
[16]
. It is easy to implement in the wide-area
monitoring and control center or locally in a numerical relay.
A new node voltage stability index called the equivalent node
voltage collapse index ENVCI), which is based on ESM and
uses only local voltage phasors, is presented
[17]
. Well-known
ZIP model has been used to represent loads having
components with different power to voltage sensitivities
[18]

and also, the choice of voltage stability index in the context of
load modeling has been suggested.
These methods assess the closeness to the critical loading
by looking into the voltage stability sensitivity indices or the
smallest Eigen-value or singular value of load flow J acobian
matrix. The performance indices are used to provide the
measure of the severity of static voltage stability problem.
However, the disadvantages with the performance indices are
that they cannot directly provide the margin of load at a load
bus with respect to its voltage collapse limit.
International J ournal of Energy Engineering (IJ EE) Nov. 2012, Vol. 2 Iss. 4, PP. 126-136
- 127 -
The proximity of voltage collapse in a power system is
indicated by the condition of Newton-Raphson power flow
J acobian matrix. At the point of voltage collapse, the
determinant of the J acobian matrix of a Newton-Raphson
power flow analysis becomes zero. As such, the elements of
load flow J acobin matrix can be used to indicate voltage
instability problem of a power system. This paper proposes a
continuation load flow for determining the load margin of a
bus with respect to its voltage collapse limit. For this purpose,
load flow J acobian matrix of an interconnected multi-bus
power system is transformed into a two by two elements
J acobian matrix with respect to a target/selected bus. The
elements of the transformed two by two matrix and the bus
voltage of the target bus are used to develop an algorithm to
determine load margin of a target bus of the system.
II. ANALYSIS OF VOLTAGE STABILITY PROBLEM FOR A MULTI-
BUS INTER CONNECTED POWER SYSTEM
It is established that at the point of voltage collapse in a
power system, the determinant of load flow J acobian matrix
becomes zero. To quantify the condition of singularity of load
flow J acobian matrix of a multibus system at the point of
voltage collapse, it is transformed into a two by two matrix
relating k, Vk, Pk and Qk for a target kth bus. The
change in voltage magnitudes and voltage phase angles of a
multi-bus inter connected power system are related to change
in real and reactive power injection at the buses as follows:

Where H, N, M, and L are the elements of J acobian matrix
used for the load flow analysis of Newton Raphson method.
Equation (1) can be expressed as:

For the convenience of representing the system (load flow)
equation, the following assumptions are made:
(i) bus Number 1 is considered as the slack bus; (ii) the
system has npv number of voltage control buses which
includes the slack bus also, i.e., the system has (npv1)
number of PV buses.
Now, using the elements of W, X, Y and Z of Equation (2)
the change in voltage phase angle and voltage magnitude of a
kth target bus can be expressed as:

The right hand side of the Equations (3) and (4) contain
change in real and reactive power terms for all buses including
the kth target load bus. To relate k and Vk of Equations (3)
and (4) to only Pk and Qk, it is required to transform the
Load Flow J acobian matrix of an interconnected system into a
two by two matrix with respect to a target bus k. The paper
presents a sensitivity analysis approach to relating change in
real and reactive power of the target bus due to change in real
and reactive power in other buses. Recognizing the fact that
the change in bus voltage angle is primarily influenced by the
change in real power injection at the buses
[19]
, the change in
real power injection at busbars can be expressed as
[20]
:


Similarly, for other buses the relation among voltage phase
angle and real power injections at the buses can be expressed
as:


Equation (7) can be rearranged as given below:

where;

From Equation (6) and Equation (8) we have,


International J ournal of Energy Engineering (IJ EE) Nov. 2012, Vol. 2 Iss. 4, PP. 126-136
- 128 -
The Equation (10) can be rearranged as follow:


[] is a row matrix, therefore, [P] values are to be
calculated using pseudoinverse technique i.e.
[20]
,

Equation (12) is solved for terms Pi for i = 2...N for
change in Pk and expressed as:

Similarly, recognizing the fact that the change in bus
voltage angle is primarily influenced by the change in real
power injection at the buses
[19]
, the change in real power
injection at busbars can be expressed as:


The change in slack bus reactive power in terms of change in
voltage phase angles can be expressed as:

Similarly, for other buses (excluding PV buses, where voltage
magnitudes are assigned as fixed value) the relation among
voltage phase angle and real power injections at the buses can
be written as:

Equation (16) can be rearranged as given below:

where,

From Equation (15) and Equation (17) we have,



The Equation (19) can be rearranged as follow:


[] is a row matrix, therefore, [Q] values are to be calculated
using pseudoinverse technique i.e.
[20]
,
International J ournal of Energy Engineering (IJ EE) Nov. 2012, Vol. 2 Iss. 4, PP. 126-136
- 129 -


Substituting values of Pi and Qi from Equations (13) and
(22) in Equations (3) and (4) we have

where,

Therefore,

At the point of voltage collapse

The element of the transformed two by two matrix for the
kth target bus are function of network state namely bus
voltages and their phase angles.
III. DETERMINATION OF LOAD MARGIN
In this section, a procedure for determination of load
margin of the target kth bus has been developed using the
elements of the modified two by two Jacobian matrix given by
the Equation (28) and the bus voltage of the target bus.
Equation (28) can be represented as two bus system with the
target kth load bus, Y-Bus elements and an equivalent source
as depicted in Figure 1.

Fig. 1 Equivalent two bus systemwith target kth bus, Y-Bus elements and an
equivalent source
Now, for this equivalent two bus system, the expression for
real and reactive power injections at the target kth bus are as
follows:

Where, Gkk, Bkk, GkG and BkG are the elements of
admittance matrix [Y] for the equivalent two bus system. But,
the values of Gkk, Bkk, GkG and BkG depend on the system
bus voltages and their phase angles. Therefore, while
developing the mathematical model for load margin for the kth
bus, parameters Gkk, Bkk, GkG and BkG are represented in
terms of the elements of transformed two by to matrix
represented by Equation (29).
The elements of J acobian matrix for the equivalent two bus
system can be expressed as

At the point of voltage collapse the determinant of the
J acobian matrix becomes zero, i.e.,

The change in determinant value of the two bus equivalent
system of the multibus system with respect to change in Vk
and k can be expressed as:

Again, the expression for JD in terms of




The terms are derived
utilizing equations from (32) to (35). It is to be ensured that
the terms contain only the
elements of the modified two by two J acobian matrix and the
bus voltage of the target bus. Because, they are the known
values for the transformed two bus system.
International J ournal of Energy Engineering (IJ EE) Nov. 2012, Vol. 2 Iss. 4, PP. 126-136
- 130 -

Now, applying values of of
Equations (40), (41), (42) and (43) in Equation (39) we have,


Similarly, the term can be expressed as:

Again, utilizing equations from (32) to (35), we have

Processing Equation (33) and Equation (34), Gkk can be
expressed as:

Processing Equation (32) and Equation (35), Bkk can be
expressed as:

Now, applying values of of
Equations (46), (47), (48), (49), (50) and (51) in Equation(45)
we have,

The terms of Equations (44) and
(52) are the elements of the modified J acobian matrix of a
multibus system represented by Equation (28) and Vk is the
bus voltage of the kth target bus. Now, to relate change in real
power injection at target kth bus to the change in determinant
of the modified J acobian matrix of a multibus system,
variables k and Vk of Equation (37) are replaced by Pk
and Qk as follows:

The system collapse occurs when J
D
becomes zero. Now, if
is the determinant value corresponding to the defined
operating condition of the system, then the required change in
J
D
of Equation (53) to force J
D
zero is
and corresponding change in injection
at kth bus will be

Therefore, load margin at kth load bus, i.e., the additional
load that can be supplied to the kth load bus to push it to the
proximity of voltage collapse is:

Therefore, predicted critical load for kth load bus at the
point of voltage collapse is:

But, P
Dk
margin
is determined using linear relation between
J
D
and Pk. As such, load margin for the kth load bus
(P
Dk
margin
) would be more than the actual load margin of the
bus. Further, for wide change in J
D
, the predicted load value
will be considerably high compared to that of actual critical
load value at the point of voltage collapse. Slight over
prediction of load margin could mislead system planner and
operator in the decision making related to allowing more load
in the load buses which are somewhat close to voltage stability
limit. Therefore, it is required to confirm the actual load
margin or critical load of a load bus through an iterative load
flow analysis using the load margin determined by the
Equation (55). To ensure convergence of the load flow
analysis in the proposed iterative procedure, it is required to
normalize the predicted critical load value P
Dk
predt
in such a
way that the modified predicted critical load value remains
below the actual critical load of the bus.
For this purpose, a reduction factor (RF) and a distance
factor (DF) are used to normalize the predicted critical load
value for the kth load bus as follows:

The distance factor is expressed as
International J ournal of Energy Engineering (IJ EE) Nov. 2012, Vol. 2 Iss. 4, PP. 126-136
- 131 -

The distance factor (DF) will be higher when prediction is
done for wide change of J
D
, i.e., far from the proximity of
voltage collapse limit. Thus, it will normalize the effect of
over prediction due to wide change of J
D
. Therefore, as the
iterative procedure approaches the proximity of voltage
collapse limit P
Dk
margin
will become very small; thus, the term
will also approach to 1. When the value of J
D

becomes negative the iterative procedure has to be terminated.
IV. PROCEDURE FOR DISTRIBUTION OF NORMALIZED PREDICTED
LOAD MARGIN TO THE GENERATORS
Equation (11) represents sensitivity relation for all the
buses with respect to the target kth bus. It is justified to supply
the normalized predicted load margin ( ) of the
kth target bus from the generators having higher sensitivity
factor values. Therefore, to distribute the normalized predicted
load margin of kth target bus, the sensitivity relation for
generator buses with respect to target kth bus are utilized.
Equation (11) can be expressed for generator buses only as
follows;



Therefore, modified generation values are


Where, P
Gi
max
is the maximum limit on generation for ith
generating station.
V. ALGORITHM OF THE ITERATIVE PROCEDURE
Reduction factor (RF) is used to normalize the over
prediction of load margin (which is determined using linear
relation between P
Dk
margin
and J
D
governed by Equation (53))
with the objective of ensuring convergence of load flow
analysis used in the proposed iterative procedure. As such,
load at a bus must be always less than actual critical load of
the bus, i.e. to say that P
Dk
crt
determined using Equation (57)
must be less than actual critical load of the bus. In case, P
Dk
crt
determined using Equation (57) becomes slightly more than
actual critical load of the bus (due to improper selection of
Reduction Factor (RF), the load flow analysis of the iterative
procedure will not converge. To take into account of such a
situation, the proposed algorithm is equipped with a step after
the load flow analysis. This step reloads the loads and
generations of the system of the previous iteration values and
reduces the reduction factor (RF) as RF =RF/3.0 and load
flow analysis is carried out again before proceeding to the
other portion of the algorithm. This step helps in changing the
value of reduction factor (RF) to ensure proper normalization
(by Reduction Factor (RF)) of critical load governed by
Equation (53). The algorithm of the proposed iterative
procedure is as follows:
1. Initialize the load flow data for the system and set
iteration count K =0, RF =(0.3 to 0.5).
2. Conduct base case load flow analysis of the system.
3. Determine P
Dk
crt
using Equation (57) and assign P
Dk
K+1

=P
Dk
crt
, Q
Dk
k+1
=tan k P
Dk
K+1
and distribute the additional
load (RF/DF) P
Dk
margin
among the generating stations based on
defined criteria subjected to generation limits of the generating
stations as described in Section-III. In case of reactive power
limit violation, a PV bus has to be changed into a PQ bus by
assigning reactive power at its limit.
4. Conduct the load flow analysis and Check for load flow
convergence criteria. If load flow has converged, then go to
Step-5.
Otherwise, reduce reduction factor RF =RF/1.5 1.5 and
reload P
Dk
K+1
=P
Dk
K
, Q
Dk
K+1
=Q
Dk
K+1
, V [i]
K+1
=V [i]
K
for i
=1...N and [i]
K+1
= [i]
K
for i =1...N, also the generating
stations output with their Kth iteration values and go to Step-3.
5. Check for |J
D
| <0.0 go to Step-6, otherwise, set K =K +
1 and go to Step-3.
6. Stop.
VI. SIMULATON, RESULTS AND DISCUSSIONS
To verify the validity and applicability of the proposed
method, simulations were carried out on IEEE 30 bus and
IEEE 118 bus systems. The aim of the simulations was to
examine the nature of change of J
D
for different load bus of
IEEE 30 and IEEE 118 bus system with respect to change in
load at the target bus. For this purpose, load at a target bus is
increased gradually manually using an interactive load flow
program and the change in J
D
values were recorded. It is found
that the J
D
reduces parabolically with increase in load and at
the point of voltage collapse it becomes zero. Figures 2 and 3
illustrate the variation of J
D
value of two bus equivalent
system of the IEEE 30 bus system with respect to change in
load at load buses 21 and 29 for load power factors 0.8 and 0.9.
Figures 4 and 5 illustrate the variation of J
D
value of two bus
equivalent system of the IEEE 118 bus system with respect to
change in load at load buses 88 and 118 for load power factors
0.8 and 0.9.
International J ournal of Energy Engineering (IJ EE) Nov. 2012, Vol. 2 Iss. 4, PP. 126-136
- 132 -

Fig. 2 Variation of J
D
for load change at load bus 21 of IEEE 30 bus system
for load power factors pf=0.8 & 0.9

Fig. 3 Variation of J D for load change at load bus 29 of IEEE 30 bus system
for load power factors pf=0.8 & 0.9

Fig. 4 Variation of J D for load change at load bus 88 of IEEE 118 bus system
for load power factors pf=0.8 & 0.9


Fig. 5 Variation of J D for load change at load bus 118 of IEEE 118 bus
systemfor load power factors pf=0.8 & 0.9
It is observed that J
D
value of two bus equivalent system of
IEEE 30 and 118 bus systems have different initial values for
different target buses, as such, J
D
value for a bus reflects the
voltage stability characteristic of the bus.
The proposed algorithm is used to determine critical load
of a bus with different reduction factors (RF). It has been
found that RF values between 0.1 to 0.5 ensures convergence
of load flow analysis in the iterative process for IEEE 30 and
IEEE 118 bus system for any target bus of the system. But,
with RF =0.3 to 0.5, the iterative procedure terminates with
less number of iterations. During iteration, the additional load
assigned on the target (k) bus was distributed among all the
generating stations based on their load contribution (subjected
to their limits) as described in Section-IV. Continuation power
flow analysis technique is also used to determine the critical
load at the target buses with step size = 0.001 to ensure that it
arrives the point of voltage collapse. In the continuation power
flow analysis technique the point of voltage collapse is arrived
when the predicted continuation parameter () becomes
negative. However, in this case predicted normalized load is
not distributed among the generators. Therefore, the proposed
algorithm is also used to determine the critical load at the
target buses without distributing predicted normalized load
among the generators.
Table 1 and Table 2 represent the simulation results for
some of the buses of IEEE 30 bus system with RF =0.4 with
power factor 0.9 and 0.8 respectively. Table 3 and Table 4
represent the simulation results for some of buses of IEEE 118
bus system with RF =0.4 with power factor 0.9 and 0.8
respectively. Tables represent the target bus selected for
determination of critical load with respect to its voltage
collapse limit, initial value of the determinant of the two by
two J acobian matrix for the target bus, initial value of
predicted continuation parameter (), initial load at the target
bus, final value of the determinant of the two by two J acobian
matrix ,final value of predicted continuation parameter (at the
point of collapse), critical load for the target bus and remarks
about the methods. The abbreviation used in the remarks
column of the tables stand for:
PMWGD =Proposed method with generation distribution.
International J ournal of Energy Engineering (IJ EE) Nov. 2012, Vol. 2 Iss. 4, PP. 126-136
- 133 -
PMWOGD = Proposed method without generation
distribution.
CLFWOGD = Continuation power flow without
generation distribution.



International J ournal of Energy Engineering (IJ EE) Nov. 2012, Vol. 2 Iss. 4, PP. 126-136
- 134 -


It is observed that predicted continuation parameter () of
IEEE 30 and 118 bus systems has different initial values for
different target buses, as such, initial value of reflects the
voltage stability characteristic of the target bus. Further, the
International J ournal of Energy Engineering (IJ EE) Nov. 2012, Vol. 2 Iss. 4, PP. 126-136
- 135 -
simulation results for the IEEE systems indicate that the predicted continuation parameter () becomes slightly
negative when the continuation power flow analysis is
terminated. It signifies that the systems are at the point of
voltage collapse. Also, the determinant value of the
transformed two by two matrix of the system becomes slightly
negative at the point of voltage collapse for the IEEE systems.
It is to be noted that sensitivity based distribution of
normalized predicted load margin of a target bus among
generators ensures higher critical load of a target bus in
comparison to that of continuation power flow analysis
without distribution of normalized predicted load margin
among generators. It is due to the fact that the distribution of
normalized predicted load is carried out among generators
using sensitivity relation governed by Equation (61). This
ensures that the higher portion of the normalized predicted
load would be assigned to the generator having higher
sensitivity factor i.e., the generator electrically closed to the
target load bus.It has been observed that the distance factor
becomes close to 1, when the
iterative process terminates. But, at the beginning of the
iterative process it appears to be high depending upon change
of J
D
used for the prediction of load margin for the load bus.
It normalizes the effect of over prediction of load margin due
to wide change of J
D
and ensures convergence of the load
flow analysis of the system during the iterative process.
To examine the effect of limits on real and reactive power
supply from generator(s) on the load margin, simulations were
carried out on IEEE 30 bus system, since; it has only six
generating stations. It is observed that Bus-29 is more
sensitive to generator located at Bus-8. Different limits on real
and reactive power supply are set this generator to examine the
value of load margin at Bus-29.
Case I: The maximum real and reactive power limits on
generator located at Bus-8 are set as PG
8
max
=0.55 pu QG
8
max

=0.34 respectively and the bus is treated as PQ bus. Table-5
represents the load margin of bus-29 of IEEE 30 bus system
with limits mention above.
TABLE 5: CRITICAL LOAD OF BUS-29 OF IEEE 30 BUS SYSTEM FOR
LOAD PF 0.8 (APPLYING LIMITS ON REAL AND REACTIVE POWER
SUPPLY ON GENERATOR LOCATED AT BUS-8)

With the limits on Generator 8, the contribution of real and
reactive power supply from the generator located at Bus-1,
Bus-2, Bus-5, Bus-11, Bus-13 are PG1 =0.578443 pu QG1 =-
0.03426 pu, PG2 =0.478458 pu QG2 =0.252048 pu, PG5
=0.653096 pu QG5 =0.153756 pu PG11 =0.287925 pu QG11
=0.179893 pu and PG13 =0.287096 pu QG13 =0.255502 pu
respectively.
Case II: The maximum real and reactive power limits on
generator located at Bus-8 are set as as PG
8
max
=0.55 pu
QG
8
max
=0.24 respectively and the bus is treated as PQ bus.
Table-6 represents the load margin of Bus-29 of IEEE 30 bus
system with limits mention above.
TABLE 6: CRITICAL LOAD OF BUS-29 OF IEEE 30 BUS SYSTEM FOR
LOAD PF 0.8 (APPLYING LIMITS ON REAL AND REACTIVE POWER
SUPPLY ON GENERATOR LOCATED AT BUS-8)

With the limits on Generator 8, the contribution of real and
reactive power supply from the generator located at Bus-1,
Bus-2, Bus-5, Bus-11, Bus-13 are PG1 =0.563591 pu QG1 =-
0.023318 , PG2 =0.477695 pu QG2 =0.274476 pu ,PG5
=0.652348 pu QG5 =0.166568 pu PG11 =0.287079 pu QG11
=0.183276 pu and PG13 =0.286284 pu QG13 =0.257491 pu
respectively.
Table 6 indicates that the load margin of Bus-29 of IEEE
30 bus system is found to be 0.312 pu without any violation on
real and reactive power supply of all the six generators of the
system. Whereas, with limits on real and reactive power
supply on generator located at Bus 8, the load margin
decreased for the bus as indicated by Table-5 and Table-6. For
Case I, the load margin is found to be 0.299 pu with limits on
Generator-8 as PG
8
max
=0.55 pu QG
8
max
=0.34 pu. But, the
load margin decreases to 0.296 pu with limits on Generator-8
as PG
8
max
=0.55 pu QG
8
max
=0.24 pu. It indicates that the both
real and reactive supply limits on generating station(s)
influence the load margin of a bus. Further, case study Case I
and Case II indicate that real and reactive supply limits on
generating station(s) influence the contribution from the
generating station(s) supplying loads within their limits.
Bifactorization matrix solution technique with optimal
ordering of the sparse load flow jacobian matrix has been
incorporated in the load flow analysis used for proposed
method and that of continuation power flow analysis. The
programs were executed on a P C with Pentium-4 processor
having processor speed of 1.5 GHz and LINUX operating
system. The simulation results show that the proposed method
requires considerably less CPU time compared to that of
continuation power flow analysis. Continuation power flow
analysis requires on an average 2-3 times more CPU time than
that of proposed method for IEEE systems.
VII. CONCLUSION
The paper proposed algorithm for a fast continuation load
flow analysis to determine load margin of a target/selected
load bus using the singularity condition of the load flow
J caobian matrix. The load flow J acobian matrix of an
interconnected multi-bus power system is transformed into a
two by two elements J acobian matrix with respect to a
target/selected load bus by incorporating the effect of all the
International J ournal of Energy Engineering (IJ EE) Nov. 2012, Vol. 2 Iss. 4, PP. 126-136
- 136 -
other buses of the system to the target bus. To incorporate the
effect of other buses to the target bus, sensitivity relations have
been formulated to relate change in injection at the 118 bus
system indicated that the determinant of two bus equivalent
system of a multibus system reduces with increase in load at
the target bus and it becomes zero (near to zero) when load
flow solution does not converge i.e., voltage collapse takes
place. Therefore, it could be concluded that when the modified
two by two matrix become singular, actual load flow J acobian
matrix also becomes singular.
But, it is observed that determinant value of two bus
equivalent system of IEEE 30 and 118 bus systems have
different initial values for different target buses, as such, it
shows that the transformed two by two elements J acobian
matrix reflects the property/quality of the target bus.
The algorithm proposed for the determination of critical
load of a bus with respect to its voltage collapse limit of a
power system works for all buses of IEEE 30 and IEEE 118
bus system, as such, it could be used for any inter connected
power system. The use of reduction factor (RF) and distance
factor (DF) ensures convergence of the load flow analysis of
the system during the proposed iterative procedure. These two
factors effectively normalize the prediction of load margin,
which is carried out using the linear relation between J
D
and
Pk governed by Equation (55). The simulation results show
that the proposed method requires considerably less CPU time
compared to that of continuation power flow analysis.
REFERENCES
[1] D. Hazarika and A.K. Sinha Power system restoration:
Planning and simulation. Int. J ournal of Electrical Power &
Energy Systems, Vol- 25, 2003, pp. 209218.
[2] Clark. H. K. New challenges: Voltage stability IEEE Power
ENG. Rev (April 1990), pp. 33-37.
[3] P. Kessel and H. Glavitsch Estimating the voltage stability of a
power system. IEEE Trans. on Power Delivery, Vol. PWRD-1,
No.3 J uly/1986, pp. 346 - 354.
[4] Gubina F and Strmcnk B Voltage collapse proximity index
determination using voltage phaser approach. IEEE Trans.
Power System, Vol. 10, No. 2 1995, pp. 788 793.
[5] A.K. Sinha and D. Hazarika. Comparative study of voltage
stability indices in a power system. Int. J ournal of Electrical
Power & Energy Systems, Vol- 22, No. 8, Nov. 2000, pp. 589
596.
[6] Y. Tamura, H. Mori and S. Lwanoto Relationship between
voltage instability and multiple load flow solutions in electrical
system, IEEE Trans. Vol. PAS-102, pp. 115-1125.
[7] Crisan and M Liu Voltage collapse prediction using an
improved sensitivity approach Electrical Power System
Research, 1994, pp. 181-190.
[8] P A. Lof, G. Anderson and D. J . Hill Voltage stability indices
of stressed power system, IEEE Tran. PWRS, Vol. 8, No.1,
1993, pp, 326-335.
[9] A. Tiranuchit and R. J . Thomas, A posturing strategy against
voltage instability in electrical power systems IEEE Tran.
PWRS, Vol. 3, No.1, 1989, pp. 87-93.
[10] W. C. Rheinboldt and J .V. Burkhardt. A locally parameterized
continuation process. 28 ACM Transactions on Mathematical
Software, Vol. 9(2), J une 1983, pp. 215-235.
[11] V. Ajjarapu and C. Christy. The continuation power flow - a
tool for steady state voltage stability analysis. IEEE
Transactions on Power Systems, Vol. 7(1), Feb. 1992, pp. 416-
423.
[12] V. Ajjarapu, P. L. Lau, and S. Battula. An optimal reactive
power planning strategy against voltage collapse. IEEE
Transactions on Power Systems, Vol. 9(2) May, 1993, pp. 906-
917.
[13] A. C. Souza and V. H. Quintana. New technique of network
partitioning for voltage collapse margin calculation, IEE
Proceedings Gerer. Transm. Distrib., vol. 141(6), Nov, 1994, pp.
630-636.
[14] A. J . Flueck and Qiu Wei A new technique for evaluating the
severity of branch outage contingencies based on two-parameter
continuation. Power Engineering Society General Meeting, Vol.
1, J une, 2004, pp. 323-329.
[15] Y. Kataoka, M. Watanabe and S. Iwamoto, A New Voltage
Stability Index Considering Voltage Limits. Power Systems
Conference and Exposition, 2006. PSCE 06. 2006, IEEE PES
Oct./Nov, 29, 2006, pp. 1878 - 1883.
[16] I. Smon, G. Verbic and F. Gubina, Local voltage-stability index
using tellegens Theorem. IEEE Transactions on Power
Systems, Vol. 21, no. 3, 2006, pp. 1267-1275.
[17] Wanga Yang, Lib Wenyuan and Lua J iping, A new node
voltage stability index based on local voltage phasors. Electric
Power Systems Research, Volume 79, Issue 1, J anuary, 2009, pp.
265-271.
[18] Saikat Chakrabarti, Static load modelling and voltage stability
indices. International J ournal of Power and Energy Systems,
29(3). pp. 200-205.
[19] B. Stott and O. Alsac, Fast decoupled load flow, IEEE
Transactions on Power System, Vol, 93, May/june, 1974, pp.
859-869.
[20] D. Hazarika and A K Sinha, Standing phase angle reduction for
power system restoration, IEE Proccedings Genr. transm.
Distrib., Vol. 145, No. 1, J anuary, 1998, pp. 82-88.

Durlav Hazarika received the degree in
Electrical Engineering from J orhat
Engineering College, J orhat, Assam, India in
1983, the Master degree from Indian Institute
of Technology (IIT) Bombay, Mumbai, India
in 1986 and the Ph.D degree from Indian
Institute of Technology (IIT) Kharagpur,
Kharagpur, West Bangle, and India in 2000.
He is currently working as a Professor
with the Department of Electrical
Engineering, Assam Engineering College, Guwahati, Assam, India.