You are on page 1of 8

IEEE Transactions on Power Systems, Vol. 12, No.

3, August 1997

1121

A SIMPLE APPROACH TO VOLTAGE STABILITY ASSESSMENT IN RADIAL NETWORKS

F. Gubina

B. StrmEnik
Student member

Member
University of Ljubljana
Ljubljana, SLOVENIA

Abstract: Analytical approach to voltage collapse proximity


determination is proposed for radial networks. Under
corresponding assumptions, a radial network with arbitrary
bus loads is transformed into a two bus equivalent. The
voltage phasors at the generator bus and at the last load bus
of the radial network are transformed to form the voltage
phasors of the two bus equivalent. The latter are further used
for assessment of voltage collapse proximity. Exact stability
limit relations for a two bus system derived from Jacobian
matrix can be exploited. Moreover, an analytical expression
is derived for calculation of active and reactive power reserve
margins for radial network equivalent. The proposed
procedure has been tested for practical examples of radial
networks with inductive and capacitive loads.
Keywords: voltage stability, equivalents, power margin

1. INTRODUCTION
Voltage instability in power networks is a phenomenon
of highly nonlinear nature posing operational as well as
prediction problems in power systems control. To obtain a
solution to the problem, dynamic and static methods have
been applied [5]. The static methods are predominantly based
on Jacobian matrix properties [6,8]. The methods are,
therefore, bound to work with a full rank of network Jacobian
matrix. Furthermore, intensive attention has been paid to
assessment of the power system reactive reserve power and
its loadability [7,9,10,11]. All the methods are subject to
relatively extensive and time consuming calculation
procedures.
However, the voltage collapse is a local phenomenon.
As a rule, it occurs at a bus within the area with high loads
and low voltage profile. The distant buses usually do not
influence the affected bus voltage stability to a great extent if
any. The idea was'to employ properties of network phasors
and parameters which are believed to contain enough
information to assess this local phenomenon. A method of
voltage phasors has been presented in [ 1,8] pointing out the
properties of phasors in a radial network.
96 SM 475-4 PWRS A paper recommended and approved by the
IEEE Power System Engineering Committee of the IEEE Power
Engineering Society for presentation at t h e 1996 IEEUPES Summer
Meeting, July 28 - August 1, 1996, in Denver, Colorado. Manuscript
submitted December 28, 1995; made available for printing June 27,
1996.

For the two bus network with U,=const, the Jacobian


matrix indicates the voltage instability when its determinant
takes the value

The above expression yields


U , = 2 . U , costi,,

(2)

where the voltage phasors are denoted as U=U ds with


indices 1 at the input bus and with 2 at the end load bus. P
and Q are active and reactive powers, respectively. This idea
could be further extended to a multibus radial network.
Unfortunately, it is not a trivial task to show analytically
under what conditions the same rule could apply to the entire
radial network.
The procedure was extended with a limited success to
meshed networks [SI. The method did not filly exploit all the
available information on the network operational status and
its parameters. It was also not evident how to assess available
reactive power reserves.
The idea of this paper is to show that a two-bus
equivalent of a radial network may be constructed utilizing
its operational parameters. For that two-bus equivalent, the
voltage stability features may be easily assessed as seen from
equation (2). An interesting equivalent of a radial network
has been proposed in [2,3] where equal input power,
transmission losses and sum of loads were assumed. In the
paper, the network equivalent equations are derived based on
modified assumptions. They enable exact transformation of
the measured voltage phasor at the end of the radial network
or at the end of a critical path to equivalent voltage phasor
employed further for voltage stability assessment.
2. TWO-BUS EQUIVALENT OF RADIAL NETWORK

The radial network fed by a generator contains a series


of nodes feeding various loads P, and Q,. The loads can
draw active, inductive, or capacitive power. Schematically,
the sketch of a radial network is presented in Fig. 1. All the
loads match the generator active Pl and reactive Ql power.
The voltage phasors at the generator and at the radial
network end node are U, and U,,, respectively. The current
phasor ll denotes current flowing into the first section of

0885-8950/97/$10.00 0 1996 IEEE

the radial power system. The currents in other sections vary


and follow the corresponding sums of powers they are
feeding along the radial network.
The input voltage U, is defined as constant describing
a generator or constant transformer voltage. An adequate
two-bus equivalent is sought which would map the real
voltage state regarding stability. The two-bus equivalent is
presented in Fig. 2. It is important to select criteria for
adequate forming of the equivalent. They should reflect the
common properties of the original radial network and its
equivalent to such an extent that the latter one could be used
for voltage stability assessment.

The voltage phasors at the input (generator) buses of both


networks are assumed to be collinear. Since P,=P, and
Q,=Q,, the phase angle between U, and I is the same as
at the input bus of radial network. Due to that fact, the
voltage magnitude U, changes in the process of
transformation to equivalent voltage magnitude U,. Based
on this reasoning, the following relation results for phasor
magnitudes:

Sections along the transmission network are defined by


impedances and loads producing active and reactive line
losses. They result in voltage drops in the radial network as
follows:

.........

Fig. 1 Radial transmission network


The proposed equivalent was modified to retain equal
e
input power,
e
load impedance magnitude at the end of the radial
network as felt by the input bus and
e
voltage drop along the transmission path
to serve our proposed voltage collapse assessment strategy.
This set of conditions serves for transformation of a radial
network with the quantities denoted without primes onto a
two bus equivalent with quantities denoted with primes.
This transformation in turn preserves the most important
features of voltage stability for both original and its
equivalent.

where CPruss and CQ,, denote sums of active and reactive


power losses along the corresponding lines of radial
network, respectively.
They may be utilized also to calculate the two-bus
equivalent voltage drops. The voltage drops of the
equivalent two-bus network may be formulated as follows:

where P,oss and Qlossdenote active and reactive losses.


R,and X,denote transmission imitances of the equivalent
network defined as follows

R; =

U,

Fig. 2 Two bus equivalent and corresponding phasor


diagram
The current 1, flows only into the frst section of the
radial network and, therefore, it cannot be utilized for the
two-bus equivalent. On the contrary, the input active and
reactive powers should be used for that purpose. The
following relation applies for the quantities of both
networks:

x;=

uiu;

%ss

fi2+e:

fi2 +e:

Another assumption is that both corresponding voltage


drops in equations (5) and (6) are equal

to produce an adequate equivalent.

U,& = UiL

(3)

The transformation of measured voltage U,, requires


additional reasoning. From the input bus with voltage U,,
the composite loads are seen as lumped and defined by U,/l,
or by a constant ratio X/R. Load voltage at the last load is
not directly accessible as seen from Figure 3.

The equivalent is thus completely determined and its


variables could be employed for voltage stability
assessment.

3. VOLTAGE COLLAPSE PROXIMITY INDEX

I
Fig. 3 Equivalent variables as seen by the input bus
The load impedance of the transmission path end bus as felt
by the input (generator) bus could be defined by the
following expression:

(9)
where it is required to match the load impedance magnitude
of the equivalent at the same time.

Since the reduction of a radial network was


analytically derived to a two-bus equivalent, the voltage
collapse proximity could be determined in a straightforward
manner. For the equivalent two-bus radial system with
constant generator voltage U,, load bus voltage C p D line
impedance Z', and load impedance ZtL,the maximal power
transmission capabilities are reached when the equality

U; =2u;cos$

is fulfilled, Fig. 2. It defines singularity point of the


Jacobian and, hence, voltage collapse point of the radial
power system. Furthermore, it indicates the point when the
magnitudes of both, transmission and load impedances Z't
and 21are equal [ 1] .
The transmission path is now visualized as a series of
buses with declining voltage magnitudes. The Transmission
Path Stability Index (TPSI) could be formed for the
equivalent of the radial transmission network as follows:
TPSI = 0.5. U; - U;

Fig. 4 Voltage phasors as seen from the input bus


By combining this result with the expression (4), the
following basic equation of the radial network mapped to a
two bus equivalent could be obtained:

u,un= u;u;
The measured value of the last load bus voltage magnitude
U,, can be transformed by means of the derived expressions
into the load voltage of the equivalent circuit U2'.
The basic relations between the radial network and its
two-bus equivalent are as follows:

The equations (4), (5) and (10) are sufficient to solve the
following set of nonlinear equations:

(13)

(14)

For TPSI=O, the power transfer of the radial network


becomes unstable, since due to losses, any increase of the
two-bus equivalent load results in a decrease of received
power at the load bus. Moreover, the voltage stability may
be determined directly using the above analytical
procedures for the two bus equivalent.
The radial network may comprise compensating
devices situated along or at the end of its path. The source
of capacitive power is treated as a load, but it changes the
voltage profile and direction of reactive power flow.
Therefore, it is necessary to establish a critical transmission
path with declining voltage magnitudes as a measure. The
critical path does not end at the bus n of the radial
transmission network but rather at the bus k with the lowest
voltage magnitude somewhere in the middle of the radial
network between the generator and capacitive load bus. The
voltage magnitude U, may serve now as terminal voltage of
the critical transmission path of the radial network and the
two-bus equivalent could be calculated for that path.
4. POWER MARGIN ASSESSMENT

The distance to the voltage collapse may be easily


determined by the proposed procedure at any operating

1124

point, thus supplying the operator in the control center with


the necessary information. However, there is no measure as
to how much more a radial network could be loaded before
reaching the point of voltage collapse. An useful idea was
presented in [9,10,11] where the authors apply the left
eigenvector to assess the network loadability margin. In
case of a two-bus equivalent, the solution could easily be
derived analytically.
Let us consider the power flow on the two-bus
equivalent. The active and reactive power inflows are given
in equation (15):

AQ = Q,,,

- Q, = -O.5Ui2 B +

B
(GP, + BQi)
B2 + G 2
ti
= 0.5Ui2G(GP, + BQi )
B2 + G 2

(20)

AP = q
,

-4

The allowed increase of injected active and reactive powers


into the generating bus take the ratio equal to:

The values of parameters G and B are not constant in


case of radial network equivalent and vary slightly with the
power system loading which reflects in changes of losses
and equivalent voltage U,as well, see equation (7a,b).
where the parameters of the equivalent are denoted as
G, =G and B, =B.
At the point of voltage instability, the equation (13)
applies yielding the following relations for maximum real
and reactive power transfer:

After eliminating the angle 6, the following expression for


maximum power transmitted into the two-bus equivalent
could be derived:

The minimum distance to power limit may be reached


from the operating point P1 and Q1 via trajectory normal to
the trajectory defined by equation (17). Another approach
could utilize optimal search for the objective function of
minimal distance to linear constraint, equation (17).

5. PRACTICAL EXAMPLES
An example of transmission path index is given for a
10-bus radial network exhibiting monotonous voltage drops
towards the end load bus. The loads have been increased
monotonously and proportionally. The input (generator)
bus voltage was kept constant at U,=lp.u. In all cases,
stability was tested against the last stable power flow
solution for increased composite load S as compared to the
base case composite load S,.
0.5
0.45

t-

--

0.2 -0.15 -0.1 --

0.05 -.
0

1.1

1.2SlSb1.3

1.4

1.5

Fig. 5 Example of 10 bus radial network


The solution to question of maximal transmitted power over
a radial network mapped onto the two-bus equivalent
variables is reached when reactive and active powers take
the following values:

,q
,

= 0.5Ui2G+

( B 4 + GQl)

B~ + G ~

The expression (19) enables direct determination of the


active and reactive load distance as seen from the operating
point Pi and Q , .

Another 10-bus example is analyzed for the case of


capacitive power injection at the 10-th load bus. The
critical transmission path with monotonously decreasing
voltage magnitudes was found at bus No. 4, where the
voltage collapse occurred after monotonously increased
loading.
In Fig. 7, another example of 14-bus radial network is
shown. As it can be seen from the figure, the proposed
procedure perfoms as expected. For the case of TPSI =0, a
minimal increase of S was required and a stable solution
could not be computed.

1125
0.4

1.2 -

0.35
0.3

0.8

0.25

0.2

I\,

0.6

.
.

0.4

~-

0.2

-~

0.15

\\

0.1

\-,

0.05
1.2

1.4SlSb1.6

1.8

%xx,

1.2

Fig. 6 Example of a ten bus radial network with the


capacitive power injection at the bus No. 10

1.4

1.6
SilSib

\-\

1.8

Fig. 8 Power margin as a function input power

0.35 -

0.35

0.3 --

0.3

0.25 .-

0.25

5
n.
I-

0.2
0.15
0.1

0.05

0 I
1

0.2

0.4

0.6

0.8

1.2

1.1

1.2

1.3
SlSb

1.4

1.5

1.6

Fig. 7 Example of a 14-bus radial network


Another point of view is presented in Fig. 8. The 14bus case is analyzed. The current power margin is plotted
against available input (generator) power Si= S, expressed
as follows:

As it can be observed, the function is quite linear at higher


input power. Therefore, maximal loading Si could be
predicted with sufficient accuracy using linear
extrapolation. Sib=&denotes base input power.
For the same case of the 14-bus radial network,
coordinated function of TPSI and reserve input power
increase in p.u. is shown in Fig. 9. Both indices exhibit
similar accuracy in the sense of indicating loadability and
voltage collapse point proximity. Apparently, either the
first or the second index could be used to assess the
proximity of voltage instability point. The characteristics in

Fig. 9 exhibits relative linearity of the curve at the origin.

Fig. 9 Power reserve margin versus stability index

6. CONCLUSION

Since the voltage instability is generally a local


phenomenon, the variables and network parameters contain
sufficient information to assess proximity to instability.
This idea resulted in direct analytical approach to voltage
instability assessment for radial networks presented in this
paper.
The approach is based on exact two-bus equivalent of
a radial network. The equivalent is built on assumptions
which enable uniform transform of selected phasor
variables of the radial network. For a two bus equivalent, a
simple, exact, and analytically proven test for voltage
collapse proximity measure could be applied. In addition,
the developed stability index has a physical meaning since
it is measured in p.u. voltage.
The approach via the two-bus equivalent enables
exact determination of load margin of the radial equivalent.
The idea presented in the paper encompasses one step
procedure enabling fast voltage stability assessment of
radial networks. It is further shown that both indices could
be exploited for voltage instability assessment.

1126

ACKNOWLEDGMENT
Financial support by the Ministry of Science and
Technology, Republic of Slovenia, under the research grant
No J2-5202-078 1/95 is gratefully acknowledged.
REFERENCES
1. C. Barbier, J.P. Barret: An Analysis of Phenomena of
Voltage Collapse on a Transmission System,
ELECTRA, July 1980;
2. G. B. Jasmon, L.H.C.C. Lee: New Contingency
Ranking Technique Incorporating a Voltage Stability
Criterion, TEE Proc. - C, March 1993;
3. G. B. Jasmon, L.H.C.C. Lee: Maximizing Voltage
Stability in Distribution Networks via Loss
Minimization, Electrical Power & Energy Systems,
June 1991;
4. M.E. Baran, F. Wu: Network reconfiguration in
distribution systems for loss reduction and load
balancing, IEEE Trans. on Power Delivery, Vol. 4,
No. 2, 1989;
5. P. Kundur, K. Morison: Practical considerations in
voltage stability assessment, Electrical power and
energy systems, Vol. 15, No. 4, 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;
7. T. Van Cutsem: A Method to compute Reactive Power
Margins with respect to Voltage Collapse, IEEE Trans.
on Power Systems, No. 1, 1991;
8. F. Gubina, B. StrmCnik: Voltage collapse location and
proximity index determination using voltage phasors
approach, IEEE Trans. on Power Systems, Vol. 10,
No. 2, 1995;
9. I. Dobson, L. Lu: New methods for computing a
closest saddle node bifurcation and worst case load
power margin for voltage collapse, IEEE Trans. on
Power Systems, Vol. 8, August 1993;

10. F. Alvarado, I. Dobson, Y. Hu: Computation of closest


bifurcation in power systems, IEEE Trans. on Power
Systems, Vol. 9, May 1994;
11. I. Dobson: Observations on the geometry of saddle
node bifurcation and voltage collapse in electrical
power systems, IEEE Trans. on CAS-I, No. 3, March
1992;
12. I. Dobson, L. Lu: Voltage collapse participated by the
immediate change in stability when generator power
limits are encountered, IEEE Trans. on CAS-I, No. 9,
Sept. 1992;

FERDINAND GUBLNA (M73) was born in Srebrnik,


Slovenia, on May 16, 1939. He received the Diploma
Engineer, M. Sc. and Dr. Sc. degrees bom The University
of Ljubljana, Slovenia in 1963, 1969, and 1972,
respectively.
From 1963 he was with the Milan Vidmar
Electroinstitute, Ljubljana. He was heading the Power
System Operation and Control Department. In 1970 he
joined The Ohio State University, Columbus, Ohio, as a
teaching associate for one year. Since 1988 he has been
full professor at The University of Ljubljana. His main
interests lie in the area of electric power system operation
and control.
Dr. F. Gubina is member of SC 39, CIGRE Paris, and
president of the Slovenian National CIGRE Committee.
BOSTJAN STRMCNIK (StM90) was born in Ljubljana on
July 4, 1965. He received the Diploma Engineer and M.
Sc. degrees from the University of Ljubljana in 1990 and
1993, respectively. In 1990 he joined Korona Power
Engineering and is currently with the University of
Ljubljana. His main interests are power systems operation
and control.

1127

Discussion

M.K. Pal (Consultant, Edison, NJ): The voltage stability


assessment method presented in the paper is based on several old
ideas the validity of which are frequently not justified. A major
implied assumption of the paper is that voltage collapse proximity
is independent of load characteristics as well as the pattern of load
changes at the buses along the network. It is now well accepted
that, depending on the load characteristics, the voltage stability
limit can occur well below the loading at which the power flow
Jacobian becomes singular. On the other hand, for certain types of
loads, load can be added even after the singularity point is reached
without experiencing voltage collapse.
The two-bus equivalent derived by the authors is valid only
at the operating point at which it is derived. It can be used for
small load changes around this point. However, since the power
flow equations are highly nonlinear, even in a simple radial
system, the equivalent would be inadequate for assessing voltage
stability limit. Also note that it does not allow for changing
loading pattern of the various buses which would greatly affect the
collapse point.
The proposed method might have served some useful
purpose in the days when modern computing facilities were not
available. It is not clear why one would accept the uncertainties of
using the equivalent when an exact loading limit, allowing for load
variations of individual buses, can be determined very quickly
using a power flow program specifically designed for this purpose.

L L Lai (Energy Systems Group, City University, Lo.ndon


EClV O H B , UK) :
The authors have presented an interesting paper on an
analytical approach to voltage collapse proximity
determination.
The discusser has the following question.

To make the approach prectical, the physical constraints on


the system, especially load characteristics, dispatch srtategy,
reactive power generation limits and transformer taps should
be considered. How complicated your simple approach will
become if the details of the above are included?

problem was the sharp changes on the index when generator Qlimits are encountered, which is not mentioned in the current
paper, but could be a problem if several generators are feeding
the system at different points along its radial structure.
The comments of the authors to these issues would be appreciated by the discusser.
[A] C. A. Caiiizares, A. Z. de Souza, and V. H. Quintana, Comparison of performance indices for detection of proximity to
voltage collapse, IEEE/PES 95 SM 583-5 PWRS, Portland,
OR, July 1995.

Manuscript received August 27, 1996.

F. Gubina, B. StrmEnik (University of Ljubljana, Slovenia):


The authors wish to thank the discussers for their interest and
discussions of the paper.
The common comments of all discussers refer to questions
of load characteristic and limits of reactive power production,
respectively. We agree that loads have significant impact on
the power systems operating point. However, the main
interest of the authors was to establish a simple analytical
procedure for voltage collapse proximity assessment based on
actual power flow measurements and voltage magnitudes. As
demonstrated, the proposed method could be applied if
voltages, line flows, and bus injected power measurements
are available. In that case, the distance to voltage collapse and
available power margins can be assessed directly without any
Jacobian computation. However, the voltage dependency of
loads is reflected in the measured value of U, which is in turn
mapped to the equivalent voltage U,, eqn. (10).
Increased loading and subsequently encountering of the
reactive power production limits impact evolution of critical
bus location in the voltage instability process. The proposed
method enables an adequate treatment of such realistic cases
as well. For that purpose, additional information regarding
reactive power transmission path have to be taken into
account in order to determine the critical bus as shown in

[WI.

Manuscript received August 22, 1996.

.PI...,

Claudio A. Caiiizares (Universityof Waterloo): A generalization of the technique described in this paper was initially proposed by this discusser and co-authors in [A]. In that paper, the
determinant of a reduced system Jacobian matrix is proposed as
an stability index, identical to what the authors suggest in the
current paper. However, no Iimitations regarding the structure
of the system are assumed in that case, so that the technique
applies to radial and mesh systems as well, which is certainly a
more realistic condition for most power systems today.
In [A], two particular problems were identified with this determinant based voltage stability index. The first one had to do
with the identifications of the critical bus in the system, as
the reduced determinant did not present a smooth, predictable
behavior for other system buses. This is obviously not a problem in the current paper, as the studied system is radial, so
that there is no need for identifying a critical bus. The second

.....,....,.. ...,.....
I
I
I
....
....

P2YB

...

.... p.Pm
....*.....

I
S+jQa

..... ......,...... ,....,.....,


I

.... ....

...

14

Pi4-~Q~a

Figure C 1: Active and reactive power flow pattern

As an example, let us consider the 14-bus radial network


fed by a slack bus No. 1 and by a limited reactive power
source at the bus No. 14. In the first stage, until the bus No.
14 is PV controlled, the critical transmission path could be
found between the bus No.14 and bus No. 8 following the
direction of reactive power. After certain increase of bus
loads, the reactive power production limit at the bus No. 14 is
reached. The most critical path is then identified between the
buses 14 and 9, since the bus 9 has taken a reactive power
drain role as seen fi-om Table C1, causing discontinuity in
Fig. C2. Likewise, the TPSI index exhibits the unavoidable
non- linear behavior in a similar way as shown in [A].

1128

Table C 1: Critical bus identification and corresponding TPSI

The same phenomenon is to be expected in cases of reaching


the transformer tap changing limits. Therefore, we agree with
the discussers that the proposed method as Jacobian based
ones do not overcome all the problems caused by power
system non-linear nature.
Dr. Caiiizares et al. derived some kind of a two bus
equivalent [A] utilising reduced and full Jacobian based on
supposition that the loads in other buses do not change. As it
can be easily seen, our approach is different. It does not
utilise the Jacobian of a n>2 bus system, since it exploits the
transmission path concept. Moreover, the approach points out
some new aspects to the voltage collapse phenomenon. Via
transformation of radial network into two-bus equivalent, it is
shown that voltage collapse is not merely collapse of the
voltage at the most vulnerable bus, but it is instability of the
reactive power transmission path which has to be identified
on the basis of reactive power flow direction as shown in Fig.
c1.
0.5

A. C. A. Caiiizares, A. C. 2. De Souza, V. H. Quintana:


Comparison of Performance Indices for Detection o f
Proximity to Voltage Collapse, IEEE Trans. On PS,
Vol. 11, No. 3, August 1966;
B. B. Strm&&, F. Gubina: On Simplified Methods for
Voltage Security Assessment, CIGRE Colloquiuni SC
39, Johannesburg, RSA, October 1995;
C. F. Gubina, B.StrmEnik: Voltage Collapse Location and
Proximity Index Determination Using Voltage Phasors
Approach, IEEE Trans. on PS, Vol.10, No.2, May
1995;

0.4

cn
a

0.3

i-- 0.2

0.1

The geometrical interpretation o f the equation (2) with its


physical meaning makes the TPSI index operator friendly".
The method enables detection of the critical path and
identifies the bus prone to voltage collapse. Unlike the indices
derived from Jacobian matrices, it gives the operator the
distance to voltage instability point in volts or in pu values.
We agree with Mi. Pal that our method, based on nonlinear relations, is currently valid at the operating point as are
the Jacobian based ones. There is no need to rely on
extrapolation in cases of load variation. Due to methods
obviously small computational effort, the suggested proximity
index may be calculated on line, based on the exact mapping
of the measured variables to a two bus equivalent. It offers a
versatile assessment of the distance to voltage stability limit
needed by a power system operator and enabling adequate
planing of preventive actions.
Derivation of an exact two-bus equivalent of radial
network could be further exploited for meshed networks
voltage stability assessment, as indicated in the above
example. The voltage collapse is not entirely local but rather
localised phenomenon although the approaches based on
Jacobian treat it as a system process. By continuous
identification of reactive power transmission path, its TPSI as
well as reactive power production reserve calculation, an
integral assessment of voltage collapse proximity may be
obtained for a meshed power system.

1.5

S JS t,2.5

Fig. C2: TPSI vs. loading for a two end fed radial network

Manuscript received November 6, 1996.

You might also like