Beruflich Dokumente
Kultur Dokumente
ROC(A)
Vol. 25, No. 1, 2001. pp. 53-62
ABSTRACT
This paper presents an efficient computing algorithm for enhancing voltage security. The algorithm uses the
genetic algorithm (GA) to dispatch reactive power sources under various system conditions. GA aids the dispatching
of reactive power sources so as to maintain the specified security level. The reactive power sources used in the proposed
genetic algorithm are transformer tap changers, static capacitors, static Var compensators (SVCs) and generator terminal
voltages. The proposed method has been tested on an IEEE 30-bus power system, and successful results have been
obtained.
Key Words: voltage security indices, voltage security enhancement, genetic algorithms, phasor measurement unit
I. Introduction Ajjarapu et al., 1994; Chang and Huang, 1998; Begovic and
Phadke, 1992; Nanba et al., 1998; Chebbo et al., 1992b).
In recent years, an instability problem, usually termed Specifically, Van Cutsem (1991) used the solution of a reactive
voltage instability, has been observed and has been found to power optimization problem as the voltage security margins.
be responsible for several major network collapses in different Ajjarapu et al. (1994) introduced a method for determining
countries (IEEE Publication, 1990; Taylor, 1994). This phe- the minimum amount of shunt reactive power support needed,
nomenon did not always occur in response to a contingency, which indirectly maximizes real power transfer before voltage
such as the loss of an important transmission line or a generator, collapse is encountered. Chang and Huang (1998) used
but rather in response to an unexpected rise in the load level, optimal multi-objective static Var compensator planning to
sometimes in combination with inadequate reactive power enhance voltage stability. Begovic and Phadke (1992) used
support at critical network buses. sensitivity analysis to control voltage security. Nanba et al.
A large amount of research has focused on voltage (1998) proposed a control method for improving voltage
instability or voltage collapse. Indeed, many researchers have security based on the concept of the voltage instability prox-
proposed voltage security margins which show how close the imity index. In addition, Chebbo et al. (1992b) described a
current operating point of a power system is to the voltage linear reactive power dispatch algorithm to reduce the risk of
collapse point (Schlueter, 1998; Canizares and Alvarado, 1993; voltage collapse.
Gao et al., 1992; Chebbo et al., 1992a; Lf et al., 1992; Van The system operators usually apply proper operations
Cutsem, 1991), which serves as an assessment of voltage to controllers to enhance voltage security when the power
security. In addition to the identification of voltage security, network has voltage instability or shows vulnerability to voltage
system operators are also interested in knowing how much collapse. Due to mismatch problems between the searched
and where external reactive support is needed and located with optimal or near optimal control decisions and practicable
regard to both security and economics. Reactive power can dispatches under these limitations, such as placement,
be dispatched effectively to maintain acceptable voltage levels expenditure, supplementary instruments etc., it is necessary
throughout the system and to reduce the overall real power that multi-solutions be selected by operators to enhance voltage
loss of the system. Therefore, many researchers have pointed security. If we use the traditional optimal approaches, which
out that the security margin of power systems can be enlarged are the linear (Hegdt and Grady, 1983), nonlinear (Billiington
to reduce the possibility of voltage collapse by providing a and Sachdev, 1983), mixed integer programming (Aoki et al.,
sufficient amount of reactive power (Van Cutsem, 1991; 1988), and decomposition (Mangooli et al., 1993) methods,
53
C.W. Liu et al.
II.Voltage Security Enhancement Scheme IF (x HR5 VL), THEN voltage security is at the
very secure level.
Voltage security indices are alternative techniques for
the detection of collapse and insecurity points. These indices IF (x HR4 L), THEN voltage security is at the secure
are scalar variables that are continuously monitored to deter- level.
mine how close a system is to voltage collapse. Simple
examples of voltage security indices are the minimum singular IF (x HR3 M), THEN voltage security is at the
value and real eigenvalue of the power flow Jacobian matrix: alert level.
the closer these values are to zero, the closer the system is
to collapse. In this paper, we use five fuzzy linguistic variables IF (x HR2 S), THEN voltage security is at the
(i.e., Very Small (VS), Small (S), Medium (M), Large (L) and dangerous level.
54
Power System Voltage Security
The system operating constraints consist of the available A. The Representation of Strings
range of active and reactive generated power, bounds on vol-
tage magnitudes, phase angle difference limits, line flow limits Each string (chromosome) represents a possible voltage
etc., which are shown as Eqs. (7) (10): security enhancement strategy, and each element (gene) rep-
resents a value of control variables. Therefore, the numerical
S min max
g S g S g g N g, (7) string in this paper is that shown in Fig. 2. The length of the
55
C.W. Liu et al.
V1 V2 ... Vg T1 T2 ... Tt 1 2 ... k to the rest of the population. Stings with higher fitness
values (Sn) have a higher probability of contributing
numerical string
offspring and are simply copied into the next generation.
Fig. 2. Representation of a numerical string of a GA. In this paper, we employ an elitist policy; that is,
the sting with the highest fitness value in the population
string is equal to the total number of control variables. Usually, is directly copied into the next generation.
the value of an element in the string is represented by a binary (2) The crossover operator recombines the extremely im-
bit. However, this paper proposes the use of a real number portant features of two strings to make offspring (child)
string. strings. Not only do they inherit some important
characteristics from their parent strings, but they also
B. Fitness Function have a chance of getting closer to the optimal decisions.
Crossover is performed on two strings that are selected
In a genetic algorithm, a fitness function is a mapping from the population randomly at one time, its frequency
which determines the fitness of each string in the population. being controlled by a crossover probability, Pc. Cross-
The GA proceeds to evolve better-fitting strings, and the fitness over can occur at a single position (single crossover)
value is the only information available to the GA. The strings or at a number of different positions (multiple crossover).
with large fitness values offer better solutions to the problem In this paper, we adopt a uniform crossover technique,
and have a higher probability of being selected. Therefore, which exchanges elements between the two selected
the adopted fitness function of the GA methodology for VSE parent strings to create new offspring strings by means
is a measure of how close a specific operating point is to of a random mask. In this manner, the elements are
the point of voltage instability. swapped when the value of the random mask is 1.
We will use a voltage security index as a fitness function On the contrary, the elements remain unchanged when
as an example. We will consider the power flow equation the value of the random mask is 0. Figure 3 illustrates
and use the strong coupling relations between the reactive such a uniform crossover technique. Uniform cross-
power and voltage magnitude in a stressed power system; then over was first presented by Syswerda (1989), and its
the relation between the changes in reactive injection and main advantage is that the convergence speed is faster
voltage magnitude can be described as Q = JsV, where Js than that in one-point crossover or two-point crossover.
= J4 J3J11J2. In addition, the singular value decomposition (3) The mutation operator is a mean used to avoid losing
(SVD) is applied to the matrix Js. That is, Js = LSRT = important information at a particular position in the
n
L iS iR Ti , where L and R are n n orthonormal matrices,
i =1
decisions. The mutation operator with small mutation
probability, Pm, is applied to all the elements (genes)
S is a diagonal matrix, and Si is a singular value of matrix that are in each offspring (child). The mutation operator
Js with the order s1 s2 s3 ... sn 0. We know from of a GA for VSE is the element added or subtracted
Lf et al. (1992) that the minimum singular value, Sn, of Js by one step (scale) of regulating devices, such as the
can be used as a voltage security margin measure; that is, the load ratio of OLTCs, the excitation control of genera-
smaller the singular value is, the closer to voltage collapse tors or VAR source installation whenever a mutation
the operating point is. Thus, the minimum singular value, Sn, occurs, as shown in Fig. 4. Therefore, we can escape
is an indicator of proximity to the voltage security limit. Thus, from the local optimal solution and search for the
we can take Sn as the fitness function of GA. Since the GA
is basically an unconstrained search procedure in the original
problem domain, to incorporate the constraints into the ob-
jective function, any string that violates the constraints is
penalized by setting its fitness value to zero. In this manner,
the GA implicitly becomes a constrained search procedure.
56
Power System Voltage Security
57
C.W. Liu et al.
Table 1. Weak Bus Ranking under Heavy Load Conditions for the IEEE Table 3. The Limits of Controllers and Bus Voltages
30-Bus System
Voltage and Tap setting limits
a ri index load demand V max
g V min
g Tmax Tmin
Rank voltage angle PMU
Bus ri MW MVAR 1.1 0.87 1.050 0.910
Capacitor / Reactor settings ( = 0.04a (p.u.)) obtained in different runs using the genetic algorithm. Thus,
VAR 1 2 3 4 5 these results of the GA were the multi-selection and near-
initial 0.04 0.04 0.04 0.04 0.04 optimum solutions, shown in Table 5. In Table 5, we only
list three near-optimum solutions as control decision selections
Transformer taps settings (T = 0.015) (CDS) for VSE in every case. The best fitness in each genera-
OLTC T1 T2 T3 T4 tion was calculated. For example, the searching process for
initial 1 1 1 1 case 3 is shown in Fig. 7. In the figure, it is seen that the
best fitness of these generations clearly increased, and that the
Generator bus voltage settings (Vg = 0.01)
oscillatory phenomenon occurred because the voltage security
Bus 1 2 5 8 11 13 index had a large jump (refer to Fig. 6) at the alert level.
Vg (case 1 & 2) 1.05 1.05 1.0 1.0 1.0 1.0
Vg (case 3) 1.05 1.0 0.95 1.0 $b 0.95
Vg (case 4) 1.05 1.0 0.9 0.9 $b $b Case 1. The test system was investigated under heavy load conditions,
a and the system was operating at the alert level. The load demand
VAR source settings in p.u. on a 100 MVA base.
b
outage contingency condition. and power supply were:
58
Table 5. The Results of VSE Using GA on the IEEE 30-Bus System
59
CDS3 4 0.3705 0.9354 1.06 1.03 1.01 1.01 1.07 1.03 1.0 0.985 0.985 1.0 0.06 0.28 0.04 0.32 0.12
CDS2 4 0.3593 0.9262 1.06 1.03 1.01 1.0 1.0 1.0 0.97 0.985 0.97 0.08 0.04 0.24 0.08 0.36
CDS3 4 0.3578 0.9241 1.06 1.03 0.99 1.0 1.01 0.985 0.955 1.0 0.985 0.12 0.24 0.04 0.32 0.08
Table 6. Comparison between the Stochastic Approach and Deterministic Pg + Qg = 348.30 MW + 157.75 MVAR.
Approach
After the GA process finished, the objective function (NSV)
Stochastic approach Deterministic approach improved from 1 (very dangerous level) to 3 (alert level); the
performance GA SA Newton method Broydens method fitness function (Sn(Js)) increased from 0.0905 to 0.2067(CDS1),
accuracy ** ** * * 0.2042(CDS2) or 0.2010(CDS3); and the lowest voltage of the
flexibility *** ** * * load bus increased from 0.7662 to 0.8986(CDS1), 0.8966(CDS2)
efficiency ** * *** ** or 0.8931(CDS3). Thus, in this case, an improvement in the
stability *** *** * **
voltage security could be obtained to prevent voltage collapse.
simplicity *** * *** ***
Notes: ***distinguished, **mediocre, *disappointing In the above test cases, we could obtain results very close
to many results of numerical tests. Therefore, these results
are nearly global optimal solutions.
The controller settings and network states are shown in Table
5. After the GA process finished, the objective function (NSV) V. Discussion and Conclusion
improved from 3 (alert level) to 4 (secure level); the fitness
function (Sn(Js)) increased from 0.2157 to 0.3808(CDS1), 0.3796 With the increased loads on existing power transmission
(CDS2) or 0.3779(CDS3); and the lowest voltage of the load systems, VSE problems have become a major concern in
bus also increased from 0.9233 to 0.9495(CDS1), 0.9480(CDS2) power network operations. Most of the early works on VSE
or 0.9464(CDS3). were based on determinism. The deterministic approach often
Case 2. The test system was investigated under heavy load conditions cannot find optimal or even near optimal solutions, and the
and was operating at the alert level. The load demand was the searching process tends to lead to divergence or instability.
same as in Case 1, but we did not use weak buses as candidate It is difficult to deal with non-continuous, non-differentiable
buses for switching to new capacitors/reactors. After the GA and complex real-world optimization problems, and it is not
process finished, the objective function (NSV) improved from easy to deal with inequality constraints or obtain algorithms
3 (alert level) to 4 (secure level); the fitness function (Sn(Js)) for VSE. Also, only a feasible solution can usually be obtained,
increased from 0.2157 to 0.3718(CDS1), 0.3711(CDS2) or which may lack sufficient robustness to cope with the con-
0.3705(CDS3); and the lowest voltage of the load bus increased tingencies encountered.
from 0.9233 to 0.9371(CDS1), 0.9363(CDS2) or 0.9354(CDS3). On the other hand, stochastic methods offer more than
Table 5 compares the results of Cases 1 and 2. one optimal solution, can deal with non-differentiable con-
Case 3. The test system was investigated under heavy load conditions, strained search and optimization procedures, and can search
and generator #11 and line #18-21 were studied under outage a complicated and uncertain area to find global optimal or near
contingency conditions. The system was operating at the dan- optimal solutions. In addition, the stochastic methods have
gerous level. The load demand and power supply were: been shown to be more flexible, accurate and stabile than
deterministic methods (Chang and Huang, 1998; Hsiao et al.,
Pload + Qload = 325.07 MW + 138.53 MVAR, 1993; Goldberg, 1989; Lee et al., 1995; Chen and Chang, 1995;
Syswerda, 1989). Therefore, the stochastic approach has
Pg + Qg = 356.90 MW + 158.45 MVAR. become the prime candidate for constrained optimization and
search problems in recent years. In particular, both the SA
After the GA process finished, the objective function (NSV) and GAs have been used in VAR planning applications (Chang
improved from 2 (dangerous level) to 4 (secure level); the fitness and Huang, 1998; Hsiao et al., 1993; Lee et al., 1995), and
function (Sn(Js)) increased from 0.1859 to 0.3680(CDS1), 0.3593
(CDS2) or 0.3578(CDS3); and the lowest voltage of the load
bus increased from 0.8773 to 0.9293(CDS1), 0.9262(CDS2) or
0.9241(CDS3). Therefore, in this case, an improvement in the
voltage security level from the dangerous level to the secure level
was obtained.
Case 4. The test system was investigated under peak load, and generators
#5 and #13 were studied under outage contingency conditions.
The system was operating at the very dangerous level. The system
was on the verge of collapse. The load demand and power supply
were:
Pload + Qload = 337.65 MW + 152.70 MVAR, Fig. 7. Searching process of a GA for Case 3.
60
Power System Voltage Security
the SA has been shown to be able to obtain optimal VSE Canizares, C. A. and F. L. Alvarado (1993) Point of collapse and continuation
solutions (Chang and Huang, 1998). However, the SA has methods for large AC/DC systems. IEEE Trans. on Power Systems,
8(1), 1-8.
drawbacks in that a tremendous amount of execution time is Chang, C. S. and J. S. Huang (1998) Optimal multi-objective SVC planning
needed to obtain a near global solution and its control param- for voltage stability enhancement. IEE Proceeding-C, 145(2), 203-209.
eters are complex and difficult to handle. GAs, on the other Chebbo, A. M., M. R. Irving, and M. J. H. Sterling (1992a) Voltage collapse
hand, have inherent parallel computation ability, so they can proximity indicator: behavior and implications. IEE Proceeding-C, 139
search for many optimum solutions in parallel and use fitness (3), 241-252.
Chebbo, A. M., M. R. Irving, and M. J. H. Sterling (1992b) Reactive power
or objective function information directly according to the dispatch incorporating voltage stability. IEE Proceeding-C, 139(3), 253-
searching directions. GAs are more efficient, flexible and 260.
simple than the SA. Therefore, we find that GAs provide a Chen, P. K. and H. C. Chang (1995) Large-scale economic dispatch by
powerful search technique, unequalled by any other approach genetic algorithm. IEEE Trans. on Power Systems, 10(4), 1919-1925.
in terms of performance based on accuracy, flexibility, Freris, L. L. and A. M. Sasson (1968) Investigation of the load-flow problem.
IEE Proceeding-C, 115(10), 1459-1470.
efficiency, stability and simplicity. A comparison between Gao, B., G. K. Morison, and P. Kundur (1992) Voltage stability evaluation
stochastic and deterministic methods is shown in Table 6. using Modal analysis. IEEE Trans. on Power Systems, 7(4), 1529-1542.
To sum up, based on the numerical examples and test Goldberg, D. E. (1989) Genetic algorithms in search. In: Optimization and
results given in Section IV, we can make the following ob- Machine Learning. Addison-Wesley, Reading, MA, U.S.A.
servations: Hegdt, G. T. and W. M. Grady (1983) Optimal Var siting using linear load
flow formulation. IEEE Trans. on Power Apparatus and Systems, PAS-
(1) Searching for control decisions for VSE using GAs 102(5), 1214-1222.
is efficient and effective. Hsiao,Y. T., C. C. Liu, H. D. Chiang, and Y. L. Chen (1993) A new approach
(2) These searched control decisions for VSE are near for optimal Var sources planning in large scale electric power systems.
optimal, flexible solutions. IEEE Trans. on Power Systems, 8(3), 988-996.
(3) Control decisions searched using GAs are accurate and IEEE Publication (1990) Voltage Stability of Power Systems: Concepts,
Analytical Tools, and Industry Experience. 90TH0358-2-PWR, IEEE
stable. Service Center, Piscataway, NJ, U.S.A.
(4) Reactive power sources can be dispatched using GAs Lee, K. Y., X. Bai, and Y. M. Park (1995) Optimization method for reactive
to enhance voltage security. power planning by using a modified simple genetic algorithm. IEEE
(5) Dispatching of reactive power sources using GAs is Trans. on Power Systems, 10(4), 1843-1850.
easy to implement. Liu, C. W., C. S. Chang, and M. C. Su (1998) Neuro-fuzzy networks for
voltage security monitoring based on synchronized phasor measurements.
(6) GAs have potential as efficient on-line tools for VSE. IEEE Trans. on Power Systems, 13(2), 326-332.
Lf, P. A., T. Semed, G. Andersson, and D. J. Hill (1992) Fast calculation
Acknowledgment of a voltage stability index. IEEE Trans. on Power Systems, 7(1), 54-
64.
We thank the National Science Council of the Republic of China Mangooli, M. K., K. Y. Lee, and Y. M. Park (1993) Optimal long-term
for financial support under grant NSC 87-2213-E-002-947. reactive power planning using decomposition network techniques. Electric
Power Systems Research, 26, 41-52.
Nanba, M., Y. Huang, T. Kai, and S. Iwamoto (1998) Studies on VIPI based
References control methods for improving voltage stability. Electrical Power &
Energy System, 20(2), 141-146.
Ajjarapu,V., P. L. Lau, and S. Battula (1994) An optimal reactive power Schlueter, R. A. (1998) A voltage security assessment method. IEEE Trans.
planning strategy against voltage collapse. IEEE Trans. on Power on Power Systems, 13(4), 1423-1438.
Systems, 9(2), 906-917. Syswerda, G. (1989) Uniform crossover in genetic algorithms. Proceeding
Aoki, K., M. Fan, and A. Nishikori (1988) Optimal Var Planning by of the 3rd International Conference on Genetic Algorithms, pp. 2-9,
approximation method for recursive mixed integer linear planning. IEEE George Mason University, Fairfax, VA, U.S.A.
Trans. on Power Systems, 3(4), 1741-1747. Taylor, C. W. (1994) Power System Voltage Stability. McGraw-Hill, Inc.,
Begovic, M. M. and A. G. Phadke (1992) Control of voltage stability using New York, NY, U.S.A.
sensitivity analysis. IEEE Trans. on Power Systems, 7(4), 114-123. Van Cutsem, T. (1991) A method to compute reactive power margins with
Billiington, R. and S. S. Sachdev (1983) Optimum network Var planning respect to voltage collapse. IEEE Trans. on Power Systems, 6(1), 145-
by nonlinear programming. IEEE Trans. on Power Apparatus and 156.
Systems, PAS-102(5), 1214-1222.
61
C.W. Liu et al.
* * **
*
**
GA
SVCs
IEEE 30
62