Académique Documents
Professionnel Documents
Culture Documents
a b
Department of Electrical Engineering , The Hong Kong Polytechnic University , Hong Kong
Published online: 19 Jun 2007.
To cite this article: H. Wu , C. W. Yu , N. Xu , X. J. Lin & W. H. Chen (2007) Analysis of Reactive Power Support
of Generators Using Power Flow Tracing Methods, Electric Power Components and Systems, 35:9, 1079-1092, DOI:
10.1080/15325000701250231
To link to this article: http://dx.doi.org/10.1080/15325000701250231
H. WU
Department of Electrical Engineering
Zhejiang University
Hangzhou, China and
Department of Electrical Engineering
The Hong Kong Polytechnic University
Hong Kong
C. W. YU
N. XU
X. J. LIN
W. H. CHEN
Department of Electrical Engineering
The Hong Kong Polytechnic University
Hong Kong
Abstract With the emergence of competitive electricity power markets, reactive
power ancillary services and reactive power markets have attracted attention from
researchers and system operators all over the world. Reactive power is an important
system support service for secure and reliable operation of power systems. Improper
management of reactive power can also hinder the operational efficiency of other
power markets. It has been recognized that the reactive power of a generator has
several roles, namely, supplying reactive demand, maintaining system security and
supporting its real power transmission. It is rational that the minimal reactive power
used to support its real power transmission should not receive financial compensation
in power markets. Hence, this component of reactive power can be regarded as the
minimal reactive power support of a generator. A reactive power optimization model
along with a power flow tracing based method is proposed in this article to tackle
this problem. The validity and rationality of the approach are verified using a simple
meshed 5-bus system.
Keywords reactive power ancillary service, reactive power support, transmission
open access
1. Introduction
Reactive power has profound effects on real energy transfers and on the security of
power systems, as it affects the voltage profile throughout the system. Reactive power
Received 22 August 2006; accepted 2 January 2007.
Address correspondence to Dr. H. Wu, Dept. of Electrical Engineering, Zhejiang University,
Hangzhou, 310027, China. E-mail: vuhao@zjuem.zju.edu.cn
1079
1080
H. Wu et al.
1081
jQg i j
(1)
i 2SG
s.t.:
Pg i
Pli
Vi
i 2 SB
(2)
i 2 SB
(3)
Qg i < Qg i < Q g i
i 2 SG
(4)
V i < Vi < V i
i 2 SB
(5)
Qg i
Qli
Vi
Vj .Gij sin ij
Bij cos ij / D 0
where Pg i and Qg i are the real power output and reactive power output of the generator
at bus i , respectively; Pli and Qli are the real power demand and reactive power demand
of the load at bus i , respectively; Vi is the voltage magnitude of bus i ; V i and V i are the
lower and upper voltage magnitude limits of bus i , respectively; ij is the difference in
the voltage angles between bus i and bus j ; Gij C jBij is the admittance between bus i
and bus j ; SB is the set of all buses; and SG is the set of all generators. Equations (2)
and (3) are power flow equations. Inequality (4) represents the generators reactive output
limits. Inequality (5) represents voltage magnitude limits. All real power outputs of the
generators are fixed except for one generator, which serves as the slack generator to make
good any losses in transmission. As a result, the real power flow pattern is obtained, and
the reactive power of a generator that is used to support its real power transmission can
be studied. The control or optimized variables are the reactive outputs of the generators.
Solving this model, the total minimal reactive power support of the generators can be
assessed. The following are some explanations of the model.
Objective Function. The objective function of the model is the summation of the absolute value of all reactive power produced or absorbed. Notice that there are no cost
related items, such as operating cost and lost opportunity cost, because the problem under study is the minimal reactive power support of generators which are not cost related.
There may be some cross-subsidies among generators. If some generators provide more
reactive power support, the others may provide less. However, under the equity principle,
each generator, no matter its capacity is large or small, should have the same priority to
provide less reactive power support in a power market. The simple summation of absolute
value of reactive power supports in (1) is an attempt to model this. It is easy to see that
when the reactive output limits are not violated, the Lagrange multipliers corresponding
to the reactive power flow equations are all 1.0 or 1.0 at the optimal solution, depending
if the generators are producing or absorbing reactive power. This characteristic can be
regarded as an indicator of equity.
Reactive Demand of the Load. The minimal reactive power support of a generator
should not include the reactive demand of the load because, in power markets, the loads
should pay for their reactive power demands and it is not mandatory for generators
to supply reactive loads. Therefore, Qli in Eq. (3) is set to zero when calculating the
minimal reactive support of generators.
1082
H. Wu et al.
Voltage and Line Current Limit. The voltage magnitude limits are regarded as compulsory constraints. Line current limits were not considered because the system operator
took the limits into account when dispatching the real power energy transactions.
Shunt Compensation. Shunt compensators are modeled implicitly in Eq. (3). They are
not modeled explicitly because it is assumed that, before analyzing the minimal reactive
power support of generators, the shunt compensators that are scheduled to operate have
been determined. One reason behind this assumption is that compensators and generators
may belong to different entities and, hence, it cannot be expected that compensators will
cooperate with generators to reduce the minimal reactive supports of the latter without
any economic incentives. Line charging reactive power can also be regarded as shunt
compensation and hence it is modeled similarly in Eq. (3).
On Load Tap Changer (OLTC) Transformer. OLTCs are assumed to be fixed in the
model. The reason is similar to that for shunt compensators, i.e., OLTCs may not cooperate with generators to reduce the minimal reactive support to the latter without any
economic incentives because OLTCs and generators may belong to different entities.
2.2. Solution Method
Models (1)(5) can be regarded as a nonlinear programming (NLP) problem with the
following general form:
min f .x/
(6)
s.t.: h.x/ D 0
(7)
(8)
1083
it separately identifies the contribution or consumption of the real or reactive power for
a particular generator or load.
As explained in Section 2.1, all reactive loads are set to zero. Therefore, the reactive
outputs of generators are used to support the transmission of real power and will be
dissipated in the network. Hence, the reactive power support of generators can be regarded
as supplying reactive losses to a power system. However in a power system, in addition
to generators, line charging capacitances and shunt capacitors can also supply reactive
losses. The reactive power tracing method proposed by Bialek [18] is used in this article
to identify the amount of the reactive loss of each transmission branch supplied by
generators. After that, a new problem of allocating reactive losses to generators is formed.
A simple and direct method of solving this problem is to allocate the branch reactive
loss to different generators according to the shares of the branch real power that the
generators account for. Again, the real power tracing method described in [18] is used to
identify the composition of the real power of each transmission branch and to determine
the real power contribution of each generator to each branch. It is assumed here that
reactive power losses are all incurred by the transmission of real power.
Just as there is no a unique method for allocating real power losses [2023], there are
also alternative methods for allocating reactive power losses. Two methods are proposed
here, namely, the proportional allocation method and the quadratic allocation method.
Their formulae are given in (9) and (10) below. Upon solving models (1)(5) and carrying
out the reactive power flow tracing procedure, the reactive loss of a branch b that are
g
supplied by all generators Qloss;b can be obtained. Assume that there are n components
of real power Pi (i D 1; : : : ; n) coming separately from n generators and denote the
gk
reactive loss allocated to generator k by Qloss;b .
Proportional Allocation. This is the most intuitive and straightforward approach that
results from the proportionality linear assumption.
gk
Qloss;b D Qloss;b
Pk
n
X
(9)
Pi
i D1
Note that the summation in the denominator is actually the real power flow of
branch b.
Quadratic Allocation. Since power losses increase quadratically with power flows, it is
also reasonable to use formula (10). A similar concept can be found in [20].
!
n
2
X
P
Pk2 C
2Pk Pi 2 k 2
Pk C Pi
i D1
gk
g
Qloss;b
D Qloss;b
(10)
!
2
n
X
Pi
i D1
This formula splits the cross term 2Pk Pi between generator i and generator k
into two different items that are proportional to the squares of these two real power
components.
Obviously, if the cross term is split equally, the result will be the same as that using
the proportional allocation method. After allocating the reactive losses of all branches,
1084
H. Wu et al.
the reactive power support of a particular generator is the sum of the reactive power
allocated to it.
2.4. Procedure to Assess the Minimal Reactive Power Support of Generators
As a summary, the following procedure to assess the minimal reactive power support of
a generator is given.
1. Solve models (1)(5) to get the state of the system at the optimal point. The
resulting reactive output of all generators is the total minimal reactive power
support of the generators.
2. Perform a reactive power flow tracing algorithm and determine the reactive loss
of each branch that comes from the generators.
3. Perform a real power flow tracing algorithm and determine the real power flow
composition of each branch.
4. Using Eqs. (9) or (10), allocate the reactive loss of each branch to generators
according to the real power flow composition of that branch.
5. Add up the allocated reactive losses of each generator accordingly. The sum is
the allocated minimal reactive power support of the generator.
3. Case Studies
3.1. A Simple 5-bus System
For illustrative purposes, a simple 5-bus system with two generators, three loads, and six
transmission lines is used. The topology, load demands, and real power generations of the
system are all shown in Figure 1. Table 1 gives the network parameters. All parameters
and values are in per unit.
Table 2 lists the results of three case studies. The differences among these cases lie
only in the values of the upper voltage magnitude limit. The upper voltage magnitude
limits of all buses in cases A, B, and C are set at 1.05 p.u., 1.07 p.u., and 1.10 p.u.
respectively, while the lower voltage magnitude limits are set at 0.95 p.u. in all three
cases.
The results show that the voltage magnitude at bus 1 is bound to the upper limit in all
three cases, and that the system real power losses Ploss and the reactive power output of
1085
Table 1
Network parameters for the 5-bus system
Line
no.
From
To
Resistance
(p.u.)
Reactance
(p.u.)
Line
charging
susceptance
(p.u.)
1
2
3
4
5
6
1
1
2
2
3
4
2
3
4
5
4
5
0.02
0.08
0.06
0.04
0.01
0.08
0.06
0.24
0.18
0.12
0.03
0.24
0.06
0.05
0.04
0.03
0.02
0.05
s1
generator 4 Qg4 decrease when the system upper voltage magnitude limit is relaxed. Qg1
,
s1
s2
s2
Qg4 , Qg1 , and Qg4 are the minimal reactive power support allocated to generators 1
and 4 using the proportional allocation method and the quadratic allocation method,
respectively. It can be seen that they are nearly the same in all case studies, respectively.
The error is within 5%. Therefore, the difference between the two allocation methods is
s1
s1
s2
s2
small. Also the sum of Qg1
and Qg4
, or Qg1
and Qg4
equals the sum of Qg1 and Qg4 .
As Qg1 is zero here, the sum therefore equals to Qg4 . Obviously, the allocated minimal
reactive power support decreases notably as the system upper voltage magnitude limits
increase, especially the amount of generator 1.
Table 3 gives the real power flow tracing results of the three case studies. Pg1 and Pg4
stand for the real power contributions that come from generators 1 and 4, respectively. It
can be seen that the real power flows are quite steady in all three cases. The differences
among them are less than 1%. However, as indicated in Table 2, the differences in the
reactive power flow are significant.
3.2.
In order to gain further insights into the characteristics of the proposed method and
the transportation of real power, the variations in the minimal reactive power support
Table 2
The results of three case studies
Variables
Case A
Case B
Case C
Variables
Case A
Case B
Case C
V1
V2
V3
V4
V5
Ploss
1.0500
1.0168
1.0296
1.0384
0.9826
0.1300
1.0700
1.0374
1.0488
1.0572
1.0039
0.1245
1.1000
1.0684
1.0777
1.0856
1.0357
0.1170
Qg1
Qg4
s1
Qg1
s1
Qg4
s2
Qg1
s2
Qg4
0.0000
0.1270
0.0689
0.0581
0.0715
0.0555
0.0000
0.1002
0.0511
0.0491
0.0532
0.0470
0.0000
0.0617
0.0271
0.0346
0.0285
0.0333
1086
H. Wu et al.
Table 3
The real power flow tracing results of three case studies
Case A
Case B
Case C
Load
no.
Demand
Pload
Pg1
Pg4
Pg1
Pg4
Pg1
Pg4
2
3
5
1.0
1.2
1.4
0.8997
0.3869
0.7538
0.1003
0.8131
0.6462
0.8994
0.3856
0.7537
0.1006
0.8144
0.6463
0.8990
0.3839
0.7535
0.1010
0.8161
0.6465
of generators under different conditions with varying bilateral transaction amounts are
studied in this section.
Assume that generator 4 and load 5 form a bilateral transaction. The system upper
and lower voltage magnitude limits are set at 1.1 p.u. and 0.9 p.u., respectively. With
the reactive demands of all the loads taken as zero, Figure 2(a) shows the changes
in the reactive power outputs and the minimal reactive power support allocated to
the generators when the amount of the bilateral transaction is varied. Qg1 is the actual reactive power output of generator 1; Qg1 -A1 is the minimal reactive power support allocated to generator 1 using the proportional allocation method; and Qg1 -A2
is the minimal reactive power support allocated to generator 1 using the quadratic allocation method. Similar notations apply to Qg4 , Qg4 -A1, and Qg4 -A2. Figure 2(b)
shows the variations in the system voltage profile against the real power taken from
bus 5.
Figure 2(a) shows that both the reactive output and the minimal reactive power
support of generators are zero when the real power taken from bus 5 is less than 1.2 p.u.,
and that the nodes voltages are not bounded to the system voltage magnitude limits.
When the real power taken from bus 5 is in the range of 1.2 to 1.8 per unit, generator 1
still does not produce any reactive power. However, the terminal bus of generator 1 hits
the system upper voltage limit. It is important to note that part of the reactive power
produced by generator 4 is used to support the real power transmission of generator 1.
When the real power taken from bus 5 is higher than 1.8 p.u., generator 1 begins to
produce reactive power and bus 4 also hits the upper voltage limit. The voltage of bus 5
drops gradually as the loading at bus 5 increases. When the power taken from bus 5
is 3.4 p.u., the voltage approaches the system low voltage limit (0.9 p.u). Beyond that
point, models (1)(5) cannot be solved.
The difference between the allocation results of the proportional method and quadratic
allocation method is small when the power taken from bus 5 is less than 2.0 p.u., but
becomes obvious when it is greater than 2.5 p.u.
Figure 3(a) shows the changes in the real power compositions of the loads when the
amount of the bilateral transaction is varied. PL2-G1 is the share of the real power of
load 2 that is supplied by generator 1, and similar explanations apply to other notations.
Obviously, as the real power of the bilateral transaction is increased, both generators
contribute more real power to load 5. At the same time, loads 2 and 3 receive more real
power from generator 4 and less real power from generator 1. Comparing with loads 2
and 3, load 5 is remote from generator 1. This explains the reason why the minimal
reactive power support of generator 1 increases although generator 1 is not involved in
the economic bilateral transaction.
1087
Figure 2. Relationship between the real power of the bilateral transaction and the (a) reactive
power support of generators and (b) system voltage magnitudes.
Figure 3(b) shows the relationship between the real power of the bilateral transaction
and the total reactive power that need to support the real power transmission of the generators. In addition to the reactive power from generators, the reactive power from line
charging capacitances is also allocated to generators 1 and 4, using a method similar to
step 4 of the procedure described in Section 2.4. Here, reactive charging power instead of
0
reactive loss of each branch is allocated to the generators. Qg1
-A1 is the total allocated
reactive power that is used to support the real power transmission of generator 1 using
0
0
the proportional allocation method. Similar explanations hold for Qg1
-A2, Qg4
-A1, and
1088
H. Wu et al.
Figure 3. Relationship between the real power of the bilateral transaction and the (a) real power
compositions of the loads and (b) total reactive power support of the generators.
0
Qg4
-A2. It can be observed that although generator 1 produces no reactive power when
the real power taken from bus 5 is less than 1.8 p.u., it gets significant reactive power
support from line charging capacitances. The reactive power support of generator 4 from
line charging capacitances is not as large as that to generator 1. As the real power of the
transaction becomes larger and larger, the total reactive support of generator 4 increases
rapidly and finally exceeds that for generator 1. Taking into account Figure 2(a), it can
be seen that generator 1 gets a significant amount of reactive power supports from both
1089
Figure 4. Relationship between the real power of the bilateral transaction and the (a) reactive
power support of generators and (b) the system voltage magnitudes.
generator 4 and line charging capacitances and, hence, can produce less reactive power.
Generator 4 not only gets much less reactive power support from line charging capacitances compared with generator 1 but also needs to support the real power transmission
of generator 1. Therefore, it needs to produce more reactive power.
Figures 4 and 5 show the results when generator 1 and load 5 form a bilateral
transaction. The system upper and lower voltage magnitude limits are 1.10 and 0.9,
1090
H. Wu et al.
Figure 5. Relationship between the real power of the bilateral transaction and the (a) real power
compositions of the loads and (b) total reactive power support of the generators.
respectively. These results are essentially similar to the results presented in Figures 2
and 3. Compared with Figure 2, it is interesting to note that generator 1 produces less
reactive power itself while gets more reactive power support from generator 4.
The above studies show that the proposed methodology can produce reasonable and
convincing results. Because of the local characteristic of reactive power, it is possible that
one generator is more efficient to producing reactive power than other generators from
the system viewpoint. Therefore, some methods of allocation must be applied to identify
1091
the contributions and responsibilities of each generator. Only on the basis of an equitable
criterion can a reactive power market possibly be set up. It is also found that power flow
due an economic transaction, following Kirchhoffs law, would distribute over different
routes that lead through the transmission network. As a result, assigning the minimal
reactive power support of a generator solely according to the real power output specified
by an economic contract may lead to misleading conclusions. Moreover, the method of
allocating reactive losses is not unique. Different methods of allocating reactive losses
may lead to slightly different amounts of reactive support of generators.
4. Conclusions
Reactive power from generators is critical to voltage security. A reactive power optimization model along with a power flow tracing based approach to assess the minimal
reactive power support of a generator that is used to support the generators real power
transmission has been established in this paper. Considering that supplying the reactive
demand of the load is a service and not a responsibility of the generators in a power
market, the reactive demands of the loads are taken to be zero in the evaluation. An optimization problem with the objective of minimizing the total reactive power support of
generators is firstly solved. Because the reactive demands are taken as zero, a correlation
exists between the reactive power support of generators and the reactive losses in the
network. Upon obtaining the solution for the minimal total reactive power support of the
generators, a reactive power flow tracing method is then used to identify the amount of
the reactive loss of each transmission branch that is supplied by the generators. After
that, a real power flow tracing method is used to allocate the reactive loss of the each
transmission branch to different generators according to two proposed criteria. Finally,
the minimal reactive power support of each generator can be evaluated by summing the
allocated reactive losses from all of the transmission branches. The calculated minimal
reactive power support of the generators can serve as a base for the clearing of a reactive
power market.
Acknowledgments
The authors gratefully acknowledge the financial support of the Research Grants Council of Hong Kong (PolyU 5215/03E) and the Research Committee of the Hong Kong
Polytechnic University (G-T618).
References
1. Trehan, N. K., Ancillary servicesreactive and voltage control, IEEE Power Engineering
Society Winter Meeting, Vol. 3, pp. 13411346, 2001.
2. Hao, S., and Papalexopoulos, A., Reactive power pricing and management, IEEE Trans.
Power Systems, Vol. 12, pp. 95104, February 1997.
3. Baughman, M. L., Siddiqi, S. N., and Zarnikau, J. W., Advanced pricing in electrical systems,
Part I: Theory, Part II: Implications, IEEE Trans. Power Syst., Vol. 12, pp. 489502, February
1997.
4. Gil, J. B., Roman, T. G. S., Rios, J. J. A., and Martin, P. S., Reactive power pricing: A
conceptual framework for remuneration and charging procedures, IEEE Trans. Power Syst.,
Vol. 12, pp. 483489, May 2000.
5. Bhattacharya, K., and Zhong, J., Reactive power as an ancillary service, IEEE Trans. Power
Syst., Vol. 16, pp. 294300, May 2001.
1092
H. Wu et al.
6. Da Silva, E. L., Hedgecock, J. J., Mello, J. C. O., and Da Luz, J. C. F., Practical cost-based
approach for the voltage ancillary service, IEEE Trans. Power Syst., Vol. 16, pp. 806812,
November 2001.
7. Lin, X. J., David, A. K., and Yu, C. W., Reactive power optimization with voltage stability
consideration in power market systems, IEE Proc. Gener. Transm. Distrib., Vol. 150, pp. 305
310, May 2003.
8. Zhong, J., and Bhattacharya, K., Toward a competitive market for reactive power, IEEE
Trans. Power Syst., Vol. 17, pp. 12061215, November 2001.
9. Zhong, J., and Bhattacharya, K., Reactive power management in deregulated electricity
marketsA review, IEEE Power Engineering Society Winter Meeting, Vol. 2, pp. 12871292,
2002.
10. Xu, W., Zhang, Y., Da Silva, L. C. P., and Kundur, P., Assessing the value of generator
reactive power support for transmission access, IEE Proc. Gener. Transm. Distrib., Vol. 148,
pp. 337342, July 2001.
11. Xu, W., Zhang, Y., Da Silva, L. C. P., Kundur, P., and Warrack, A. A., Valuation of dynamic
reactive power support services for transmission access, IEEE Trans. Power Syst., Vol. 16,
pp. 719728, November 2001.
12. Da Silva, L. C. P., Zhang, Y., Xu, W., and Da Costa, V. F., Investigation on the dual functions of
generator reactive power support, IEEE Power Engineering Society Summer Meeting, Vol. 3,
pp. 16161620, 2001.
13. Xu, W., and Wang, Y., An investigation on the reactive power support service needs of power
producers, IEEE Trans. Power Syst., Vol. 19, pp. 586593, February 2004.
14. Chicco, G., Gross, G., and Tao, S., Allocation of the reactive power support requirements in
multi-transaction networks (republished), IEEE Trans. Power Syst., Vol. 17, pp. 12831289,
November 2002.
15. Quintana, V. H., Torres, G. L., and Medina-Palomo, J., Interior-point methods and their
applications to power systems a classification of publications and software codes, IEEE Trans.
Power Syst., Vol. 15, pp. 170176, February 2000.
16. Wei, H., Sasaki, H., Kubokawa, J., and Yokoyama, R., An interior point nonlinear programming for optimal power flow problems with a novel data structure, IEEE Trans. Power Syst.,
Vol. 13, pp. 870877, August 1998.
17. Kirschen, D., Allan, R., and Strbac, G., Contributions of individual generators to loads and
flows, IEEE Trans. Power Syst., Vol. 12, pp. 5260, February 1997.
18. Bialek, J., Tracing the flow of electricity, IEE Proc. Gener. Transm. Distrib., Vol. 143,
pp. 313320, July 1996.
19. Wu, F. F., Ni, Y. X., and Wei, P., Power transfer allocation for open access using graph
theoryfundamentals and applications in systems without loopflow, IEEE Trans. Power Syst.,
Vol. 12, pp. 95104, February 1997.
20. Exposito, A. G., Santos, J. M. R., Garcia, T. G., and Velasco, E. A. R., Fair allocation of
transmission power losses, IEEE Trans. Power Syst., Vol. 15, pp. 184188, February 2000.
21. Gross, G., and Tao, S., A physical-flow-based approach to allocating transmission losses in a
transaction framework, IEEE Trans. Power Syst., Vol. 15, pp. 631637, May 2000.
22. Conejo, A. J., Galiana, F. D., and Kochar, I., Z-bus loss allocation, IEEE Trans. Power Syst.,
Vol. 16, pp. 105110, February 2001.
23. Galiana, F. D., Conejo, A. J., and Kochar, I., Incremental transmission loss allocation under
pool dispatch, IEEE Trans. Power Syst., Vol. 17, pp. 2633, February 2002.