Professional Documents
Culture Documents
On Multiobjective Volt-VAR Optimization in Power Systems
On Multiobjective Volt-VAR Optimization in Power Systems
1
P1, Q1
Fig. 2. Transmission system model
TS Optimize: F
0
Subject to: G(G, V, Qc) = 0 (1)
Qci = k Qc0 k = 0, 1, 2 …
where:
4 3 1 2 F - vector of objectives
G - set of power flow equations
G - vector of voltage angles
V - vector of voltage magnitudes
10 8 5 6 Qc- vector of reactive support
Qci - reactive support applied to the bus “i”
Qc0 - incremental reactive support step
kVA/phase. The optimization results are presented in This problem can be solved using the same genetic
Fig. 5 and Table 2. Figure 5 shows the Pareto-optimal algorithm as the distribution feeder. The cost of the
front of the solutions, while Table 2 provides numerical transmission reactive support is assumed to be PT =
explanations for only a few of the possible solutions. 10 $/kVA. The incremental reactive support step is
Due to the linear relationship between investment and assumed to be Qc0 = 1 MVA/phase. The optimization
the amount of reactive support, the latter is shown as an results are presented in Fig. 6 and Table 3. Figure 6
objective in Fig. 5. shows the transmission system Pareto-optimal front of
solutions, while Table 3 provides numerical
6
explanations for a few of the possible solutions. Due to
the linear dependence between investment and the
5 amount of reactive support, the latter is shown as an
objective in Fig. 6.
Rea ctive supp ort (M V A )
4
90
80
3
70
Reactive support (M V A )
2
60
50
1
40
0
270 290 310 330 350 370 390 410 430 450 470 30
Optimal feeder schedule: BUS1 - 50%, BUS3 - 50% initial solution (86MVA support
1.1 to transmission system)
no reactive support
The transmission system was supported with 14MVA, 1 solution that minimizes loses
while the overall feeder support was 48MVA (ten
0.9
feeders compensated with 4.8MVA each). The fact that
0.3
5. Impact on voltage stability margin
0.2
1 1.1 1.2 1.3 1.4 1.5
The influence of the capacitor allocations discussed in Loading factor (p.u.)
the above example on the voltage stability margin is Fig. 8. System’s PV curves for different
investigated. Continuation power flow, applied on the capacitor scenarios
system model without reactive support, yields some
alarming data. The critical loading factor of the system Decomposition of the transmission model from the
is λ = 1.108 (10.8 percent load increase before voltage distribution feeders is proposed as a necessary step to
collapse). Applying reactive support to the system is reduce the complexity of the problem. After the
expected to be beneficial, but it is not known whether decoupled parts of the system are solved independently,
the transfer of reactive support from the transmission to a suitable interface is designed for simultaneous
the distribution portion of the system would worsen the optimization. Custom designed genetic algorithms are
voltage stability loading margin. The answer appears to used as multiobjective optimization tools that rely on
be negative. Transferring reactive support to the sensitivity analysis to reduce the search space and allow
distribution network decreases the reactive load of the implementation for large system models.
transmission system and increases the system voltage
stability loading margin. Figure 8 illustrates this 7. Acknowledgment
observation. The following PV curves are shown in
Fig.8: Financial support of the National Electric Energy
• No reactive support to system (λ = 1.108) Testing, Research and Applications Center (NEETRAC)
• Entire ($860k) reactive support applied to the used for part of the work presented in this paper is
transmission network (critical loading margin gratefully acknowledged. The authors also would like to
increased to λ=1.294) acknowledge Dr. Damir Novosel, with whom they had
• Minimal loss scenario (critical loading margin many fruitful discussions about multi-objective
further increased to λ=1.508) optimizations.
6. Conclusions 8. References
The problem of simultaneous optimization of reactive [1] D Goldberg, Genetic Algorithm in Search, Optimization
resources on the transmission and distribution system is and Machine Learning, New York: Addison Wesley,
solved by decoupling the analysis of the transmission 1989.
and distribution networks. The investment resources are [2] B. Baran, J. Vallejos, R. Ramos, U. Fernandez "Reactive
assumed to be limited and known. Under this Power Compensation using a Multi-objective
constraint, a number of optimization objectives can be Evolutionary Algorithm" IEEE Porto Power Tech
chosen (distribution losses, distribution feeder power Conference, Porto, Portugal September, 2001
[3] J.T. Ma, L.L. Lai “Evolutionary Programming Approach
factor, voltage profile for conservative voltage
to Reactive Power Planning” IEE Proceedings –
reduction, transmission losses, transmission capacity, Generation, Transmission and Distribution, Vol 143, No.
voltage stability, etc.). This paper addresses the various 4, July 1996
issues that need to be resolved to solve this problem.