# Optimal Capacitor Placement

I have 50 caps to be distributed in a 5000 bus network. How to choose the optimal capacitor placement to improve voltage profile using PSS/E?

edit retag close merge delete

Sort by » oldest newest most voted

A QV study under contingency will identify buses with reactive power deficiencies. Grouping those buses within a geographical area, wil make the sink area(s) in a PV study (against the rest of the system) to verify/identify voltage collapse conditions under contingency. Modal analysis of the worst scenarios would verify if the number of buses participating in the voltage collapse is large (system voltage collapse) or small (local voltage collapse). Capacitors can be place at those buses with higest participation factor (from modal analysis) and their sizes will be enough to eliminate the voltage collapse outcome or sufficient to provide additional voltage stability margin.

To place dynamic reactive devices, similar tests can be performed with dynamic simulations under fault/contingency.

more

Probably the best he could do but remember such trial-and-error approach is never optimal. Also note that his objective is to improve system voltage profile not maximising voltage collapse margins.

( 2015-05-25 22:37:58 -0500 )edit

It's might be not the optimal solution, but I think you should consider buses with low voltage profiles first.

Then, according to your size and type of capacitors, you determine the position to place them (like in the load-end buses or buses with many interconnections). Load flow solution will show how effective your placements are.

more

This is a very difficult problem to solve. Theoretically the best solution would be to employ an OPF algorithm with its objective being voltage profile improvement, and its control variables being the locations of capacitor placement.

However I don't think any of the available OPF objectives in PSSE is suitable for your use. Even if there is one, control variables still won't suit your problem.

Rewriting an OPF algorithm using Python for the dimension of your problem (5000 bus) is probably out of question as it involves deriving complicated Hessian matrices.

You could run numerous load flows to check impacts of different placements, i.e. trial-and-error type approach, but it won't be optimal.

more