Genetic algorithms for optimal reactive power compensation planning on the National Grid system

A modified integer genetic algorithm (IGA) is presented. The algorithm is applied to the full reactive-power compensation planning (RCP) problem. The algorithm solves both the siting problem and the operational problem, which it does simultaneously for the intact/base case and any contingent cases, for practical sized systems. The objective of the optimisation is to produce a power system which (in terms of the cases considered) does not violate any system or operational constraint and is optimal in terms of the voltage deviation from the ideal and the cost incurred though the installation and use of reactive power compensation devices. This multi-objective problem is solved through the use of Pareto optimality. The algorithm is tested on a practical sized problem which was developed with the cooperation of the UK National Grid. It is compared with the results generated by the package SCORPION: a RCP program currently being used by National Grid. This paper shows that GAs can be successfully applied to practical sized systems when multiple operating states are considered; in addition to siting devices, the system performance is optimised via adjustment of tap settings and controller characteristics; and both fixed and dynamic compensation devices are being sited.