A two-phase approach to provide decision support in the design and selection of load control actions in electricity networks is described. The first phase consists in the use of a multiple objective genetic algorithm, coupled with physically-based load models, for the design and selection of a set of potential non-dominated solutions (load control actions). The second phase performs the selection of the control action, to be implemented over groups of electrical loads, based on the analysis of indifference regions associated to each alternative in a parametric diagram.