Three multi-objective optimization algorithms have been developed to tackle a direct load control problem in electricity distribution networks. In this paper these algorithms are presented and compared: a hybrid multi-objective algorithm based on Greedy Randomized Adaptive Search Procedure (GRASP) and Simulated Annealing (SA) called GRASP+SA, an evolutionary algorithm (EA) with an embedded construction phase (EGRASP), and an EA with incorporation of preferences (EvABOR-III). Preferences are incorporated into the three algorithms using an outranking relation to sort the non-dominated solutions in categories according to the preferences elicited from a decision maker.