Dispatch of interruptible loads using binary particle swarm optimization: A comparison of constraint-handling methods

Interruptible loads are consumers who agree to be interrupted, or whose consumption may be reduced by the utility, in order to maintain system security or to reduce market prices. Incorporating interruptible loads in day-ahead markets could maximize their potential in providing ancillary services because it enables them to plan their operations, and for them to offer reliable curtailment capacities and competitive bids. This paper investigates the use of binary particle swarm optimization to schedule a set of interruptible loads over a 16-hour period. The scheduling objective is to minimize the total payments to the interruptible loads and the frequency of interruptions, and to satisfy the required hourly curtailments and operational constraints imposed by the interruptible loads. The constraints in this multi-objective optimization problem were handled using four constraint-handling methods: using static and adaptive penalty functions, tracking only feasible solutions, and using repair algorithms. The suitability of these constraint-handling approaches to the problem at hand were investigated and compared.