An optimization algorithm for load-shifting of large sets of electric hot water tanks

Large sets of Electric Hot Water Tanks (EHWT) found in numerous countries appear as very relevant for load-shifting applications. An EHWT heats water over relatively long periods of time, for later consumption. The simple time-of-use pricing strategy applied to most consumers increases the overall electricity consumption in the beginning of the night, while hot water is used much later, in the next daytime. With the development of home automation, more advanced strategies could enable improved cost-reductions for both users and the electricity producer. An efficient and robust improvement is to reschedule each EHWT heating, while keeping the heating periods undivided (to minimize the malicious impacts of complex dynamic effects taking place inside each EHWT), so that an objective load curve for the whole set of EHWT is reached. The first contribution of this article is to formulate the rescheduling of EHWT heating with comfort constraints as an optimization problem. The second contribution is a heuristic to solve this problem with less than 1% optimality loss for representative test cases with large sets of EHWT (typically hundreds of thousands).