On the Use of Parallel Programming Techniques for Real-Time Scheduling Water Pumping Problems

Abstract

Most of the energy consumed by a water company is used for pumping systems. The electricity has a hourly pricing policy, therefore finding in real time the optimal schedule to operate those systems drastically reduces the power bill. Scheduling pumping problem with three main elements: pumping system, tank and water demand to be satisfied is analyzed. The mathematical programming model and techniques used to solve the problem, considering as known the water demand are exposed. Parallel programming paradigm is proposed to solve this problem when introducing stochastical programing techniques (scenario tree evaluation) and multi site problem. Mixing classical mathematical programing techniques and parallel tools (MPI and OpenMP), numerical experiments on parallel computers are designed and completed. As a result parallel programing strategy is experimentally proved as a useful technique to improve the real-time pumping scheduling problem solving.