A Novel Approach to Avoiding Technically Unfeasible Solutions in the Pump Scheduling Problem

Optimizing pump operation in water networks can effectively reduce the cost of energy. To this end, the literature provides many methodologies, generally based on an optimization problem, that provide the optimal operation of the pumps. However, a persistent shortcoming in the literature is the lack of further analysis to assess if the obtained solutions are feasible from the technical point of view. This paper first showed that some of these available methodologies identify solutions that are technically unfeasible because they induce tank overflow or continuous pump switching, and consequently, proposed a novel approach to avoiding such unfeasible solutions. This consisted in comparing the number of time-steps performed by the hydraulic simulator with the predicted value, calculated as the ratio between the simulation duration and the hydraulic time-step. Finally, we developed a new model which couples Epanet 2.0 with Pikaia Genetic Algorithm using the energy cost as an objective function. The proposed method, being easily exportable into existing methodologies to overcome the limitations thereof, thus represents a substantial contribution to the field of pump scheduling for optimal operation of water distribution networks. The new method, tested on two case studies in the literature, proved its reliability in both cases, returning technically feasible solutions.