Analysis of diverse optimisation algorithms for pump scheduling in water supply systems

Abstract

Nowadays, the major expenses with water supply systems (WSS) correspond to energy consumption. The number of scientific works dealing with operational optimisation in WSS has been increasing over the past years, demonstrating significant reductions on energy costs and consumption. Pump stations usually represent the major portion of total energy costs in WSS. Consequently, in this work, it is pretended to give a contribution for energy efficiency improvement in pump stations. Generally, in WSS, the pumps are switched on when the reservoirs, responsible for supplying certain populations, reach their minimum levels. These pumps are only switched off when the reservoirs reach their maximum levels. The introduction of an operational pump pattern adapted to the energy tariff variation and the consumption patterns of the populations can optimise pump stations operations, minimising energy costs significantly. However, the process of finding the best pattern can present difficulties due to the complexity of some WSS (multiple pumps, multiple reservoirs, nonlinear behaviour of the systems, etc). In this work, an interface was developed with the aim of applying different optimisation algorithms for pump scheduling in WSS. The interface makes an automatic connection between a hydraulic simulator (EPANET 2.0) and the different optimisation algorithms selected, providing, after multiple iterations and evaluations, an optimal pump pattern for a certain water supply network represented. Two different examples of water supply networks are introduced in this study in order to validate the developed methodology. For both WSS, classic and meta-heuristic optimisation algorithms are tested and analysed.