M Baymani-Nezhad - Faculty of Engineering, Queen’s Building, University of Bristol, Bristol, BS۸ ۱TR
D.A Savic - Center for Water Systems, College of Engineering, Mathematics and Physical Sciences University of Exeter, Exeter,UK, EX۴ ۴QF

A Water Distribution System (WDS) is required to convey fresh water from the main water sources to consumption nodes. Depending on the topography of an area, water typically needs to be pumped to elevated parts of the network or service reservoirs to ensure that sufficient pressures are maintained in the WDS. Pumping stations contribute most to the overall energy consumption of a water utility as well as to its greenhouse gas footprint. Pumping stations consist of a number of pumps that run together or individually during different times of the day. This study presents an optimal pump scheduling strategy based on the two local search strategies to minimize pumping costs in a WDS. In the first stage of the model, a linear formulation of the pump scheduling problem is defined in the Microsoft Excel spreadsheet. An initial pump combination was found using Microsoft Excel Solver (ME-Solver) which is an optimization add- on to Microsoft Excel. Compliance of the solution obtained by the ME-Solver with all of the hydraulic constraints was confirmed using the EPANET hydraulic simulator. An initial feasible solution was improved by a local search algorithm derived from the L1 algorithm originally proposed by Tolson et al. 2009. The performance of the new hybrid algorithm was compared with the ME-Solver which is based only on linear programming optimization. The results show that, the newly developed hybrid algorithm can reduce the pumping cost more than using just the ME-Solver. The newly proposed hybrid algorithm, which combines linear programming with another local search method, can be applied to optimize pump schedules in real-time.

Water Distribution Systems, Real-time pump scheduling optimization, Microsoft Excel Solver, Hybrid algorithm