Optimizing pumping system for sustainable water distribution network by using genetic algorithm

Considerable portion of water distribution cost is related to energy usage. Pumps are often the greatest energy consumer in water distribution systems. Optimizing the working schedule of these pumps has the potential to significantly decrease operation costs of these systems. It is also possible to optimize for environmental concerns and arrive at the least polluting solution of pump operation that meets system-wide hydraulic demands. As with water usage, electrical usage varies over the course of the day. Electrical utilities must satisfy the daily power demand which, in turn, may result in using different types of electrical generators - some of which vary in their environmental impact. By scheduling energy consumption at times that correspond to the least polluting generators on-line, water distribution systems can become more environmentally sustainable. Optimizing energy consumption and working schedule of pumps is a multi-objective nonlinear problem that contains considerable amount of constraints. The solution space of this type of problem even for a small water network can be very large and finding the boundaries associated with the solution space is quite difficult. Evolutionary optimization methods, such as genetic algorithm (GA), are well suited for solving this kind of problem. In this paper, two methods for describing the pump optimization problem within a GA solution framework are considered. Each leads to different methods for conducting crossover and mutation steps of the GA. Advantages and disadvantages of each method are discussed. In terms of sustainable computing, the effect of each method on computer storage and computing requirement is investigated. Novel GA coding methods which circumvent infeasible solutions that may arise with this specific type of problem are discussed as well as other associated issues.