Multiobjective optimization of the designs of water distribution networks using Evolutionary Algorithm

MAX DUARTE; DARIO ALVISO; PEDRO GARDEL; BENJAMIN BARAN

SAN JOSE

Conferencia; CLEI XXXIII Latin America Conference on Informatics; 2007

Universidad de Costa Rica

This work implements an innovative evolutionary algorithm approach to optimize the multiobjective water distribution system design. The main idea is to design a water distribution system considering not only the demand for a given point in time, but also projected water demands for different times in the future. This way, instead of the traditional methods that predefine the utility time desired for the network, a whole compromise set of solutions considering different utility periods may be calculated before decision makers decide which alternative best fit their interest. This way, the problem is reformulated as a multiobjective optimization that simultaneously tries to minimize network cost and the pressure height deviation at the nodes (considering the desired pressure height at each node), while maximizing the utility servicetime. To solve this new formulation of the problem, the Strength Pareto Evolutionary Algorithm (SPEA), a well known second generation evolutionary algorithm, was implemented. In order to validate the proposed method, the Hanoi water distribution network, a very well known instance, was used. Experimental results confirm the efficiency of the proposed method.