On solving water distribution network design problems with stochastic search optimization techniques

摘要

In spite of twenty years of research efforts, optimal design of looped water distribution systems still remains a challenge. The physical behaviour of a looped network is described by a set of equations, some of them non-linear. Although numerical methods can simulate the hydraulic behaviour of large size looped networks, finding the optimal network design is, even for simple networks, a complex task. If we are to represent reality adequately, this problem must be formulated in combinatorial form defining extremely difficult models, whose resolution in real-world conditions, via classical optimization techniques, is often impossible. However new developments in the field of stochastic optimization techniques allow their solution to be faced with some hope of success.rnIn this paper a simulated annealing algorithm is proposed to solve this kind of problem. Simulated annealing is a stochastic search optimization method that can work well for large-scale optimization problems that are cast in discrete or combinatorial form. The authors present some work on the application of this heuristic method to three well-known case studies (Alperovits and Shamir, Hanoi, and New York networks). The results obtained show an improvement both on the optimal solution and on computer running time when compared with those from the literature. This fact was encouraging, leading us to deal with more complex systems in this paper. The results found through this study can be considered very promising.