Optimization of pipe diameters in a water distribution network through simulated annealing algorithms

Abstract

To set pipe diameters in the least-cost design of a water distribution network is a strong non linear restricted problem with multiple local optima and its solutions space has many unfeasible regions. The heuristic algorithm of optimization, called Simulated Annealing (SA), is a global method that has been used to make stochastic searches in the problem's solutions space bettering the performance of other methods. This paper proposes a problem formulation with penalty functions, which lets SA algorithm, among another advantages, that the stochastic walk done by it could be less sinuous, crossing unfeasible regions. This approach improves the algorithm efficiency, for the same error level, with respect to a classical restricted formulation