Article ID: | iaor20161397 |
Volume: | 67 |
Issue: | 3 |
Start Page Number: | 187 |
End Page Number: | 198 |
Publication Date: | May 2016 |
Journal: | Networks |
Authors: | Srensen Kenneth, De Corte Annelies |
Keywords: | distribution, water, programming: nonlinear, combinatorial optimization, heuristics, heuristics: local search, design |
The optimal design of water distribution networks is a complex non‐linear combinatorial optimization problem. It consists in finding the least‐cost pipe configuration that satisfies hydraulic laws and customer requirements, using a limited set of available pipe types. In a previous paper (De Corte and Sörensen, Eur J Oper Res 228 (2013), 1–10), we have argued that state‐of‐the‐art optimization algorithms proposed in this domain are unduly complicated and poorly tested. The main contribution of this article is a straightforward, fast, transparent, and effective iterated local search (ILS) algorithm that has at least equivalent performance when compared to the best approaches in the literature, but has a much simpler algorithmic structure. A full‐factorial experiment is conducted to obtain the heuristic's best parameter settings. Contrary to existing algorithms, the ILS algorithm is additionally shown to perform well on a broad set of much more challenging HydroGen (De Corte and Sörensen, Water Resour Manage 28 (2014), 333–350) test instances.