Benchmarking study of water distribution system solution methods

Abstract

In recent years, a number of new water distribution system (WDS) solution methods have been developed. These methods have been aimed at improving the speed and reliability of WDS simulations. However, to date, these methods have not been benchmarked against each other in a reliable way. This research addresses this problem by using a newly developed software platform as a fair basis for a detailed comparison of the performance of these methods under different settings. In this work, efficient implementations of four solution methods, namely the global gradient algorithm (GGA), the GGA with the forest-core partitioning algorithm (FCPA), the reformulated co-tree flows method (RCTM), and the RCTM with the FCPA, are compared using eight case-study benchmark networks containing between 934 and 19,647 pipes and between 848 and 17,971 nodes. These simulations were carried out under both a once-off simulation setting and a multiple simulation setting (such as occurs in a genetic algorithm). Timings for these benchmark runs are decomposed into stages so that the performance of these methods can be easily estimated for different settings. The results of this study will help inform the choice of solution methods for given combinations of network features and given design settings.