A new algorithm for water distribution system optimization: Discrete dynamically dimensioned search

Abstract

The Dynamically Dimensioned Search (DDS) continuous global optimization algorithm by Tolson and Shoemaker (2007) is modified to solve discrete, single-objective, constrained Water Distribution System (WDS) design problems. The new algorithm is called Discrete Dynamically Dimensioned Search (DDDS). DDDS characteristics parallel those of DDS, namely that it is a simple, parsimonious and efficient global optimization algorithm. This paper evaluates DDDS in relation to Ant Colony Optimization (ACO) and Genetic Algorithms (GAs) for WDS optimization. The first implementation of DDDS, called DDDS-v1, was developed and then applied to the Hanoi (HP) and New York Tunnels (NYTP) benchmark WDS optimization problems without algorithm parameter-tuning and with a simple parameter-free penalty function approach. DDDS-v1 results are good for the NYTP in comparison with published ACO and GA results. DDDS-v1 identified the best known solution to the NYTP in 5/20 optimization trials. For HP, DDDS-v1 generated better average results than any ACO and GA results available from a previous study. Importantly, DDDS-v1 had no trouble finding the feasible region and returned final solutions from this region that were on average improved relative to other algorithms. Overall, findings suggest that DDDS shows good potential as a new tool for WDS optimization.