Water distribution system design to minimize costs and maximize topological and hydraulic reliability

Many surrogate measures for water distribution system (WDS) reliability have been introduced in the last three decades. This study investigated the differences between designs based on topological and hydraulic reliabilities. The former considers network structural redundancy and connectivity, whereas the latter considers system performance under uncertain future conditions. Two topological reliabilities are proposed based on the network theory: the average node degree ratio (AND_r) and meshedness coefficient ratio (MC_r). The capacity reliability and robustness are classified as hydraulic reliability. The Pareto optimal pipe sizes and configuration were found for a real medium-size grid-type network to minimize the total cost and maximize AND_r, MC_r, capacity reliability, and robustness individually. The nondominated sorting genetic algorithm II was used for the optimization, and the uncertainty of the nodal pressures was quantified using the first-order second-moment approximation method. The results were compared in terms of the configuration, pipe sizes, and four reliability values to develop guidelines on selecting a reliability metric for WDS design.