Abstract This paper describes an extension to the Optimal Power Use Surface (OPUS) methodology, which consists in applying a metaheuristic post-optimization process after a network has been designed through the hydraulically-based OPUS algorithm. Once a set of diameter sizes is obtained by means of this hydraulic-based design methodology, it is used as the system's initial configuration for it to be post-optimized using different nondeterministic techniques. The metaheuristics tested in this study include 3 different types of Genetic Algorithms (GAs), Harmony Search (HS) and Simulated Annealing (SA). The proposed methodology is tested on three benchmark problems (Hanoi, Balerma, Taichung) and a fourth network called R28, which is introduced herein. When compared to the results obtained through other methodologies, this algorithm stands out for allowing designs with constructive costs very close to the lowest found in other investigations. However, the improvements with respect to the OPUS algorithm are very small, while the number of iterations required increases by more than 2 orders of magnitude. Moreover the Resilience Index (RI) increases in most cases as well. This extension to the OPUS methodology proves that following hydraulic principles allows obtaining near-optimal results, whose improvement demands a considerable number of iterations, providing minimum benefit as the reduction in cost is only of 1% at the most.