Comparison of the efficiency of seven meta-heuristic algorithms in leak detection on water distribution networks

Abstract Leakage in water distribution networks causes water loss, pressure drop, increase in fixing costs, and the possible entrance of pollutants into the networks. Therefore, detecting the un-reported leaks and fixing them is one of the main challenges of water utilities. Since the common leak detection methods are generally costly, time-consuming, and require expert personnel, the use of hydraulic simulation methods to localize leaks has been paid considerable attention in the in recent years. In this study, leak detection is carried out by using the hydraulic model calibration method, defining the objective function based on minimizing the difference between field data and simulated data, and finally optimizing the defined objective function. In the present approach, optimization is performed using meta-heuristic algorithms. Since the more accurate the used algorithm is, the fewer error on the leak detection results will be, it is necessary to compare the algorithms to introduce the best algorithm in hydraulic and leak detection problems. Seven powerful and well-known algorithms in three water distribution networks (WDNs) were analyzed by defining 45 different scenarios of random leakage. Due to the multiplicity of output data, the Chess Ranking System for Evolutionary Algorithms (CRS4EAs) was used for a comprehensive comparison of the studied algorithms. The obtained results showed that the Equilibrium Optimizer is superior to its competitors in terms of the optimal solutions, the reliability of single runs, and the run time. Therefore, EO is suggested to be used for leak detection in WDNs with maximum reliability and minimum calculation time.