A simulation-optimization model for proper safety management in urban water distribution systems

Water security is a vital issue that may threat by contaminating drinking water of a city, for example through a terrorist attack. In the absence of a proper managing safety in water distribution system, the number of victims of such terrorist attack can reach to whole of city population. In a passive defense approach, the first step for any integrated public alert system and contaminant controlling is rapid detection of contaminant source. Observations from a set of monitoring points are used to detect the contaminant source. The optimal number of required monitoring points and their pattern is a challenging task that closely depends on the potential injection points, the amount and characteristics of injected pollution. The best monitoring system corresponds to the pattern that minimizes (a) the required number of monitoring points, (b) the affected population at detection instant, and (c) the undetected cases. In this study, the fast multi-objective evolutionary algorithm, NSGA-II, is coupled with the simulation model, EPANET, to find a trade-off curve among the conflicting objectives. Monte Carlo simulation is used to incorporate uncertainties in injected pollution. The Nash bargaining solution is utilized to find the equilibrium point of the obtained trade-off curve. In order to demonstrate the efficiency and capabilities of the proposed methodology, the water distribution network of Lamerd City is modeled and the best monitoring points are determined.