Multi-objective optimization response modeling to contaminated water distribution networks: Pressure driven versus demand driven analysis

Implementation of management strategies following contamination detection in water distribution networks may extensively change operational mode of nominated valves and hydrants. The commonly used demand driven network solvers may fail to realistically represent system’s performances of new topology due to possible pressure-deficient condition. Realizing their drawbacks, this paper integrates a Pressure Driven Network Solver (PDNS) with multi-objective Non-dominated Sorting Genetic Algorithm-II (NSGA-II) in a simulation-optimization scheme. It is illustrated that the two commonly used objective functions, namely minimization of consumed contamination mass and number of polluted nodes, may be in conflict when an operational strategy is implemented. A trade-off is developed to help decision-maker compromise between restraining spatial spread of contaminant and its risk to public health. Decision variables in this optimization model are valve closure and hydrant opening. Each trial solution developed by the NSGA-II addresses a new system topology by changing operational modes of the nominated valves and hydrants. The PDNS determines the nodal pressures and refines the nodal withdraw for trial solution. To illustrate the performance of the proposed methodology, Net3 from EPANET 2 is employed. The results show that the pressure-driven analysis is more realistic and appropriate in comparison with demand-driven analysis in operational conditions.