Optimal Design of Water Distribution Network under Hydraulic Uncertainties

AbstractThe uncertainties in projected water demands and changes in pipe roughness values over time are common causes of hydraulic failure of water distribution networks. Such hydraulic uncertainties are usually handled through probabilistic approaches. However, these uncertainties are nonprobabilistic in nature and are mainly characterized as such due to imprecise knowledge; therefore, they can be appropriately modeled through a fuzzy approach. A fuzzy optimization model is formulated for the least cost design of a water distribution network considering uncertainties in water demands and pipe roughness values. The membership function for the nodal pressure head requirement is modified to account for various levels of reduced acceptability of pressure. The constrained fuzzy optimization model is converted to a deterministic model by considering the relationship between demands and pressure head requirements at nodes and solved using a genetic algorithm (GA). The methodology is illustrated through two example networks from the literature. The cost of the network decreased with an increase in the tolerance level in the pressure head requirement. The proposed GA-based methodology is quite simple and provides a solution that meets the pressure requirements in a worst case situation.