Optimization Model for Water Distribution System Design

Abstract

A new methodology is developed for determining the optimal (minium‐cost) design of water distribution systems. The components that can be sized are the pipe network, pumps or pump station, and tanks. In addition, the optimal settings for control and pressure‐reducing valves can be determined. This methodology couples nonlinear programming techniques with existing water distribution simulation models. Previous methodologies have typically simplified the system hydraulics to be able to solve the optimization problem. This new methodology retains the generality of the hydraulic simulation model so that the problem is only limited by the ability of the simulation model rather than the optimization model. The methodology uses a generalized reduced gradient model to solve a problem that is reduced in size and complexity by implicitly solving the conservation of mass and energy equations using the hydraulic simulator and an augmented Lagrangian approach to incorporate pressure head bounds in the objective function. Since the network equations are solved implicitly any number of demand patterns can be considered, including steady state loads, extended period simulations, or both.