No abstract

No abstract

We develop and discuss an operational planning model aiming at minimizing production and distribution costs in large drinking water networks containing buffers with free inflow. Modeling drinking water networks is very challenging due of the presence of complex hydraulic constraints, such as friction losses and pump curves. Non-linear, non-convex constraints result from the relationships between pressure and flow in power terms. Also, binary variables are needed to model the possibility of free inflow or re-injection of water at reservoirs. The resulting model is thus a non-convex Mixed-Integer Non-Linear Program (MINLP). A discrete-time setting is proposed to solve the problem over a finite horizon made of several intervals. A commercial solver, BONMIN, suited for convex MINLP models is used to heuristically solve the problem. We are able to find a good solution for a small part of an existing network operated by the Vlaamse Maatschappij voor Watervoorziening (VMW), a major drinking water company in Flanders.

Abstract:In this paper we propose a mathematical programming model for a large drinking water supply network and discuss some possible extensions. The proposed optimization model is of a real water distribution network, the largest water supply network in Flanders. The problem is nonlinear, nonconvex and involves some binary variables, making it belong to the class of NP-hard problems. We discuss a way to convexify the nonconvex term and show some results on two case instances of the actual network.

No abstract