Network Analysis, Control Valve Placement and Optimal Control of Flow Velocity for Self-Cleaning Water Distribution Systems

In this paper, we consider the proactive control of flow velocities to maximise the self-cleaning capacity of the drinking water distribution systems under normal operations both through a change of the network topology and through an optimal control of pressure reducing valve (PRV) settings. Inspired by line outage flow distribution in electrical networks, we show how a fast network graph analysis of link closures can be used to estimate the potential changes in flow velocities, which are then used to determine the most favourable pipes for closure. Where closing of pipes cannot be used because of other conflicting objectives, we consider the optimal control of PRVs to maximise self-cleaning at peak demand periods. We formulate a novel optimisation problem to maximise the network operations for increased self-cleaning capacity, while satisfying hydraulic and regulatory pressure constraints at demand nodes. A new smooth objective function approximation for cleaning capacity of the network is proposed along with a scalable sequential convex programming method to solve the resulting valve optimization problems. We use a published benchmark network as a case study to show the efficacy of these new approaches.