Control of water distribution networks using convex approximations

Abstract

The status quo of control in water distribution networks (WDN) relies on simple, rule-based control strategies to operate pumps and valves. For example, pumps are switched on/off if a tank's water level---a local measurement---is below/above a certain threshold. As an alternative to localized, rule-based control, network-driven control that utilize WDN topology and measurements can be beneficial. Unfortunately, network-driven control of WDNs is very difficult as hydraulic models are nonconvex, valve and pump models form non-trivial, combinatorial logic, and water demand patterns are uncertain. Prior research on control of WDNs addressed major research challenges, yet mostly adopted simplified hydraulic models, WDN topologies, and rudimentary valve/pump modeling. This Work-in-Progress showcases the potential of network-driven control of WDNs. The proposed approach amounts to solving a series of convex optimization problems that graciously scale to large networks. Simple case studies are included showcasing early results, and applications of the proposed optimization methods to other WDN applications are discussed.