Major-Minor Mean Field Control for Adaptive Control of Urban Water Distribution Systems

Water distribution systems (WDSs) have so far been controlled using either optimized pump scheduling or real-time control of pumps and valves. Both approaches focus on WDSs as technical systems, yet energy saving requirements and increasing risk of disruptions call for more comprehensive methods. This paper presents a method for integrating consumers into the control strategy, conceiving WDSs as sociotechnical systems. To assess this approach, two scenarios are compared: a base scenario, where the system operator controls the system with given time-dependent consumer demand, and a cooperative scenario, where both operator and consumers jointly pursue the control objective: energy efficient demand fulfillment. To ensure scalability while modeling consumers on a household scale, Major-Minor Mean Field Control (M3FC) is employed. Both scenarios achieve 10% energy savings compared to conventional two-point pump control. In the event of a partial pump outage, consumers in the cooperative scenario collectively reduce their demand, achieving fairer water distribution compared to the base scenario.