Safe and Robust Cloud-based Predictive Control for Water Distribution Networks

Water distribution networks with storage equalize the operating pressure in the system. Apart from providing pressure, elevated reservoirs also allow for exploiting fluctuation in electricity prices. In this work, we use an Economic Model Predictive Control scheme for planning the filling times of storage in water distribution networks. This controller needs access to electricity prices via an Information Technology (IT) connection to a power-trading system. Therefore, we propose a control architecture, where a Global Controller is deployed on a cloud server, whereas a locally placed data-acquisition and supervisory controller provide the safety guarantees in terms of an Operational Technology (OT) layer. The approach is demonstrated numerically on a real-world network architecture where real data is used to verify the robustness of the method.