Dr Edo Abraham

Edo Abraham

Dr Edo Abraham (M.Eng in Electrical and Electronic Engineering, Imperial College London, 2008; Ph.D. in Control Systems Engineering, Imperial College London, 2013) is Associate Professor of  Water & Control Systems at TU Delft, Faculty of Civil Engineering and Geosciences. His research is personally driven by global societal challenges that relate to water within socioeconomic development and climate change adaptation. Most of his research deals with techniques of integrated water systems. Edo’s research in infrastructure planning for water, energy and agricultural systems develops and applies novel modelling, control, machine learning and optimization approaches for making integrated water-energy-agricultural systems of the future more efficient and resilient. Some research projects include water allocation and optimising irrigation scheduling using agro-hydrological models and remote sensing, integrated planning and operational optimisation of multipurpose cascade reservoirs, national planning of WEF systems, the control of flood defence infrastructures for smart electricity grid regulation, and optimising nascent water and energy infrastructure using water-food-energy-land nexus models. Edo has co-authored over 40 peer reviewed journal publications across these subjects that broadly lie within the subject of Hydroinformatics. Edo is also an Associate Editor for Energy Systems (Springer) and Journal of Hydroinformatics (IWA).


Title: “Model Predictive Control of Water Systems: modelling to control under uncertainty”

Model Predictive Control (MPC) is increasingly used in the management of water resources and infrastructure systems. The approach is attractive due to its ability to explicitly represent physical, safety and performance constraints, as well as its flexibility to incorporate forecast uncertainty into the real-time optimal control problem, and its inherent feedback through a receding horizon implementation. More recently, new approaches are increasingly needed to deal with hydro-climatic, water demand and market uncertainties in water-energy system management. Another important requirement is explicitly dealing with the multi-objective nature of optimal control aims in interconnected water systems. We will present practical examples encompassing open canal pumping control, hydropower reservoir management, irrigation scheduling and wastewater treatment to highlight these key needs. Using the case of large-scale pumping into the North Sea, we will demonstrate how water infrastructure can be active in Demand Response services for the electricity market, where a stochastic MPC framework is developed for risk-aware decision-making. In another application, a parameterised Dynamic MPC strategy for multi-objective reservoir control is used to incorporate changing and pragmatic operator preferences in real-time reservoir flood control. We will highlight ongoing work on the use of piece-wise affine models and learning-based control that integrate machine learning approaches and MPC.