Model predictive control in urban systems

Jan Lorenz Svensen: The drainage systems of the urban environments, has in the recent decades, been challenged by an increase in the number of larger rain events.

This increased frequency has caused the amount of rain water to exceed the capacities of the drainage systems on a more regularly basis, than the systems was defined and legislated for. Exceeding the system capacities results in either flooding of the streets, or overflow of polluted wastewater at the treatment plant into the neigh by environment, nature or recreational; an expense which in 2011 costed insurance companies more than 4 billion DKK, from one rain event. Updating the drainage systems to the new weather conditions would cost Trillions of DKK and the construction work would take decades at best. Better utilization of the current drainage system is therefore of interest to society.

The utilization of the drainage system is generally controlled by conventional strategies. Conventional strategies, such as PID, are either given operating set points manually by the operators or based on constructed set of rules (Rule-Based Control) for the given conditions of weather, cost and legislation. In those cases expert knowledge, from for example trial and error, are needed ether to operate or to define rules for the system. This makes the conventional control strategy more complex and time-consuming, when the weather pattern, legislation and system costs are changing. For these reasons this project focuses on replacing the conventional control strategy with an optimization based control strategy, in particular model predictive control (MPC).

The benefits of MPC is the ability to include system constraints in the solution; the solution is then the optimal solution obtainable for the given sets of constraints. The MPC strategy is a Real-Time optimization and control method; finding the optimal solution for time k, while the system is running and therefore taking into account the changes to the uncontrolled inputs occurring during the running.

The aim of this project “Model Predictive Control in Urban Systems” is the research and development of MPC methods, concerning the prevention of floods, bypasses and overflows in Urban City water system during large weather events, for example: cloudburst.

PhD project title: Model Predictive Control in Urban Systems

Effective start/end date 15/09/2017 → 14/09/2020

DTU supervisor is Niels Kjølstad Poulsen from the section for Dynamical Systems at DTU Compute.

PhD project by Jan Lorenz Svensen

Research section: Dynamical Systems

Principal supervisor: Niels Kjølstad Poulsen

Co-supervisor: Anne Katrine Vinther, Henrik Madsen & Hans Henrik Niemann

Title of project: Model Predictive Control in Urban Systems

Project start: 15/09/2017 → 14/09/2020


Jan Lorenz Svensen
DTU Compute
+45 45 25 39 10


Niels Kjølstad Poulsen
DTU Compute
+45 45 25 33 56