Sewer orientated framework for ensemble‐based chance‐constrained model predictive control

In this work, we present a framework for ensemble‐based (E) chance‐constrained (CC) model predictive control (MPC) in sewer systems. The framework considers the availability of ensemble forecasts and the difficulties with propagation of distributions; through distribution estimation. Utilizing a case study of the sewer network of the city of Aarhus in Denmark, the performance of the ECC‐MPC framework is evaluated through simulations. The evaluations were based on linear models of the case study and compare the ECC‐MPC performance with the performance of CC‐MPC. Based on the simulations, it was found that the ECC‐MPC performed comparable to the performance of the CC‐MPC, not only in the context of overflow and outflow but also with respect to behavior in response to changes in different aspects of forecast uncertainties. Regarding the aspects, it was found that expectation offset biases in the forecast were affecting the performance of the CC‐ and ECC‐MPC the most. While other aspects only had a reduced effect on the performances, within the ranges tested. With the comparable performances, it was found that ECC‐MPC would work as an alternative approach to CC‐MPC. Ensemble‐based Chance‐Constrained (CC) Model Predictive Control (MPC): An alternative to CC‐MPC given ensemble forecasts and difficulties with propagation of stochastic distributions.