Aggregation-based model predictive control of urban combined sewer networks

The flow control problem of urban sewer networks to minimize overflows is cast in the framework of model predictive control. The predictive control algorithm is developed based on the discrete maximum principle, employing nonlinear constrained optimal control concepts. By introducing element models, the mathematical model of a sewer network can be formulated. Since the function of the reservoir overflow is not differentiable, a new type of smooth function is proposed instead so that the gradient-based method can be developed. In the model predictive control algorithm, an aggregation scheme is proposed to improve the computational efficiency. A detailed study for a particular large-scale multireservoir sewer network and a comparison of the computation time are illustrated to demonstrate its efficiency.