Learning Safe and Optimal Control Strategies for Storm Water Detention Ponds

TL;DR

This paper presents a formal method-based approach to automatically synthesize safe and optimal control strategies for storm water detention ponds, improving safety and efficiency over traditional static controls.

Contribution

It introduces a hybrid Markov decision process model and uses Uppaal Stratego to derive control strategies that optimize pollution reduction while ensuring safety.

Findings

Synthesized strategies prevent emergency overflows effectively.

Control strategies improve sedimentation during low rainfall periods.

Approach outperforms static control methods in simulations.

Abstract

Storm water detention ponds are used to manage the discharge of rainfall runoff from urban areas to nearby streams. Their purpose is to reduce the hydraulic impact and sediment loads of the receiving waters. Detention ponds are currently designed based on static controls: the output flow of a pond is capped at a fixed value. This is not optimal with respect to the current infrastructure capacity and for some detention ponds it might even violate current regulations set by the European Water Framework Directive. We apply formal methods to synthesize (i.e., derive automatically) a safe and optimal active controller. We model the storm water detention pond, including the urban catchment area and the rain forecasts, as a hybrid Markov decision process. Subsequently, we use the tool Uppaal Stratego to synthesize a control strategy minimizing the cost related to pollution (optimality) while guaranteeing no emergency overflow of the detention pond (safety). Simulation results for an existing pond show that Uppaal Stratego can learn optimal strategies that prevent emergency overflows, where the current static control is not always able to prevent it. At the same time, our approach can improve sedimentation during low rain periods.