Pipedream: An interactive digital twin model for natural and urban drainage systems
Matthew Bartos, Branko Kerkez
Abstract
Faced with persistent flooding and water quality challenges, water managers are now seeking to build digital twins of surface water systems that combine sensor data with online models to better understand and control system dynamics. Towards this goal, this study presents pipedream—an end-to-end simulation engine for real-time modeling and state estimation in natural/urban drainage networks. The engine combines (i) a new hydraulic solver based on the one-dimensional Saint-Venant equations and (ii) a Kalman filtering scheme that efficiently updates hydraulic states based on observed data. Using sensor data from a real-world watershed, we find that the simulation engine is effective at both interpolating hydraulic states and forecasting future states based on current measurements. By providing a complete, real-time view of system hydraulics, our software will enable rapid detection of flooding, improved characterization of maintenance and remediation needs, and robust real-time control for both small-scale (stormwater) and large-scale (river/reservoir) networks.