Real-Time Data Integration and Model Performance Assessment for a Digital Twin of a Permeable Pavement Stormwater Control Measurement

Abstract Submission: Integrating Low Impact Development (LID) with real-time monitoring, modeling, and control offers significant potential to improve several water quality and quantity goals of urban water management. This integration can be achieved via the Digital Twins (DT) concept, which includes near real-time modeling, data transmission, storage, assimilation, visualization, and predictive analysis. Despite advancements in permeable pavement models, current approaches often lack the ability to assimilate real-time data. This limitation creates a gap in using adaptive, real-time models to forecast future system performance accurately. This study addresses this gap by adapting the MicroPollutants in RainGarden (MPiRe) model—a "three-bucket" approach traditionally used for bioretention systems—to model permeable pavements and incorporating an Ensemble Kalman Filter (EnKF) for real-time data assimilation. By assimilating collected outflow data into the model using the EnKF, the model's performance and forecasting capabilities are improved. The study compares the performance of the MPiRe model with the permeable pavement module in the SWMM while incorporating real-time data assimilation to enhance model accuracy. Four types of permeable pavements (permeable concrete, permeable asphalt, plastic grid, and permeable interlocking concrete pavers) are evaluated, each covering an area of 190 m² and installed at the same site in San Antonio, Texas. The MPiRe model uses a one-dimensional approach based on Darcy’s Law and a conservative mass balance to simulate water movement across the ponding, unsaturated, and saturated zones. The Nash-Sutcliffe Efficiency (NSE) and Root Mean Square Error (RMSE) metrics were used to compare the MPiRe and SWMM output results for 12 observed events.

Learning Objectives/Expected Outcome (Optional) :