Author(s): Alejandro Gonzalez Barbera; Delia Trifi Rufino; Jaume Luis Gomez; Oscar Prades Mateu; Guillem Monros Andreu; Paloma Barreda Juan; Javier Climent Agustina; Rosario Arnau Notari; Raul Martinez Cuenca; Sergio Chiva Vicent

Linked Author(s):

Keywords: Digital twin; ML; CFD; WWTP

Abstract: This research presents a hybrid digital twin (DT) framework for Wastewater Treatment Plants (WWTPs), integrating Computational Fluid Dynamics (CFD), Machine Learning (ML), and Industry 4.0 principles to enable future decision making tools. The architecture employs a Docker-based co-simulation environment composed of three interconnected containers. OpenFOAM is utilized to simulate multiphase flow and biochemical transport phenomena, while U-Net surrogate models are trained on CFD outputs to accelerate inferences of hydrodynamics. Leveraging spatial features such as 3D coordinates and Euclidean distances to diffusors, and employing a 2D slicing subsampling strategy, the system achieves accurate emulation with only 30 CFD cases and 32-hour training runtimes. This DT demonstrates the potential of ML-enhanced hydroinformatics to bridge physics-based modeling and real-time operational intelligence, advancing the digital transformation of urban water systems toward resilience and sustainability.

DOI:

Year: 2026