CONTAMINANT DISPERSION SIMULATION IN A DIGITAL TWIN FRAMEWORK

Abstract

Within a digital twinning process, we aim to establish a highly automated workflow to simulate air contaminant dispersion based on up-to-date urban geometries, seamlessly retrieved from open-source geo-referenced databases. The final declared scope is providing assistance and informed real time decision making processes in the context of evacuation scenarios. This entails generating an initial simulation space and refining its characteristics as needed to gain an adequate level of accuracy. In the preliminary phase of the project, high-fidelity wind flow simulations are coupled with contaminant concentration assessments based on advection-diffusion processes. A second more advanced phase aims at leveraging the workflow efficiency by introducing a Reduced Order Model (ROM) for the wind field evaluation, together with the investigation of scientific machine learning methods to merge physics with data, with the final objective of defining a reliable predictive model for a hybrid digital twin.