Generative AI for Physics-Aware Barrier Optimization for Air Pollution and Odour Mitigation

Short description of the innovation study

SHIELD-GEN develops a generative AI framework for physics-aware design of mitigation barriers that reduce air pollution and odour in urban and industrial environments. The project combines high-fidelity CFD with conditional generative models to learn joint representations of geometry, operating conditions, and mitigation performance, enabling non-intuitive novel barrier concepts. Trained on large synthetic CFD datasets and calibrated with selected real-world measurements, the model synthesizes and ranks buildable barrier configurations conditioned on site-specific scenarios and mitigation objectives. The approach is extensible to related barrier problems such as noise mitigation, wind comfort, and flood protection, establishing a physics-aware foundation model for environmental mitigation.

Organisations involved: 

Main SME: Simulation Zero, S. L.

Technology expert: Sociedad de fomento agricola castellonense S.A.U. (FACSA)