- Academic lead He Wang (Computing)
- Industrial lead Bernardo Vazquez (BuroHappold)
- Co-supervisor(s) Amir Khan (Civil Engineering)
- Project themes Energy and Transport, Environmental Flows, Underpinning Methods for Fluid Dynamics
Have you ever felt uncomfortable or suffocated in crowds in a shopping mall? Do you know how fast an airborne contagious disease can spread in the atrium of a hospital? The indoor air quality is crucial and sometimes life-threateningly important. Being able to understand and predict airflows in enclosed environments not only lowers physical dangers in extreme situations, but also has wider and long-term impact on safety, thermal comfort and energy efficiency.
Computational Fluid Dynamics (CFD) is the major tool for airflow simulations. It has been used for optimising the design of new buildings or improving existing ones. However, few people model airflows with crowds together. CFD is already computationally expensive and slow, and crowd simulation/prediction itself is difficulty too. Coupling two complex systems with different dynamics would impose a great challenge.
This is the challenge the project attempts to address. The potential solution lies within data-driven methods. Recently, machine/deep learning has had massive successes in different domains, on a variety of problems, previously modelled based on deterministic approaches (such as fluids and crowds). This project will look into solutions bypassing traditional methods in CFD and crowds by leveraging cutting-edge deep learning approaches.