Jonathan Bull
Division of Scientific Computing
Department of Information Technology
Uppsala University
Uppsala
It is now well-established that computational fluid dynamics (CFD) is an invaluable tool for analysing turbulent fluid flows in environmental and engineered systems. For example, 'quick and dirty' CFD simulations can be used to reduce the parameter space of a concept aircraft design, after which a few options can be tested as scale models in the wind tunnel. However, when a high level of quantitative accuracy is required, CFD is often inadequate. Methods that work beautifully in carefully-selected academic test cases do not look so good in real-world problems featuring geometric complexity and high Reynolds numbers. Moreover, turbulence models tend to require tuning to suit the particular flow conditions and the required mesh resolution may not be known a priori.
The author's research is directed at tackling the inaccuracy and unreliability of CFD for complex turbulent flows on several fronts:
Robust general-purpose subgrid and near-wall models,
High-order accurate numerical schemes,
Unstructured mesh adaptivity.
This talk summarises the author's efforts on these topics as a PhD student in the Applied Modelling and Computation Group (AMCG) at Imperial College London and as a postdoctoral scholar in the Aerospace Computing Lab (ACL) at Stanford University.