Much of my research is about understanding and modelling turbulence so that we can accurately predict how a car, bike or plane behaves as it moves through the air and importantly how to improve its design so that it uses less energy doing so. This desire to make greener, more efficient vehicles is crucial for the UK and global economies as the automotive and aerospace industries account for a major part of their GDP.
However, both sectors face new challenges to meet stricter emissions targets over the coming decade. For major manufacturers this can mean a requirement to reduce CO2 emission by as much as 45%. The aerospace industry also faces challenges to meet its own ambitious targets of reducing CO2 emissions by 75% and noise emissions by 65% by 2050.
One of the key factors to reduce CO2 and noise emissions is the aerodynamics of the car or plane. A more aerodynamically efficient vehicle has lower drag and thus uses both less fuel and requires a smaller engine to achieve similar performance. A key part of the aerodynamic design process is Computational Fluid Dynamics (CFD). CFD can be thought of as a virtual wind-tunnel, where computational methods based upon the solution of the Navier-Stokes equations simulate the flow of a fluid passing over an object such as a wing, engine or car. These simulations can then provide information on the flow, such as the velocity and pressure. As these are entirely computational models, no manufacturing is required and thus many variants can be tested at a much lower cost compared to a traditional wind tunnel.
The current limitation of CFD is however a lack of consistently high accuracy and a lack of software that can efficiently make use of supercomputing resources. This means that CFD is often not trusted enough to make major design decisions.
My research using largely open-source software, such as the CFD package (OpenFOAM) together with the latest generation of supercomputing hardware aims to deliever more accurate, faster solutions by improved CFD methods which automatically adapt to the solution in real-time.