Robust aerodynamic optimisation
The aim of Robust Design is to try and reduce performance variation through consideration of sources of error, uncertainty and variation. We are interested in the difference between the performance predicted by an entirely simulation-based design process and that which is achieved in practice; in essence, we are interested in design risk management. Work in this project is focusing particularly on two areas: error quantification arising from the use of variable fidelity Computational Fluid Dynamics simulations in optimisation; optimisation for aerodynamic components operating under unknown conditions.
In the first case, robust design techniques are being used to try and quantify the error that arises when variable fidelity simulations are used in optimisation procedures (necessary due to the computational cost of high fidelity CFD simulations). Thus we seek put bounds on the risk that a particular design does not meet its stated performance targets. A related area of research is the effect that different parameterisation schemes have on the design space for a given problem - a scheme that produces a particularly multi-modal or constrained space may be seen as increasing the risk that an optimisation algorithm fails to find good solutions.
In the second case, we consider aerodynamic optimisation problems where the operating conditions are unknown - this occurs frequently, for example at the combustor exit and at the fan inlet in a gas-turbine. We then attempt to optimise the design both for maximum performance and minimum performance variation in face of this uncertainty, requiring the use of multi-objective optimisation algorithms.
This research is very much driven by industrial need where commercial risk must be balanced with the technical risks associated with complex design generation using variable fidelity and computationally expensive simulations.
