General model for automatic design optimisation of aerodynamic components. Wind tunnel case study

Abstract

Trial and error is still the current approach for the design of 
many complex aerodynamic components, although extensive 
research is being carried out in automated optimisation 
methods. Thereafter these designs are computer simulated 
using Computational Fluid Dynamics (CFD). Some optimisation 
tools are used, at most, for the final fine-tuning. This paper 
proposes the use of a robust, efficient and automated 
optimisation methodology throughout the whole design 
process. The use of these methodologies can yield improved 
yet not conventional designs, while reducing the design cycle 
time. This paper presents the development of an original, 
general methodology based on a multi-attribute, structured 
optimisation, following a so-called Hybrid Direct Search 
(HDS), which combines genetic, gradient and swarm search 
intelligence. An example case study of a wind tunnel shape 
optimisation is presented. The main contributions of this paper
are the exploitation of the concepts of variable hierarchy 
and variable value change (i.e. optimisation phases) and the 
HDS optimisation method, which allows for an intelligent 
and efficient direct search in complex aerodynamic problems 
in which the use of surrogate based optimisation is not 
accurate enough. This optimisation methodology is applicable 
to advanced aerodynamic design in cars, aircraft, high-speed 
trains, etc.
• Keywords: aerodynamic optimisation, wind tunnel design, 
optimisation methodology, multi-attribute, structured 
optimisation, hybrid direct search, swarm search, variable 
hierarchy, geometry parameterisation, computational fluid 
dynamics