Comparison of two approaches for the direct optimisation of a car engine intake port

Abstract

The shape optimisation involving complex flows, and based on direct CFD approaches, has now become very possible within industrial contexts, thanks to recent advances in optimal design technologies. However, one key remaining issue is related with the way the shape deformations are modelled. This paper presents the development of such an optimisation procedure, applied to a car engine air intake port, and compares two alternative methods for the shape deformation modelling. Global port variations are defined through a set of shape parameters. The first option consists in building a CAD model (Catia v5) to allow deformations, followed by a grid generation stage with tools enabling to generate and adapt the mesh to shape variations (Gridgen). The second option consists in starting from a grid done for a given shape, and defining parametric deformations on this grid, with a morphing tool (Sculptor). Then, after both modelling methods, the flow calculation is carried out with a CFD solver (Fluent). The resulting fields are processed by a flow post-processor (Fieldview) and by a Matlab procedure to extract the global criteria involved as optimisation objectives: the flow rate and the tumble. All this process is bundled within the modeFRONTIER multi-objective design environment. Similar optimisation approaches were led on both cases, involving mixed response surfaces – direct calculation evaluations of the designs, and a multi-objective genetic algorithm. The paper presents the main results, highlights the advantages and shortcomings of both modelling methods, and provides guidelines for their use and improvements, in view of practical industrial applications.