Gradient-based Optimization of Parameterized CAD Geometries

Timothée Leblond (1, 2) , Pierre Froment (1) , Paul de Nazelle (1) , Reda Sellakh (2) , Philippe Serré (2) , Gaël Chevallier (3)

1 IRT SystemX
2 LISMMA - Laboratoire d'Ingénierie des Systèmes Mécaniques et des MAtériaux
3 FEMTO-ST - Franche-Comté Électronique Mécanique, Thermique et Optique - Sciences et Technologies (UMR 6174)

This paper presents a novative method for gradient-based optimization with regard to CAD parameters. This method allows to respect manufacturing, design and surface quality rules, particularly required in computational fluid dynamics (CFD). It can be used with any objective function available in adjoint solvers (both in structural and CFD). To prove the efficiency of the method, the workflow schedule was apply on an air-conditionning duct so as to maximize the eigen frequency and then to minimize the pressure drop.

Mots clés: CAD parameters; Sensitivity; Optimization; adjoint solvers; CFD - Computanional Fluid Dynamics

Domaines: Ingénierie assistée par ordinateur; Mécanique des fluides [physics.class-ph]

Timothée Leblond, Pierre Froment, Paul de Nazelle, Reda Sellakh, Philippe Serré, et al.. Gradient-based Optimization of Parameterized CAD Geometries. 11th World Congress on Structural and Multidisciplinary Optimization, ISSMO, Jun 2015, Sydney, Australia. ⟨hal-01429321⟩