RBF-based mesh morphing improvement using Schur complement applied to rib shape optimization - Archive ouverte HAL Access content directly
Journal Articles International Journal of Numerical Methods for Heat and Fluid Flow Year : 2019

RBF-based mesh morphing improvement using Schur complement applied to rib shape optimization

Abstract

Purpose This paper aims to improve the radial basis fuction mesh morphing method. During a shape optimization based on computational fluid dynamic (CFD) solvers, the mesh has to be changed. Two possible strategies are re-meshing or morphing. The morphing one is advantageous because it preserves the mesh connectivity, but it must be constrained. Design/methodology/approach RBF mesh deformation is one of the most robust and accurate morphing method. Using a greedy algorithm, the computational cost of the method is reduced. To evaluate the morphing performances, a rib shape optimization is performed using the NSGA-II algorithm coupled to kriging metamodels based on CFD. The morphing method is then compared to a re-meshing strategy. Findings The authors propose a method, based on Schur complement, to speed-up the greedy process. By using the information of the previous iteration, smaller linear systems are solved and time is saved. The optimization results highlight the interest of using a morphing-based metamodel regarding the resolution time and the accuracy of the interpolated solutions. Originality/value A new method based on Schur complement is addressed to speed-up the greedy algorithm and successfully applied to a shape optimization.
No file

Dates and versions

hal-03184692 , version 1 (29-03-2021)

Identifiers

Cite

Franck Mastrippolito, Stephane Aubert, Frédéric Ducros, Martin Buisson. RBF-based mesh morphing improvement using Schur complement applied to rib shape optimization. International Journal of Numerical Methods for Heat and Fluid Flow, 2019, 30 (9), pp.4241-4257. ⟨10.1108/HFF-06-2018-0309⟩. ⟨hal-03184692⟩
39 View
0 Download

Altmetric

Share

Gmail Facebook X LinkedIn More