Skip to Main content Skip to Navigation

Simulation tridimensionnelle du remplissage de corps minces par injection

Abstract : This study provides new tools for injection moulding simulation of thin pieces. This work is based on an existing code dedicated to describe the filling of 3D parts. Finite element methods in an eulerian description of the flow are used. At first, a new anisotropic meshing method is proposed. Issue from locals optimisations methods, this new tool is based on metric concept. It enables to generate or to adapt unstructured anisotropic meshes in 2D or 3D (triangle or tetrahedral). In the second part, we focus on the error estimation of the flow calculus using anisotropic meshes. We propose a new a posteriori anisotropic error estimate for Stokes problem. It is shown that this estimate is able to detect the anisotropy of meshes but without being optimal. Finally, using anisotopic meshes could not be convenient to describe precisely the flow around free surfaces and flow fronts inducing too large diffusion. In order to prevent this problem, we developed a new mesh adaptation dedicated to front tracking. This method can be seen as a weighted barycentric algorithm and allows to dynamically adapt the mesh at flow fronts and to smooth it toward and backward. Applications of these new features are presented in the last chapter of this study. It is shown that the 3D simulation of large thin parts like car fender is now affordable using anisotropic meshes.
Complete list of metadata
Contributor : Magalie Prudon Connect in order to contact the contributor
Submitted on : Wednesday, April 23, 2008 - 11:33:12 AM
Last modification on : Wednesday, November 17, 2021 - 12:28:11 PM
Long-term archiving on: : Friday, May 28, 2010 - 5:19:44 PM


  • HAL Id : tel-00275305, version 1


Erwan Bigot. Simulation tridimensionnelle du remplissage de corps minces par injection. Mécanique []. École Nationale Supérieure des Mines de Paris, 2001. Français. ⟨NNT : 2002ENMP1073⟩. ⟨tel-00275305⟩



Record views


Files downloads