Benutzer: Gast  Login
Dokumenttyp:
Masterarbeit
Autor(en):
Rosa Lebkowski Jiménez
Titel:
Optimal Coarse Quad-Layout for Multiple Polycubes in CAD-Reconstruction
Abstract:
The output of a topology optimization process is a triangle mesh that represents a surface that has to be reconstructed in a CAD system. The developing workflow in-house at BMW for CAD-Reconstruction uses polycube maps to generate a coarse quad layout to represent a triangle mesh and then obtain a smooth geometry by means of subdivision surface. The quality of the output can be improved by tackling the quad mesh quality and the param- eters of the mesh. For that reason, this project profits the alignment of a polycube to its own coordinate system to improve the quad mesh quality with a more structured quad mesh, and finally modifies the result of the coarse mesh by means of shape optimization. First, this project develops a method to generate coarse quad layout a surface given by mul- tiple interconnected polycubes, using lattice grids, mapping functions and solving topology problems. It gets individual meshes for each polycube that are mostly quad-dominant, ex- cept for the areas where the meshes are interconnected in the geometry, which are left as triangular to preserve the relevant edges. The meshes on the polycubes go back to the geometry space with the inverse of the mapping functions. The result is a quad-dominant mesh that coarsely represents the geometry of the triangle mesh from topology optimization. This coarse mesh acts as control mesh for subdivision surfaces technique, supported by CATIA V5, CAD-module used widely in the automotive industry. Furthermore it makes modifications to the mesh obtained, with a constrained shape opti- mization problem with subdivision parametrization, that minimizes the deviation measured by the metric and maintains or improves the quality. The gradient based method to reach this solution is the Relaxed Gradient Projection. The resulting coarse mesh can be the input in CATIA V5 with the module Imagine & Shape or continue the process of generative design in further form optimization or assembly to a bigger geometry, with the advantage that it is not a dead surface but a parametrized geometry.
Betreuer:
Moustafa Alsayed Ahmad, Ihar Antonau
Jahr:
2022
Monat:
Mar
Sprache:
en
Hochschule / Universität:
Technische Universität München
Fakultät:
TUM School of Engineering and Design
TUM Einrichtung:
Lehrstuhl für Statik
 BibTeX