Topology Control and Simplication of Reality-based Meshes for Finite Element Analysis

Reverse Engineering has been exploited for several application, e.g measuring and inspection, customt design, cultural heritage. The result of this process is a 3D representation of the targeted object, usually encoded in an unstructured triangulated surface fashion (e.g. an STL le). Going beyond the mere visualization purpose and focusing on engineering products, prediction of mechanical performances is often required, especially for items critical from a structural perspective. Finite Method Analysis (FEA) is a de-facto standard to evaluate mechanical performances, both in academia and industry, but the meshes derived from reverse engineering techniques are not suitable for direct use in FEA solver. In those cases, a proper degree of accuracy, precision and reliability throughout the whole pipeline have to be guaranteed, from the acquisition of the geometry to the analysis of the nal result. Regardless the reverse engineering technology chosen, the standard pipeline requires the acquisition of the chosen object and the post process of the 3D model obtained in order to have a closed 3D mesh. The most important and thorny phase is the one related to the topological check of the 3D mesh and its simplication to obtain a proper model, suitable to be then translated in a computational grid for FEA. The problem is not new in literature [4] but researches have been focused more on large items (e.g. buildings). Firstly, the order of magnitude of the relevant feature is di erent: talking about architectural items, small features, i.e. less than 1mm, are negligible. Secondly, similar researches conducted so far don't evaluate FEA results systematically. This paper aims at comparing dierent methodologies in order to prepare computational mesh of geometries derived from reverse engineering technologies. A benchmark case, i.e. a structural-steel parallelepiped, has been chosen in order to have complete control over the variables involved in the process (both during the reverse engineering and the FEA). The test object has been acquired with a laser scanner and post processed in order to x artifacts. Once the mesh is closed and error-free, two dierent methodology for simplication have been used: a triangular simplication and retopology. The acquired geometry, before and after the simplications, has been compared with the reference model (i.e. a 3D geometry created using CAD geometric primitive): mean and standard deviation between the baseline model and the acquired geometries has been tracked. Finally, a tensile test has been simulated making use of a FEA software and the results have been compared with the theoretical solution