Automated tetrahedral meshing from manifold tessellated optical scan data is investigated to determine its viability as an approach for finite element analysis. This approach avoids the costs of constructing a volumetric representation of the scan data that can be meshed with conventional grid generation approaches. This paper demonstrates an auto-meshing algorithm for inserted airfoil and integrally bladed rotor hardware. These automatically generated models are compared to experimentally obtained frequencies and mode shapes for validation.
In an effort to compare the fidelity as well as the effect of mesh density on analytical convergence rate, manually generated all-hexahedral models are compared against the auto-meshed tetrahedral finite element models. CPU time, solution accuracy, and mesh convergence are evaluated to determine the viability of automatically generated tetrahedral meshes versus the standard approach of manually generating hex-dominant meshes. This paper demonstrates that given the power of modern CPUs, automatically generated all-tetrahedral meshes can serve as a viable alternative to manually generated hex-dominant finite element models, especially when these meshes can be refined for solution convergence within the auto-mesher. This new approach effectively solves both the mesh convergence problem while demonstrating that models based on as-measured geometry can be rapidly built with virtually no human interaction.