Automated Preparation of Geometry Models for Computational Applications

In this proposed effort we seek to develop and integrate CAD repair algorithms that greatly decrease the time spent preparing complex geometry for grid generation as part of numerical simulation problem setup. Highly detailed CAD models are critical for design and manufacture of all sorts of geometries, from antennas to vehicles, bridges and buildings. However, in many cases small features in the CAD model are not needed for the numerical simulations carried out using those models, such as those involving structural mechanics, fluid dynamics, or electromagnetics. De-featuring, or simplifying CAD models and performing associated clean-up of the geometry entities remains a large bottleneck in the simulation process. In addition, deficiencies exist in exchanging data among CAD tools. This proposal outlines methods for greatly reducing the CAD preparation time using existing technology developed at HyPerComp, open source solutions where available for data exchange, and the development of new and novel capabilities for importing geometry and performing CAD clean-up.