Robust Algorithms for Managing Large and Complex Multi-Volume Assembly Meshes

High Performance Computing has enabled computer modeling of everlarger and more complex problems. For some of these problems, models utilizing allhexahedral meshes are preferred due to better accuracy and efficiency. However, algorithms for automatic generation of all hex meshes in general geometries have so far eluded researchers. Currently, these meshes are typically created by subdividing the original geometry into simpler subsets which are meshable using traditional algorithms. As models grow in complexity, more and more subdivision work is required for successful meshing, and it becomes successively more difficult to insure that all conditions are met in each sub part to allow successful meshing. Research has shown that up to 80% of the total modeling time can be taken up in this meshing step. In the absence of a fully automated algorithm, effort has been directed at aiding the user in subdividing the model. Algorithms for automatic scheme selection greatly assist this effort. These algorithms typically try to emulate an experienced user and can use very sophisticated algorithms to select a correct meshing scheme for each sub volume. However, the selection is often somewhat ambiguous, and with hundreds or thousands of sub volumes, even one incorrect selection will prevent successful meshing. Further, even a correct local selection can be invalidated by global conditions in the model far removed from the sub volume being considered. Taken together, these problems make it very difficult for the computer to assist the user with this task. Better algorithms are needed to make construction of valid allhex meshes faster and easier for the user. The Trelis mesh generation software from csimsoft, based on Sandia National Laboratories Cubit, will be used as the test bed to first improve the automated scheme selection algorithm. Since conditions are often ambiguous, the algorithm will be modified from an expert system which tries to select a single correct scheme for each volume to an exhaustive algorithm that checks all possible solutions and reports successful permutations. The algorithm will also be enhanced for more robustness. Additionally, with a robust method to verify that a geometry is meshable, the algorithm can also employed to enable a more robust automated subdivision algorithm to further aid users. Commercial Applications and Other Benefits: The algorithm improvements will be available in Trelis, a commercial pre and post processor from csimsoft. This enhanced system will allow faster and easier generation of allhexahedral meshes in complex assembly models, which in turn will facilitate faster simulation and modeling of engineering systems using High Performance Computing.