Mesh Generation for High Performance Computing: Current Trends and Applications

Mesh Generation for Massively Parallel and High Performance Computing-Based Analysis

Instructor: Timothy J. Tautges, Ph.D.

Audience: Contains mostly intermediate/advanced material. Some background in mesh generation and high performace/massively parallel computing will be assumed.

Advances in massively parallel and high performance computing have made possible computational simulation at much higher fidelity and finer resolution. This has only widened the bottleneck posed by mesh generation for finite element analysis. This tutorial will present background on tools and mesh generation techniques for analysis on massively parallel computers.

Mesh generation starts with a geometric model, often in a CAD system. The analysis model is derived either by simplifying an existing CAD model or by building a new one with a CAD model as reference. This stage is common to all types of meshing. Massively parallel analysis applications are typially incorporating more detail than before, rendering the previous rebuilding approach inadequate. This tutorial will introduce new techniques for preparing geometry for high-fidelity analysis. This will involve using existing (CAD) tools as well as new techniques developed for ASCI-class meshing applications.

The meshing process for massively parallel and high performance computing can be quite similar to the current state of practice. For example, many of the same difficulties encountered in massively parrallel applications are also encountered in standard procedural applications. Mesh generation for massively parallel-based analysis also introduces problems with model complexity, large meshes, and data handling not usually present in smaller analyses.

Some issues specific to larger analysis will also be introduced, including techniques for handling model complexity, team-based approach to generating meshes, tools for generating meshes in pieces and assembling the pieces into a larger mesh, and preparing the mesh for input to MP-based analysis

The tutorial will also include exercises and examples from recent mesh generation applications for ASCI-class massively parallel computers