Jason F. Shepherd
CUBIT Mesh Generation Toolkit
|
![[Sandia National Laboratories]](http://www.sandia.gov/images/logo.gif){kind=logo}
|
I've worked on the CUBIT meshing toolkit for Sandia since 1997. At Sandia I've been funded mostly through DOE's ASCI initiative, and MICS. (check out the following website for other interesting MICS projects at Sandia.)
In CUBIT, I've been most active with the Multisweeping algorithm for generating all-hex meshes of 2.5 D volumes containing multiple target surfaces. I have also done a fair amount of work with interval matching algorithms, sweep verification, mesh grafting, and auto-scheme selection.
Mesh sweeping is taking a surface quad mesh and extruding it to produce an all-hex mesh. One-to-one sweeping takes a single surface mesh, known as a source surface, and extrudes the mesh through one or several layers to a single "target" surface. Many-to-one takes several "source" surfaces and extrudes the mesh to a single target surface. Many-to-many sweeping, or multisweeping, creates an all-hex mesh in volumes containing multiple source and target surfaces. The difficulty with multisweeping is guaranteeing mesh conformity through the volume and often requires matching the target surface topology with the source surface mesh.
The multisweeping algorithm in CUBIT was originally authored by Mingwu Lai as his doctoral dissertation at BYU. His code was integrated into CUBIT by David White. David also wrote the multisweep surface partitioning algorithms utilizing the virtual geometry capabilities already within CUBIT. I have been in charge of making the tool ready for production meshing by improving the robustness, efficiency and capabilities of the algorithms used to multisweep a volume. Our latest paper on the multisweep algorithm in CUBIT, entitled "Methods for Multisweep Automation" is found in the Proceedings of the 9th International Meshing Roundtable.
Scott Mitchell and I have been working on an algorithm for guaranteeing and forcing the sweepability of a volume. The algorithm, known as SweepVerification, is also discussed in "Methods for Multisweep Automation". Scott has recently written the SweepVerification as a linear program, and will be using it as a basis for some of his future research. The algorithm works by setting vertex and curve types, and to some extent, the surface meshing schemes, on the volume to guarantee, or force, the sweepability of the volume. The algorithm has proven to be very useful for us, and has simplified the meshing process and reduced the amount of time to mesh several difficult volumes and assemblies.
While I was finishing my graduate work at BYU, Dr. Steven Benzley, Steve Jankovich, and I developed an algorithm for creating transitional meshes into other volumes. We call the algorithm "mesh grafting" and have filed for a patent on the algorithm. This algorithm was presented at the 8th IMR and the paper "The Graft Tool:An All-Hexahedral Transition Algorithm for Creating a Multi-Directional Swept Volume Mesh" can be found in the roundtable proceedings. The algorithm uses some techniques discovered by Scott Mitchell in his work in Whisker Weaving. (see Scott's paper on pillowing hexes)
My master's thesis was entitled "Interval Matching and Control for Hexahedral Mesh Generation of Swept Volumes. This is an algorithm designed to formulate interval constraints for swept, and eventually submapped, volumes which are not caught by the surface meshing constraints. The algorithm, in a nutshell, uses a graph algorithm formulated from the volume's topology, to find directed paths and cycles in the volume topology, and then extract interval equations which can be supplied to Scott Mitchell's surface interval matching linear program. This research was presented at the 2nd Symposium on Trends in Unstructured Mesh Generation in 1999. The paper is published in a special edition of the International Journal of Numerical Methods in Engineering entitled "Volume Interval Assignment for Swept Volumes with Holes".
