Publication Details




Integration of Parallel Data Refinement Method with Advancing Front Local Reconnection Mesh Refinement Software


Kevin Garner, Thomas Kennedy and Nikos Chrisochoides.


Published in Virginia Space Grant Consortium (VSGC) 2017 Student Research Conference, Williamsburg, VA, April, 2017




A procedure is presented for parallelizing an industrial-strength sequential mesh generator called Advancing Front Local Reconnection (AFLR) which has been under development for the last 25 years at the NSF/ERC center at Mississippi State University. AFLR is currently used at NASA (including NASA/LaRC) and other government agencies as well as in the aerospace industry such as Boeing. The parallelization procedure for AFLR presented in this paper is called Parallel Data Refinement (PDR) and it consists of the following steps: (i) use an octree data-decomposition scheme to break the original geometry into subdomains (octree leaves), (ii) refine each subdomain with the proper adjustments of its neighbors using the given refinement code, and (iii) combine all subdomain data into a single, conforming mesh. This approach is stable (i.e., guarantees the same mesh quality as the sequential mesh generation), it is robust (i.e., it can generate meshes for the same type of geometries AFLR can), and it can speed up execution time to a point where it can be 10 to 20 times faster on a 32-core node (PETTT Year 1 report sent to U.S. Department of Defense) (Kennedy, Chernikov, & Marcum, 2016). Finally, given earlier experience of an implementation of PDR with TetGen, a similar open source mesh generation code, the expectation is that one can modify about 5% of the AFLR code, achieving high code re-use. By the completion of this project, the robustness, stability, and end- user productivity aspects of the parallel mesh generation will be delivered.




  [PDF]          [BibTex] 



[Return to Publication List]