Skip to main content
U.S. Department of Energy
Office of Scientific and Technical Information

Unstructured mesh relaxation

Conference ·
OSTI ID:6686315
The relationship between the human asking questions and the code providing answers is being revised currently; the trend is toward a situation wherein each partner does what he/she/it does best. The human excels at processes requiring thought, such as analyzing the results of a simulation, and organizing lists of things to do and guidelines for doing them. Currently, the computer excels at following instructions and doing tedious, repetitive tasks without error. Thus, the human should ask the questions -- set up geometries -- specify boundary and source conditions -- and analyze the results. The code should decide, dynamically, during the evolution of the problem, where to refine the mesh and which algorithms to use in which regions of the problem in order to provide both accuracy and robustness. The ultimate code will be based on an unstructured mesh. It will involve the ability to do Free-Lagrange and ALE (Arbitrary-Langrangian-Eulerian) and it will dynamically adapt the mesh to suit the problem at hand. It will be capable of adopting other hydro algorithms and of including other physics options. At its heart will be a mesh manager that will decide which option is appropriate for each region of the problem. Free-Lagrange methods, ALE on a structured mesh, and adaptation are existing technologies. ALE requires a method for relaxing the mesh in regions of high fluid distortion and we do not yet have a robust method for relaxing an unstructured mesh. This paper addresses that problem.
Research Organization:
Lawrence Livermore National Lab., CA (USA)
Sponsoring Organization:
DOE/DP
DOE Contract Number:
W-7405-ENG-48
OSTI ID:
6686315
Report Number(s):
UCRL-JC-104630; CONF-9006195--8; ON: DE90015614
Country of Publication:
United States
Language:
English