An Efficient Error Approximation Technique for Use with Adaptive Meshing

This paper on "An Efficient Error Approximation Technique for Use with Adaptive Meshing" was presented at the International Conference on Quality Assurance and Standards in Finite Element Analysis - 22-25 May 1989, Stratford-upon-Avon, England.

Summary

The expansion in the utilization of finite element analysis in recent years has been accompanied by an explosion in the uncertainty about the accuracy of any analysis. Error control has been traditionally left to the analyst. who must be able to evaluate the adequacy of a mesh by visually reviewing the mesh and/or results.
Quantitative error evaluation theories are available (Ref. a1-a5, b1), but have not been widely implemented due to the relative cost of determining an error. This paper presents an implementation of the error approximation of Zienklewicz and Zhu (Ref. a6) in the general purpose finite element program ANSYS that allows for efficient error approximation, both in magnitude and location. By keeping the resources required to approximate the error to less than 10% of the original analysis, it is now a practical tool for engineering and analysis. With knowledge of the approximate location and magnitude of the error, an adaptive meshing procedure can be established.