Content area
Full text
The advent of automatic tetrahedral (tet) and semi-automatic hexahedral (hex) meshers has sparked a controversy regarding which type of element is most suitable for parts built with solid modeling programs. But the real problem lies in finding an efficient way to perform FEA for complex solid models without straining computing power beyond its limits. According to recent tests, a technology that takes both hex and tet lower-order meshes and forms selective higher-order elements during the solution stage may offer the answer.
Consider solid regions containing thin-walled areas, concentrated loads, and connecting assemblies made of different materials. Such regions stretch the capabilities of any automatic mesher and the accuracy of the elements, regardless of their type. They also tend to cause problems related to large numbers of degrees of freedom (POF). In general, any element will converge if its shape functions contain complete polynomials, have continuity across boundaries, and satisfy the patch test.
Many engineers accustomed to using only hex elements for analysis claim that tet elements are inaccurate and yield questionable stress results, and that tet meshers produce models with too many DOF for economical computer use. And there is some truth to this. Using four-node tetrahedra, for example, is less likely to yield a converged solution than using well-shaped 20-node hexahedral, unless a highly refined tet mesh is used. Of course, the computing power needed for...