Development of Finite Element Analysis Lab Module in the Second Course of Solid Mechanics

Abstract

Many universities have started introducing Finite Element Analysis (FEA) at an earlier point in the curriculum. However, there is a wide diversity of university backgrounds, course content and sequence, pedagogical objectives and approaches, etc. This paper describes the development of FEA lab modules in the second course of solid mechanics in our specific context. Students in this course were introduced to FEA earlier in the first course of statics and solid mechanics. They had learned the basic steps in FEA for axially loaded and planar truss structures. In the second course, the FEA was extended to the planar cases. One of the objectives was to make the students aware of the descretization and numerical errors of the FEA. Hence there was a particular focus on element displacement fields and how they influence element behavior in comparison with an actual structure behavior. The lab modules were designed to be complementary to the class room learning. Approximate nature of the FEA was taught via the lab modules on descretization errors and numerical errors. The descretization error was demonstrated in the first part of the lab wherein different types of elements for planar problems were compared. One cantilever beam problem was solved with different types of elements and the results were compared with the theoretical value. Numerical error was studied in the second part of the lab wherein the effect of the element shape quality on the results was studied. A systematic study of the effect of mesh distortion was undertaken. ANSYS Parametric Design Language (APDL) macros were developed to change the mesh distortion quickly in a controlled fashion. A study of convergence of the results followed in the third part of the lab. A reasonable convergence was obtained for a plate with a central hole for which the theoretical results are known. Once the students grasped the need of convergence, a real life problem was attempted in the fourth part of the lab. The actual results are not known in the real life and a reasonable convergence needs to be established for acceptable results and for subsequent analysis and design. Design of a seat belt buckle was undertaken. A Pro/E CAD model was imported into ANSYS. The students used the subset of the CAD model to build their FE model considering only the relevant part, the symmetry and the mid plane. At the end of the semester, the students used the FEA tools for a real life design problem with a firm grasp of the approximate nature of the method.