Welcome to the final entry in our finite element analysis (FEA) blog series, in which I’ll discuss a little about analyzing and evaluating FEA results. Over the course of this blog series, I’ve covered tips on setting up the model in CAD and in the FEA program, configuring the analysis, and generating results. The final step in the process is the analysis of the results, both to get the answers to whatever problem inspired the FEA simulation and to ensure that the final results are accurate and representative of the system’s actual physical behavior.
These analysis tools and graphics generators aren’t just useful for generating pretty reports. One of their most important purposes is that they let the engineer easily check their own work. Since FEA software is very sensitive to incorrect inputs or problems in setup, it’s vital that the engineer make sure the model has solved in a way that makes sense. There are a few simple things to check that can show if results are accurate, or if the mesh needs to be refined and the simulation re-evaluated.
Checking the largest values for the maximum and minimum principle stresses is the first quick check that an engineer can perform (these can by highlighted in Autodesk™ Simulation using the toolbar shown in Figure 3). These should have reasonable values and be located at points that make sense to the engineer; otherwise, a value far past the failure point for that material or a stress concentration in a single unexpected spot might indicated that the mesh was not generated accurately. For a higher level sanity check in systems involving deformation, it can also be useful to check the sign on the principle stresses across the model, indicating whether sections are in tension or compression. It’s usually easy to predict whether an area should be in tension or compression, so this is a fast way to confirm that the loads and boundary conditions were applied correctly.
For instance, since our concrete slab example was bowing under its own weight, we would expect that the top surface is generally being compressed, while the bottom surface is under tension as it is stretched out. Since our simulation results matched our predictions, we were confident that we had created an accurate model. The shear stress for our model is shown in Figure 1.
Fix Errors in the FEA Process
It takes a lot of experience with FEA programs to get a simulation right on the first try. Even if a simulation converges to a solution without aborting or running out of system memory, there’s a good chance that upon viewing the results the user might find unexpected or incorrect mechanical behavior. Depending on the problems discovered, the user will have to work backwards through the process, ensuring that the analysis parameters are correct, that the mesh was generated appropriately, and that the CAD file does not contain any interferences, errors, or extraneous complexities. After iterating through changes to the setup and review of the results, one can be more certain that they have the most accurate simulation for the situation.
The Whole Process
Getting all the way from a CAD model to a set of accurate and useful FEA results can be a long process, and frustrating for an inexperienced FEA user. One must be thorough as they go through the process, since finding a mistake in the results might mean many more hours of mesh generation and analysis.