What are FEM interview questions

10 golden rules for validating your FEM simulation

Thanks to the increasingly easy-to-use software, finite element (FE) calculations are increasingly being carried out by design engineers.
If FEM results are available, they must always be validated. This is important because every FE calculation is always an approximation.
This article uses 10 golden rules to show you how to ensure the plausibility of your FE results.

If you have created an FE model with a lot of effort, the loads have been correctly applied and the simulation results are in front of you, you need to check them for plausibility. There are 10 golden rules to follow. I have divided these into qualitative, i.e. not verifiable by numbers, and quantitative, i.e. calculable rules:

  1. Rule (qualitative): What should be evaluated?
    First of all, think about which load size should be evaluated. In the case of structural-mechanical problems, this is usually the elongation, tension or temperature. Assuming you are interested in voltage, then you will be looking at voltage gradients. Always have the contour plot and the FE mesh displayed. With the contour plot you do not have any “blurred” voltage curves displayed, these often lead to misunderstandings.
  2. Rule (qualitative): Evaluate places of high tension
    In the run-up to an FE calculation, you have already thought about what to expect from the FE calculation. This means that you already have an expectation of which areas in the component are highly stressed. Check qualitatively whether your expectations match the results. If this is not the case, find the cause! This can be of various types. It is possible that you did not take into account all the conditions in your preliminary considerations or you actually applied a faulty load.
  3. Rule (qualitative): Is the deformation of the component traceable
    Observe how the component deforms under load. Does your calculation match your experience / expectation?
  4. Rule (qualitative): Check the stress curve (e.g. stress curve)
    Optically good-looking voltage curves are an indication of a sufficiently fine FE mesh. Make sure that there are no jagged gradients in the evaluation area and that there are as few color jumps as possible (i.e. voltage differences) per element. Rule of thumb: a maximum of one color per element.
    Important: if you want to evaluate stress gradients, you have to mesh much more finely!
  5. Rule (qualitative): avoid evaluations in singularities
    Singularities are areas in the FE mesh where the voltage continues to rise as the mesh is further refined. I.e. not converged. This is always the case on inner edges without a radius, force introduction points. Evaluate far enough away from these points.
  6. Rule (quantitative): Are the networks sufficiently fine?
    There are essentially two element types for 3D elements. Tetrahedron (pyramids) and hexahedron (cubes). Make sure that the elements in the model do not differ too much from their original shape, i.e. are distorted. Some FE packages provide you with criteria for this and mark places with critical elements. Refinement of the network often helps. Also check whether the stresses to be evaluated do not increase significantly if the network is refined further.
  7. Rule (quantitative): Evaluate the level of stress analytically.
    Often one does not have a feeling for the level of stress. Use a rough manual calculation using analytical methods from the fields of strength engineering, technical mechanics or dynamics to check the amount of stress to be expected.
  8. Rule (quantitative):
    If you have a second FE program available, run the same calculation on both programs and compare the results. There should be no differences here for standard calculations such as linear structural mechanics calculations.
  9. Rule (experiences): Use your experiences
    Compare calculation results with experiences you have made with previous products. Was there a similarly high level of stress here? Were the high stress locations similar?
  10. Rule (attempt):
    The most reliable method of validation is trial and measurement. Compare forces, accelerations, strains or stresses that you have measured with the simulated values. Also compare the failure location during the testing with the simulations. This gives you the greatest security, but at the price that it is time-consuming, expensive and possible relatively late in the development process.

The FEM method is a powerful tool. By following the golden rules mentioned above, you will ensure that you are using this tool properly.
Always remember:
FE result "only" reflects the engineering expectation! This, however, with high accuracy.