Answer:
Yes, you can very quickly estimate whether an FE mesh is fine enough by comparing the averaged and unaveraged results.
In the following, you see a surface with a free rectangular load and lifting line supports. The upper window shows the first principal stresses, smoothed continuously within the surface, and the lower window shows them constant in mesh elements. The meshing increases in the models from left to right.
As you can see at first sight, the correlation improves as the mesh refinement increases. Compared at a selected point, you can see the deviations of the results in the following table.
| Mesh | 0.500 m | 0.250 m | 0.100 m | 0.050 m | 0.010 m |
| Uniform | 17.657 MPa | 16.614 MPa | 17.015 MPa | 17.031 MPa | 17.045 MPa |
| Continuous | 14.009 MPa | 16.104 MPa | 16.863 MPa | 16.992 MPa | 17.043 MPa |
| Deviation | 26.04% | 3.17% | 0.90% | 0.23% | 0.01% |
A similar behavior can be observed for the strains, as shown in the following image.
This rough evaluation cannot replace a detailed convergence analysis. However, it is recommended to use this procedure as a first estimation.
It should also be noted that this does not work for so-called singularity locations, as the stresses there increase to infinity for an infinite mesh refinement.