Dealing with stress singularity and, evaluation of max equivalent stress

[bl2404]

Hello everyone,
I’m engineering student, and I have to perform analyze of stress of street lamp pole. It has octagonal cross-section and rectangular hole on bottom part. I modeled this lamp as a surface body.
Due to geometry I’m facing the problem with stress singularity and I can’t obtain convergence with mesh refinement, and my results depends on mesh density. I’d like to ask you what should I do in this case. I guess results on picture no. 2 are ok, but to get to know exact value of stress in corners I should perform analyze on exact geometry with rounded corners?

 

[rickfischer51]

This is what stress singularities do. The finite element stress at the sharp corner is infinite. The stress goes up as you refine the mesh because the distance between the integration points and nodes is smaller, making them closer to the corner, so you are averaging two numbers that get progressively closer to infinity.

Your options are:
1. Add a radius and mesh it accordingly. You will not get a perfect sharp corner in the real part. There will be a small radius. Take a look at a part under magnification and estimate the radius. The ease of this will depend on the inspection equipment you have available. This is a workable solution if the radius is not too small. The radius will be determined by how the hole is made, i.e. milling, punching, plasma cutter, etc.
2. Fill in the slot and run an analysis. Take the stress where the corner would be and apply a stress concentration factor. The best resource for those that I have found is Peterson's Stress Concentration Factors. Depending on your loading you may have to interpret your results to determine the stress type (tensile, bending, torsion,tensile+bending, etc) so you can match it up the the correct stress concentration chart.

Rick Fischer
Principal Engineer
Argonne National Laboratory