no convergence at contact location

[mbuller32]

I'm new to FEA so if I leave out important info. required to answer my question, please let me know.

The max von Mises stress isn't converging @ the contact location of the pin and hole.
I performed H-adaptive and stress increases on every iteration.

Courses of Action
I can either
(1) Figure out how to make the point converge or
(2) ignore this singularity and determine if stress else where is converging.

I don't know how to do either, please give guidance.


Picture of Stress Singularity

https://files.engineering.com/getfi...-a2cf-48c2-b770-a97371930629&file=Capture.JPG

Picture of Mesh

https://files.engineering.com/getfile.aspx?folder=479bd8eb-d963-42d0-b1eb-54b702b813bf&file=mesh.JPG

 

[NRP99]

1-Recommend to mesh it with hex (brick) mesh with sufficient density (mesh element size) at contact location. Then increase/decrease the density at contact location which is called mesh convergence study.
2-The mesh you done is rather coarse Tet mesh with lot of variation from contact point towards the other areas. Maintain smooth transition for at least surrounding areas. Are you using linear tet? Use quadratic tet and check mesh convergence study.
3-Normally stress singularities are hard to get converged. But you can try finer mesh. If the contact node is not converging, then take node next to it and check for convergence study.

Mesh convergence study - Plot Element count/node count vs output value (Stress, strain displacement etc). When the values are within +-3% (you can take judgment value of up to 10% but less is good) of max stress for typical size then mesh is converged and you can use that value as converged value.

Check these links -
Link
Link

 

[FEA way]

You didn’t mention the name of the software but I recognize SolidWorks Simulation. Recent versions have a tool to detect stress singularities, it can be useful here. Unfortunately, this software has limited capabilities in terms of meshing. Hexahedral elements are not available here and you can only try local refinement of a tetrahedral mesh.

Add plasticity to this analysis and you will see more accurate stress distribution. This will also let you determine the size of the yielding zone, it might be small enough to ignore it.