Mesh Convergence Test

[AlanSYH_]

Hello all,

I followed a youtube video & made a 3 point bending test simulation for my concrete beam. Was trying to find the optimal number of mesh to use so I tried a few simulations with different mesh size & got different stress patterns and more stress concentration at contact point instead. May I know if this is supposed to happen & how would you do the mesh convergence test then? Images are attached for references

 

[SWComposites]

Is it supposed to happen? Yes.
Is the FEM reality? No
You are attempting to model a singularity.
In reality,
The loading bars are not rigid.
There is very local material nonlinearity at the contact points.
Von Mises stress is probably meaningless for concrete (its a yield criteria for ductile metals)

Look up contact stresses in a text book.

 

[AlanSYH_]

Hey, thanks for the response. I will look into it. But what explains the different stress patterns on the concrete beam itself?

 

[SWComposites]

The color scales are different.
And its partly an effect of the stress averaging to create the contours.

 

[AlanSYH_]

So how would you find the optimal size for the mesh then?
I'm just confused what to do from here

 

[FEA way]

Check the following chapter of the Abaqus documentation: Benchmarks --> Analysis Tests --> Static stress/displacement analysis --> The Hertz contact problem. There you will find a similar case. Not 3-point bending but here it's also more about contact between the indenter and larger sample.

Is the loading pin modeled as a discrete rigid part ? Analytical rigid would be better.

 

[AlanSYH_]

Hello,

Yes, it is a discrete rigid part. I will try the analytical rigid part instead. Thank you for the direction!