Singularity detected at grid component Error as soon as I change my interference from 0.006" to 0.017" in a double shoulder pin-box connection.

[vinay_bhargavGRELT]

Hello Everyone,

I am running a make up torque static non-linear analysis on a double shoulder pin-box connection with threads modelled as grooves. FEA was running good and producing high quality results when I had my interference of two connections set at 0.006" but as soon as I increased my interference to 0.017", it started giving me "Singularity detected" error E5049 and does not converge starting from very first subcase even at 0.030% of load case. Even after 12 hours of run, it bisected with max iterations reached and did not converge at all. 

I believe it is something to do with my surface contacts but not sure how to fix it. I am sharing my screenshots from 0.006" interference study that is working for contact setup and subcase setup. Please advise.

[vinay_bhargavGRELT]

more screenshots

[John_Holtz]

Hi @vinay_bhargavGRELT 

The maximum activation distance on the contact dialog (0.05) is very small. What is the mesh size on the contact face?

The maximum activation distance needs to be large enough to account for the element size and the interference (or gap). The rule of thumb is maximum activation distance = 1.1*square root(mesh size^2+interference^2).

Or maybe 0.017" interference is creating such a large stress that the analysis is not able to converge. You could set the "Nonlinear Setup > Intermediate Output > All". Hopefully that would show the results from some of the iterations so that you can determine what is happening.

Note: Subcase 1 should not have any loads applied. The solver is just trying to solve for the pressfit in subcase 1.

Let us know what you find out.

John

[vinay_bhargavGRELT]

Hello John,

Thank you for your response.

The mesh size around these contact faces were 0.03" while the interference is 0.0171". Based on the formula, the maximum activation distance should be 0.038" while I have it set up at 0.05".

The stresses produced can be large and while I agree with you on intermediate results output set to "All" (I have that set up too), it is not even converging for subcase 1 increment 1, thus not producing any results whatsoever.

I did manage to solve it by turning the maximum activation distance "OFF" (AUTO) and very fine mesh of contact shoulders to 0.001". However, it did not produce any quality results as I see no stress on the shoulders which should not be the case.

I am not sure what else to try. Please advise.