Learning Forum

[Jordan Dutch]

Hi,

For my project, I am currently doing a 3-point bending test to compare to experimental data obtained.

In the Ansys model, i have set the top roller to a -4mm displacement for a set amount of time. When I remove the displacement back to 0mm, I get errors saying that the solution failed to converge at all time points. I think this occurs due to the top roller becoming out of contact with the beam.

All contact points are frictional with a value 0.2.

Was just wondering if anyone knew any fixes to allow the solver to converge at all points, not just when the roller is in contact with the beam?

[peteroznewman]

Here is a relevant discussion.  Read it to the end to locate a link to another relevant discussion.

[Jordan Dutch]

Hi, my model is a little different from the linked discussion, my supervisor told me to do it as a 3D model not using symmetry to replicate the experiment conducted as seen in the photo below.

All the supports use flexible stiffness behaviour. I was thinking of changing this to rigid but the face split on my top roller makes the geometry invalid. Would it be a good idea to get rid of the face-split?

All the contacts are using auto-detection pinball region and adjust to touch interference treatment.

For my analysis settings, I have set the "displacement" of the top roller to -4mm in the y-axis over 120 seconds with 30 substeps. This seems to work fine. I then hold it at this displacement for 600 seconds with 30 substeps again and again seems to work fine. The problem comes when I take displacement in the y-axis back to 0mm over 10 seconds using 30 substeps and then the model comes out with an error saying the solution failed to converge over all time steps. The results then are only solved up until approximately halfway through this last sub-step as shown in the picture below, I assume this is when it loses contact. Is there a way to see results after losing contact? I want to see the permanent deformation in the y-axis after the displacement is removed.

I think the error is due to the contact region between the beam and the top roller. The other two supports are "remote displacements" that are fixed in all axes but allowed to rotate as in my real-life experiment they could. Any help would be much appreciated as I have been struggling for a while.

[peteroznewman]

This discussion has some similarity between the punch retracting and the center roller retracting in a nonlinear 3-point bending analysis.

[Jordan Dutch]

Hi, 

Is it similiar if i have to hold it at a certain displacement for time, therefore I cant do a 2-step analysis for the spring-back. my supervisor also wants me to do it as a 3D Full Model.

[peteroznewman]

You won't do a 2-step analysis, you will do a 3 or 4-step analysis. One of my replies in that other discussion mentions adding a last step to raise the punch above the plate.