Learning Forum

[Daniel Koch]

Hi, for a student project we are trying to model to start with a simple car upright, however we have a stumbled upon some issues with a singularity, in the meeting point of the two contact points (90 degree angle). When refining the mesh for a convergence analysis, the stress keeps increasing:

The meeting point is for the cars upright and shaft. As this meeting point will be bolted, we tried both gluing the two volumes together, which results in the singularity and also a few notes away from the meeting point. We also tried a bonded (always) contact model, however this results in stresses higher than the yield strength of Aluminium even when this case is for steady state of the vehicle.

I read here on the forum about the bonded (always) condition, however I'm sligthy confused as to why it would like a friction coefficient incase of using this. As it simply says there will be no friction in the model then.

As of now we are considering making a fillet, to make this corner less sharp. However a real life situation wouldn't have such fillet sticking out? Is there anything else that could be done, to obtain a better model?

Btw. the following model is set up in Mechanical APDL, as that is what we have been provided from the university and learned.

[Armin]

Hi Daniel,

You are right and adding a fillet would reduce the stress concentration. Please see the video below where the topic of artificially high stresses is discussed in detail.

Understanding and Dealing with Artificially High Stresses — Lesson 3 - ANSYS Innovation Courses

Regarding the contact with bonded behavior, as far as I know, no friction coefficient is needed because the parts are fully bonded and thus no sliding is allowed.

[Daniel Koch]

Hi Armin, thank you for responding back so fast. Just got through the video your sent and as of now it also seems like the easiest fix for the given situation, to add a fillet/chamfer.
Sorry to ask this, but wouldn't it be bad to add a fillet that you in a real life situation, might not be seeing on the product? I do understand that this will fix the FE problem with singularity on the contact, but just seems odd to me?

[Armin]

No problem Daniel. As you noted, these are simplifications and may or may not represent the actual scenario. I'm not familiar with how the two parts are connected in your case but you can also look into different "joint" options available in Ansys Mechanical.

[peteroznewman]

Daniel,

Graphs should always have axis labels including units. Since you are talking about a stress convergence plot, I assume the Y axis is Equivalent Stress (MPa) and the X axis is Element Size (mm). The curve in the graph does not show a singularity. The stress does not increase by a factor of 2 when the element size is reduced by a factor of 2.

[Daniel Koch]

 

Hi Peter, thank you for your answer and yeah you are right I didn’t even realise that. Since the value is still rising with smaller element size, then how would you go about having a conclusive value? Would extrapolation down to zero in element, be a good way to do it or is there a better way to go about this?

Also since we are only interested in the cars upright for results, we have decided to make the shaft slender and also adding a fillet to that slender region. Keep in mind I haven’t actually done a convergence analysis on this one yet.

Once again, thank you for your respons!

 

[peteroznewman]

Yes Daniel, you can extrapolate to zero element size to estimate the peak stress.

[Daniel Koch]

Alright, thank you so much for your quick reply ?