Press-fit contact analysis with no load--unable to run

[gpommeranz]

I have a simple contact analysis that I would like to run.  I have a gear that is press-fit onto a shaft.  Instead of enforcing a displacement (the distance the gear in pressed onto the shaft), I wanted to start with the gear already in it's final location and simply run the analysis with the press-fit box checked and the maximum interpenetration set to my press-fit distance (0.038mm in this case).  However Simulate gives me an error that I can't define the analysis without either a load set or enforced displacement.  Is there a setting I have incorrect?  I thought I could run a press-fit without needing a load set or enforced displacement.  I've attached my assembly with the setup I'm having issues with.  

[346gnu]

As well as a zero load Stephen mentions, (which simply fools the software into thinking there is a load set so that it runs a static analysis),

you will need a light spring or two to prevent the female part translating axially or rotating ... contacts are by default frictionless and the model will likely fail insufficiently constrained. If it doesn't fail it may take a much longer time to solve.

Deselect the Press fit (initial interpenetration) check box, this is irrelevant in your case and just confuses.

bfn 

[gpommeranz]

Thanks, I wonder if the engine error is due to the default frictionless setting.  I changed the setting to infinite and it ran fine.  Adding the small load took care of my other problem too.  

[346gnu · Accepted solution]

Stephen,

I ran it as was upoaded using one of the predefined studies in the zip, it immediately failed (as expected) as there where no loads.

I made the following single change:

ADD a global temperature load at assy level with zero ref temp and zero model temp. This gives me the zero load - or as I refer to it, a 'dummy' loadset to make the s/w carry out the static study.

Running the model : The model failed insufficiently constrained (as expected) the female part rotates/translates on the frictionless contact. The residual norms in the .pas file show that something is moving somewhere.

So I added 2 light springs to prevent spinning/axial transltion of the female component on the shaft. I changed nothing else at all and it ran no problem in about 2 minutes.

NOTE I am viewing MPa

It took me far longer to write this than to fix and run

I don't understand why you are getting internal engine errors. What version of the s/w are you using?

This was Creo3.0 M120

Charles

[Domen]

Hi Charles.

Can you give a little bit more detail, how you defined the Springs?

Simple, Advanced or To Ground?

What are the references - center point? Did you add any weighted links?

Regards,

Domen

[346gnu]

Domen,

The springs are SIMPLE - POINT to POINT.

I used convenient vertices. But sometimes it is better to create datum points.

The springs only have axial stiffness in this case (the configuration of springs deals with both axial and rotational DOFs)

No weighted links required

Considerations are:

• Use points on parts. Do not use ground springs. The movement of one part relative to another is small and relative to ground is just wrong
• Preventing axial movement with a simple spring will work if the spring is parallel to the axis of rotation BUT it may still fail by spinning as the spring cannot exert any tangential force (and vice versa). Therefore ensure your system of springs is able to apply both axial and tangential components.
• Choose the stiffness of the springs so as not to affect the results - which causes users problems. So .... Remember that the springs only have to be 'light' and have to overcome numbers that should be zero in a stiffness matrix somewhere but are actually small eg. 10e-13. But also, don't choose a very light spring eg. 0.0001N/mm. The 'negligable tangential and axial forces may turn out to me moderately big e.g. 10N (which can be ignored when compared with forces in the model and that disappear into the noise of errors) and with such a low stiffness the displacement is massive and it fails insufficiently constrained again (or takes ages to solve). I can't remember the diameter of the shaft in the model but say it's 60mm then its bending/torsion/tension stiffness can estimated by hand and choosing a spring 1% this value is reasonable. My springs were 1000N/mm (guess); and is a good choice for most of the structures we deal with for small/negligable relative (see first point) movements that are measureable at the microstrain level : microstrain x 1000N/mm = small force.
• When adding springs always add spring force measures to confirm that in the grand scheme of things the springs carry ignorable loads leaving your freebody calculations unaffected. If you have time add component measures to each spring. (I really wish PTC would automatically put these in a file because its a time consuming, error prone faff otherwise)

Hope this helps

[skunks]

example attached

regards

paul

[gpommeranz]

Thanks everyone for your help.  I appreciate it.