COSMOSWorks - Rigid Connection

[lehighluke]

I have a question with an answer that I cannot find in the CosmosWorks Help. I am trying to apply a rigid connection restraint in a part analysis, rather than an assembly. Any ideas?

The specifics.
I am doing FEA on a vehicle frame, in which the motor/transmission is used as a stressed member. For my purposes, I am assuming it to be rigid. The motor mounts and tranny mounts are some distance apart, and I would like to prevent any deflections relative to eachother. I do not wish to model the drivetrain and make an assembly, my computers' resources are at their limit with the frame FEA alone.

Thanks!

 

[ShaggyPE]

If the rigid link is not an option when doing an fea on a part, you could insert your part into an assembly file and perform your fea there. Your assembly will only consist of the one part. You will have no contact or any of the other things that make assembly feas slow.

 

[lehighluke]

Thanks, but that didnt work. I am using a shell model, and apparently Cosmos cannot process a shell model in Assembly mode, it would not let me create a study...it gave a "no solid body to process" error

I wonder if there is some way to insert a "rigid bar" or some other simple structure in a part

 

[tar]

I don't understand what you mean by "I am using a shell model".

I just made a part in SWX and did a shell to hollow it out and it had no problem making an FEA study or meshing it.

 

[lehighluke]

By shell model, I mean that the entire vehicle frame is modeled with only surfaces...no solids (The actual production model is solid...i made a shell version soley for FEA)

It runs analysis just fine. My Q is how to make rigid connections w/in the same part. Oh, SWX didn't like shell models in ASSEMBLY mode. Parts worked fine

dunno *shrug*

 

[macPT]

Things that I'm thinking of, that need to be validated with some trial and error:

1 - Maybe you can simulate your study invertig the model, that is, making the restraint imovable in the motor mounts. The success depends alot on the model and other restraints/forces.

2 - Model a very stiff conection element like a profile(this should be attained by geometry, since you cannot assign a stiff material). The results for this profile should be ignored. This is not possible if you are considering the weigth

3 - Run the problem without any restraints in the motor mounts. Mesure the relative displacement between mounts.
Rerun the problem inserting displacement restraints that oppose the relative displacement. Test for correct results.

Good Luck

 

[lehighluke]

macPT... good suggestions

actually, w/ shells, even in part mode, you can make each shell a different material. I have like over 300 diff. shells, i had to cut it up alot to get it to mesh right. I could model a simple shape, assign it a near-infinite stiffness, and near 0 density. Only problem is that you would see the structure in the results, and I would always have to be explaining to lay-people what that is.

Anyway, I have given up on this, there is not alot of deflection between the points, so any bracing effect that the drivetrain may have will just add to my safety factor, I guess.

Thanks for the help guys

 

[macPT]

lehighluke

You're right, you can assign different materials to different shells. Duuuhhh!

Given this, the process you described of having a stiff material of almost 0 density it's correct and that's how I would do it. But the stifness, I would consider it in a value of 3 to 4 times the value of the other material, since big differences between material stiffness can cause numeric errors in the calculations.

About the explanations, well you have technical valid reasons to do it. If they don't understand them, then they don't understand FEA and there's no point in any explanation.

Be careful about your evaluation of deflections. Don't forget that, if you have a stiff part, preventing small deformations can cause big stresses! I think you should validate your model and study using that stiff connection part (check the deviations on the results of the two studies and make your own conclusions).

Regards

 

[MadMango]

lehighluke, you might also want to post this question in the FEA forum.
forum727

[green]"But what... is it good for?"[/green]
Engineer at the Advanced Computing Systems Division of IBM, 1968, commenting on the microchip.
Have you read faq731-376 to make the best use of Eng-Tips Forums?