FEA die crimp non linear quasi static 1

[MechEngineerNT]

Hey Guys,

I'm running a quasi-static non-linear FEA in Ansys trying to simulate crimping a primer into the cartridge using a die crimp tool. In short, the crimping operation works in real life (80 units tested), to which I've decided to try to simulate. However, the results of the simulation aren't matching up with the real life results. I've put in all the right material properties of a 7075-T6 aluminum, and have also put in the bi-linear curve to simulate strain hardening. In the attachment, you can see I'm above the ultimate strength of the material with von mises equivalent stress criterion, although it is in compression. Why does it work in real life, but the simulation isn't showing it? This has been giving me hell!

Thanks!

 

[L_K]

How is the simulation not matching with measurements/real life? What are you comparing?
Is your model 2d axisymmetric? Are you applying a force or displacement?
One option could be to model the material using a multilinear
model instead of bilinear.

 

[MechEngineerNT]

Thanks for the reply, L K. I'm very concerned about the stresses the crimped material exhibits when the die comes down and pinches it to the primer. Some portions of the material are above it's Von Mises Maximum ultimate tensile strength when it undergoes the crimping process (as shown in red). This would certainly indicate possible failure, and I would not pursue this option-- however, all of the cartridges and primers were tested at 80C with 80/80 units meeting functional requirements (not blowing out) and igniting propellant.

The simulation is 2D axisymmetric with an Isotropic bi- linear curve and a displacement is placed on the Crimp Die. There is a fixed support on the very bottom of the cartridge. I used a displacement boundary condition for this quasi-static simulation because applying a force resulted in rigid body motion, and it was near impossible to get the damn thing to converge. The stepped displacement works pretty well in this situation (achieves a solution) ....that I'm not happy with!!!!

I'm going to input a multi linear and see if that helps at all and I'll touch back.

Thanks!

 

[L_K]

Getting the surfaces to touch initially (move the die tool lower and set interface treatment: adjust to touch) and decreasing
time step size will help convergence when using force.

Did you define your material using real stresses and strains instead of engineering stresses and strains?

How many elements are showing stresses over ultimate stress?

 

[MechEngineerNT]

Yeah, I'm trying to understand what benefit applying a force would have as opposed to a displacement value?? They should achieve the same result. And yes, I'm using True -stress strain curve, not the engineering stress.

It is a substantial amount of elements, however, they are very marginal-- abou .88-.99 FOS. I'm wondering if the multi linear curve will do the trick.

 

[MechEngineerNT]

If you take a look at the video I posted, all the elements in red are above .88-.99 ult FOS to von mises failure criterion.

 

[L_K]

I guess none if you know the real displacement of the tool and not the force.

 

[MechEngineerNT]

Is it possible the pieces in real life are being work hardened by the crimp, and thus attaining a higher ultimate strength?

 

[L_K]

I doubt if it will change results a lot but you could run the analysis with large
deflections (if you have not already). What friction coefficient did you use?

 

[MechEngineerNT]

Yeah, have already done that from the start (large Deflection). Simulation would crash if large deflection wasn't activated. I put in a friction coefficient of around .2 (aluminum to brass), however it didn't make any real impact.

--Alex

 

[L_K]

I encourage you to post this thread to FEA or Ansys forums. They might have good insights on this subject.

 

[WARose]

Your link doesn't work for me....and I am not familiar with the process....but sounds like part of the problem may be the contact stresses. Contact stresses are (of course) hard to accurately estimate with FEA because most FEA programs don't accurately account for the localized deformation among other things. (I've typically used hand calcs for that.)

Another problem may be the stresses from permanent deformation. That gets into non-linear behavior and you have to start thinking about a permissible strain as much as stress.

EDIT: By the way (don't take this the wrong way), you know what stress concentrations are right?

 

[3DDave]

This looks like a duplicate post. It's not a structural topic.