Impact Attenuator Report - Solidworks?

[Alexisonsmith]

Hey,

Does anyone know if I can test my solidworks model of my nose cone in solidworks and getting away with testing it on solidworks instead of having it built?

Alexisonsmith

[t21jj]

No some form of actual test must be done.

[Adambomb]

Yikes, have you been to competition before? The rules are quite clear on this. This isn't like those other student design competitions, where you "get away with" BS.

[Adambomb]

Also...pretty much every impact attenuator I've seen relies on at least two of these things which Solidworks Simulation won't do:

1. Plastic deformation
2. Anisotropic materials
3. Self-contacting surfaces
4. Strain-rate effects

[Kirby]

The only package I have used (albeit, briefly) that might even be able to model a modern foam, composite or metal impact attenuator is Ls-Dyna.

If you want to go ahead, do an FEA analysis, the judges will still want physical validation.

Previously, we've had enough trouble convincing judges of repetitive manufacture quality of physically tested attenuators.

God knows you'd never do it just with a computational model.

[Tech Guy]

I agree with t21jj, Adam and Kirby.

Firstly the Rules say "test data" and secondly, any experienced engineer will tell you that ANY modeling, whether it be an FEA, GT Power or WAVE, is only as good as the validation. In other words, to trust a simulation or a model, you have to validate the model first with some actual testing. Too many student teams, especially some from outside North America, do some modeling and believe everything the computer tells them! The old saying is "garbage in, garbage out".

I believe there is an FAQ that says you do not have to do a dynamic test, but a crush test on an Instron or suchlike, is acceptable, as long as you interpret the results correctly.

[Adambomb]

Originally posted by Kirby:
The only package I have used (albeit, briefly) that might even be able to model a modern foam, composite or metal impact attenuator is Ls-Dyna.

If you want to go ahead, do an FEA analysis, the judges will still want physical validation.

Previously, we've had enough trouble convincing judges of repetitive manufacture quality of physically tested attenuators.

God knows you'd never do it just with a computational model.

I worked with a company doing a computational model of impact testing, they used ABAQUS. It was a true dynamic study, included pressure waves, strain-rate effects, multi-body dynamics and lots of contact. The purpose of it was to help pre-validate models before they went into production (real physical testing was naturally still done; this was just supposed to help identify failures before a full production model was built).

It was an 18 month project headed up by an experienced CAE engineer with a masters in engineering mechanics (specializing in FEM). Run time was on the order of about a week on a 10 node cluster.

[alumasteel]

I worked with a company doing a computational model of impact testing, they used ABAQUS. It was a true dynamic study, included pressure waves, strain-rate effects, multi-body dynamics and lots of contact. The purpose of it was to help pre-validate models before they went into production (real physical testing was naturally still done; this was just supposed to help identify failures before a full production model was built).

It was an 18 month project headed up by an experienced CAE engineer with a masters in engineering mechanics (specializing in FEM). Run time was on the order of about a week on a 10 node cluster.


Good lord, a week on 10 nodes? How many cells/node points did the models have? I've been working on my thesis research with a guy who just finished his PhD, and he studied sliding wear at the micro-level by modeling asperity impacts in ABAQUS. I don't think his took nearly a week to solve and he had a fairly fine mesh. I'm doing the same work, only at higher sliding velocities and I have to use a hydrocode to solve the problem since finite elements crap out at the speeds I'm studying. I have a fairly fine mesh (160,000 cells), but it only takes 1-10 minutes on 8 nodes depending on how many other jobs are in the queue.

Back to the original post: If you know what you're doing and how FEA works, you can develop your attenuator effectively. I'm not sure how great SW is for that. It isn't the most robust FEA package out there. It's a modeling program first, and the physical simulation is more of "rough pass" than anything. Might be a better idea to use a program intended for FEA for doing design iterations. Once you get a design you like, you should definitely test it and validate your model.

Before you get to in-depth with doing any analysis, check to see what constitutive equations you have available in the FEA software. As mentioned, you'll need something that can handle high strain rates and high levels of strain. If you don't have anything available that is developed for that purpose, then you'd be wasting your time anyway.

[Adambomb]

Originally posted by alumasteel:
Good lord, a week on 10 nodes? How many cells/node points did the models have?

Don't remember off hand, although lets just say it was a pretty big model.

[alumasteel]

Yeah, I guess so...

Another thing for the original poster... you'll need to check what kind of elements are available to be used. If you're planning to use a sheet metal, or carbon fiber, box type of attenuator, then shell elements would be helpful. Not sure off hand, but I'm pretty sure SW doesn't have that ability. I could be wrong, though...