crush analysis 4

[gio1]

Hello

I am doing some dynamic analyses (with Abaqus/Explicit) to correlate the crushing of an extruded aluminum cylindrical tube with experimental results.

The tube is mounted around a steel rod (assumed rigid) with no clearance. It is supported rigidly at one end and is impacted at the other end by another steel tube (assumed rigid) of a certain mass and speed.

I have managed to (almost) match experimental results in terms of forces, acceleration and displacement, but there is a fundamental difference with the experiments.

Experiments show that the crushing causes the aluminum to fold several times over itself around the tube ONLY ON THE END WHICH IS SUBJECT TO THE IMPACT. The crushing stops halfway through the length of the tube and the remaining part of it remains perfectly straight.

Analyses show that as the crushing progresses the tube folds alternatively AT BOTH ENDS and eventually a straight section is left between the folded ends.

I have tried analysing the tube starting from several different pre-buckled meshes and also from an underformed shape (perfectly cylindrical), but I always come to the same conclusions.

Given the symmetry of the problem I would expect analysis results to be correct. So why the experiments show that the deformation takes place only on the end which is subject to the impact??

Thanks

Gio1

 

[jhardy1]

My guess is that in real life, the boundary conditions at the two ends of the test piece are not the same. It sounds as though the fixed end may be partially clamped by the snug fit around the steel rod, effectively preventing buckling deformations at that end, whereas the impacted end of the tube is less restrained and therefore able to deform more freely. Have you modelled the restraints at both ends of the tube identically, or is one end freer to deform than the other?

Hope this helps!

 

[GregLocock]

Your analytical result contradicts common sense. Your experiments don't.




Cheers

Greg Locock

SIGlease see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[VoyageofDiscovery]

gio1,

Conduct experiment without impact.

VOD

 

[GBor]

Why do you assume that the other cylinder is rigid? For an analysis of this magnitude, I would model the additional cylinder and get the boundary conditions as far away from the physical problem as can be reasonable. You may be getting some "crushing" on the end with the rigid cylinder because as the impacted cylinder tries to displace, the rigid cylinder holds part of it in place and buckles an area that normal would flex.

 

[gio1]

At the moment both ends are modelled in the same way, i.e. the only restraint is contact with friction. I will now try to clamp the non-impacted end as suggested by JulianHardy.

I had also already planned to run the experiment quasi-statically, as suggested by VoyageofDiscovery. Of course will re-run the analysis to match these conditions

I am assuming the other cylinder to be rigid because it is made of hard steel; i.e. much stiffer than Al. Perhaps though assuming it to be rigid is a bit too much. Will try this one too GBor.

Will update with results as soon as I get them. I put my money on JulianHardy's tweak, although don't know how much realistic his assumption would be

Thanks

Gio1

 

[btrueblood]

Are you also modelling a distributed friction due to contact with the steel rod? I.e. as the tube buckles, points of contact induce friction loads; the effect should dissipate the load along the tube, making it more likely to buckle near the moving impactor, less likely to buckle towards the stationary b.c.