Three point bending

Phillip

In 3d static with loads and restraints being applied to a three point bending meshed model the results are as obtained experimentally and as standards imply. Now if I try to model contact between a cylinder (2 mm diameter and 5 mm in length) and the rectangular 2x 2 x25mm specimen I can not seem to find a solution that approaches the 3d solution above.
I have tried dynamic nonlinear 3d, nonlinear static 3d and any other button I could press. I am hopeful of not having to edit the calculiX input files. (For I will struggle). If that is what I have to do then will do so.

At present the cylinder which is applying the load blows up like fireworks and displaces the nodes all over the place mostly sub bar. It is very pretty. I have tried bonded contacts, just contacts and even frictionless surfaces. The loading cylinder seems to nearly always go through the three point bending bar even if I enclose the cylinder in a low modulus material. It goes through even if I place restraints on the cylinder.

It just does not seem to recognise the contact between the bar and the loading cylinder (or loading bar if I place the cylinder inside a bar) ((even when it is enclosed in a soft contact moduli material or the same material and with bonded interfaces (bonding contact does work in other places in the model just not this mid bar loading contact and the supports for that matter contacts)). I have increased the contact faces by making element sizes smaller, I have reversed master and slave contacts.
I have played around with parameters but seem to always either get no solution (did not converge) or the fireworks.
I am not an engineer so I suspect it may be me that is the problem. However I only have me to work with. Any help as to at least what buttons I should press to have a chance of a solution. Would be appreciated.

thank you Phillip

Victor

3-point bending with contact is a bit difficult because the bar and the loading cylinder can fly away if they're not constrained. But you can do it with both the free objects having all 6 of their degrees of freedom each constrained by soft springs and constraints.

Other than that, maybe the contact stiffness (slope or pressure-overclosure curve) is too low (penetration) or too high (no convergence, random displacement field in the non-converged solution)?

Could you post the .liml file?

Phillip

Thanks Victor, I have tried various contact stiffness but high or low did not seem to make much difference and in the middle was not great either.
Attached is the mecway files.
The first is the last one in which I have surrounded the cylinder in Stainless steel so bonding could occur and to see if I could achieve contact recognition. The supports are not in operation as yet so there is support from points.
The first was the basic 3d and it produced the deflection as expected.
Thanks
Phillip C

Victor

The main problem is that the two SSteel materials have Youngs modulus = 200 Pa. I changed them to 200 GPa and it converges.

I also turned off quasi-static since there aren't any time dependent loads to make use of it.

Phillip

thanks