Elastic-plastic material in a static torsion test

[Nlgon]

Hi Community,
I'd like to discuss with you all an issue I've been facing for a week. I have some data, taken in static torsion tests on tubes. These tubes
- are shaped with cold forming (they are drawn): there have lobes.
- are telescopic: one goes into the other

Before to test the two tubes couples, I simulating the single tubes mainly because the tests are performed on single tubes. My idea is to simulate the torsion test in a static step, by fixing (encastre) one end and by imposing a concentrated moment on the other. I impose the load on a RP, contrained via Coupling to the end face.
I'm using Hex element, Element library Explicit and linear order.

I really need to catch the nonlinearity of the phenomenon, so I must impose at least a bilinear material model. I have not particular information about the material. Of course I know what material is, but the only info I have on a sample taken from the tube is the Rm (engineering value).

For yield stress and plastic strain I have no reference. I tried to iterate in order to match the experimental data (a sort of reverse engineering), but still I have problems. Also accepting the error coming from the simplification of the bilinear model I find always curves that differ from experimental ones. To get closer to the experimental curves I must input in Abaqus non-realistic values for material true Rm (>2000). The problem I have is that the Torque-angle curve gets flat (complete yielding of the section) before the real case. If I increase the yield stress then I lose accuracy on the transition out from linearity (elastic limit). The only solution for matching seems to increse the Tangent modulus (i.e. the true Rm by keeping the same true plastic strain at Rm).

Can you please give some suggestions about this problem? Am I mistaking in simulation settings? Are there some better methods to get the right definition of an unknown material? Is it possible that cold forming process on tubes gave to the component some different properties?

Thank you in advance for your help and attention

Regards,
Nlgon

PS I hope I respected the posting policy.

 

[FEA way]

So you can't perform regular tensile testing on this material ? In such case it might be difficult to match results from simulation and experiment but start from making sure that all settings are correct - double check units, loads and interactions. Also take a closer look at the mesh and consider refining it as well as using different type of element (there are many element types in Abaqus, suitable for various problems).

 

[Nlgon]

Thank you FEA way for the rapid answer. I think I'm fine with units and checks on the simulation. I consider that fact that I always get results compatible with the experiments; besides, for elastic field I find almost perfect match. Problems start when passing in plasticity.

Do you think it's crazy to increase the tangent modulus to get a good match? I would not do it because it's like blind manipulation of parameters, ignoring the physics behind. On the other hand my goal is now to well reproduce the behaviour of the tubes to build a reliable model of the couple. Is this enough to manipulate the tangent modulus?

Thank you again
Nlgon

 

[FEA way]

It might be a good idea to adjust plasticity model by changing its parameters until you get results similar to experiment. In fact, it seems like the only option in this case (when you don’t have material plasticity data).