3D Cyclic loading on unit volume element

[coolFL]

Hello Everyone,

I am doing one very simple 3D simulation in abaqus. It includes unit volume element which is loaded with three normal stress components along three coordinate axes and one shear stress component (in XY plane). The stress amplitudes vary periodically as function of time over the step. I have defined smooth variation of stresses over step time. I am trying to simulate one complete cycle of stresses. But somehow my simulation is not converging after step time of 0.4. I have attached inp file of my model (in zip format). Can somebody tell me where is the issue with my model? Why solution is not converging after step time of 0.4? Any input will be appreciated,
Please note: units for Force is Newton and distances are defined in microns.
Thank you in advance,

Nik

 

[Mustaine3]

I see only loads and no boundary conditions. Do you expect that all loads are self-balancing at all times in a numerical analysis?

 

[coolFL]

Hey,
Thanks for your reply!
Yes all the loads are self balancing during entire analysis. In the model, I have applied loads as stresses/pressure. Same pressure is applied on the opposite faces. meaning S11 of same magnitude is applied on the faces perpendicular to X axis. Similarly, S22 is applied on the faces perpendicular to y axis. Also, S33 is applied on faces perpendicular to z axis. Shear stress is applied as traction on planes perpendicular to X and Y axes. Therefore, tension and compression is applied by the same amount on all the opposite faces. Therefore, I expect external forces/moments to remain in equilibrium all the time.

 

[Mustaine3]

Think more about the last part of my question: "[...] in a numerical analysis?"

 

[coolFL]

Hey,
Since the solution is not converging, I guess forces are not in balance at elemental/nodal level. But at global level I have applied equal and opposite forces in all directions, therefore I thought my solution should converge. Ideally there are no boundary conditions to this volume element because it is a part of larger solid. I just want to see effect of cylindrical carbide particle on the surrounding stress field at local scale. Therefore, I have used unit volume element approach commonly discussed in mechanics classes for stress equilibrium illustrations. Do you think this approach is wrong? or any other modification is required in the model?
I have also attached inp file of another model which has slight correction over the previous one. In this model I have used Direct cyclic step rather than Static general step which I used earlier. For this new model solution is converging but no load is applied. I am getting stress field of zero everywhere. Don't know why? Can you see any errors/issues in this new model?

Thank you in advance for ur interest in this analysis,

Nik

 

[Mustaine3]

Issue 1: You assume that load 1 is exactly opposite to load 2, so 1 vs. -1. That might be the case in theory, but at the end of a numerical solution process the ratio is probably 1.002 vs. -0.999.
You can try to use stabilization or maybe inertia relief to get artificial forces that prevent the rigid body motion.

Issue 2: I'm not much familiar wit direct cyclic. Check the .msg if you have convergence. Also make a simple test and don't define any output request. Then the default should be used.