the model I'm working on is loaded in compression (buckling is NOT critical though,I checked it). When I plot the Min Principal Stresses of my linear run (aluminum), I see a wide region below the -(Tension Yield Stress).
I don't know if I can consider that my model is getting plastic or not. Some references I've seen argue that the compression stress-strain curve is the symetric of the tension curve, some other argue that the part in compression can go way further than the tension yield stress.
Should I run a non-linear material analysis with the tension stress-strain curve? Or can I just consider that my part is fine and safe and won't go plastic?
Solved! Go to Solution.
If I understood the email you run linear analysis and you see large region exceeding the yield stress. If the large stresses are coming from:
If the stresses are “real” and the whole section is exceeding the yield probably the structure is close to its limit and you will have convergence problems. Plotting the curve load-displacement you can see the actual buckling load limit
I hope it is helpful .
If you used the simplest setting for SOL106, plasticity will only occur if the Von Mises stress exceeds the specified yield stress. The most common situation where Min Principal stress could be "very large", without plasticity occurring, is in high hydrosatitc loads. For example, an item placed at the bottom of the deepest ocean is subject to full hydrostatic stresses. The Min Principal stress (and Max Principal and Intermediate Principal) could be, say, -100MPa, whilst the Von Mises stress will be zero. The same can happen in tension as well. So if your stress state is quite "hydrostatic", then the Von Mises stress will be much less than the worst (absolute) principal stresses.
So, if you had high negative min principal stresses (more in absolute value than the yield), then you are probably getting no plastic strain becasue the VM stress is less than yield. You would need to use an alternate yield criteria, and you may observe a difference. I think there are a few yield criteria which separate the hydrostatic aspects of a stress state.
The von mises criteria takes plasticity as a number . It does not depend on compression or tensile. So the curve defintion is enough with the tensile part. At least in other solvers.
If you are using sol106 it seems to be that you are not reaching von mises stress. As EndZ explained it.
Maybe you should provide some images of tension (von mises and min) to see the problem
Thank you EndZ,
I knew about the hydrostatic component and how the VM criterion doesn't fit with this stress state. I would have zero VM stress is my part was, for example, a cylinder completely filled with material. In the case of a hollow cylinder (my case), the walls are bending under pressure, and VM stress appear.
I'm done some reserach on alternative Yield Criteria, there are a few (Tresca, Mohr-Coulomb), but all based on shear (which makes sense, metals are failing by plane sliding).
Bottom line, I think it's ok to consider VM stresses as a criterion in my case, and not consider Min principal Stress.
Analyzing the information:
Questions about the model
Ok. I know this problem. No time no powerful machine... Meshing with hexaedral takes time.
From my experience pure bending needs at least 3 elements through thickness. 4 better of course.
Take into account that pure bending the maximum stress is at the outer surfaces. And the gauss points do not locete at these locations
Great if the posts were helpful