Hi,
I am using VUMAT to write my own material model and simulate undrained triaxial compression tests in Abaqus/Explicit.
For dense samples, everything is perfect.
For loose samples, which will suffer static liquefaction, the single-element simulation is perfect; however, the multi-element simulation shows weird errors.
Then I plotted the strain increments of one element in these multi-element simulations, as shown in Fig. 1. Because I am using axisymmetric elements, there are only four strainInc components, At first, the strain increments show significant oscillations, and the simulated results are correct.
However, these oscillations keep decreasing. At about t = 186 s, these oscillations nearly disappear. Then, something weird happens, and the simulated results are getting wrong.
Fig. 2 is the enlarged view of Fig. 1.
Fig. 3 is the volumetric strain of the same element in Fig.1. Similar characteristics are observed.
My questions are:
- Is this volumetric locking?
Because this is the undrained simulation, the material can be regarded as nearly incompressive (the bulk stiffness of water is much larger than material's). At first, I thought this might be volumetric locking. But none of CAX4R, CAX3, and CAX6M elements can solve this problem (I think reduced integration should solve the volumetric locking issue). Since ABAQUS/Explicit doesn't have Hybrid elements, I cannot test if hybrid elements can solve this issue.
2. Why are these strain increments always oscillating? Why did these oscillations keep decreasing?
If this is not volumetric locking, I am thinking what is wrong with this multi-element simulation. I know this question is very case-based, which might be caused by my VUMAT subroutine (the tricky thing is that the single-element simulation didn't show this issue).
I would appreciate any suggestions and opinions very much!
Thanks a lot!