Triaxial test for loose sand

Hi,

I am trying to understand CD triaxial testing for loose sand samples. I have a target porosity (before consolidation) of 0.47 and a target confining pressure of 25 kPa. To obtained this, I prepared a sample using REM and by introducing rolling resistance plus adhesion (later turned off) and damping during sample preparation along with a higher contact friction angle using triaxialStressController. The damping is then reduced to almost negligible (0.05) during shearing.

When I shear this loose sample, it undergoes significant volumetric contraction of almost 10% up to 2% axial strain, before any increase in deviatoric stress. After about 2% axial strain, the volumetric strain changes slow down, and deviatoric stress starts to rise.

This brings me to a few questions about porosity and confining stress in DEM triaxial modeling:

Target porosity for preparation:

Should I prepare the DEM sample at the initial porosity measured in the lab before consolidation, or at the porosity after consolidation?

If I prepare at the initial porosity (0.47) under the target confining pressure of 25 kPa, the sample will undergo substantial volumetric contraction before mobilizing strength, since damping is reduced to 0.05 during shearing.

On the other hand, if I prepare at the post-consolidation porosity measured in the lab, large initial volume changes will be avoided, and the deviatoric stress will start increasing earlier with axial strain.

Alternative approach:

Would it be better to prepare the sample at a low confining pressure (e.g., 5 kPa) with the initial porosity of 0.47, then isotropically consolidate it to 25 kPa before shearing?

This would essentially create a two-stage process: initial loose preparation, followed by consolidation, then shearing.