Loose specimen and hard_particles

Hello everyone!

My objective is to simulate a Rectangular Drained Triaxial test of particles of radius between 4 and 8mm. But I am have problems simulating loose soils. The test consist of a fixed based, four pistons (x and y directions) applying a constant stress and a vertical piston (z) moving at a constant speed.
The particles are being inserted with "insert/rate/region" with different values for the friction coefficient and rolling friction coefficient. I start with values of 0.1 and go up to 1.1, generating 5 different specimens. I obtain samples with varying number of grains between (from 19k to 17.2k).

The problem is that even with the loosest specimen the strain obtained shows first a contractative than dilatative behavior. Here are some of he parameters for the execution part :

#material
variable Young equal 1e8 #Young modulus
variable V equal 0.25 #Possoin coef
variable u equal 0.57 #friction coefficient
variable e equal 0.5 #restitution coefficient
variable rf equal 0.3 #rolling friction coefficient
#timestep
variable dt equal 2e-5 #timestep value in seconds
#piston
variable CpSpeedV equal 0.02 #speed of vertical pistons in Compression part
variable KP equal 1 #value to control piston advancing speed on manual mode
variable KD equal 1e-6

#Material and interaction properties
#1 for grains 2 for walls
fix m1 all property/global youngsModulus peratomtype ${Young} ${Young}
fix m2 all property/global poissonsRatio peratomtype ${V} ${V}
fix m3 all property/global coefficientRestitution peratomtypepair 2 ${e} 0.1 0.1 0.1
fix m4 all property/global coefficientFriction peratomtypepair 2 ${u} 0. 0. 0.
fix m5 all property/global coefficientRollingFriction peratomtypepair 2 ${rf} 0. 0. 0.

fix walls1 all wall/gran model hertz tangential history rolling_friction epsd2 mesh n_meshes 6 meshes mBse sPy1 sPy2 sPx1 sPx2 sPz

So I wanted to try higher values for the Young module (E) parameter to see if this was the problem. But whenever I apply the command "hard_particles yes" that allows E>1e9 the servos stop working properly. The only values I change (besides the amount of run steps) are the following :

hard_particles yes
variable Young equal 1e10 #Young modulus
variable dt equal 2e-6 #timestep value in seconds

The Servos no longer obey the parameters and appear unable to move backwards. I wanted to have a maximum stress of 100kPa on the X and Y directions, but I was getting up to 250kPa. Since the begging of the simulation they would move only “forward” in relation to their movement direction.

Does anyone know how "hard_particles" work?
Can anyone give me any other tips on how to run Loose soils?
Can anyone find any mistakes on my files?

Thank you very much for your time,

GONCALVES Joao,
PhD Student at INSA de Lyon
joao.goncalves-de-oliveira-chueire@insa-lyon.fr

** EDIT
It seems that when the servo Young Modulus is bellow 1e9 they work well. So I am able to launch hard_particles simulations by applying only the high E on the particles. But I still am not able to get a loose response.