Monopile Stability Analysis during Installation

Dear all,

I am in the process of doing an analysis of the stability during installation of an offshore monopile in PLAXIS 3D. For this analysis, I am using the HSsmall model for the sand layers. 
For this analysis, I have the following setup:

- A monopile of a total length of 70 meters embedded 10 meters into the ground.

- Watertable at the bottom of the model (- 50 m) to perform an effective stress analysis using the submerged weight of the soil

- A rigid body at 35 meters above ground level in the center of the pile to perform a translation in x-direction (Displacement condition)

The analysis has the following phases:

1) Insert the pile using a "Wish in place" method

2) Push the pile at the rigid body position to 1 meters in x-direction

3) Pull the pile  at the rigid body position to -1 meters in x-direction

4) Push the pile back to the original centre position

Below you can see an image of the set-up 

After performing the calculation, a horizontal force-displacement plot is made. The result indicates a "hardening" response of the soil when pushing the pile back to the centre position during phase 4 (More force required to push the pile back during phase 4 than phase 2). This is shown in the figure below. To me, this seems unrealistic. I would expect that the soil becomes disturbed and pushing the pile back would require less or the same amount of force.

To further analyse the soil response, I inserted an extra phase of "push-pull-push" to the pile. Unfortunately, the model fails at the start of the second cycle of loading.

My question is, how can I avoid this hardening response of the soil and get a more realistic response? Is this perhaps a limitation of the HSsmall model? 

Kind regards