Implimentation of SANISAND constitutive model in ABAQUS/Standard for a soil-monopile interaction

Dear all,

I am trying to model soil-monopile interaction (3D) in ABAQUS/Standard in order to perform subsequently a cyclic and dynamic analysis.

For the soil I am using the SANISAND umat constitutive model given herein by David Mašín and for the parameters I am using those given by Dafalias and Manzari (2004) for Toyoura sand.
For the monopile it is steel with an elastic constitutive model (E=210 Gpa and density=7850 kg/m3).
For the soil-monopile contact I am using the surface to surface contact pair formulation with small sliding that is found in ABAQUS/Standard.

In ABAQUS I am performing first of all a geostatic step in which the monopile and the (soil-monopile) contact pair are desactivated by using the the “*Model change , Remove” for the monopile and the “*Model change Type = CONTACT PAIR, remove” for the interaction.
Simultaneously , in the same step I am activating a soil part (having the same geometry as the monopile) having the same SANISAND material properties as the whole soil domain to replace the monopile.
A predefined initial geostatic stress field and an inital uniform void ratio field (= Parameter 19) are applied to the soil. The weight of the soil is also applied in this step in order to have equilibrium with the initial predefined stresses.
The geostatic step finishes succesfully in a short time and at the end of this step I am obtaining relativly very small displacements thus verifying its succesful implimentation.
The next step (Step 2), is a static general step in which the weight of the monopile is applied. In this step, the monopile and the soil-monopile interaction are reactivated using “*Model change, add” for the monopile and “*Model change, Type = CONTACT PAIR, add” for the interaction. Also, the soil part which has been added previously in the geostatic step to replace the monopile is removed in this step by using the “*Model change, remove”.
Unfortunatly, STEP 2 is giving me errors upon the second increment.
The error is an Overconstraint checks. meaning that there are multiple constraint that are applied directly or chained constraints that are applied indirectly at various nodes of the soil.
By asking ABAQUS to print the constraints in the message file, I was able to identify that these overconstraints are from one side due to the tie of the soil nodes with the removed soil nodes, and from the other side due to the tie of the soil nodes with those of the monopile.
It should be noted that the soil part that has been added in geostatic step and then removed in the static step is removed by using the Model change, remove.

I tried to check very well the different interactions, to verify that the problem is not a modelling problem. I am sure that is not like that.

It should be noted that the analysis run successfuly shen using the Mohr-Coulomb constitutive model founf in ABAQUS.

Does anyone have any idea, or have encountered the same problem when dealing with the SANISAND umat ?
Dear David, do you have any idea how the SANISAND umat is dealing with the model remove and add ? {because in ABAQUS documentation it is stated that elements with a user-defined material type can be removed and reactivated}

Can the problem be at the level of the input parameters for the SANISAND model ?

or Maybe at the level of the type of interaction ?

Thank you all in advance
Best regards,
Philip