Convergence problems (Cohesive surface & XFEM)

[StevenJS]

Hi all,

I'm currently working on a model where i want to model cracking of a fibre-epoxy interface.
I did model the fibre-epoxy interface with a cohesive surface.
Ater initial debonding between the epoxy and fibre, the cracking is expected to continue in the epoxy material.
The cracking of the epoxy is made possible with a XFEM-crack.

The situation is as in the following image (where "vezel" = fibre).

However the model has lot's of problems with the convergence.
i do use the following stabilization parameter:
- Damage stabilization (10^-5)
- Maxs tolerance (0.05)
- Interaction damage viscosity coefficient
- Automatic stabilization (0.0002 ; 0.05)

As for the interaction damage viscosity coefficient, when i use a realy high value, say 0.01 the simulation is able to be solved.
However cracking occurs not at the interface and completely in the epoxy (thus this gives inaccurate results).
Therefor i do not want to use a high interaction damage viscosity coefficient.

Does anyone has any suggestions how to solve this model, without getting these big inaccurate results
Any suggestions are welcome!

I added the .inp file of the model in the attachement

Thank you all in advance!
Steven