If I'm understanding your problem correctly, your model is 3D. Thus, you should select solid in Part Module, use solid elements to model the bulk of the fiber and the matrix, and cohesive COH3D8 elements for the cohesive layer.

Dear Luca,

Thank you for your reply. As you said my problem is a 3D one, this brings another question to my mind that since the fiber diameter is 7 um, Abaqus allows to pick a minimum thickness of 2.5 um which seems to be thick with respect to the fiber dimensions. Can this issue have an effect on the results? Moreover, in the bilinear CZM model, I have found max shear stress and G(j/m2) for the interface while in the property module, "Maxs damage" traction separation law , I need to define 3 nominal stresses. Could you please let me know which of them is related to the shear and how can I calculate the other two stresses.

Thank you,

Yes, the initial thickness has an effect on the result. If you use cohesive elements with finite thickness, you're modeling a thin layer of adhesive. If instead, you use zero thickness, you're using them just to model the decohesion and not the presence of a real material. I guess you're interested in the latter. Thus, you should check that nodes defining the two sides of the cohesive layers are distinct but coincident. Initial thickness will thus be zero.

As for the material properties, I think the best way is to look for analogous studies in the literature for similar materials. You can either use the same values and then compare your results with those of the reference paper, or at least get a feeling of the ratio of shear to transverse interface strength and thus make your own estimate.