I am modeling one-fourth of a tubular lap joint utilizing ABAQUS Explicit. The outer tube is metal, and the inner tube is CFRP. The outer tube is connected to the inner tube by an adhesive (cohesive element). I used cohesive surfaces between each layer (5 total) of the CFRP to observe delamination. Since my model is one-fourth of a tube, I have used x-symmetry and y-symmetry constraints. The end of the CFRP tube is fixed while the end of the metal tube is displaced (see picture).

When I run my simulation, the highest Von Mises stress appears on the bottom row of elements (see pictures) in the CFRP and does not propagate to other elements. Because of this, the failure occurs in that bottom row of elements, which is not correct. The stress should be uniform around the tube because it is symmetric, with the high-stress intensity region located around the edges of the overlap region. When I reduce my mesh size to something much more coarse, the problem is solved, but then I receive excessive distortion in my elements.

I ran this same model in Abaqus Standard and did not have such a problem, but the model was very unstable and did not converge, which lead me to try Abaqus Explicit due to mass scaling.

I believe the problem is due to my meshing, but I am not sure what would cause the stress distribution to propagate in a single row of elements only. I have utilized sweep meshing and specified the sweep path to be through the thickness of the CFRP.

Any suggestions on what could be causing the inaccuracies within my model's stress distributions?