Modelling cohesive zone model to capture debonding failure in steel-cfrp interface

Hello everyone,

 Basically I am using ABAQUS/Explicit to perform quasi-static analyses on steel members reinforced with cfrp strips. I have obtained the cohesive zone model parameters and I am trying to apply them to my model. I am assuming that failure will occur in the adhesive, as observed experimentally.

In terms of the model I have 3 parts: the steel part, the adhesive and the CFRP. The parts are connected using *TIE surface constraints and I have further defined contact between the CFRP and the steel, to be activated throughout the simulation, in order to prevent penetration of both elements.

The adhesive is modelled using cohesive elements COH3D8. They are assigned the following material rule and the section controls defined as follows:

 *MATERIAL, NAME=EPOXY
*DENSITY
1200e-9
*ELASTIC, TYPE=TRACTION
Knn, Kss, Ktt

*DAMAGE INITIATION, CRITERION=QUADS
sigma_nn, sigma_ss, sigma_tt

*DAMAGE EVOLUTION, TYPE=ENERGY, MIXED MODE BEHAVIOR=POWER LAW, POWER=1, SOFTENING=LINEAR
Gnn, Gss, Gtt

*SECTION CONTROLS, NAME=COHESIVE_FAILURE, ELEMENT DELETION=NO

I went ahead and tried to use my experimental data to obtain the different parameters. I am not too concerned with the values of the parameters, but I am observing penetration of the CFRP in the adhesive, even though I expected the cohesive element to be able to undergo compression. Moreover, if I don't define contact between the CFRP and the steel, unreal deformation spikes occur in the CFRP and equilibrium is not reached.

Since I can't really narrow down where my model is failing, I was hoping someone out there with some expertise on the subject could possibly give me some hints on what can be happening...

Thanks in advance to everyone.