Hi everyone!

I am working on ABAQUS model which consists of a reinforced concrete wall and has a cantilevered W-shaped steel beam embedded (see attached picture). In order to make it computationally efficient, I am modeling just half of the full-size specimen. The steel beam is subjected to a cyclic shear loading whereas the wall has its base fixed. I have built the model and am trying to match the output with the experimental results. The stiffness of the connection is more or less matching, but the shape of hysteresis loop is quite different. The connection is failing at the same loading stage as that observed in the experiment. In summary, every other thing looks good but just hysteresis loops are different. When I looked more in detail, I found that results from loading and unloading branches are crossing for some cycles. (Plot for one of the cycles is attached herewith.) I don't know what the reason for this might be. CDP is used for concrete and combined hardening is used for steel for defining plasticity.

I am defining cohesive interaction between steel beam and concrete. (Parameters for cohesive interaction that I am using are: Knn=0, Kss=Kst=32 kips/in2, nominal stress along normal direction=0, along shear1 and shear 2= 0.42 ksi, plastic displacement = 0.1 in.).

Can you please suggest me what may be the reason and what parameters I should focus on to fix these issues?

Thank you in advance.