Notched tensile specimens and force-displacement response.

 Dear forum members,

I'm trying to simulate a tensile test (of ductile material) using smooth and notched specimens. I've started with a simple model without damage (isotropic von Mises plasticity implemented in ABAQUS) to compare the force vs. displacement response between simulations and experiments.

In the first step (smooth specimen) everything went smoothly. I used data from the experiment and I've got perfect agreement between force-displacement curves. Then, keeping that material data, subsequently notched geometries were defined. As a result, the value of displacement was correct but the reaction force was lower for output history in each case. Approximately 10 to 20% in comparison to the experiment (depending on the notch radius). The boundary conditions were attached like in the experiment (e.g. one side of the sample has boundary condition as YSYMM and other movement in Y direction), no restriction on the notched section. I've tried different ways but the results were always the same. It doesn't matter if it was Abaqus/Standard or Abaqus/Explicit. Axisymmetric and 3D model gave the same results. The mesh was refined in the necking point and in places where reaction forces act. Numerical simulations were stable (without unexpected fluctuations in calculated variables), etc..

So, the question arises in my mind. Is it possible that the higher stress triaxiality (or simply hydrostatic pressure) generated by the notch in the critical section may affect the results (i.e. reduce the reaction force acting under deformation)? I think it's unlikely. But, I'm running out of ideas...

I'll be grateful for any suggestions.

Kind regards,

Lukasz