Hi,
 

I hope all is well. 
 

I am working on a metal milling simulation using the Johnson-Cook model. In my first simulation I had issues with Abaqus’ visualization module and the reaction forces in the z and y directions were quite accurate, while the forces in x were substantially higher than my experimental measurements. In my new simulation, the forces in x and y are accurate, but the forces in the z direction are considerably low. The new simulation shows a higher predictive power than the old one in terms of residual stresses when compared to my experimental data.
 

Here are the plots that show the difference between simulations:

The difference between the old and the new simulation is the mesh in the transition zone (from finer mesh to coarser mesh) and some additional refinement in the lower layer. In the first simulation I used linear tetrahedral elements (C3D4) and the material top layer was just one part. In the second simulation I wanted to use linear hexahedral elements (C3D8R) so I created multiple parts joined by tie constraints. 
 

These are the workpieces:

I have tried multiple approaches to try to increase the forces in the z direction in the new model without success. Some of the approaches were:
- Modified the old simulation to use hexahedral elements in the transition zone. This caused very high stresses far from the cut, so I discarded it. However, the forces in the z direction matched the experimental values.
- Changed the boundary conditions on the sides of the plate, holding both surfaces completely. This yielded no change in forces.
- Used fully integrated elements everywhere except the cutting zone, hoping to capture the additional stiffness. This did not cause meaningful changes in the reaction forces.
- Adjusted the size of the plate, but saw no substantial differences.
- Increased the displacement at failure from 0.05 to 0.07, which raised the z-direction forces (though still low), but also wrongly increased y-direction forces and residual stresses.
- Tried a hybrid mesh with tetrahedral elements on the sides of the transition zone and hexahedral elements in the rest of the volume, which still resulted in low z-direction forces.

Unfortunately, I haven't been able to solve my problem, so I am reaching out to ask you for help. Do you have any suggestions on how I could modify the simulation to achieve higher z-direction forces in this new setup?

Thank you very much.