Here is the link to your model that behaves differently to the one I uploaded. That archive included results so was 145 MB and beyond the size allowed for using the Attach button here. You can either post a Google Drive link such as I did in order to include the results, or you can Clear Generated Data on the Model or Mesh item, which will clear the solution also. That archive file size will be the smallest possible.
If you study my model carefully, you will see that my displacement BC is only on the bottom edge of the tool, which allows the rest of the tool body to be flexible.
Your model applies two displacement BCs to the entire tool body, which prevents the tool body from deforming in any way, so no strain develops and no elements fail. This is equivalent to making the tool body rigid, only a lot less computationally efficient. This explains why my model erodes on the tool body and your model doesn't. Is this what you want?
I don't understand why you are paying attention to the stress in the specimen. You are cutting it. That means you are taking the stress in the material to a point at which failure occurs. The current definition of failure is when strain > 1.5 and since stress = E times strain, and you provided the value of E as 2822 MPa, you can calculate the stress at failure as 4,233 MPa. So what? Why do you care?