It looks like the ANSYS model has excellent agreement into the initial post-yield response.  It is only well past yield that it doesn't track the experimental data.

In the experimental samples, is significant necking occurring?

What does the sample look like?

What kind of geometry flaws exist in the shape of the sample?

How is the sample held in the tensile testing machine?

How is the deformation measured?

What does the ANSYS model look like?

Did the ANSYS model have the same kind of geometry flaws that exist in the sample?

How is the model supported and loaded?

What does the mesh look like? 

Were the results shown above plotted after a mesh convergence study?

If the sample exhibits necking, does the ANSYS model?