Simulations of nonlinear materials should have an experimental input of the nonlinear material behaviour. In my project I did some 4-point-bending-tests and used the values of force/displacement resp. stress/strain as input values in the Ansys simulation. Without the experimental data of the softening behaviour there is only the possibility to use theoretic models, but these will never represent the reality exact.

High quality simulations need high quality input data.