How to get good curve fitting for finding of chaboche parameters

Dear all members,

I have tried to find Chaboche parameters using curve fitting tool in ansys workbrench as shown here. Unfortunately, bad curve fitting is resulted here. What is the problem here? If someone who has experience please help me? Thank you.




  • This involves a bit of trial and error but I'll provide some tips:

    From your screen shot I see that the units of stress are in Pa so the numbers are pretty high. In curve-fitting we typically work with least-square function where the number is squared. As a result, the number may be too high and might run into truncation issues. I recommend changing the working units to (tonne, mm, ...) so the units are in MPa to avoid this issue.

    A few other tips include:

    Increasing the number of kinematic models increases the nonlinearity and can help in finding a better fit.

    Switch between absolute and normalized errors and see if it improve the quality of fit.

    You can provide an initial guess (or seed value) in the top left table for each parameter and see if it helps.

    Also, go through the notes provide in this link from documentation.

    Although it's for APDL interface, any tips on selecting initial seeds provided here are relevant for curve-fitting in Engineering Data too.

Sign In or Register to comment.