criterion

Dear all,

my problem is dealing with a fatigue crack propagation analysis using XFEM (linear elastic case) with a prescribed initial imperfection in the modeled solid. What is the criterion to exit from such a kind of analysis?

Thank you in acvance!