Newton solver convergence for differential materials Young's modulus E

Hi all,

I have a question about newton solver convergence problem. I am simulation a domain with two different material properties. The first part is soft material, and Young’s modulus E_s is 3MPa, the left is hard area, where E_h is 800MPa, when I run it in dolfinx, the algorithm doesn’t converge.

But when I set the hard material Young’s modulus E_h = 80MPa(1 order low than the original parameter), and soft area E_s = 3MPa (doesn’t change). Then the model converges.

Could someone tell me why and how I can fix the problem?

You may want to try with a continuation method. Solve the nonlinear problem for (Eh = 80, Es = 3), and then use the converged solution for solving the nonlinear problem for (Eh = 85, Es = 3). Go on like this increasing Eh by 5 (or whatever number you may want to try) step by step until you get to (Eh = 800, Es = 3).


Additional to continuation, see Default absolute tolerance and relative tolerance - #4 by nate.

1 Like