First, I believe there’s a typo in your formulation, since, in the line Wint = ..., it would be more standard for uz to be the Function and w to be the TestFunction, and for the nonlinear solve to have uz as its unknown solution. Second, if w is interpreted as a displacement, grad(w)[1,1] is not the normal stress in the y-direction, it is the extensional strain. (In general, strain would be the symmetrized gradient of displacement, but symmetrization doesn’t change diagonal components.) You do not expect strain to be continuous. The stress tensor would be Landa1*sigma(uz) or Landa2*sigma(uz), depending on which subdomain you’re in (assuming the code is changed so that uz is the displacement, rather than the test function).
1 Like