Hi,
I have been trying to do deep indentation using the demo_hyperelasticity.py for a generalized neo-hookean constitutive model. Specifically, I am applying a Gaussian traction distribution so that I can approximate a point load using a smooth function which is given as
t = -A*exp(-pow( ( x[0] / l),2 ) ),
where l is chosen to be 0.1 times length of the block. I compress the material by increasing A. I want to achieve a maximum displacement or an indentation depth of 10l but I am able to go achieve an indentation depth of only 3.2*l. I am using the method nonlinearvariational problem specifically,
# Non linear variantional problem
problem_u_nl = NonlinearVariationalProblem(F, u, bcs, J)
# Non-linear solvers
solver_u = NonlinearVariationalSolver(problem_u_nl)
# Solver parameters
prm = solver_u.parameters
prm['newton_solver']['linear_solver'] = 'cg'
prm['newton_solver']['preconditioner'] = 'hypre_amg'
prm['newton_solver']['absolute_tolerance'] = 1E-5
prm['newton_solver']['relative_tolerance'] = 1E-5
prm['newton_solver']['maximum_iterations'] = 500
prm['newton_solver']['relaxation_parameter'] = 1.0
I am applying the traction in multiple steps on a fine 3D square domain. Typically for large deformation problems, updated Lagrangian formulation is performed but i am not sure if I need to an equivalent formulation in Fenics. I would appreciate any advice or pointers on this regard.
Thanks, Kaushik.
