Re: [Getfem-users] Adding an initialized fem data in a hyperelastic problem

[David Danan]

Dear Kostas,

as promised, please find enclosed the result of the original simulation without the active contraction term.

Also, note that in the end (just to be sure), i had used the routine provided by the benchmark to generate the fiber data (even in my first email), even though i had obtained the same data on my own.

Thanks in advance, once again,

David.

2017-10-06 23:44 GMT+02:00 David Danan <address@hidden>:

Dear Yves, Dear Kostas

thanks for the simplification, i was looking for this notation.

Sure, i had to decrease the internal pressure (15kPa to 4 kPa) to reach convergence in both cases with the classical newton solver (and i fear the continuation method might take some time, even if it converges) so, meanwhile if it helps, please find enclosed the solution obtained with the contraction term decreased by the same amount (60kPa to 16kPa) (Fibre_f4Ta16.vtk) and without it (Fibre_f4Ta0.vtk).

Find also enclosed the simulation with the contraction term (Fibre_f4Ta60.vtk) where the active constrained is the same as in my first email, just the internal pressure was decreased.

Thanks in advance,

best regards,

David.

2017-10-06 18:05 GMT+02:00 Konstantinos Poulios <address@hidden>:

Dear David,

Could you also send us a result of your simulation without the active contraction term, only with internal pressure? This is to estimate how large is the contribution of T_a compared to the contribution of the applied pressure p.

Best regards

Kostas

On Fri, Oct 6, 2017 at 3:07 PM, Yves Renard <address@hidden> wrote:

Dear David,

I do not see any mistake. You can of course simplify the _expression_ in

"T_a*((Id(3)+Grad_u)*(address@hidden)):Grad_Test_u"

which should be faster.

Yves.

Le 04/10/2017 à 18:04, David Danan a écrit :

Dear Getfem users,

i would like to take into account the fiber direction in my model;
to do so i computed the vector associated to each point of the mesh in the order given by mf.point_of_basic_dof, stored it in a vector, used an initialized fem data

  model.add_initialized_fem_data("fibre",mf_u,Param_fibre);

and included it in the model (basically, add the component T_a*fibre*fibre^T to the second Piola-Kirchhoff tensor in the reference configuration)
      getfem::add_nonlinear_generic_assembly_brick(model, mim,"T_a*(Id(3)+Grad_u)*[fibre(1)*fibre(1),fibre(1)*fibre(2),fibre(1)*fibre(3);fibre(2)*fibre(1),fibre(2)*fibre(2),fibre(2)*fibre(3);fibre(3)*fibre(1),fibre(3)*fibre(2),fibre(3)*fibre(3)]:Grad_Test_u");

(this is where i have some doubts)

The results i obtained are qualitatively correct but a visible difference still remains between the reference solution and my own solution.

It corresponds to the last problem described in this article
http://rspa.royalsocietypublishing.org/content/471/2184/20150641#sec-15

the first two were already validated, so such a difference should (a priori) come from the fiber i guess.

I have tried naively to interpolate the fiber data computed on a P3 mesh on the original P2 mesh but it didn't change anything.

Some help will be really appreciated, i may have overlooked something.

Thanks in advance,

David.

PS: Please, find enclosed the reference solution (deformed.vtk) and the solution obtained (Fibre.vtk)

-- 

  Yves Renard (address@hidden)       tel : (33) 04.72.43.87.08
  Pole de Mathematiques, INSA-Lyon             fax : (33) 04.72.43.85.29
  20, rue Albert Einstein
  69621 Villeurbanne Cedex, FRANCE
  http://math.univ-lyon1.fr/~renard

---------

Fibre_f15Ta0.vtk
Description: Binary data