Re: [Getfem-users] boundary conditions

[Yves Renard]

Dear Egor,

The difference between mdbrick_generalized_Dirichlet
and the add_generalized_Dirichlet_condition_* functions is that the first one applies
to the old brick system of getfem and the functions apply to the new one.
So, if you use the old brick system, you can use mdbrick_generalized_Dirichlet
but it is not still documented in the up to date documentation (a little bit in the
matlab interface).
Of course, it is recommended to prefer the new brick system.


Yves.


Le 09/09/2013 09:38, Egor Vtorushin a écrit :

Dear Yves

Could you please explain(provide a hint)

When they should use mdbrick_generalized_Dirichlet class and when 

one of 

add_generalized_Dirichlet_condition_with_multipliers(md, mim, varname,
                                         multname, region,
                                         dataname, Hname);


add_generalized_Dirichlet_condition_with_multipliers(md, mim, varname,
                                         mf_mult, region,
                                         dataname, Hname);

add_generalized_Dirichlet_condition_with_multipliers(md, mim, varname,
                                         degree, region,
                                         dataname, Hname);


add_generalized_Dirichlet_condition_with_penalization(md, mim, varname,
                                          penalization_coeff, region,
                                          dataname, Hname);

as it described in 

Generalized Dirichlet condition brick

part of documentation

What i'm trying to implement is to solve the problem below

The point here that i use strain-displacement pair instead of classic stress-displacement. That does yield some boundary conditions issue. In stress--displacements formulation we pose boundary conditions on surface vector of a stress tensor field  (g may equals to zero so so we call it stress free condition) that occurs naturally.

When we go to strain--displacement formulation we need to rewrite boundary conditions for surface vector of a strain tensor . It maybe a slightly difficult to account this conditions since we can't just

fix components of tensor value by value since the expressions of pair of components turn to prescribed values. So it may look like Lagrange multipliers technics using rather than system dof reducing.

Dear Yves, could you please advise me which way i should go? "Bricks" or "Conditions?"

Egor

2013/6/17 Yves Renard <address@hidden>

Dear Egor,

Some mixed element are already implemented in Getfem++. An exemple of utilisation is in the file

contrib/mixed_elastostatics/mixed_elastostatic.cc

of the distribution. The first step to implement new elements is to define them at the reference element level. For intrinsic vectorial element, you can take example on the RT0 element which is defined in the file  getfem_fem.cc (line 1107). As this kind of element is not "affine equivalent" you have to use the post-transformation matrix (P1_RT0_::mat_trans method for the RT0) which allows a supplementary linear transformation which is real element dependant. See the documentation at the page:
http://download.gna.org/getfem/html/homepage/project/femdesc.html#finite-element-methods-description


Yves.


Le 13/06/2013 12:56, Egor Vtorushin a écrit :

Dear Yves 

I've used Getfem++ some years ago and since that moment i'm thinking that this library does pretty much cover FEA needs.

Right know i'm solving some dynamic elasticity problem with redundant requirements of strain smoothness. The decision has been made to proceed with mixed elements solution when strains are from H_div space and velocities are from L2. It led me to strains approximations that looks like this(mostly i took it from  BECACHE, JOLY, TSOGKA 1999 year work)

(same pic attached)

It looks like something in the middle between RTQ1 and QK(2,1) i think. 

Could you please  provide me with a hint how to start implementing this. I've searched the mail-list but was not able to find good materials could help me

Thank you, 

Egor Vtorushin | Scientist

Baker Hughes | Novosibisk Technology Center | Int. Research for Drilling & Evaluations

_______________________________________________
Getfem-users mailing list
address@hidden
https://mail.gna.org/listinfo/getfem-users

-- 

  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

---------

_______________________________________________
Getfem-users mailing list
address@hidden
https://mail.gna.org/listinfo/getfem-users

-- 

  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

---------