Roman Putanowicz <[email protected]> a écrit : > On Tue, Jul 13, 2010 at 02:59:35PM +0200, Christian Fischer wrote: >> Dear All, >> >> I recently thought about using the following structural elements: >> >> * 1D: Euler-Bernoulli or Timoshenko beam >> * 2D: axially symmetric continuum elements >> * 1D: axially symmetric flat plate (like an axially symmetric beam) >> >> Before I start working on any of these elements. Maybe there is somebody who >> already implemented any of these. (Like the truss element in a >> recent email.) >> Or somebody has a few hints for me e.g. how to implement axial >> symmetry in general. >> >> Best wishes >> Christian Fischer >> > Dear Christian > > I am glad for your posting. I will gladly discuss implementation of the above > elements as any such discussion greatly enhances documentation of GetFEM > internals. > > Recently I have implemented 2D frame element. I have attached some source > code (the implementation of frame brick is not complete in the sense that > I do not account for element stiffness). It is sparse of comments but I can > serve any info. > > From my understanding of GetFEM (Yves can correct me for sure :) > implementation > of any of your above problems does not really require adding new elements > to GetFEM. Everything can be implemented on the assembly level.
Normally, yes. Except if one wants to make connections between the elements which links the degrees of freedom of different dimensions (traction compression and bending for instance) In that case, there is some possibilities to do this with the reduction matrix of the mesh_fem object. But may be the simpler should be to define a special brick dealing with such elements. Yves. > For beam elements there already Hermite element (in case we agree to skip > axsial displacement - if not it should be possible to handle it via separate > mesh_fem). Of course having true structural elements would be nice to. > > I have not implemented axisymmetric problem but If I recall correctly > in some post I have seen comment that functions like Lame coefficients > can be implemented via mesh_fem_global_function. So again it should suffice > to write right assembly routines. That is my five cents. > > Regards, > > Roman > -- > Roman Putanowicz, PhD < [email protected] > > Institute for Computational Civil Engng (L-5) > Dept. of Civil Engng, Cracow Univ. of Technology > www.l5.pk.edu.pl, tel. +48 12 628 2569, fax 2034 > _______________________________________________ Getfem-users mailing list [email protected] https://mail.gna.org/listinfo/getfem-users
