1 (edited by juanjose.espadas 17-10-2016 11:48:32)

Topic: New user: PBCs, PCG solver, List of results, Coordinate system...

Hello OOFEM community,

I recently looked at OOFEM and it really looks a very interesting code. However, I'm wondering if currently is capable to do some stuff:

- Is it possible to select the method to impose periodic boundary conditions (e.g., Lagrange multipliers, static condensation?). Since with the first, a direct solver cannot be used, while with the second is possible. This leads me to the second question...

- For a linear elastic problem, is it possible to select between a direct sparse solver (parallel?) or in particular the iterative method: preconditioned conjugate gradient?

- Is it possible to export the results in text format: Displacement of nodes (e.g., as list), stresses in each gauss point (and location) in each element, in order to be able to post-process in e.g. matlab?.

- Is it currently to rotate automatically the material coordinate system (like fibre direction, for composite materials?) O you need to rotate the Cijkl (stress) tensor by your self and then input the rotated elastic constants?.

- Is it possible to use parallel computing under OpenMP scheme (shared memory) instead of MPI (Dist. memory?).

I am thinking about using OOFEM for numerical homogenisation of fibre composite materials.

Most likely if all of this is possible and you have a simple example it will be highly appreciated,

Thank you and looking forward to hear from you!.

Best,

/JJ.

Re: New user: PBCs, PCG solver, List of results, Coordinate system...

1. There exists several premade specialized boundary conditions for homogenization, and a manual b.c. where you can tie together DOFs using lagrange multipliers. Since you work with homogenization, I'm happy to tell you OOFEM has excellent support for this (full fledged FE^2 support with all the standard BCs + more).

2. There are many many solvers to pick from. Everything from PETSc, plut both direct and iterative built in solvers.

3. There is a matlab export module that does exactly this.

4. This is supported by all shell/beam elements by default. Several solid elements support this as well (in fact, it might actually be all continuum elements that support this).

5. We support OpenMP for assemble. For linear sparse solvers, you have the selection available in PETSc, though most of the direct solver require you do use MPI.

Re: New user: PBCs, PCG solver, List of results, Coordinate system...

Dear Mikael, thank you for your answer.

I've seen that people from Chalmers is using/contributing to OOFEM quite actively. Here at Uppsala, nobody uses it as far as I know (only Matlab), but would be nice to use it/contribute to it as well...

Will take a look on how to implement such things that I want.

Thanks again,
Best,

/JJ.





Mikael Öhman wrote:

1. There exists several premade specialized boundary conditions for homogenization, and a manual b.c. where you can tie together DOFs using lagrange multipliers. Since you work with homogenization, I'm happy to tell you OOFEM has excellent support for this (full fledged FE^2 support with all the standard BCs + more).

2. There are many many solvers to pick from. Everything from PETSc, plut both direct and iterative built in solvers.

3. There is a matlab export module that does exactly this.

4. This is supported by all shell/beam elements by default. Several solid elements support this as well (in fact, it might actually be all continuum elements that support this).

5. We support OpenMP for assemble. For linear sparse solvers, you have the selection available in PETSc, though most of the direct solver require you do use MPI.

Re: New user: PBCs, PCG solver, List of results, Coordinate system...

Hi JJ,

More developers is great. My old research group at Chalmers does/did a lot of work on computational homogenization. For this, we have developed the homogenization framework in OOFEM extensibly (strongly periodic, weakly periodic, neumann, dirichlet, and more coming). We also have a research group on composites, in particular on cohesize zone modeling, XFEM for delaminations and cracks, shell models.

I can also say openly, that I'm always interested in collaborating with research papers (in particular with development of new models/methods in OOFEM).

Re: New user: PBCs, PCG solver, List of results, Coordinate system...

Thank you for your answer Mikael, it sounds really nice and thank you for being open in that sense. I've seen that you have many interesting works!... It is nice to have you as a contact.

Best,

JJ.










Mikael Öhman wrote:

Hi JJ,

More developers is great. My old research group at Chalmers does/did a lot of work on computational homogenization. For this, we have developed the homogenization framework in OOFEM extensibly (strongly periodic, weakly periodic, neumann, dirichlet, and more coming). We also have a research group on composites, in particular on cohesize zone modeling, XFEM for delaminations and cracks, shell models.

I can also say openly, that I'm always interested in collaborating with research papers (in particular with development of new models/methods in OOFEM).

Re: New user: PBCs, PCG solver, List of results, Coordinate system...

Hi Mikael,

you said its possible to assign material rotations for composites. I've been trying to do so, but I get weird results. I have 18 layers and only if the rotations are 0 or 90 degrees I get stress values for every layer. If I assign 45 degrees rotation I get output only for some sof the layers. But if i put 45 degrees for the first layer, the software collapses. What is happending there?  I'm attaching my input file.

Thank you very much!

Post's attachments

oofem2.in 28.76 kb, 3 downloads since 2017-03-20 

You don't have the permssions to download the attachments of this post.

Re: New user: PBCs, PCG solver, List of results, Coordinate system...

Hi, I think there is a bug, I am working on it.

8 (edited by nitramkaroh 22-03-2017 12:01:13)

Re: New user: PBCs, PCG solver, List of results, Coordinate system...

I pushed the fix.
Can you, please, test it? It would be great if you prepare an example. Something simple, like a plate with two rotated layers, which can be added to basic tests.

Re: New user: PBCs, PCG solver, List of results, Coordinate system...

I haven't used the git  updates yet. I don't quiet understand, did you also compile a new executable file with the changes? I prepared a small test though.

Post's attachments

rotated_layers.in 964 b, file has never been downloaded. 

You don't have the permssions to download the attachments of this post.

Re: New user: PBCs, PCG solver, List of results, Coordinate system...

You can find details about git here http://www.oofem.org/wiki/doku.php?id=git-tutorial