I am actually working on the stress output in FEMaster right now. I got the shells working in linear and nonlinear cases. Technically i can evaluate at any point but for now ill just output at the top and bottom as for non-laminates you expect the highest stress to be there.
3 Likes
Regarding the shells, I managed to get a pretty good version going for shells. I took the existing MITC4FRT shell by Rao (thats where I got that from), and got rid of the projection onto a reference plane to allow curved shells. I also extended it to shells with 3, 6 and 8 nodes. So the difference is:
- Introduced MITC3FRT, (MITC4FRT), MITC6FRT, MITC8FRT
- Introduced properly curved shells with non-constant normals / directors.
- All shells should work with the abd shell sections and the normal shell sections.
- Normal shell sections integrate through the thickness, so nonlinear material models should work here in the future beside Neo-Hooke.
- You can rotate shell sections with the given coordinate system and choose the axis which is projected onto the plane.
I also found a few big speedups, especially in multithreading context. If you want, you can try it out here: Release FEMaster 2.6.0 · Luecx/FEMaster · GitHub
Ill be working on contact next I guess although I already see the frustration coming.
5 Likes
I’m very interested in non-linearities with shell elements (for the models I use, it’s unthinkable to use solid elements)… I find this to be Calculix’s biggest limitation at the moment… If the arc-length method could be integrated, that would be really cool, but I suppose I’m asking for a bit too much 
..
Anyway, brilliant work with the solver. I can’t wait to try it out in Prepomax
Arc-Length IS integrated 
I ran nonlinear stability analysis with the shells and it worked pretty good. Sometimes it may do weird things, its not yet production ready but technically working i think. Although with arc-length I had a problem with prescribed displacements. Like saying that the displacement is like 10mm, this only works with load-control at the moment… gotta fix that soon.
3 Likes
hi,
interesting solver as alternate since its new and use modern cpp language. Maybe FEMaster can share code with similar project of SuanPan for more feature of nonlinear material and dynamic analysis. Beam elements have advanced ones also, i had small conversation with the author in their forum.
best,
I’ve also heard that Kratos has probably the best and most modern architecture, but it would be hard to integrate it, so some tool to convert from .inp format may suffice.
not to integrated, i mean sharing code between each other solver may possible since it uses the same programing language. Like advanced beam element i’ve mention and discussed is much influenced from openSees code.
meshio can help with that: meshio · PyPI
I found difficult to start with kratos but looks very promising. Would be nice to have Prepomax supporting it as well. Kratos is supported by the commercial (but inexpensive) GiD pre/post by CIMNE. Still under significant development but I consider it one of the best open source option now available.
I have noted that there are several solvers that use a similar sintaxis than Abaqus and run also in Windows and from what I understand they use as postprocessor mostly Paraview.
Maybe then Prepomax could setup some facilities to use them, such as the new feature that allow to edit or supress internal CCX cards is great, because then can be used to prepare the mesh and groups of nodes/elements/faces, and the custom cards writed directly in the actual input deck editor (same as I did for FEMaster).
Can Prepomax open Paraview results files? Maybe that is the next step for allowing integrating new solvers, the same as implementing a custom solver tab in the Settings windows (not for FEMaster but for any compatible solver) with a way to set up the job parameters that those new solver would need (for example FEMaster need a custom parameter in the command line to setup de number of cpus)
1 Like
PrePoMax has no support for ParaView files. Result files generated by CalculiX (.frd) can be converted to ParaView format. But the thing about all those open-source solvers is that postprocessing is not an issue for most of them, as they typically rely on ParaView for that. The most difficult part is preparing the input deck since they usually have no GUI for that, and Gmsh, Salome, or other general tools have to be used.
CalculiX is still likely the best choice for most structural analyses, but indeed there are cases where it could be supported by other solvers for specific tasks. That’s what several other preprocessors (including FreeCAD FEM) do.
That´s the point, as you say the most difficult part is preparing the input deck (meshing, creating groups and defining bc), Prepomax could be used if they accept the mesh and groups in Abaqus/CCX style as FEMaster, but also Paraview is a pain to use compared with Prepomax, that´s why I ask if that reader could be implemented. I know that there are several flavours of Paraview results, but if the main ones are implemented in Prepomax then the developers of those solvers could modify it to write results in a more conventional format knowing that they will get instant support from a well stablished prepocessor tools and comunity as Prepomax. I´m pretty sure that Prepomax is the best and more used open source prepocessor available in the Windows comunity these days.
ParaView is a really powerful general-purpose tool for the visualization of even huge scientific datasets from various sources, but it lacks FEM-specific features such as transformation to local coordinates. FreeCAD FEM relies on it for most of its postprocessing features. The concept of pipelines and filters has some significant advantages, but also isn’t very user-friendly.
1 Like
Yeah, it has all, but you need to setup a lot of filters to get something basic that now is very well implemented in Prepomax. Separating components in Paraview is a pain if the solver don´t make that separation, creating a simple x/y plot is the same, can be done but with lot of efforts that every time you want to do you need to setup those filters again and again (well, it can saved as a template and opened with different results… but rarely works).
And not to mention that the user must install and learn a third program (pre, solver and post) for making the job.
Nowadays, I mostly use ParaView for some more advanced operations on the results (such as integrating variables or getting various statistics / operating on raw data) and to generate symbol plots. Other things are already available in PrePoMax. But ParaView is also suitable for CFD postprocessing (e.g. when using OpenFOAM together with PrePoMax). It has some nice customization of rendering aspects too.
1 Like
I think a tool - a separate executable - could be used to import Paraview results and export .frd results. In this way, it could be added not only to PrePoMax but also used independently.
1 Like
I get vtk imported into one version of ppm. But ppm doesn’t support plugin, too hard to maintain the importer.
2 Likes
Do you have that version of Prepomax available to test?