I’d love to hear your opinions on why/why not.
I’d love to hear your opinions on why/why not.
Many people (including large companies) do it but it depends on what you need. CalculiX has many features and very intuitive syntax (especially for those who know Abaqus) but it has some bugs and limitations (especially regarding shells and beams but also nonlinear buckling or explicit analyses).
That begs the question then: When would you NOT use it?
When some of the aforementioned (or other) limitations make it hard or impossible to use CalculiX for particular purpose.
I use as my main solver, but using Mecway as preprocessor, and in the very few times that I need beams elements or some complex shells I can use the internal Mecway solver. We build medium size gas compressors for oil ang gas industrie. I use it in the same way as freelance consultor.
Yes, and we do. Using Calculix coupled with Mecway has allowed us to put FEA on many desks, as commonplace as Excel vs. having a small set of licenses for an elite group. Some highly non-linear systems are probably more trouble in CCX than ANSYS, but not enough to alter our strategy.
What companies do you know of that use it?
Within the first paragraph on http://calculix.de/ there is a large company mentioned. And below that there is a list of professional users.
For me and may the others too, using two FE solver inside one GUI only raised to doubt and confusion.
even FE is a general purpose, still one of them are slightly different betwen each others. In solid element, small deformation and elastic material they both solver should give identical or exactly the same results. But it’s not similar in case of nonlinearity, did the load follower in large deformation? Are both solver have similar capabilities in nonlinear material type? etc
In beam element, even for classical formulation being used still there may have different workflow. An example of simple supported beam, common FE required the beam to be meshed but another is not. Only one element is done to get maximum moment and deflection at midspans. Some are account rigid zone offset at beam column intersection, another is not.
Mostly FE solver use classical shell element formulation for single layered or multiple composite. CalculiX use expanded approach and generates solid element during the solver runs.
It’s not easy to trace any discrepancy occurs when someone switch the FE solver. It seems better to be separated rather than mixed in a single GUI due to not all have similarity.
One benefit of the ccx shell expansion to solids is you can see composite failure in the zz direction. I’ve been seeing that as the failure mode a lot. Whereas, with traditional shell elements you can not get that output at all. So the traditional composite shells can look fine, when in fact, they may fail in the zz direction.
We primarily use the CCX solver, and we have a lot of history of validation with the work we do. We don’t make use of beam/shell elements very often. In the end, I think your choice has to be based on your own needs.
CalculiX is at best in working with solid element since all analysis type become supported. also, the reasons all common 1D and 2D element being expanded to solid for uniformity,
although classical beam and shell available trough user element but still limited.
i’m personally still interesting in CalculiX unique feature of expanded element due to it advances and challenging.
regarding confusions of mixed a solver inside one interfaces, maybe it’s a problems for general user also. for examples switching between solver CalculiX, Code_Aster and/or OOFEM in a single interfaces. did all the solver have similarity and capabilities? that’s what i mean, need to be separated not mixed ones.
I am using it in work for clients of my employer. Mainly in composites. Both sandwich constructions and bonded constructions. I tend to explicitly create solid elements, also for the composite materials. I haven’t been able to get tying shells to (solid) cores to work.
Since both the precprocessor and the solver process text files, it is relatively easy to use e.g. Python to automatically set up and process a host of parametric models to do design optimization.
One of my “helper” programs can automatically create orientations for composite material he20(r) elements based on their geometry. It makes use of the fact that lamina are generally much smaller in one direction (thickness) then the others.
I use ccx for model testing phases, to refine the model. I carry out linear calculations directly with ccx, for complex analyzes we rely on Abaqus.
I have been using Mecway and Calculix for non-linear analysis to help with development of standards for concrete bridgerails and adjacent bridge decks under quasi static extreme impact loads for a US state government. We don’t model the vehicle impact, using results of other research for the loads on the rail deck system. We also use crash tests, expert analysis by professional crash test centers, comparisons against other similar rails, and hand and spreadsheet calculations. The current code provisions (actually appendix with provisions as well as guidelines) are somewhat outdated and crude, dating to prior to 1974.
i’m working on construction industry and mainly in steel structure. CalculiX has been used to validate specific cases of analytical calculation where it’s limited by approximation and assumption. I can not imagine how hard in designing decision without assisted from CalculiX nonlinear analysis results. Here, commercial packages such as Ansys or Abaqus many offices seems they can not afford.
Note I would have been willing to personally spend $$$ one time for a program. That would probably have been LsDyna, origionally a free program, as that fits my current focus on bridge rails, if they had not changed to the leasing model per core, as I figured on taking about a year learn and then using from time to time for the remainder of my life. The leasing model has an unbounded cost over a long period of time, suited to business with a short horizon. The people I work for have a time horizon for the life of the bridge. I still have occasion to reanalyze bridges I have designed or repairs I worked on 45 years ago. I would still be using software I wrote in the 1980’s if microsoft had not changed to making the use of 16 bit programs difficult (some of these I had converted from my 8 bit apple or other now long gone computers they were in or I had converted to Fortran 77).
EXDIN Solutions (POLAND), FEA analysis in CalculiX for:
Approximately 80% of the FEA analyzes were performed in the CalculiX, the correctness of the results was confirmed by actual tests and measurements:
Are there any reports that show this?