During my testing so far of the User element I worked with Prepomax. Just try it.
2 Likes
great enhancement, many thanks. Perhaps, it could be interesting to introduce basic design module of steel members and displaying result of demand capacity ratio in CGX. This can attract another user contribution to extent for specific code regulation they are familiar.
Dear Xyont,
I have to admit, adding codecheck is on my to do list, as it will open to civil engineering CCX,
I have already created a tool for Freecad (and in the end would like to use it with CCX)
Beam FEM Workbench - FreeCAD Forum
But i am still not fully convinced that it should be par of the CCX code, as:
- many different code
- high maintenance
- need to follow the rules that constantly update & have some national parameters.
- need to add many properties to item, useless for FEM resolution (Buckling length, stiffener spacing etc…)
It would need an army to cover all, and a second army to check it
Some structural software calculate it at the post processing phase (RFEM for instance), so i would rather make a good interface with SDC verifier for instance, or any equivalent open source UC check soft.
1 Like
hi @p.viguier
Indeed, design code is complex enough to implemented so there’s user parameter overwrites for selected members is available usually. It seems needed of separate module in dynamic linked library as Mfront does, may these approach can made a feasible workflow in further of extending and maintained.
An introducing how some design modules interact with database results and displaying in post processing can help much another user for starting to contribute.
best,
another structural analysis programs such as STAAD and GTSTRUDL perform design at input files directly, an example below.
*edited, add
related to 1D beam element itself, a capability to account rigid end joint offset can be useful also.
This is really great job! There are not too many programs for beam elements for structural engineers. I know frame3dd (very old: https://frame3dd.sourceforge.net/) and this one (freeware): https://www.gerritwolsink.nl/
This was always a weak point of CalculiX - not having fully-featured beam element, always ‘under development’, despite the fact that Guido is structural engineer by his education.
It would be nice to have ability to pre-process it on cgx (make U21 beams visible).
here is the linux executable for modified ccx (“linux_ccx_2.23.zip”):
Testing:
UB21 3D Frame Tower Module - Frequency Analysis
Issue: frd file - DISP only for the 1st form
2 Likes
Would it be a possibility for this to be bundled with the next release of Calculix?
They had discussed it above - this new element looks very complicated for CalculiX. Open-source community is usually very conservative and follow the modular principle. The size is growing from 10 to 16 MiB (libraries??) Also this implementation will require additional work for pre-postprocessor too… But anyway - this is the good and important feature - for some engineers it can be more useful than all other CalculiX features. What is probably possible is to make some improvements on existing UB (common section) - provide internal forces at output, maybe hinges and offsets. Some kind of light implementation. But nothing prevents to compile the additional binary with new feature for every new ccx build.
f i can understand properly this is not really new element, only an enhancement of existing beam element (U1) and latter also supports mass matrix for dynamic analysis. Probably, add to the next release is not a problem in compatibility of other CalculiX feature due to limitation itself i.e. contact and plasticity.
even many advanced beam and shell element exist, a classical formulation still being used today for both in education and practice.
1 Like
Dear all,
I am still on holidays without computer, so it is complex to check all question/implementation.
First I thank you for your interest, it is really motivating to see that the community is interested in it.
@xyont I agree this implementation of code is mainly tweaking what had already been done to implement the user element. So once the code is stabilized it will be possible to merge. I am still unhappy with some input/output. For next issue I will provide compiled version for both windows and Linux.
@xyont can you detail me a bit more what is the rigid end offset?
Offset has already been added beams at both end
@calc4fem Thanks for catching that bug with the Frequency analysis. I will investigate the output loop to ensure all extracted modes are correctly written to the .frd file. I’ve logged this as an issue.
@calc4fem Regarding CGX visualization, it remains a known hurdle (currently results only activate on one face). If anyone here is familiar with the CGX source code and wants to help patch the visualizer, your help would be incredibly welcome
Regards
Pierre
hi @p.viguier
in models beam and column are node-to-node of center intersection, actually it has overlap and shortening the length can be significant in results. An old reference book in structural analysis just simple use it is clear length in formulation, but another using rigid links at overlaps joint region. Result of beam at column face intersection is required for design purpose also.
best,
image source from PCA-Frame
1 Like
@xyont member end releases can be set up for each side individually according to x,y and z.
Therefore rigid offset is already included. I still need to make some test as the calculation as it is may not fully capture the comportement with 2 member offset with a big difference of X value.
For a better control of joint fixity UCONN can be setup for non linear relationship between joints
1 Like
@p.viguier rigid end offsets usually for continuous member intersection i.e rigid connection without any release assigned even though it’s possible.
however, i read again the documentation of UB21 and it have local coordinates ‘x’ offset. So, a feature of rigid end offset is already done, and it should be capable indeed.
btw, is this element can be use with equation to create rigid diaphragm behaviour? or beam nodes connected and share common with shell element of US3.
1 Like
Dear @xyont
That was I was making some test with the rigid body, which seems to work to link ub21 to U21, but the test was not extensive enough.
I will investigate this point once I get back home
Regards
Pierre
1 Like
Basically the purpose of the user element is giving the template for implementation of user element. This is self-explanatory. Abaqus has Timoshenko Beams (B21, B31, B22, B32), Euler-Bernoulli Beams (B23, B33) and Warping Beams (B31OS, B32OS). CCX somewhat is trying to follow this main structure, and it already has some of the beam elements (boxed and etc). From my understanding the main issue from the beginning was that CCX is developing from top to bottom, upside down - i.e. from volumetric elements to beams. And math of beam elements mimic the the volumetric ones. If UB21 is based on the classic approach, it is trying to break the whole program structure. Maybe it is not correct, just my intuitive opinion. Anyway, CalculiX development team has to have clear roadmap for implementing future improvements. They can’t just replace educational code of user beam with new complicated code. For example I do compilation of CCX with old university Mohr-Coulomb model, as user material, just for my personal use. But they just developed permanent Mohr-Coulomb material recently.
CGX definitely requires improvements for 1D elements, not only beams. For example it is possible to represent diagrams better than it is now (picture, attached). Maybe local beam axes is good to have too, to verify orientation of section. And etc.
1 Like
also, it could be a good idea to provide the documentation in TeX format to prepare when approved. Perhaps, CalculiX teams are welcome to this patch of enhancement since documentation and validation is complete enough, thanks again.
1 Like
An alternative improvement is a combination by area parts by b31 and b32 elements.
A double T area is combined by 3 parts with I parts.so the elements are here tripled by same nodes.
What is here needed? A composition editor in CGX or other preprozessors by the basic area elements with I, Pipe, channel.
Mostly this is enough to make all profile forms with enough precision.
Perhaps there are some problems in internal numerics by stresses in area at contact lines at the combined profile. But this is solvable with internal automatic contact elements at 3D-expansion.
But deformations and frequencies should be better than the wished and limited elements here.
B31 and B32 are here in advantage with internal 3D expansion.
So my conclusion? Extend the actual very good elements with a profile editor with combination of basic profiles to,a beam 3D „super“ element.
1 Like
Dear Stefano,
Please note that the current user element is not a replacement of B31 element, but an addition to CCX Kit, that people are free to use, so why limit enthusiasm???
I do agree that B31 element are already good for Stress and displacement analysis, however, tends to provide weird results for buckling and Modal analysis (B31R seem to be a bit better than B31 for thin section).
Please note that i create what i use the most as an Building and construction structural engineer.
Structural Engineer usually tend to limit as much a possible Plate & Solid element because:
- model are usually big (many dofs) → Calculation of an offshore structure would take ages ( and fill your RAM).
- Linear plate element generate hotspots → justification tend to raise question
- non linear behavior of section( Up to 5% plastic strain) is included in codes. If used want to use linear material, then additional justification with Neuber rule need to be done.
- buckling check is covered by analytic check, no need to perform buckling analysis + reduced stress methology (which would became a nightmare to organize, or report), or complex analysis MNA, GMNA, etc…
and the list continues.
Structural engineering for building and offshore is built on one main paradigm: calculation need to be simple, with code check to cover complexity, and bring resilience of structure. Only the new revision of Eurocode will add more information on FEM calculation and if you look at it, it is a nightmare.
For this reason, and for my pleasure, i try to implement a solution, that is efficient and can be used in your daily workflow as an engineer. The beauty of Open source is to allow people from different field to try thing, fork software, and if community enjoy the fork, sometime it is merge, else, people can still use the fork.
My aim in a few year is to merge all the existing bricks to perform full offshore analysis with open source software.
You may have seen on Freecad the pre/post processing tool have created, right now limited to pynite for now (early alpha/test phase, Making sure you're not a bot!)
If you do agree, lets close the discussion (on continue in private, not to pollute this post) on the interest that a user creates a user element, at it is the purpose of a user element is to be a custom element, that may not be used by all users.
1 Like
For you it is a nightmare to extend calculix b31 or b32.
For others it is a night mare to implement this new oldstyle element with their limitations.
See report by adapt with comparison of a simple double T beam and their frequencies also by beam elements in abacus, code aster and sesam sestra.
Great differences in many shapes and their frequencies in first and second order to second order shell or solid elements.
So beam elements in other codes are not in „gold“ standard after more than 60 years beam FEA in rocket technology programs.
An other alternative is to make for beam a 3 D section and combine this after a snmall 3D example calculation to condensed super elements with reduced formulations but better matrix than beam elements with normal theories.
This super elements are in problems in nonlinear solutions.
1 Like
when i follow CalculiX development since early, all element of 1D, 2D or 3D are generalized in solid element approach. Later, an improved classical formulation of beam and shell element available by MTU and it seems not to debate whichever is better or needed to replace. A classical beam element proves to be fast in design practice of trial error to fulfill demand/capacity ratio, solid element is well known to be good in nonlinearity (geometry, material & boundary/contact) and stress triaxiality even for a thin part.
Some regulation guides to use 1D element approach for nonlinearity also, but it may not relevat in the future since large memory in 32gb or 64gb is available on the market today, quadratic element with coarse mesh and reliable results can reduced computaional cost significantly.
However, one of three models of beam in CalculiX, i personally will chose a composed beam approach due to its capability in nonlinearity. An improvement in section force output, reduced knot problems and provide predefined common section shapes, it will eliminate complexity of input by large duplicated element and offsets.
1 Like