Castellated Beam

[HarryvL]

I experimented quite a bit and find that the following color setting offers optimum clarity ... a bit 70's though

Castellated_Beam_Opt_Color_Scheme.png (335.07 KiB) Viewed 1443 times

red = compression
green = tension

[paul18]

hi

I would like to say that we are here faced to a basic ‘case study’ :
- how many symetries can we deal with? considering both geometry and loading(s)? probably half or even the 1/4th
- basic patterns can be used and duplicate accordingly (maybe with ‘loop’ keyword in gmsh or ‘copy’ in cgx for example - firthermore you can build a fully parametric model)

In addition, are you using linear elements or quadratic ones; you've been using shell elements but it's well known that triangles and quite stiff compared to quads

Remember that decades ago models were optimized to reduce their size, the memory cost and the calculation duration

Paul

[bernd]

You just sait it ... decades ago ... Today what costs most is time and it takes just a quater of an hour to set up the model without all this smart stuff ... I really like it ... Sure for a real case study one would do it fully parametric, best just all in Python in FreeCAD FEM

But dventually the question comes up, what should be parametric, the cross section, the span, the load, the material ....

[paul18]

Today what costs most is time and it takes just a quater of an hour to set up the model without all this smart stuff

Exploiting the symetries does not take more time, and it's just the good practice.

My feedback: I've ever seen models with million(s) of elements, on a server with hundred Go of RAM, that crashed because such basics were not respected

But eventually the question comes up, what should be parametric, the cross section, the span, the load, the material ....

as well as the dimensions or the number of patterns among other things

[HarryvL]

hi

I would like to say that we are here faced to a basic ‘case study’ :
- how many symetries can we deal with? considering both geometry and loading(s)? probably half or even the 1/4th
- basic patterns can be used and duplicate accordingly (maybe with ‘loop’ keyword in gmsh or ‘copy’ in cgx for example - firthermore you can build a fully parametric model)

In addition, are you using linear elements or quadratic ones; you've been using shell elements but it's well known that triangles and quite stiff compared to quads

Remember that decades ago models were optimized to reduce their size, the memory cost and the calculation duration

Paul

Paul, the main point here was to show the capability and application of principal stress plots. The rest was me just fooling around (enjoying myself) with the insights gained. Harry

[bernd]

It would be great if we could standardise all dialogues as per the last method.

If it comes to deleting items from those lists, things get even more fancy
With some dialogues a context menu arises on stage

FreeCAD-FEM:
It's not a job, it's a an adventure

solving these issues is a matter of finding someone who is willing to spend his spare time with GUI development.

[UR_]

spare time

... spare time! ...
... spare time? ...
I have a really vague memory of this.
Do you know any details?

[cristian_ag]

Harry, great job.

Could you share with me your part file? I'm trying to replicate your example meanwhile I'm learning how to use FreeCAD, but I'm convinced that I'm modelling wrong the beam because when I run Calculix I have the following error.

"Nodes, but no results found in frd file. It means there only is a mesh but no results in frd file. Usually this happens for:
- analysis type 'NOANALYSIS'
- if CalculiX returned no results (happens on nonpositive jacobian determinant in at least one element)
- just no frd results where requestet in input file (neither 'node file' nor 'el file' in output section')"

But after some new tries, I continue obtaining the same error and I don't know how to solve it. If you can tell me some tips about the correct way to model the castellated beam, I'll be very gratefully.

Thanks.

[NewJoker]

@cristian_ag Getting the original model from Harry is one thing but it would be also good to find the cause of error in your model. Can you share the file ?

[HarryvL]

Harry, great job.

Could you share with me your part file?

Thanks.

I lost the original file, but recreated it:

Castellated_Beam.FCStd

(416.57 KiB) Downloaded 37 times

It takes several steps to get to this model. I created two sketches (see file), then with Part>Shapebuilder I turned them into faces. I then used Draft>Array Tools to copy this across 10 times in x-direction. I then used Draft>Downgrade twice to the Array to get simple faces. After this I stitched all faces together with Part>Fusion. The Fusion can then be used in Fem workbench to do the rest.

To be honest, there are 100 ways to get it wrong and only a few to make it work.