Stress Result in a cylinder Shell

About the development of the FEM module/workbench.

Moderator: bernd

User avatar
HarryvL
Veteran
Posts: 1285
Joined: Sat Jan 06, 2018 7:38 pm
Location: Netherlands

Re: Stress Result in a cylinder Shell

Post by HarryvL »

jake77 wrote: Fri Jan 18, 2019 7:56 am I guess switching to Linux would be a reasonable step at this point. Quite a step to take to get FEM results ;-)
@jake77, I switched to Linux because it is a dream in terms of installing and compiling FOSS, but if you don’t need this Windows should work equally well.
jake77 wrote: Fri Jan 18, 2019 7:56 am The inverstigated structure is the pressurized part of a pressure vessel that is to be checked for stability according to DIN EN 1993-1-6 for external pressure (full vacuum in the vessel). The vessel itself has two such compartments, that are placed above each other. From the outside it looks like a giant pipe (height ~22 m, diameter 3.6 m) with the helical halpfpipes (one on the upper, one on the lower part) around it. The speciality is that the cylinder is thin (9,5 mm) so the helix acts as a stiffener. However, the helix pitch is quite large so the formulae of the norm may not be applied (although they yield correct results). Hence the fight with the FE model.
Why can’t you sweep the “half-tubes” along a helix path?

PS: I started to explore FreeCAD, Calculix and Paraview a year ago and although the learning curve is steep the options to “hack” are limitless. As I said somewhere else they are an amazing trio. You can do some pretty amazing stuff when you are willing to invest the time and effort and the forum community is really really helpful to get you there. I have worked with Abaqus in the past and found you always hit some limitation you would wish you could hack your way out of. In a professional setting where money is less of an issue anything can be done of course, but in a private (or small consultancy) environment you are on your own. Finally, I we define “Value for Money” as the ratio of value and cost, then clearly FC value for money is infinite. In fact I am starting to understand that infinity comes in shades of grey, because I find that the value of freecad is higher than most FOSS out there :D
User avatar
bernd
Veteran
Posts: 12849
Joined: Sun Sep 08, 2013 8:07 pm
Location: Zürich, Switzerland
Contact:

Re: Stress Result in a cylinder Shell

Post by bernd »

jake77 wrote: Fri Jan 18, 2019 7:56 am The inverstigated structure is the pressurized part of a pressure vessel that is to be checked for stability according to DIN EN 1993-1-6 for external pressure (full vacuum in the vessel). The vessel itself has two such compartments, that are placed above each other. From the outside it looks like a giant pipe (height ~22 m, diameter 3.6 m) with the helical halpfpipes (one on the upper, one on the lower part) around it. The speciality is that the cylinder is thin (9,5 mm) so the helix acts as a stiffener. However, the helix pitch is quite large so the formulae of the norm may not be applied (although they yield correct results). Hence the fight with the FE model.
stress linearization plots might be interesting for you, since it seams to allow pressure vessel design according to ASME and EN codes and standards. See https://forum.freecadweb.org/viewtopic. ... 10#p147289


jake77 wrote: Fri Jan 18, 2019 7:36 am For me FreeCAD is not production-ready yet (something I suspected but still gave a try),
You exactly said it the way it is. FreeCAD might be not production-ready for your current knowledge of FreeCAD and the tools FreeCAD uses. Ansys is not production ready for me, because I used it 20 years ago and I lost all the knowledge about it. Keep learning and FreeCAD will be production ready for you. Harry said it FreeCAD and its tools have endless possibilities. You can do stuff with it no comercial software can do. But on the other side ther are a lot of processes you could do very much faster with some proprietary software.


jake77 wrote: Fri Jan 18, 2019 7:36 am but I forked the code and will look into the whole on toy problems first.
YEAH that are the lines I'd like to read. :mrgreen:



BTW: Your meshing problem is not a problem of FreeCAD, it is a problem of Windows version of gmsh bundled with FreeCAD.

bernd
jake77
Posts: 19
Joined: Thu Jan 17, 2019 6:23 am

Re: Stress Result in a cylinder Shell

Post by jake77 »

bernd wrote: Fri Jan 18, 2019 8:58 am
jake77 wrote: Fri Jan 18, 2019 7:56 am The inverstigated structure is the pressurized part of a pressure vessel that is to be checked for stability according to DIN EN 1993-1-6 for external pressure (full vacuum in the vessel). The vessel itself has two such compartments, that are placed above each other. From the outside it looks like a giant pipe (height ~22 m, diameter 3.6 m) with the helical halpfpipes (one on the upper, one on the lower part) around it. The speciality is that the cylinder is thin (9,5 mm) so the helix acts as a stiffener. However, the helix pitch is quite large so the formulae of the norm may not be applied (although they yield correct results). Hence the fight with the FE model.
stress linearization plots might be interesting for you, since it seams to allow pressure vessel design according to ASME and EN codes and standards. See https://forum.freecadweb.org/viewtopic. ... 10#p147289


jake77 wrote: Fri Jan 18, 2019 7:36 am For me FreeCAD is not production-ready yet (something I suspected but still gave a try),
You exactly said it the way it is. FreeCAD might be not production-ready for your current knowledge of FreeCAD and the tools FreeCAD uses. Ansys is not production ready for me, because I used it 20 years ago and I lost all the knowledge about it. Keep learning and FreeCAD will be production ready for you. Harry said it FreeCAD and its tools have endless possibilities. You can do stuff with it no comercial software can do. But on the other side ther are a lot of processes you could do very much faster with some proprietary software.


jake77 wrote: Fri Jan 18, 2019 7:36 am but I forked the code and will look into the whole on toy problems first.
YEAH that are the lines I'd like to read. :mrgreen:



BTW: Your meshing problem is not a problem of FreeCAD, it is a problem of Windows version of gmsh bundled with FreeCAD.

bernd
Well, TBH I did face lots of other issues besides meshing, some of which you also saw, and I consider these more discouraging as difficulties with a 3rd party program: loads gone missing, no Analysis object after loading a saved file etc. Sure, going the hacker way to utilize the full capabilities can make miracles, but in a production environment like mine (only engineer of the company) I need some reliability, too. This is what I miss, and IMHO this is what should be fixed first.
User avatar
bernd
Veteran
Posts: 12849
Joined: Sun Sep 08, 2013 8:07 pm
Location: Zürich, Switzerland
Contact:

Re: Stress Result in a cylinder Shell

Post by bernd »

jake77 wrote: Fri Jan 18, 2019 9:44 am loads gone missing, no Analysis object after loading a saved file
These both does for sure not happen for me. What does not mean they do not happen at all. I just do not have them, so I can not fix them ... I would consider them a release breaking bug. Means they have to be fixed before release. Would it be possible to post examples to be able to reproduce them.

bernd
User avatar
HarryvL
Veteran
Posts: 1285
Joined: Sat Jan 06, 2018 7:38 pm
Location: Netherlands

Re: Stress Result in a cylinder Shell

Post by HarryvL »

bernd wrote: Fri Jan 18, 2019 11:21 am These both does for sure not happen for me. What does not mean they do not happen at all.
+1

For sure there are issues, but I have not experienced many as significant as you raise.

Anyway, this is not commercial software and I think some patience and a willingness to highlight issues one at a time with detail, such that the developers can address them, is required. It’s quite amazing how quickly bugs get fixed and new features developed based on user feedback if you keep in mind that this is all done in the “evening hours” ... and then there is the fact that all code is open for users to modify themselves, if they feel inclined so. I think we need to use a different yardstick for judging FOSS than we use for commercial software.
User avatar
bernd
Veteran
Posts: 12849
Joined: Sun Sep 08, 2013 8:07 pm
Location: Zürich, Switzerland
Contact:

Re: Stress Result in a cylinder Shell

Post by bernd »

In the same folder as FreeCAD.exe you will find gmsh.exe. This starts gmsh GUI. Export the geometry from FreeCAD as brep (use brep not step!), import in gmsh and start to play. If exact your preferences are taken the gmsh GUI crashes too, screen attached. Gmsh 4 should be tested here. If this happen too, a bug report to gmsh developer is highly appretiated. ATM I have no idea why gmsh crashes, but there are a lot of logs. I am not a meshing expert, but playing with the parameter might help. Export the mesh as unv file, import unv in FreeCAD and drop the mesh in the analysis is what I would do.

cheers bernd

gmsh.jpg
gmsh.jpg (394.82 KiB) Viewed 807 times
jake77
Posts: 19
Joined: Thu Jan 17, 2019 6:23 am

Re: Stress Result in a cylinder Shell

Post by jake77 »

bernd wrote: Fri Jan 18, 2019 12:01 pm In the same folder as FreeCAD.exe you will find gmsh.exe. This starts gmsh GUI. Export the geometry from FreeCAD as brep (use brep not step!), import in gmsh and start to play. If exact your preferences are taken the gmsh GUI crashes too, screen attached. Gmsh 4 should be tested here. If this happen too, a bug report to gmsh developer is highly appretiated. ATM I have no idea why gmsh crashes, but there are a lot of logs. I am not a meshing expert, but playing with the parameter might help. Export the mesh as unv file, import unv in FreeCAD and drop the mesh in the analysis is what I would do.

cheers bernd
I tried this as well, but could not get a fine mesh, just a very crude one. I'll check on this later back.
User avatar
bernd
Veteran
Posts: 12849
Joined: Sun Sep 08, 2013 8:07 pm
Location: Zürich, Switzerland
Contact:

Re: Stress Result in a cylinder Shell

Post by bernd »

jake77 wrote: Fri Jan 18, 2019 12:41 pm I tried this as well, but could not get a fine mesh, just a very crude one. I'll check on this later back.
If I had the problem I would gmsh 4 give a try ... or use netgen gui ...
User avatar
bernd
Veteran
Posts: 12849
Joined: Sun Sep 08, 2013 8:07 pm
Location: Zürich, Switzerland
Contact:

Re: Stress Result in a cylinder Shell

Post by bernd »

bernd wrote: Fri Jan 18, 2019 7:18 am
jake77 wrote: Fri Jan 18, 2019 7:04 am
bernd wrote: Thu Jan 17, 2019 7:23 pm runs here too apart from the pressure problem (which is a FreeCAD limitation ATM for curved shell geometry and CalculiX DLOAD) ... has quite funny mode shapes ...
Could you elaborate what exactly is the limitiation? I changed the way the external pressure is applied. It is on the cylinder now, not on the outer faces, meaning, pressure load is applied only on singly curved surface. This should be OK I hope.
It has to do with the work flow. In FreeCAD we gone define the constraints on the geometry but the analysis is done with the mesh. This pressure does not work on curved shell geometry ATM. I need to have a closer look what exact the problem was. See https://github.com/FreeCAD/FreeCAD/blob ... ls.py#L712


@harry we gone had this topic already. You have been asking this when you came to the forum. There should be a topic around which exactly describes the problem. I am quite in a hurry and do not have the time to get into this. Last work day before ski holidays and there is a lot of snow in the european alps ATM :D
issue #3050 and https://forum.freecadweb.org/viewtopic.php?f=18&t=21872
User avatar
HarryvL
Veteran
Posts: 1285
Joined: Sat Jan 06, 2018 7:38 pm
Location: Netherlands

Re: Stress Result in a cylinder Shell

Post by HarryvL »

Sorry I missed your question. Yes I did ask about that in one of my firsts posts. Can’t find it anymore. Now I have done the curved 6-node interface element and with a better understanding of the code I could have a look and see how much effort it is to determine the nodal forces from pressure on a curved shell. I can do the maths but am not sure where shell nodal loads get calculated.
Post Reply