[ANSWERED]Help learning FEM

About the development of the FEM module/workbench.

Moderator: bernd

bzb.dev001
Posts: 102
Joined: Sun May 05, 2019 3:08 am

Re: Help learning FEM

Postby bzb.dev001 » Mon Jul 22, 2019 9:07 pm

Thanks for the help. I've followed your instruction according to https://www.freecadweb.org/wiki/Part_Ch ... _BOP_check to configure my FreeCAD.

I got it to work finally.

Something of note:
1) Fusing all solids at once fails BOPCheck.
If I were to fuse the legs array, table top spacers and the tabletop all at once and set property refine=true, it would fail the BOPCheck
2) Fusing solids a bit at a time works.
If I were to fuse the legs array and table top spacers. Then fuse the resulting fusion to the tabletop and then set the resulting final fusion property refine=true, BOPCheck succeeds.

Is this how you fuse solids in FreeCAD? ie. small bits at a time and not all solids at once. Is this a bug?

Thanks.

openBrain wrote:
Mon Jul 22, 2019 5:35 pm
bzb.dev001 wrote:
Mon Jul 22, 2019 3:56 am
I tried using a different solid to perform FEM on and it seems to work albeit there was a warning message. I wonder if there could be something wrong with the original solid.
Definitely there are. I took your OP example file and ran a geometry check with BOP enabled. There are some errors that probably lead to your final problem. They come from your array that has intersecting shape.
The good news is that this isn't a big issue, and everything get solved by setting "Refine = true" in the object properties.
Then gmsh still complains (AFAIK it always complains :lol: ) but it generates a good mesh that can seamlessly be used for FEM analysis.
Eg. below I fixed the feet and pushed 50 kg it's too much :D ) :
fem_table.png

I'll attach the FreeCAD file but I have to clean it so it is accepted by the forum. ;) => EDIT : I had to remove the results from the FEM analysis !
Beware that refining the shape will mostly break the existing FEM constraints and material attachment.

HTH
bzb.dev001
Posts: 102
Joined: Sun May 05, 2019 3:08 am

Re: Help learning FEM

Postby bzb.dev001 » Mon Jul 22, 2019 9:22 pm

Regarding https://www.freecadweb.org/wiki/Part_RefineShape
I find this function is no more helpful when operating on a fusion of all selected solids. No different than setting such fusion with property refine=true.
openBrain
Posts: 4473
Joined: Fri Nov 09, 2018 5:38 pm

Re: Help learning FEM

Postby openBrain » Tue Jul 23, 2019 8:50 am

bzb.dev001 wrote:
Mon Jul 22, 2019 9:07 pm
Something of note:
1) Fusing all solids at once fails BOPCheck.
If I were to fuse the legs array, table top spacers and the tabletop all at once and set property refine=true, it would fail the BOPCheck
2) Fusing solids a bit at a time works.
If I were to fuse the legs array and table top spacers. Then fuse the resulting fusion to the tabletop and then set the resulting final fusion property refine=true, BOPCheck succeeds.

Is this how you fuse solids in FreeCAD? ie. small bits at a time and not all solids at once. Is this a bug?
Here both ways (all at once, or successive steps) works OK. Maybe it depends on the OCC version.
bzb.dev001 wrote:
Mon Jul 22, 2019 9:22 pm
Regarding https://www.freecadweb.org/wiki/Part_RefineShape
I find this function is no more helpful when operating on a fusion of all selected solids. No different than setting such fusion with property refine=true.
Actually it's different, as the copy is independent from the original shape. ;) => :!: :!: :!: This changed 15 days ago, so depending on your version, you'll get either an independent or linked copy. ;)
There has been a discussion about this feature usefulness. But ATM I'd vote to keep, as in some cases there is no other way to refine.
User avatar
bernd
Posts: 10523
Joined: Sun Sep 08, 2013 8:07 pm
Location: Zürich, Switzerland

Re: Help learning FEM

Postby bernd » Tue Jul 23, 2019 9:51 am

Usually the fusion should result in only one Solid, but a fusion returns a Compound. In rare cases gmsh has problems with the Compound. To rxtract the solid out of the compound Part_CompoundFilter can be used. Same for CompSolids in multi material analysises.
openBrain
Posts: 4473
Joined: Fri Nov 09, 2018 5:38 pm

Re: Help learning FEM

Postby openBrain » Tue Jul 23, 2019 11:20 am

bernd wrote:
Tue Jul 23, 2019 9:51 am
Usually the fusion should result in only one Solid, but a fusion returns a Compound.
:?: :o @bernd, could you explain a bit ?
Because AFAIK, a fusion (which in doc is named Part Union) always result in a single solid and never in a compound. ;)
I never saw a case where a compound was created. This is also documented this way.
Finally if you try a compound explode on the fusion of the OP file, it will return only one item as an evidence it is a single solid. ;)
User avatar
bernd
Posts: 10523
Joined: Sun Sep 08, 2013 8:07 pm
Location: Zürich, Switzerland

Re: Help learning FEM

Postby bernd » Tue Jul 23, 2019 1:10 pm

The other way around ... a boolean union, cut or intersectiondoes return a compound in any case! Just make a union of two intersected boxes and use check geometry from part wb.

If you gone make a union of two intesect solids you should get a compound with one solid, but if they do not intersect you will get a compound of two solids. If you mesh this you may get a not connected mesh. Another example would be if the union return one solid and a two shells because of problems. Than may be some part of your union is not solid at all and will not be meshed with solid elements.

But if the solid is extracted by "Part Compound Filter" it is sure what is solid and what not.

bernd
openBrain
Posts: 4473
Joined: Fri Nov 09, 2018 5:38 pm

Re: Help learning FEM

Postby openBrain » Tue Jul 23, 2019 4:49 pm

bernd wrote:
Tue Jul 23, 2019 1:10 pm
The other way around ...
Oh yes ! Thanks for taking time to upgrade me on that point. ;)