Multi Material Tutorial

[ceremcem]

I'm trying to determine correct material type and structure geometry for my open source caravan project. I need to create assemblies with different materials. Is FEM module capable of doing that? If so, how can I declare "ABS" for wheels (pink) and "Aluminum" for chassis (olive) and "Steel" for axle (blue)?

Screenshot_2019-08-10_16-13-25.png (173.93 KiB) Viewed 2388 times

[Jee-Bee]

I added multiple materials as example

I noticed that it is impossible to pre-select shapes before adding. I no there is spoken about that before... but one thing i noticed when creating a force the select is direct active while it is not with the material selection...

I have an older version but not that old...

OS: OS X El Capitan (10.11)
Word size of OS: 64-bit
Word size of FreeCAD: 64-bit
Version: 0.19.17175 (Git)
Build type: Release
Branch: master
Hash: 507c40669d48ae4732268324cd0140c80cf68222
Python version: 3.7.3
Qt version: 5.9.7
Coin version: 4.0.0a
OCC version: 7.3.0
Locale: English/Netherlands (en_NL)

[ceremcem]

(to be removed, see below post)

[ceremcem]

(to be deleted because it was an incorrect error report)

[ceremcem]

I added multiple materials as example

I understood how to declare multiple materials for multiple solids/geometries. However, I can see no complete Analysis (no mesh, no Calculix results) in your example file.

I tried creating separate meshes for each materials but Fem threw the error: "Fem: Multiple mesh in analysis is not yet supported". So I created a compound: Selected all solids (chassis, axle, wheels) and clicked Part->Compound -> Make compound, renewed material and solid relations, then created a mesh (with Netgen) by using this compound. Clicking "Write .inp file" threw the following error:

Code: Select all

FEM mesh has no volume elements, either define a shell thicknesses or provide a FEM mesh with volume elements.

How should I continue at this point?

[Jee-Bee]

You need to create a single mesh from all objects at the same time.
When creating the .inp file for Calculix than there is made a distinction between what nodes belongs to what materials.

[shIxx]

You have to make a single mesh from all bodies combied in a compsolid. To do this select all bodies and make a boolean fragment and then switch from standard to compsolid under mode in the property of this boolean fragment. Than you can add a compound filter to remove the last compound and stay with only one compsolid existing of 6 solids (see the screennshot). Then you should mesh this with Gmesh because Netgen is not supporting multiple materials at the moment. It is important that the mesh is sharing the nodes between the different faces (see second and third screenshot).
It should worked now but I still got an error that calculix can't write the .inp file

Code: Select all

Error in get_femelement_sets_from_group_data -- > femelements_count_ok() failed!

I thought it was because the of the geometry. I found out that the axle was sticking into the wheels (see screenshot)
After I fixed this, everything has worked well.
Watch this, unfortunately in German but it shows the concept well https://youtu.be/T-ggqH3tXHs?t=492

[shIxx]

I attached the .FCStd as well without the mesh for smaler file size

Edit:
BTW if you want the axle to go through the wheels, than you have to cut the axle out of the wheels. This should also work.
In summary, it can be said that its important to:
- have no overlapping geometries!
- use compsolid instead of compound
- and mesh with Gmesh

[shIxx]

I found a very important comment on the video I posted above. I translated this comment in English for all English-speaking people here (by google translate):

With several materials with different E-modules, the stress evaluation at the contact surfaces is wrong! This is because in the FEM method, from the integration points of the elements, the stresses at the common nodes are averaged. With different moduli of elasticity, these stresses are higher with the equation Sigma = E * epsilon and lower with the high modulus of elasticity. These are then averaged together at the nodes. I therefore recommend only evaluating the strains epsilon. If someone needs the tensions, the strains themselves must be multiplied by the respective modulus of elasticity.

[shIxx]

sry for the double posts but I have repeated the analysis with Netgen and it looks like Netgen does support multiple materials from now on *thumbs up*
I think this was not possible in .17, but it seems it is implemented now in .18.