Simulation of Non-layered Composite

[Searc]

Hi there,

The composite I am looking to simulate has long fibre cast into a matrix (metal-matrix-composite), and is not layered up like an epoxy-carbon. The fibres are sometime uni-directional and sometime follow a 3D path.

The 'traditional' approach for composite FEA is to model the plies on top of a surface but this does not seem very appropriate to my composite.

What would be the best approach to simulating this non-layered composite?

I am interested in finding the best approach but so far I have considered (please suggest if there is a better method):

1. Just use the 'traditional' method in Ansys ACP of surfaces but with thick layers - least favoured.
2. Use the solid extrude feature in Ansys ACP, though I will still need to use layers.
3. Use the solid extrusion guide in Ansys ACP, though I will still need to use layers.
4. Model as a solid using orthotropic material properties in Ansys Mechanical, and use the orient element to align each element with the desired fibre direction at that point. Though I have the issue of applying a failure theories in Ansys Mechanical (best method is to use the composite damage tool).

Thanks!

 

[FEA way]

Maybe you could use a discrete model of the reinforcement - bar/beam elements embedded in solid elements. I don’t work in Ansys but I know that this approach is possible in other software so there’s a great chance it can be done in Ansys as well.

 

[Searc]

Thanks for the reply FEA way.

Hmm, I think method is good if you have say up to 20 fibres through a volume, but in this case it is thousands plus tows. Additionally, though I may know the general placement and path of a group of fibres, each fibre may deviate. I have test data for the composite as a whole, so I know the macroscopic response of the fibre-matrix together.

 

[FEA way]

It depends what level of accuracy you need. Macroscale approach (homogenous equivalent material with orthotropic properties for the whole composite) is also often used in such cases.

 

[Searc]

Please could you go into further detail about the macroscale approach you mentioned? My current approach to this method is to model and mesh in normal 3D, then orient each element to have the macro-orthotropic properties in the correct direction for the fibre direction at that point in the volume. My my issue with this method is applying composite failure theories - there are some 3D versions, but my understanding is they are know to be not that accurate. Additionally, the only way I have found to implement failure theories in Ansys mechanical is through the composite damage tool.

As a related question, am I correct in thinking the this approach also the best hand analysis method for non-layered 3D composites? That is, using the macro-isotropic properties of the part, finding stress at each point in the part, then use 3D versions of the composite failure theories to determine the location and mode of failure. Similar to the FEA method above, this method is only as good as the 3D versions of the composite failure theories.

 

[FEA way]

The method that you use is what I mentioned before as macroscale modeling but with one significant difference - in the case of the approach that I described (standard macroscale composite modeling) material orientations are the same for all elements. Since you know the properties of the whole structure in each direction then you can treat the composite as a homogenous material with global orthotropy.

 

[Searc]

Not sure I understand because as the fibre direction changes over the part the isotropic properties stiffest direction changes too. So unless you were modelling a part with no fibre direction change you would need the mesh element directions to change over the part volume.

 

[FEA way]

Yes, but the method I described is more simplified (in fact it's the most simplified approach to composite modeling). It assumes that there's no variation of mechanical properties within the composite and treats it as a uniform material that has global orthotropy.

Can you share some picture or scheme showing this composite with visible fibre orientation ? Maybe it will be easier to suggest other method for this particular case.

 

[Searc]

So I am looking at the method in general, but I will try to give you an idea of the type of parts if could be used for - these are just generic examples.

https://st.mascus.com/imagetilewm/product/hso/volvo-link-arm-volvo-l180e,99295602_1.jpg

The fibre could be UD down the length of the arm and flow in a circle around the eyes.

https://www.douglasvalley.co.uk/images/21000/20215/fancybox_IMG_1401.JPG

Similar to the above apart from the fibre would split at the fork and extra fibre would follow the large curve at the fork.

https://www.usedpartsfast.com/36370...ght-cx236037ab-ref1316-jaguar-xf-x250-22d.jpg

The fibre placement would be along the path of stress flow from the bottom attachment points to the top.

As mentioned, for all these examples the fibres are just imbedded in the matrix - no layering/plies.

EDIT: In terms of what the fibres look like in the matrix it is similar to the following image:

https://images.app.goo.gl/V72L7ZZw1exyqa8a6

 

[SWComposites]

None of the failure theories in Ansys are likely to be valid for your material (they don't even really work for typical carbon/epoxy composites).

 

[Searc]

Hi SWComposites, thanks for the reply.

Is there another approach you would suggest, instead? Perhaps in another piece of software? Or is it a problem without a current solution which can only be solved with testing & validation?

Additionally, if the failure theories in Anaya are not applicable, could Ansys still be used to determine accurate stress/strains at each point in the MMC?

 

[SWComposites]

Yes, test it.

Ansys could calc accurate stresses if you can accurately model the material and if the input material properties are accurate.

 

[FEA way]

When it comes to modeling of this material, it’s very likely that you could use a software like Moldflow, Moldex3D or Fibersim to design and model complex fiber layup. They allow you to export it to FEA software (including Ansys) as material orientations and orthotropic properties for each element. These programs are meant for plastic injection molding but I think that it won’t be a problem to switch polymer matrix to metal one by adjusting its mechanical properties at some point of the setup.

The choice of failure theory might be a problem here but you should be able to find some workarounds in scientific articles. Accuracy may not be very high but the whole thing is an approximation anyway.

 

[Searc]

Hi FEA way and SWComposites - thanks for the help btw.

I'm having a look at Moldflow, Moldex3D and Fibersim.

I think the best current method I have available now is to either use a low number of thick plies and extrude into a solid elements within Ansys ACP, or use oriented elements with orthotropic properties all within Ansys workbench - to at least get stress-strain results.

Having a look through the MMC analyst section of Composite materials handbook. Volume 4, Metal matrix composites (Military Handbook - MIL-HDBK-17-4A Composite Materials Handbook, Volume 4 - Metal Matrix Composites) - it references Maximum stress, Maximum strain, Tsai-Hill, Tsai-Wu, Hashin, Puck and LaRC03 criterions. So I would think these are applicable.

 

[SWComposites]

Also look at Digimat for predicting fiber angles.

And I don’t know how those failure theories got into CMH-17 Vol 4, but I have never seen them validated for MMCs. Cripe another section of CMH-17 to have to go rewrite.