Finite element modeling of a composite monocoque and critical areas

[Flavio Galetti]

Hey guys,

There are dozens of topic here on the forum about FEA, and of composites and for the last days I've been looking through it to find answers to some problems, but failed to find. If you know a particular topic that can be helpful and could point it out for me I would appreciate it.

Anyway...

I have some questions regarding the FEA model. The primary idea is to mesh the geometry using shell elements and apply the material properties of the laminate into the elements. This is maybe good for static stress and deflection analysis.

But when it comes to the strength analysis I don't know. The idea is to make a submodel of the critical area and make a ply-by-ply analysis, but I don't know how to do it. There is no "von Mises equivalent", so an area that is more lightly loaded on average could be the critical one instead of another area where the loads are higher but in a higher strength direction. How do you guys model the different lamina and make the bonding in the model, and then how do you identify the critical areas?

Thanks a lot

[The AFX Master]

FEA with the appropriate elements would do the job for most in plane loading. But only if you provide extensive testing to back it up

The material properties MUST be obtained with real testing. Getting properties from diagrams and tables is just wrong when is supposed that your manufacturing techniques are far from ideal. (after all, we are students)

FEA formulation in ANSYS let you define each ply of the laminate independently (using Shell99 or Shell 181 elements), or by entering the laminate stiffness matrix (A B D Matrix) directly. You also can define sandwich structures this way (providing the core as another ply or with the full ABD matrix of the sandwich structure)

In CosmosWorks (now simulation), you only can set individual plies and materials (No matrix input). The sandwich structures can be done only assembling the outer skins to the core, This procedure gets akward when the geometry is far more complicated than a simple plate, and you'll need to assign all the contact pairs between facesheets and core as bonded. CosmosWorks results for most simple problems are comparable with hand calculations and ANSYS (i've tested myself). Of course, this approach solely sustain itself from a theoretical point of view

Regarding failure criteria, Hill-Tsai and Tsai WU are generally acepted theories for IN PLANE FAILURE OF PLIES, and can be checked with FEA and some hand calcs. Delamination or Buckling are more severe failure modes that are backed by far more complex theories. I wouldn't recommend any way other than real testing on the critical locations of the monocoque like suspension attachments.

The main concern here is how to relate the REAL testing with the FEA counterpart. For sure, things as interlaminate shear, internal non linearities, boundary condition setup and material quality (regarding how close is the theoretical FEA input data to the real laminate data), will make experiments to differ from FEA.

Two books about this:

Composite Structures : Laszlo Kollar
Compostite Materials : Daniel Gay, and Tsai

Final toughts: TEST TEST TEST!, And adapt the FEA to your TEST results, not the other way.

Final Edit: READ!, These aren't isotropic materials, VOn Mises, Principal Stresses, and such, DO NOT APPLY with composites. They're a totally different kind of materials. You need to learn quite a bit of the theoretical basis behind composites BEFORE attempting any FEA.

Read, and learn first.. FEA is Garbage in = Garbage Out

[Flavio Galetti]

Hello AFX

I think I should have been more clear about what I'm trying to do. I've been studying mechanics of composites for some 7 or 8 months now (yes, I am a newbie) on every scale (micro, macro and just starting structural), and I have a semester project which is the preliminary FEA of a composite monocoque structure. I got less than 3 months to do it, and the goal right now is not to manufacture the monocoque.

Rather, as you said, FEA is garbage in and garbage out. I guess what my project is about is analyzing the thing without garbage, but not when it comes to the properties of the material employed but, given a set of material properties (calculated, not obtained from testing), -what is the correct approach to modelling it and what useful results do you get from this analysis-

I understand this is usually not the way to go for an FSAE approach, but as I said it is not my scope.

Thanks for the tips. As this thing develops here I'll post some news. Maybe its useful to someone.

[Flavio Galetti]

Hey again

Figured out most of the stuff that I was in doubt before. Got one little question that maybe somebody here got it figured out already.

It's about the Transverse shear stiffnesses on ansys, called there by E11, E22 and E12, which you can override using the section controls (I'm using SHELL181). Apparently it defaults to zero, which is rubbish, but can somebody confirm that to me?

Also, if somebody has more info regarding transverse stiffness (things regarding the solver) in ansys I would really appreciate, since the help file for shell181 isn't clear enough and doesn't use the usual variable names for transverse stiffness.

Thanks