Modeling laminates; how to ensure consistent model?

About the development of the FEM module/workbench.

Moderator: bernd

Post Reply
FVerbaas
Posts: 13
Joined: Fri Mar 03, 2023 8:52 am

Modeling laminates; how to ensure consistent model?

Post by FVerbaas »

I need to model a laminated glass plate of 750x500 mm. The laminate consists of 2 plies of 12 mm glass (E-modulus 70,000 MPa) held together by an interlayer, which is a sheet of PVB, 1.5 mm thickness, with E-modulus 3.6 MPa.
I tried modelling the glass plies and the interlayer as boxes and combine them in a compound. Meshing the compound did not lead to lamination. They were, structurally, 3 different bodies that were connected only at some points. Defining ties between the surfaces just lead to CalculiX stopping the calculation because too many vertices in the tied surfaces not having a partner. To eliminate any differences between the surfaces (as far as differences are possible in the surfaces of a box) I tried defining the boxes as boolean difference objects, no avail .
I tried merging the three boxes, materials assigned, to find confirmed that merging objects does not retain the material info of the constituting solids.
I have the impression that during the meshing the geometric limits of the constituting solids are reflected in the elements generated, subject to the sequence of addition, but then the interface does not let me assign a material to elements within a geometric boundary.
The difference in stiffness is a factor 20,000; too large to be modeled in a geometric way.

So, what else can I do? Obviously I can dive into Python and write a laminate mesh builder, but that is only last resort.

Any advice is welcome.
User avatar
NewJoker
Veteran
Posts: 3015
Joined: Sun Oct 11, 2020 7:49 pm

Re: Modeling laminates; how to ensure consistent model?

Post by NewJoker »

1) If you are using any boolean/compound operations, make sure that they worked properly - check the mesh - it must be continuous between the layers, that's the whole point. Frequency analysis with scaled deformation will show any gaps.
2) Mesh should be sufficiently refined regardless of the approach that you use. Single element through the thickness is usually unacceptable, also in terms of accuracy. At least 3 elements should be used even for thin layers (that's why hexahedral and shell elements are typically used in such cases).
3) Tie constraints in FreeCAD should be defined for parts separated by small gaps, otherwise, they don't make sense since the mesh is (or should be if everything works properly) continuous.
4) Adjustment and tolerance for tie constraints are very important parameters - the first one allows the nodes on the slave surface to be moved on the master surface, thus establishing the connection. The second one controls the search distance - nodes further away won't be connected.
Post Reply