Honeycomb core shear at support points 1

[motan]

Hello,

I aam designing an assembly of all-aluminum honeycomb panels, for which I created an FE model based on 2-D representation (PCOMP/NASTRAN). The panels are mainly subjected to pressure loading. Given that each individual panel in the assembly is are attached to the surrounding panels by discrete connections (i.e. brackets or , in some cases, fasteners penetrating the edge, namely fastener axis is located in the mid plane of the panel), the core shear (YZ,XZ) tends to be quite high around such points which act as load attractors. I wonder if I can base my core shear margins on the stresses in the fringe plot at the hot spots represented by panel connectivity points. In such case, the panel would likely need rectangular inserts of higher density core at such locations. Since I do not have any previous experience with honeycombs, I would appreciate any feedback, possibly including a simplified way to assess core shear stress for panels supported at dicrete locations. N.B. No panel testing is scheduled. Thank you.

 

[PanelGuy]

There are a couple of common solutions you can find in the online resources at MC Gill or Hexcel. However...

The more you can distribute the load in a hole in a lamination, the better. You can do this by laminating with a higher density core in sert or even a monolithicinsert...Aluminum puck or such. It would be interesting to see if with opimization you end up with square or round inserts.

As well, it is common to bond in an insert that is oversize but to also rabbet the core inside to allow for the potitng agent to lokc in and further distibute the loading.

Do not be afraid to explore using a backing plate on the top or bottom to essnetially create an external "local" reinforcment.

Anything you do will create its own loading stresses at its boundaries, which may lend itself to using a higher density honeycomb to ease the load transistion.

Hope this help a little. If you search online for "bonded honeycomb inserts" your fist page will get you some good papers but also two sources for inserts.

Good Luck!

 

[motan]

Thanks for suggestions. I might want to use the edge treatment as a support for attaching points, by milling the attachment feature into the edge treatment (a C-beam), rather than going for individual inserts. My dilemas are more related to interpreting the stress plots vs. the actual realization of the panels supports. Let's say that, given the size of the mesh/model, one would attach the panels at a few nodes per edge (while in reality, the machined edge may contain piano hinge type of connection on one edge and relatively closely spaced latching attachments on the opposite side.Other panels may well have a number of discrete attachment points all around the perimeter, each of them being fasteners threading into the edge treatment milled boses. In all such panels, the model shows high stress spots around the attachment points located on the edge treatment (C-beam). The issue is what core shear stress to assign to such points. I am not confident that I can apply the insert stress calculation methodology without some testing, and I am looking for some advice of how I can go by with analysis without testing.

 

[PanelGuy]

I see the issues. I really don't know how to attack it without some testing. Perhaps you can model it as you are and in a separate model put together a localized example of the attachment feature and apply the stress profile from what you already have...dunno, made sense when I started typing...

 

[SWComposites]

"looking for some advice of how I can go by with analysis without testing. "

Its a difficult problem; you can either make conservative assumptions in the analysis; or if you don't like that answer, then conduct some tests to validate/calibrate the analysis.

Attaching the FEM at only a few nodes per edge is not realistic and is going to lead to conservative peak stresses. If your mesh is small enough, you probably could ignore the stresses from all of the elements connected to the connection node.

 

[Hansmeister]

The two most common ways to handle point loads on a sandwich panel is to pot in the core locally, or, as mentioned, splice in a higher density core at the load point.

For attachment to the panel, NAS 1836 post-potted inserts are commonly used. Typical strength values are about 400 pounds in tension, and about 900 pounds in shear. Torsion usually is not a problem.

Recommend coupon testing to verify all numbers and design data.

 

[motan]

Yes, I thing I will need some "local" potting. Given the discrete attaching points, I would think the "local" would actually be a narrow strip of higher density core. This alleviates the shear core problem (based on what the model predicts).
I would also want to ask you about the core layout direction practice for a generic rectangular panel (actual ratio side "a"/side "b" of approx. 2.0, but the question relates to any typical rectangular panel loaded transversally). While I see the benefit of aligning the ribbon (L) direction of the honeycomb core with the long side of the panel, I could not find anywhere a clear instruction of how the core layout process should be in this respect. Namely, is the ribbon aligned along "a" (long side of the panel) in all cases? Are any advantages (e.g. lower core shear) by aligning the ribbon along the short side of the panel? Thank you for your support.

 

[PanelGuy]

hmmm...Depnds on what you are doing for stiffness. Generally the ribbon is run parallel to the width of a rectangular panel because the slice is rectangular in nature.

If you analysis shows that the ribbon must be parallel to the length then that is how it should be.

If it does not matter just use the geometry that works without a splice. If you must splice, just remember that in the lay up you want the cut edges of the cells to interlace as far as possible.

If doing higher density inserts I would recommend that you do discreet inserts (2" round etc) rather than a strip. Again, better material usage and no core splicing.

DFM - Know your process!