loads in the edge band of honeycomb panel

[gokmavi]

Dear All,

I would like to ask your opinion about an issue for HC panels.
We have a typical honeycomb panel with orthotropic face sheets . At edges of the panel there is an edge band region (monolithic). This HC panel is attached to a metallic backup structure from these edge band area with only one row of fasteners.
A uniform negative pressure is acting on the panel.
Below I add a simple sketch.

^^^^^^^^^^^^^^^^^^^^^^^^^
IIIIIIIIIIIIIIIIIIIIIIIII
--------------------------------
T \ / T
\ /
---------------------

This pressure load is reacted by heel-toe forces between fastener line and free edge and pull-thru load at fastener line.(If you have Niu's stress analysis and sizing book you can see an example at pages 216 thru 219). OK, up to here no discussion.

1) Are there any bypass loads at the edgeband ? (My opinion is no because there is only one row of fasteners)
2) In Niu's book, without any explanation, %60/%40 rotational restraint is assumed for ramp & fastener ends, respectively. As engineers, we love this magic number (60/40) but what is the justification of it?

Thanks in advance.

Regards.

 

[40818]

If the panel has an applied pressure transforming the panel into a deformed shape, then you will have a degree of shear load applied to the structure. You could propbably use (conservativly?), a case of a beam restrained against horizontal displacement at the ends. Roark 7th page 245 (Table 8.10) gives some cases.
Case 3, ends pinned to rigid supports, uniformly distributed transverse load on span is:

(ymax + (A/4I)ymax^3) = (5wl^4/4piEI), ans solve for ymax

transverse load P:

P = (pi^2EA/4l^2)ymax^2

Of course a quick model will give better results, though if you do one, then watch the corners, as they will try to deflect the opposite way.

As for Mikes 60%/40% ratio, i dont know, maybe it was a rule of thumb.

But i dont think the method used is conservative (probably more accurate though), as if you just freebody the end joint, you would end up with higher fastener tension loads then you do if you work his example.

Maybe SWComposites might give a clue?

 

[FEMdude]

Bending moment would be 50%/50%, if both ramp and fastener ends of the edgeband were rigid (M=FL/2). Niu assumes fastener end to be more flexible than ramp end.

 

[RPstress]

1) No. (Also, bypass loads are typically only taken into account with in-plane loading. In the case of secondary structure we would usually only go as far as out-of-plane analysis, essentially through-thickness shear, bending (includng local instabilities, etc.) and fastener pull-through.)

2) Accounting for every number used in a hand calculation is quite hard. Niu is almost a primary reference these days. If he says 60/40 then it's probably good enough. I have used 50% fixity at the fastener before now; Niu's 40% is a bit more conservative than that.

Evaluating the degree of fixity is quite hard, as it depends on the bending stiffnesses of the attach flange and the edge band, the rotational constraint from the fastener head and probably a few other things (it is the degree of fixity at failure which counts most here; even with brittle composite this is unlikely to be the same as an elastic analysis would give. On the plus side, this might make friction effects negligible).

If you absolutely had to come up with something, then a simple stick model with FE beams might be the simplest way to get a numerate answer. Of course, it might not bear that much relation to reality. A marvellous months-long 3D non-linear analysis programme would probably yield a different answer, probably with even less relation to reality...

If the degree of fixity really, REALLY mattered, then a small test program might actually be necessary. NB: whenever sandwich panels are tested to destruction the results are always better than predictions (sometimes by a factor of two, depending on what failure mode is being predicted).

Now, just please don't ask how to analyse the ramp region...

 

[gokmavi]

Thanks for all the replies.

My questions are arisen from some discussions done with colleagues from a major aerospace company.

We performed bearing-bypass analyses by only considering in-plane loads at the edgeband. As quoted in the 2nd reply, we simply do not consider any bending. What they claim is that equivalent in-plane loads due to bending (at the edgeband) have to be calculated and they have to be combined with in-plane loads. This approach seems non-sense for us but we can not convince them.

Second issue is about rotational restraint. We simply followed the same approach in Niu's book to determine max moment in the edgeband and considered 60% of it while analysing the edgeband laminate. As RPstress said, I did a simple stick model idealizing unit width of the panel and fastener with spring elements. Then I calculate bending moments at fastener location & middle of the ramp area, I got very close values to given in Niu's book. 50% seems very reasonable because if we consider edgeband as a beam with guided end (ramp side) & fixed end (fastener side).

Thanks once more for all the information you shared.

gokmavi

 

[RPstress]

Is the panel then also under in-plane loading (perhaps from induced strain due to wing bending or similar)?

The main way in-plane loads can come from pressure load is due to membraning, which is usually ignored for less important structures, especially ones like sandwich panels that are stiff in bending compared with in-plane.

You must of course combine the face stresses for in-plane and pressure loads.

NB: in-plane load/strain can increase the bending due to pressure, although the offset of the fastener line from the neutral axis can ameliorate and even reverse this. It depends on the ratio of the pressure to the in-plane loads and the panel NA position. This is beam-column behavior and is non-linear.

Sometimes even more non-linearity occurs: IF the in-plane load is due to something like wing bending, then it will come after the pressure load, as the whole wing takes longer to bend than the panel. If the panel has bent significantly under the air then the response to the in-plane load can be different than if both out of plane and in-plane loadings are applied at the same time.

 

[gokmavi]

Dear RPstress,

I post an attachment, which is a pdf file showing a simple sketch. If there is only pressure load (no additional in-plane load), I think, it is not possible to talk about any bearing-bypass analysis. An I wrong?

Best Regards.

 

[40818]

Govmaki, as i gave in my earler reply.

The simplifications existing with handcalcs tends to ignore the lateral constraint forces due to the loadin g set-up you show.

If you did a fem of this you would end up with the lateral restraint forces, why? because they do exist. Though it is ignored for all intents and purposes for the analysis.
Conservative to include it though, as if the loads are high you would reduce the fastener RF due to combined shear-tesion interaction.

 

[gokmavi]

40818, Sorry but there is a misunderstanding. My question is about bearing-bypass interaction not pure bearing. Lateral restraint force (if it is not neglected) that you mention is considered as pure bearing load, according to my opinion. It has nothing to do with bypass because we have only single row of fasteners at the edge.
Regards.

 

[RPstress]

If you have a single row of fasteners with some in-plane load then it is usual to use a "bearing-bypass" analysis, but the bypass load is zero.

The reason you still need a "bearing-bypass" analysis is because of the chance of the panel failing in tension between the fasteners. This will only happen if the fasteners are excessively close together or the material is a bit weird. If doing the analysis by hand, then you'd just do a check on bearing and another on tension between fasteners.

Remember also to check for shear out if considering in-plane load.

If you have a compressive in-plane load with a single row of fasteners then bearing is the only check necessary. No load goes past the fastener line.

The tension between fasteners will need to be added to any bending stresses from the flange bending. For a honeycomb panel the bending will still dominate. NB: if you have to take account of in-plane load due to bending, then, on the plus side, the flange bending will be reduced a bit.

For your situation the in-plane load due to membrane bending should be low. If you really must consider it, then in the absence of initial curvature, calculating it will need a non-linear plate analysis. ESDU can help with that, or a non-linear FE run (if doing FE remember to put in fastener stiffnesses and model the attach flange as well as the panel).

 

[RPstress]

NB: I said to add the in-plane tension due to membrane bending to the flange bending.

However, I don't think that Niu's 40% bending should be included in the "bearing-bypass" calculation. The fastener head should prevent significant bending occuring beside the fastener hole where the high stresses from the in-plane tension occur.

 

[gokmavi]

RPstress,

Sorry for the late response.
Thank you very much for your explanations.

I meant bearing-bypass interaction not bypass alone.

Yes, I check shear-out & pull-thru as well.

My main question was exactly about what you mentioned in your last sentence of your last quote. I agree with you.
I think confusion comes from in-plane load concept.
When I said in-plane load, I meant equivalent in-plane load due to bending. It is not a physical in-plane load.
It is an fictitious value, which will result the strain equivalent to the max strain, which is obtained from a pure bending moment to any section.
They calculate this value because their bearing-bypass interaction computer codes work with in-plane loads only (not with the stresses).
I think we can close this thread now.

Thank you again and Merry Christmas

gökmavi