Effective skin width 4

[Trajano]

This is my first post: hello everybody.
I am working on the skin (aluminium) of a wing. I am studying the stringers run out. Between the end of the stringer and the supporting rib there is a distance of about 1.6 inches. This region of the skin must carry the vertical shear load of the stringer. One effect of this load is a moment on the skin.
Which is the effective width of skin that I should consider? Any reference?
Thanks in advance.

 

[rb1957]

before discussing effective width, i'd challenge your assumption that the skin carries the "vertical" load of the stringer; assuming that vertical means out-of-plane. skins and stringers react in-plane loads (pretty much only, 'cause that's their stiffest direction).

if "vertical" means axial, i'd assume that the stringer end-load is reacted by the entire skin bay, rather than visualising an effective piece of skin continuing the stringer. the reason is that the skin panel will probably buckle through this unsupported skin.

i assume you're using FE results. you'll have stresses in the skin panels (axial, transverse, and shear) and endload in the stringer. i'd assume that the endload in the stringer is sheared into the skin bays on either side of the stringer, increasing their shear stresses. you could combine the two bays together, apply the average shear stress (including the component from the stringer) and compare to the buckling allowable of the combined panel; i think this is conservative (possibly overly so) but i think it represents the scenario of a buckle across the unstiffened skin the best.

in my experience stringer run-outs extend only slightly into their last bay ... which would naturally lead into the above analysis.

of course, all this is with a static stress analysis view. a DT view would be different, almost the direct opposite (just to be conservative). i'd analyze the last rivet, allowing it to transfer almost as much load as it can; with some concern over secondary bending of the skin.

of course, the stringer run-out is very carefully tapered; the stringer cap is first tapered, then gradually the stringer web, and in parallel the skin thickness is increased.

 

[Trajano]

Thank you for your reply. I beg your pardon for the vagueness of my explanation: instead of “vertical” load I should have said “out-of-plane” load.

The load that worries me is not the end axial load of the Bar elements of the stringer, but the load out-of-plane (shear force of the Bar element, according to Nastran terminology, to be precise). Due to this load, a moment appears in the skin. And the question is: what is the effective skin width for this moment?

 

[rb1957]

since i don't think the stringer would develop an out-of-plane load, i think you should remodel the stringer with rod elements. this could have an impact on the skin elements, if there are skin nodes that use the bending stiffness of the bar to support their out-of-plane freedom.

if that is a problem, then there are ways to fix that problem.

 

[crackman]

Trajano

You are correct about the out of plane load. Depending on the actual design configuration of the stringers and wing panel, there can be an out of plane moment. This is typically accounted for in FEM with Z-offsets for the neutral axis of the stringer(there is a good writeup on this in the Boeing FEM Techniques Handbook). The stringer runout problem you are speaking of is a common one. This is a typical area where fatigue cracks occur (AFAIK most airplanes have had cracking in runouts and runouts have been a topic for WFD issues). Normally, if the designer has done a good job, the stringer has been tapered (reduction in height reduces NA) prior to reaching the end of the runout so that it is shedding load to the adjacent stringers and/or spar cap so that the end moment is reduced. As for the effective width for bending, prudently you could use anywhere between 30t and the flange width of the stringer. By the way, if the panel is integrally stiffened, I would stick to the 30t and pay attention to the end radius as it will have a big effect.

 

[Trajano]

Thank you both for the reply.

In order to clarify the problem lets assume a ref. system : x-> chord wise; y-> span wise; z-> normal. Then xy is the skin plane.

Crackman, when you say “there can be an out of plane moment” should I understand:
-1. “there can be an in plane moment (Mx) due to the out of plane load (Fy)”, i.e. due to the offset of the NA?
-2. or you mean: “there can be an in plane moment (Mx) due to the out of plane load (Fz)”?
-3. or another thing?

I’m asking for the second: the Fz load that the stringer carried, and that Rb1957 reckons it shouldn’t be there. By the way: I cannot change the Global FEM to put rods instead of beams, I’m limited to do my best with the current FEM. Crackman, do you also reckon that a proper model should have rods, not beams?
Cheers

 

[crackman]

Trajano

I was speaking of Item 2 as you have listed. As far as Fz loads on the stringer, the only way I concieve this is that if you are referring to the typical stability support which ribs/frames provide to stringers via rib/frame clip attachments. Remember, the main reason for the stringer to be attached to a frame or rib is in order to provide stability, ie beam column issues. There is no other Fz type loading. If your stringer runs out before being able to be connected to a rib, then you have stringer (ie column) with a shorter supported length and actually an unsupported end. This is not a bending stress issue but a stability issue. The only bending stress issue AFAIK is related to Item 2.

As far as proper modelling, the Boeing FEM handbook I believe demonstrates a dozen or more ways to idealize the stringers and skins on a wing panel. Really has much to do with how you idealize your overall design and the type and format of data you want to obtain. For example, the types of stringers models have alot to do with whether or not you model the skin at the OML or the cg of the panel.

 

[rb1957]

i agree with crackman, that there is small moment due to the axial load in the skin being offset from the the plane of the skin (ie, at the stringer's neutral axis). this is minimised by carefull design of the run-out, as both crackman and i have posted.

but there shouldn't be any out-of-plane load at the stringer run-out; yes, there can be some between the ribs, as the stringer is clipped to them, but this should still be very small.

it sounds like you're using someone else's model, i pity you 'cause that usually leads to this type of problem (if it was my model, i'd pretty much ignore the Fz load, noting that it wasn't practical and pretty much just an artifice of the model).

still a bit puzzled where this load is coming from, out-of-plane pressure loads are pretty small. i hope your skins are modelled as membranes (without bending stiffness), but i suspect that they are plates. there are modelling reasons to use plates, but there are also alternatives; but you're stuck with the model you have !

at the end of the day, i'd use 30t as an effective width. i suspect that there's a machined (or bonded) pad at the stringer run-out, you could use 15t either side of this pad-up (as well as the pad-up).

good luck !

 

[Trajano]

Many thanks for your opinions, gentlemen.

Yes, I am using someone else’s model. The skins are plates: the FEM “experts” argue that it is in order to avoid too-many-DOF problems, and that the skin and stringers actually do have that bending stiffness: not a too big stiffness, but they say that it is the real one, i.e. the skin-stingers is not a “pure” shell, but has some “small” bending capacity.

Most of the stringers lack of clip attachments to the rib: the stringer is connected the rib only by means of the skin -a strip of skin 1.6in length-.

Best regards

 

[rb1957]

the stringers aren't clipped the the ribs ???
i guess your column allowables are pretty low.

as for skins as plates, i hope "they" have done something about the skin/rib/stringer intersections. by having bending freedoms on all the elements you are welding everything together. this is not particularly realistic (rib normal bending is welded to stringer torsion, as an example), particularly if you don't have stringer clips; but adding these releases is usually too much work (!) so you end up messing around with loads that really exist (producing lovely looking reports, with +ve MSs, that substantiate squat)

 

[Andries]

Thanks for a very interesting thread. I have a question for Crackman: You refer Trajano to the Boeing FE manual. Is this document available outside Boeing? I would expect it to be restricted to Boeing personnel.

Regards

Andries

 

[crackman]

Andries

As with most OEM data, it probably is but it doesnt have a restricted notice that I can recall. I have seen it at several places I have worked at and it is a very thorough handbook (as all Boeing manuals are) with tried and proven airframe methods for building FEMs. It provides very specific approved modeling methods for wings, fuselages and provides the specific details of how to model stringers, skins, ribs, frames, etc. Only manual I know of with this quality and detail level. Most FEM manuals are generic and of little use in building airframe FEMs. I spent several years building coarse grid FEMs and it was a very useful learning tool along with working with older experience stress engineers (who actually understood load path).