Diagonal Semi Tension Field Analysis

[Kristina Sornikova]

Dear Frieds, I try doing according part 2 of section C11 in Bruhn of a fuselage skin shear panel type analysis, there is a hole and edges have rods.

I have shear panel average shear stress from f06, and edge rod axial stresses also. I only use the stringer axial stress, not the hoop direction stiffeners/bulkhead caps.

I have all otehr values, like Long Edge, short edge, radius of curvatures etc.

But when I use all formulas, I get Rc values much more than 1.0 and also much less than 0.0, same story, for Rt value some I get negative values, but higher than -1.0.

Bruhn says Rc is always less than 1.0 and Rt is always greater than 1.0.

So I confused. Am I doing wrong?

 

[TimMcK_Aero]

Check the negative sign for B/A is inside the brackets.

Tim

 

[rb1957]

maybe the panel is no good ?

another day in paradise, or is paradise one day closer ?

 

[Kristina Sornikova]

Hello Tim, yes the -ve sign inside brackets, formula in excel: [-C2/C1 + sqrt((C2/C1)^2+4)]/2

C1 C2
Fccr/Fscr fc/fs

Problem is in some rods, axial compressive stress much higher than panel shear stress. So C2 much high -ve, then Rc is also much high -ve.

I also use the sign of CSHEAR element average shear stress as it is in the f06 file. So some positive some negative.

But why some elements have average shear stress -ve same number as max shear stress?

Example:

S T R E S S E S I N S H E A R P A N E L S ( C S H E A R )
ELEMENT MAX AVG SAFETY ELEMENT MAX AVG SAFETY
ID. SHEAR SHEAR MARGIN ID. SHEAR SHEAR MARGIN
40227 1.081132E+01 -1.023650E+01 40228 7.187690E+00 -7.187690E+00

rb1957, I don't know it may be no good.

 

[rb1957]

have you tried to calc thru an example you know works ? or a textbook case ?

if your algorithm is correct, and your inputs are correct, then the conclusion is correct (and the panel fails).

I don't know your shear question ... maybe try some simple test cases ? maybe read the user manual ?

another day in paradise, or is paradise one day closer ?

 

[Kristina Sornikova]

rb1957, can you kindly give me a simple model with some calculations? I still learning so not sure where to start.

 

[rb1957]

bruhn has several ... and someone should have a solution set to the problems.

Or where ever you are now, have they done a similar analysis in the past ?

another day in paradise, or is paradise one day closer ?

 

[Kristina Sornikova]

OK I will study them rb1957, thank you.

 

[VN1981]

Hi Kristina,
If you figure out how to do Diagonal Tension analysis using Nastran, please do share it here if you can.

 

[rb1957]

I think it's simply taking the NASTRAN stresses and plugging into the hand calc.

I don't think FEA can manage the load redistribution after shear buckling of the panel.

another day in paradise, or is paradise one day closer ?

 

[Kristina Sornikova]

This type analysis looks like most challenging, specially with skin and stringer with cutout type. Still learning, but difficult to understand.

 

[Kristina Sornikova]

OK so here is one example:

I have a fuselage skin shear panel element (CSHEAR), with rods at the top, left and right, no rod at bottom. In this case, large compressive stress is there in rods, but panel shear stress small.

Panel A (long edge) = 10.753"
Panel B (short edge) = 9.881"
Panel Radius of curvature R = 44.628"
Thickness t = 0.016"
Panel Limit Max Shear Stress (from f06 CSHEAR output) fs_lim = 215 psi
Panel Ult Max Shear Stress fs_ult = 322 psi

Average Limit Rod Axial Stress fa_lim = -4840 psi
Average Ult Rod Axial Stress fa_ult = -7260 psi

Poisson's ratio μ = 0.33
Young's Modulus E = 10.7e6 psi

Z = (B^2)/(R*t)*(1-μ^2 )^0.5 = 129.08

A/B = 2.0 (cutoff at 2.0 for A/B>2.0)
R/t = 2789.25

From charts:
Shear Buckling Factor Ks = 90.0
Compr. Buckling Factor Kc = 25.0

Ultimate Shear Buckling Stress Allowable Fscr (psi) = [(K_s*π^2 E)] / [12(1-μ^2)]*(t/B)^2 = 2330.47 psi

Ultimate Compr. Buckling Stress Allowable Fccr (psi) = [(K_c*π^2 E)] / [12(1-μ^2)]*(t/B)^2 = 647.35 psi

C1 = Fccr/Fscr = 0.28
C2 = fa_ult/fs_ult = -22.55

Rc = [-C2/C1 + sqrt((C2/C1)^2+4)]/2 = 121.80

Rt = 0 (as stringer stress is compression)

Critical Combined Shear Shear Stress:
Tau_cr = Rc*Fscr = 283846.89 psi

Limit Buckling Ratio = fs_lim/Tau_cr = 0.00 (<1.0)

This mean no diagonal tension check need, as the panel not buckle.

But does it make sense? Rc should always be less than 1.0 but I have +121.80?

 

[rb1957]

part of your problem is (I think) using a CSHEAR, particularly with no CROD on one edge ... no compression effective skin ??

I understand some of the reasons for using CSHEAR elements (ie you only want it to carry shear loads and have lumped effective skin in the edge members). But this creates a problem with diagonal tension ... maybe use the average of the two edges as a panel compression stress ? Also, I don't believe the analysis likes biaxial loadings, not sure the typical response.

I'd think that a 322psi ultimate shear stress would mean than the panel is none buckling. That's some Ks you have ! Why is A/B = 2 when it is clearly closer to 1 ? And check the effective skin for compression (including irb). I don't think you have a problem, <8 ksi on a stringer or any respectable compression member should be ok (for crippling and long column).

C2 ... it's a long time since I read Bruhn but it looks to me as though you're using a negative since for compression stress, and I think compression is assumed so maybe you should use a +ve value ? just a thought, if his worked examples follow what you're doing then ok.

another day in paradise, or is paradise one day closer ?

 

[Kristina Sornikova]

Thanks much rb1957, I can use average of three rod axial, but some rods have tension and some compression.

Lot of times there is no stringer between two cshears, maybe the curvature is too big for one element along curve of skin. So this case I have only three sides rod. FWD and Aft rod maybe bulkhead caps, or rings or frame caps, and top maybe stringer or stiffener or longeron. Not sure why it was done like so, old model.

Regarding sign of compressive stress, Bruhn does not say -ve, it only says fc, so don't know.
If I use +ve for fc, I still get Rc -40.58?? It is 'bonkers', I know. But I thought 0.0 <= Rc <= 1.0 always? Or can it be large negative numbers also?

You are correct, for this A/B was 1.09, I copied wrong one.

 

[rb1957]

you may want to check the behaviour of CSHEAR without complete set of edge members; I suspect it may get "unhappy" or worse, NASTRAN may AUTOSPC the problem away and there may be large forces leaving your model (via the SPCs). look into the manual, run a test case (take a 3x3 model of CSHEARs with CRODs all round, each element, put it is pure shear, you should get the right result; then remove one of the CRODs.

I think Bruhn says compression stress applied is 32ksi (not -32ksi, since he's assuming compression).

If you have a large panel then
1) use CQUAD, or
2) use dummy rods (which'll probably "explode").

Make sure you're analyzing the structure and not the model. if you have a large panel modelled as many elements, it's the panel that's important (not the elmeents). Figure not what the elements are telling you of the panel (like average shear stress), and stress the panel geometry.

another day in paradise, or is paradise one day closer ?