'secondary bending'?

[Solari]

hello,
does anyone know something about "secondary bending" or "secondary tension load"?

for instance if you have a so called lap joint configuration with two splice plates fastened with 3 rivet rows and with a tension stress at the ends of the plates:

i'm asking myself how to calculate or at leat to estimate the resulting secondary bending moments in the plates around the rivets in order to calculate the bending stresses there...

because it is a non-linear problem i expect some differential equations but i hope that there is a more simplyfied and convenient way to estimate the moments and bending stresses, isn't it?

i hope anyone can help me, 'cause i couldn't find anything about it in my mechanics-books...

best regards
solari

 

[JStephen]

What is your application? (IE, old steel structures, aircraft, or what?)

As far as I know, back in the days when they did a lot riveting on steel, no one ever got into that much detail.

You'll notice for example, that you show a 2-dimensional version of what is a 3-dimensional problem. You omit the rivet heads, and likely omit the initial tension in the rivets. And quite often when there were three rows of rivets, they would be offset, not in line. My point is that if you solve your problem exactly, you still have a very appoximate solution to the actual physical situation.

If this is a cylindrical shell, check Timoshenko's Theory of Plates and Shells or other more modern references. Roark's Formulas for Stress and Strain also has formulas for bending of cylinders.

 

[desertfox]

Hi Solari

If it is just a lap joint and not cylinderical as J. Stephen
also states then you wouldn't consider bending in the joint and it would be analysed purely on the basis of shearing the rivets, bearing failure of the parent metals connected with the rivet and tearout of the rivet through the edge of the connected parent metals.

regards

desertfox

 

[GregLocock]

True enough, but just because that's the way we've always done it, and it works, that is no reason not to investigate further.

Roughly the moment will be Ft, which from M/I=sigma/y will cause stresses of the order of t/2*Ft/(1/12bt^3) whereas the axial tension will cause a stress of F/(bt)

so the fractional error is of the order of 6*t^2*F/(bt^3)/F*bt = 6*t^2*F/t^2/F, ie 6

hmm that is an awful lot!



Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[GBor]

If you want a computer-based application, you may want to check out

http://www.materials-sciences.com.

They have a program called SPLICE (

http://www.materials-sciences.com/matlMinds/splice/index.html)

that works EXTREMELY well.

 

[Kenneth]

In single lap shear construction in the aerospace industry this is often referred to as joint "rotation." A web search using those terms my help in your efforts.

 

[Cockroach]

Interesting problem. I found some related information in textbooks on structures and connections. This topic is a bit out of my comfort zone, but try Standard Handbook of Fastening and Joining, 3rd Edition, Parmley. There is a whole section on industrial riveting and related connections. Maybe something there?

Kenneth J Hueston, PEng
Principal
Sturni-Hueston Engineering Inc
Edmonton, Alberta Canada

 

[prost]

There was a dissertation by Fawaz at Delft that discussed this very thing. In addition to just accounting for pure bending (lengths, skin thickness, etc.) you must account for rivet rotation, which in principle is a simple thing to conceptualize, even if it isn't a simple thing to determine a priori. We put together a Visual Basic spreadsheet to compute secondary bending stress for 3 rivet lap joint; if anyone is interested, I can mail it to you, or put it on our ftp server.

 

[panduru]

@prost:
You will be doing a great favor by putting up the file on an ftp server and post the url here for the benefit of interested members of the forum.
Thanks a lot.

 

[Solari]

hi folks,
first of all thank you all very much for your comprehensive answering!

and prost, i'm very much interested in your visual basic spreadsheet! it would be really nice if you put the file on your ftp server!

thank you very much in advance!
solari

 

[prost]

Oh rats, if you want the Visual Basic, you have to pay for a license to some third party company. However, I was able to give you the same thing in spreadsheet that uses the matrix functions in Excel to do all the matrix manipulations. Sorry.
ftp://ftp.apesolutions.com/pub

I put a small picture there to show you some of the nomenclature.

If you don't mind paying $100 for a VB library of LINEAR1 and other numerical routines, then I have instructions for getting that code. But you won't need it, if you don't mind using the Excel matrix functions.

 

[GregLocock]

The units of P are wrong, I think, shouldn't it be lbf/row of rivets, not per inch?



Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[GBor]

There may also be an issue of "load distribution". I know for two rows of bolts, the 'front' row takes about 65% of the load, not 50% like classic analysis might suggest.

 

[prost]

Checking what "axial stress" value is, you see it is "P" divided by the t-outer to obtain the axial stress. Kind of a goofy way to define it, to be sure. The "length" in the "force/length" is the pitch (that is, the distance between fastners in same row), which is set to 1 inch (pitch=1 in.). You're right GregLocock, it would have been better to have it more flexible and included a box for 'pitch'.

 

[prost]

In your model of a two row lap joint--if the front row of a two row lap joint takes more than 50%, where is the other load transferred? Friction? By symmetry the back row must take the same as the front row. The load has to go somewhere.

 

[prost]

the hardest thing to choose in the secondary bending spreadsheet I have put on our ftp site is the 'betas', the plastic rivet rotation (not my turn of phrase, the original authors'). The spreadsheet as provided uses 1 degree (0.017 radians), because that seemed to work well for us. If you have some lap joint experiments of your own, maybe you'd like to use something else.

 

[GBor]

prost,

Oops...I'm remembering a certain fiber-reinforced composite material...not isotropic, homogeneous, beautiful metal!

Garland E. Borowski, PE
Borowski Engineering & Analytical Services, Inc.
Lower Alabama SolidWorks Users Group

 

[GregLocock]

Oh sorry, prost, thanks for posting that, it'll be an interesting comparison with hand calcs etc.





Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[rb1957]

going back to prost's post (18:35 3rd May) ...
if you do a compliance model of load going into a doubler then the outer row carries more than 50% (65% is a reasonable) of the load transferred into the doubler, and the 2nd row carries the remaining load (35% in this case).

but you're right if you have a two splice joint, each row has to carry 50% by symmetry (same result if three rows, again by symmetry) assuming the thicknesses of both splice pieces is the same (if not then the joint isn't symmetrical).