Metallic Joint Bearing Strength 4

koopas,

You have just asked the question that plagues engineers constantly. My best suggestion would be to use a fastener for which you do have published allowables, or at least geometrically and physically matches/exceeds those of a published one.

The failure modes of countersunk fasteners are significantly different from protruding head ones, which is why Fbru*t*d is unconservative.

There are methods available to approximate the failure load of countersunk fasteners, but they are only be valid for preliminary design. Any values used would need to be substanciated by test at a later time.

Regards,

jetmaker

Koopas, clearly you've been reading what I e-mailed you the other day. The statement at the top, that it is for "preliminary sizing only", is true.

One of the most annoying problems in my job, that I can't get rid of, is the absence of public data on NAS1097AD rivets. Mil-Hdbk-5 has data for MS20426AD and NAS1097E only. When repairing thin-skinned aircraft with little countersunk rivets, you don't have lots of choices.


Steven Fahey, CET
"Simplicate, and add more lightness" - Bill Stout

Ok SparWeb,

You have my curiousity peaked. After a little research, I found no allowables in our databases to justify the use of NAS1079AD rivets. So how do you do it?

Regards,

jetmaker

Jetmaker and SparWeb

Thanks bigtime for your posts.
I bet thousands of hours have been wasted by engineers looking for "simple" design information that does not exist.
It is nice when experienced design engineers state "the facts" so the design can continue by alternate information/methods or very large safety factors used to "cover up" this industrial ignorance - allowed by companies that are quite willing to sell you their product to make more profit but not willing to spend one cent to provide elementry but wide ranging design information on their products.

Jetmaker,

Thanks for your quick reply. I wasn't specifically referring to CTK fasteners. Since you wrote

"The failure modes of countersunk fasteners are significantly different from protruding head ones, which is why Fbru*t*d is unconservative.",

are you implying that using Fbru * d * t is reasonable to determine the bearing allowable in protruding head fastener applications in single shear joints?


Sparweb,

If you have access to the Douglas SRM's, they include single shear joint allowables for NAS1097AD rivets in most aluminums. I am looking in the 717 SRM, 51-40-07, figure 3 (10 sheets). Hope this helps.

Alex

koopas,

Yep, do it all the time. I believe that the charts in Mil-Hdbk-5 are based on a simple calc of Fbru*d*t. However, I have not looked at them, so please don't assume this to be correct, verify it for yourself. I know Niu uses a Stress Intensity Factor (or similarly called) for single shear joints, but if the hole is a tight fit, and respective stack-up ratios are used, I feel this factor is not necessary, albit conservative.

One exception to this rule is when rivets are used. There, for D/t ratios greater than approximately 2.5, there is a reduction taken.

Regards,

jetmaker

Hello Jetmaker,

After doing a bit of reading tonight, I've clarified a few of my own confusions...

IF

1. fastener is solid, not hollow or multi-piece
2. fastener has protruding head
3. fastener has tension protruding head for bolts
4. t/D > 0.18

(ref. Mil-hdbk 5J, 9.7.1, pp 9-142)

then, and only then, you can use Fbru * d * t to determine the joint's bearing strength. Jetmaker, in that respect, you're absolutely correct in that the Mil-Hdbk 5 / SRM joint tables are based on that simple bearing formula.

If the criteria above are not met, then, you need to rely on test data, because the joint may not develop the full bearing strength of the material.

The Mil-Hdbk 5 joint allowable tables are primarily for fastener types that require the use of test data. The SRM has both.

In any regard, to sum up the discussion above: if the joint meets the four criteria above, then, it's calculated via Fbru*d*t. If the joint doesn't meet the criteria, (i.e. flush rivet/bolt), then, the bearing allowable is found by testing. That means that blind fasteners, which do not meet criterium # 1, always have their allowables found by testing. That also explains why the SRM's do NOT allow for extrapolation outside the published tables for flush or blind fasteners (interpolation within all published tables is allowed though).

So, if your fastener meets the four criteria, and its joint allowable is not published, you should get excellent results via the bearing formula.

If you're dealing with a countersunk fastener, a blind bolt, or a low thickness to diameter ratio, then you're out of luck. However, I found some notes that allow you to determine an unknown bearing allowable using the joint allowable of a similar configuration and ratioing it by the respective Fbru's. Do you guys/gals use this technique? (btw, Sparweb, do you use the CTK bearing calculation you sent me in repair substantiation?)

Anyway, it's almost midnight and I've rambled for quite a bit. I guess I've cleared up some of my confusion by typing...in fact, I can say that I've learned quite a bit tonight, provided I am not too far off base.

Your comments are welcome.

Good night,
Alex

koopas,

Great job on the research.

The ratioing deal is only appropriate for different bearing materials (i.e. 2024 to 2224). However, it is conservative to use the lesser properties since the actual failure mode might be slightly different. For example, using a material which is substancially weaker in the ST direction yet stronger in the L direction. You might end up with a different failure mode (say fastener head pull through, vs bearing).

Keep up the good work.

jetmaker

Good day Jetmaker,

Do you mean that the ratioing deal is only valid for materials within their alloy group (i.e. within the 2000 aluminum group)? In other words, you couldn't use the ratioing principle to determine the unknown allowable in 7075 using a known allowable in 2024?

I am not sure I understand when you say "it is conservative to use the lesser properties since the actual failure mode might be slightly different". What "lesser properties" are you referring to? I am also not understanding well your example...

I would be grateful if you could re-explain. (I am not the fastest learner).

Oh..one last thing...what's fastener head pull through? Is it when the sheet metal lifts up and "pulls through" the fastener head? So it's a sheet material failure? (i.e fastener and fastener head remain intact?)

Thanks a bunch,
Alex

koopas,

It would be best to ratio within the ally group. The reason being that there can be significant differences between the behaviour of different ally groups. But even for similar alloy groups, I would check the stress-strain curve to make sure the materials have similar behaviour.

What I was trying to get at, is that it is better to show the design good with the chart values rather than ratioing them up by the Fbru values. The reason is that a plate material might have much higher Fbru values than sheet, but the shear strength may not be much better. If the failure mode determined from testing was anything else besides bearing, then you might have lower actual strength than what the Fbru ratioing calculates. It just requires a little thought on why you are allowed to use the Fbru ratio as an acceptable means to improve allowables.

Correct, fastener pull thru is just as you described.

Keep up the good work,

jetmaker

Koopas, et al.,

I've been away (long weekend here in Canada), so I haven't kept up. Excuse me if I sound like a johnny-come-lately.

I don't have a Douglas SRM. Mind scanning a few pages (the NAS1097AD ones, specifically)?

In the past, I've used the Mil-Hdbk-5 allowable for NAS1097E rivets, scaled them proportionally to the F[sub]su[/sub]'s of E and AD materials (7050-T73 and 2117-T4, respecitively) and the allowable I get looks clearly on the small side. Like Jetmaker recommends, this shaves the allowable down, not up, but I am still wary of the "knife-edge" line on these charts. I'd be interested to see how my short-cut measures up to the Douglas data.



Steven Fahey, CET
"Simplicate, and add more lightness" - Bill Stout

Hello Sparweb,

Welcome back!

I presume you wanted to determine the bearing allowable of the NAS1097AD rivets in aluminum.

Why did you scale the "E" rivet's bearing allowable to the Fsu's of the rivet materials instead of the Fbru's of the sheet metal materials? Wouldn't it make more sense to scale to Fbru since you're concerned with bearing strength? The Fsu's of the rivet materials would correlate to the shear strength of the rivets, both of which are known.

Alex

Regarding the 1097AD rivets,

I have an old Grumman reference with test data for these particular rivets. I'd be happy to scan and send what I have.



--
Joseph K. Mooney
Director, Airframe Structures - FAA DER
Delta Engineering Corporation

Koopas, JoeKM,

Thank you both for the data you provided. It is immediately practicable to a doubler I'm working on.

Now the fun begins: The allowables in the Douglas SRM do not agree with those in the Grumman manual. Not even close.

For example, a NAS1097AD4 in 0.040" Clad 2024 is allowed 359 pounds/rivet in the Douglas SRM, but only 292 pounds in the Grumman test data.

The Grumman data is hard to question: It's a photocopy of a report with Grumman's name on it, and references to the test report at the bottom of the page. About as bullet-proof as it gets. The Douglas SRM came to me as some kind of cut-and-paste Word file with a few errors in it. Hmmm.

Which would you trust? And yet, my margins are so low that I would rather use the Douglas data.

PS: Koopas, I used the relative F[sub]su[/sub]'s because there is no relative F[sub]bru[/sub]'s when you're comparing a joint in 2024-T3 sheet with data applicable to a known joint in 2024-T3.

Steven Fahey, CET
"Simplicate, and add more lightness" - Bill Stout

Sparweb,

You're right. Once again, I should think a little more before typing on the keyboard.

Ratioing the Fbru's would only be applicable when you want to determine the bearing allowable of a joint for which no test data is available, by multiplying the known joint allowable of the fastener in a similar alloy by the ratio of the Fbru's (Fbru of unknown joint allowable's material / Fbru of known joint allowable's material).

In your case, you wanted the allowable of an AD rivet in 2024-T3 based on the known allowable of an E rivet in 2024-T3. Even though the latter available is your "known allowable", the Fbru's ratio to unity since the material is the same, rendering the calculation useless.

I am guessing that you used the ratio of the rivets' Fsu's because it would give you a "knock-down" factor to reduce the known "E" rivet allowable, thus accounting for the reduced shear strength of the "AD" rivet.

Am I understanding your process correctly? Could you use another "knock-down" factor (one that's not based on the rivets' Fsu's) to account for the AD rivet's reduced strength?

Take care,
Alex

PS: I can assure you of the genuineness of the Douglas data. Cutting & pasting reduced the number of pages to be sent. BTW, the OEM confirmed a value of 300 lbs. for the allowable that's listed twice. It is indeed strange that the Grumman reference lists allowables that are almost 20% lower. What could cause such disparity?

Published and accepted data (i.e. SRM) is valid data. It is always nice to know where the data comes from, or how it was calculated, but if it accepted data, then one must have faith in it's backing, otherwise, you'd need to perfom experiments to calculate gravity everytime you work with dynamics.

Thanks all.

jetmaker

Koopas,

"Knock-down" is a good term for it. In light of the Grumman test data, the Fsu ratio is definitely conservative.

As for the disparity in the results, I have a few comments, but no conclusions. First: I don't have a definition for the "drivematic" rivets, but if they are referring to blind rivets, then they're not knocked down enough. Second: lining these numbers up beside the MS20426AD numbers for the same sheet in Mil-Hdbk-5, plus the Grumman test data, and something just doesn't look right. Grumman looks very conservative, compared to both of the other two.

I've heard many comments about smoking rivet heads, and many have been directed at these particular NAS ones. The small head may be compromising the "pull-through" resistance or the "pull-up" tension that you expect with other universal-head rivets. IMHO, Douglas set up their test differently; say, a double-shear load test, for example, which could have yielded the more generous results.


Steven Fahey, CET
"Simplicate, and add more lightness" - Bill Stout

Sparweb,

I calculated the AD4 allowable in 0.040" Clad 2024-T3 by multiplying the "E" rivet allowable by the ratio of Fsu's. I am coming up with the following:

NAS1097E4 in 0.040" Clad 2024-T3: 437 pounds (Mil-Hndbk 5J)
Fsu of "AD" rivet: 30 ksi (Mil-Hndbk 5J)
Fsu of "E" rivet: 43 ksi (Mil-Hndbk 5J)

437 x (30/43) = 305 pounds

It would appear that ratioing the Fsu's yields an allowable close to the Grumman number of 292 pounds.

In your last post, you mention that "In light of the Grumman test data, the Fsu ratio is definitely conservative."

Isn't the Grumman test data the most conservative of all, followed by the Fsu method, with the Douglas data being the least conservative?

Oh, and what's a "smoking rivet head"?

Thanks,
Alex

Touché. I didn't calculate and compare all of the allowables as thoroughly as I should have.

The Delta-F[sub]su[/sub] method and the Grumman data exchange places a few times.

see
Steven Fahey, CET
"Simplicate, and add more lightness" - Bill Stout