Bolt failure thoughts hardcore engineers please!

WOS Sean,
If you mentioned witness marks before I missed it. There were NO witness marks on my crown gear bolts, nor was there any indication that the crown gear had slipped against the differential case, if there had been I would have replaced all of them.

One thing you should be aware of, the torque spec on the crown gear nuts was significant reduced somewhere between the GT4 manual & the QV/328 manual. So far I haven't found the change in the FSBs. Since you run 2Vs, It may be that the reason the torque was reduced was because F* was seeing some spontaneous bolt failures & concluded that the torque spec was too high.

It does seem to me that, if you've got enough metal for them on the housing face, that 2-3 10mm or so diameter pins between the crown gear & the case would be cheap insurance!

 

Finnerty,

We in the diy'er catagory are not privy to bolts built to our specs. We rely on what are considered normal bolt torque ranges based on fastener size and material grade so for example carrol smith in the rookie's guide to bolts says a 7/16-20 AN bolt should be about 40 ft/lbs and a 10mm 12.9 class bolt should be at 46 ft/lbs. So are these established quoted numbers on the 50% preload side or the 90% preload side? Then what happens to what you should have in torque when threads have locking compound (locktite) or oil on them? Often car makers do not list wet or dry torque specs. Is there a rule of thumb like back off 10ftlbs for wet threads or something like that? And does measuring bolt stretch nullify the effects of lockers and oil on threads so that you can disregard the actual ftlbs torque on the nut all together?

 

FATBILLY ----


I hear ‘ya… And, I couldn’t agree more.

In the Aerospace world, we have the luxury of access to mounds of test data and lab studies used to develop specs and procedures which allow us to torque and pre-load very precisely and with excellent repeatability. Not to mention that we do our wrenching under tightly controlled conditions with tools and fasteners that are meticulously characterized and of the purest pedigree. Decades of history have given us accurate torque tables for a wide variety of fastener types, thread conditions, materials, and surface treatments (dry, lubed, with thread sealant, with thread locking agents, etc.). So, we can confidently go to 90% of allowable yield in any configuration.

In the Automotive world as well as in the “real” world (in my garage and in yours ), it’s impossible to have that degree of precision. So, you will find that nearly all the torque guidelines you will come across have been derived with the goal of achieving values aimed more at the 50% of yield target. And, the carmakers generally use that 50% target for specifying their torque values as well. It gives plenty of clamping force (if the joint design is correct) and reliability for auto applications, and it allows for a healthy margin of error in the installation process. Exceptions are in the cases of specs for assemblies subjected to a lot of heat cycling, big temperature changes, and mating components and fasteners made of dissimilar metals (like exhaust manifolds and aluminum cylinder heads) ---- where the manufacturer will call out lower torque values than normal.

Also, all the standard torques are assuming “dry” and “clean” fasteners and threads. When you add any lubricant or thread-locking agent (when Loc-tite is wet, it provides lubrication during assembly), it requires a different torque to get the same pre-load as a dry fastener. When a factory manual calls out to apply lube (such as anti-seize) or thread-lock to a specific fastener, it should also be calling out a unique torque in that instance.

As far as torqueing fasteners go, the biggest problem the DIY’er or even the professional mechanic runs into is that they are usually dealing with used / dirty parts and fasteners. Attempting to effect an accurate pre-load by applying a specific torque (regardless of what torque spec is used) to a dirty, corroded, or badly worn / damaged set of threads is a straight waste of your good time.

For critical joints --- and, I would include the differential assembly here in question in that category ---- it’s vital that you take the time to clean or restore all threads before re-assembly. Or better yet, don’t hesitate to install new fasteners, so long as the cost is practical. Assemble them dry (unless the manual says otherwise), and torque to the factory provided specs if available. If not, use whatever published standard torque values you trust --- they'll all get you to reasonably the same target.

If you want to add locking compound (a very good idea for any rotating assembly, by the way ), a good rule of thumb is to go with the standard torque reduced by 15-20%.

 

Finnerty,

Thanks for the very good information!

 

Yes. When using any fastener that has a built-in method of indicating its tension (stretch marks, strain gauge, etc.), this supercedes monitoring the applied torque or accounting for any surface treatments or contaminants.

As an example, for hyper-critical joints in Aero instruments, we often use "NSI" bolts which have integrated strain gauges within the fastener body. This allows us to apply true pre-load and stress with accuracies to within 1/10th of a percent of any target value. The torque specs on the drawings simply read, "tighten until the correct strain value "X" is obtained" .

 

...my pleasure.

Cheers

 

Very interesting to read how these problems are tackled in the aerospace industry. Apart though from the obvious differences between the 2 applications, there is one that I haven't seen mentioned, which is weight. Critical in aerospace, and looking at how a 308 is constructed, not so in the roadgoing Ferrari of the 80-ies. This thing is really over-engineered, from where, I suppose, stems its' longevity and reliability.

What would need a closer look is the changed torque spec between the 308 GT4 and the 328 diff mentioned earlier. I don't have these specs, but if there is a significant difference with otherwise identical parts it begs a question.

Also, if losing bolt-heads is a relatively common problem, what about a simple solution like using a bolt with a larger radius from head to shank and rounding the top of the holes somewhat? If the torque spec has been changed dramatically, the clamping force must be more than enough, and one might conclude that the choice/design of bolt is wanting?

 

The torque was reduced by about 25% from the GT4 to the QV:

308 GT4 spec: Fixing crown gear to differential: 10.4 kg-m (76 ft-lb)308QV & 328 WSM pg E25: "...tighten the nuts(2) at the prescribed torque of 8kg-m (58 ft-lb) fig 20.

By comparision, the M10x1.25 half shaft flange bolts are also torqued to 8kg-m.

BTW, the WSMs are very clear that the nuts are NOT to be re-used. This is consistent with what I saw, the nut threads will start to yield under the torque.

Why would you necessairily decide that the bolt was 'wanting' just because the spec was reduced? Since roughly similar bolts in another high stress application have been being torqued to the lower spec all along, I'd tend to suspect that the GT4 torque spec was excessive.

As for a bolt with a larger head-shank radius, that's easy to hypothesize, but have you ever tried to source such a bolt? I couldn't even find an alternate source for the OEM bolts, the M10x1 thread pitch & length is not an off-the shelf bolt. Oh, yes, the 19mm head hex width is not the std head for a M10 bolt either! I did find 4 or 5 custom bolt mfgs in the US that would make bolts to the specs. They all quoted NREs of around $1k + around $100/bolt!!!

Engineering wise the larger radius is a good suggestion, & I have reason to believe that Ferrari intended the bolt to have such a radius: The exerpt from the Ferrari drawing I obtained from Ferrari.uk clearly showed a radius (but no numeric value for it) between the body & the head, also the drawing specified that the head have a washer face.

However, all the crown gear the bolts that I've seen all have a pretty sharp corner between the bolt body & the head, and NO washer face. I've speculated that the aftermarket bolts are being made by cutting down the bodies of M12 bolts & rethreading them.

I also determined that the top of the bolt holes was already rounded. I was looking at cutting down & rethreading longer body M10 socket head cap screws that do have a radius between the head & body. I actually modified a M10 socket head cap screw this way & tested it to failure by overtorquing it. (See the thread I referenced earlier). The nut threads stripped out, but while the cap screw threads noticably deformed, they were a long way from failing.

If I were worried about the bolt heads popping off, I'd make my own bolts out of M10 socket head cap screws. They're class 12.8, the steel is really tough, but can be threaded with a sharp carbide point.

 

While it certainly wouldnt be "correct", what about crossing over to an appropiate sized AN bolt?

 

If the original drawings call for a radius and same is not present at the bolts used, we may be on to something.

I like the idea of the socket head (allen?) bolts, the only thing that would worry me is the size of the head. The bolt relies on the friction of the head, reducing the headsize may not be a good idea. Also, when torque'd to the same spec, the specific pressure of the head on the carrying surface increases dramatically, which may lead to surface distortion and subsequent failure to function.

 

Paul,
What is are AN bolts? I can't find any specs for them in Machinery's Handbook.

f308jack,
The differential flanges & ring gear have a substantial amount of steel around the bolt holes, I don't think distortion due to increased presssure at the hole would be an issue.
Since the wheel flanges use 10mm socket head cap screws with the same 8 kg-m torque spec, I'd be comfortable using them for this application. IIRC I researched the cap screw specs in some detail when I was looking for alternatives to the std differential bolt & this application was easily within their specs.

 

But they are by far no more 12.8. class after modifying. 12.8 is not only achieved by the strength of the steel. On class 12.8 bolts the thread is generally NOT just cut, but rolled after threading and heat treatment, what has a kind of forging effect and what contributes significantly to its strength. Therefore cutting through that ' forged ' structure is not a good idea.
One should generally stop modifying bolts with a class higher than 8.8 for above reasons.

Best Regards from Germany

Martin

 

AN bolts...More commonly referred to as MIL-spec

 

And one of the core gearboxes I tore down had 12.8 bolts installed in half of the locations. This gearbox had suffered some serious trauma. Something literally stripped the drive teeth off of one of the gears. I have no clue how this happend, but it might of been quite a shock on the crown gear.

Did I mention you can see how every factory bolt has an indentation from having the hardness checked at the factory? Some of them 2, like they might not of been drawn back far enough after heat treat so were re-processed.

I'm starting to think this bolt failure is like the axle shaft failure ... very very rare. When I tq'd those bolts up I could really feel them stretching. If someone had been in V's g'box with the broken bolt I'd quit worrying ... unless someone else comments that they've seen this before or "they" tq'd the bolt that broke and know that it tq'd up fine.

Cheers

 

ALL ---

If you want to get your mitts on some "real" bolts, go to this source (or some other, I think even 'Fastenal' can get them),

http://www.awfasteners.com/


and order to the NAS1351 or NAS1352 spec. --- these bad boys are available in strengths up to 180KSI. And, they are made with some actual quality control measures and testing standards ---- so, they won't be defective

 

Martin,
The point that the rolled threads are much stronger was raised when I was investigating alternative bolts. One of the reasons I tested my cut down socket head cap screw prototype bolt to destruction was to see how strong the single pointed threads were.

I lubricated the nut & bolt with silver anti-sieze and then torqued the test bolt until the fastener failed at ~2x Ferrari's torque spec:

~125 ft-lb/170 N-m

At that point the 6 complete threads in the nut stripped out not the bolt! The bolt threads were visibly deformed(see pix in post referenced below), but had not failed, & a nut would still fit on them.

Rember, This was a lubricated test!!! Equivalent 'dry' torque would have been significantly higher!

Assuming lubricated torque is 75% of dry torque for the same deformation, the dry torque equivalent would be about 220 N-m!!!! This is over 2x the Machinery's Handbook max recommended dry torque for a 12.9 10mm fine pitch bolt!

Details of the test including pictures are in post #61 of this thread:
http://www.ferrarichat.com/forum/showthread.php?t=43305

Admittedly this was a sample of one.

I'm not claiming the bolt's threads met the full grade 12.9 spec, just that they were significantly stronger than the threads in the mating nut!

Finnerty,
Do they make metric bolts with custom body & thread lengths? The differential bolts have a much longer body & shorter amounts of threads than the ISO/DIN std 10mm hex head cap screws.

 

AN = Army Navy. They may fall under mil spec, but it was my understanding they were classed differently?

 

Yes and yes.

For metric sizes, the equivalent standard is "NA0069" (and its associated flavours). The material strength for these screws is not quite as good --- but still rated at 160ksi.

The lengths of threaded vs. un-threaded portions of all these screws are governed by standards as well. But, custom threaded lengths are available --- though minimum buys will usually apply to customs. You may have to dig around to find some leftover custom stock from someone else's buy ---- unless you want 1,000 of 'em!!

Otherwise, I'd just go with the standard (which will probably have a longer threaded portion than what you are replacing). Remember that if the joint is proper, there won't be any bearing against the threads anyway .

 

Alternatively, you could use the NAS1351/52 screws if you are willing to have a "bilingual" differential assembly...

Nominal diameter for the 10mm is 0.3937", so you could drop down to 3/8" (0.3750"). Or, if the clearances are generous enough, you could go up to 7/16" (0.4375"). Assuming these parts have no local heat-treating / hardening around the holes, maybe you could even consider opening the holes up to take the 7/16"?

FWIW

 

Thanks to all you guys for the posts. Especially you, Dave. I am certainly not an ME, but this is really interesting and informative stuff. It's great to read and learn from people who know their sh_t!!

 

FERRARI OWNERSHIP ---- "we're all in this mess together" --- LOL