Saturday, April 14, 2012

Improper Installation May Yield the Fastener


U-lock bracket must remain in this position

This post emerged from my efforts to get the u-lock bracket in the photo above to stay where it should, in order to not impede the locking of the lock itself. As long as a gap remains between the bracket and the lock, as in the photo, the key turns with ease, and all is well. However, over time, the bracket was scootching up the lock and making it hard to turn the key. These keys have a reputation for breaking if you turn them with too much force (appropriate to this post, as you shall see in a moment), and plus, I just like the silky smooth effortless feel of a key turning easy in the lock, so I had to figure out how to get the set screw in this bracket to do its apparent job.

Which requires, when you get right down to it, a) positioning the bracket correctly, then b) applying the required torque to the set screw. And away we go.

How much torque is appropriate for a given fastener? It sounds like such a simple question. In the sense that, there should be a simple answer for the home bicycle mechanic. Yet, I've found out myself and posted here previously about it, there often seems to be a blurry and hard to locate sweet spot between too loose to do the job, and so tight that the bolt or fastener fails, or breaks. A real hint of what you can get up against comes from the experience of the surprise when the friction of a lag bolt going into wood becomes greater than the failure torque of the bolt, even before the head of the bolt contacts the surface of the wood. How does this even happen? And what chance do you have of preventing it?

With some late-night Internet browsing and reading, I found some good resources that provided insight into the real world engineering and factors related to these questions. I will list the better ones below, but to cut to the chase, I found a brilliant and thorough reference which set me straight, and gave me the grim truth, unadorned: the Fastenal Technical Reference Guide s7028 pulls no punches. It's a bit technical, in that it's a technical reference guide giving information about the proper usages of bolts and fasteners, but it's exactly what I was looking for. I read almost all of it, and couldn't put it down once I started. I was like a man whose eyes were opened with knowledge. When that happens, you don't always like what you see, on the one hand, but on the other, you come out the other side better equipped to deal with reality.

To sum it up, here's a direct quote which was eye-opening:

No two bolts respond exactly the same to a given torque. There are numerous “real world” complications. Things such things as dirt in a tapped hole, damaged threads, hole misalignment, and numerous other factors can absorb a large amount of the input torque and will result in a substantial loss in the preload which was determined. Some of the other common variables affecting the K factor may include, but not limited to:

• Hardness of all parts
• Types of materials
• Class of fit
• Plating, thickness and type
• Surface finishes on all parts
• Manufacturing processes, such as cut or rolled thread
• Washers, present or not
• Type of tool used for tightening
• Speed of tightening
• Which is torqued, the nut or the bolt
• The number of times the fastener was used
• The type, amount, condition, method of application, contamination, and temperature of any lubricant used


Then a follow-up quote which in one sentence hit me right between those open eyes:

However, even perfect input torque can give a variation of preload by as much as 25 - 30%.

Why was that sentence so enlightening? Because based on what I had learned up to that point, I had concluded that the various experts were telling me that proper torque spec tends to be around 65 to 70% of the failure (breakage) torque, or, 70 to 75% of the yield (bolt stretched beyond the point of permanent deformation) torque. 

Put another way, torque it until it breaks, then back off 30%. The problem which quickly becomes apparent, is that the uncertainties, even if you have a manufacturer-provided torque specification, exceed the 30% stated target force reduction. That is, even if you know how much force should be used, the uncertainties using the tools and materials available to the home bicycle mechanic are so great (on paper anyway) that going by measurement alone may result in either breaking the fastener anyway, or not getting it tight enough. And from personal experience, breaking a bolt with a torque wrench feels pretty absurd, when it happens. So, one more quote from the excellent Fastenal reference, then I will wrap up this post:

The main concern is to what extent must the fastener’s preload be known. Numerous methods are available to the public, which, directly or indirectly, measure preload. If bolt tension is desired with a high degree of accuracy, the torque wrench is not the answer. The following, which will be discussed later, are highly accurate and expensive answers:
• measuring the actual elongation of the bolt
• hydraulic tensioners
• strain gages
• ultrasonics

A micrometer is the simplest tool used to measure bolt stretch. This tool can be used only if there is access to both ends of the bolt before and after installation. Also, since both ends will probably not be parallel, several measurements, at different points around the circumference, must be made and the average taken for the final measurement.

The point to remember with the use of the stretch control method is that it is a very precise tool available for evaluating bolt tension. We are trying to measure a change in the length of the bolt of only a few thousandths of an inch. Tremendous skill will be required to determine the exact length dimensions.


Wow. Seriously. That quote shot truth into my head so directly that I couldn't sleep after I read it. It gave me momentum to plow through the mind-blowing sections about washers that indicate bolt stretch. So THAT'S what's going on here. Leading to the OSG Fastener Engineering Dept finding for home bicycle mechanics:

The uncertainties inherent in fastener installation lead to the conclusion that failure is always a significant possibility, and there's not much that a home mechanic can practically do to prevent failure, either fastener breakage or field failure, all the time. Even when following manufacturer specified torque with a calibrated dial indicator torque wrench, real world complications and factors can still cause you to break the bolt with too much torque, or fail to create enough tension to do the job, with too little torque. The best you can probably do is to tighten the bolt as close to specs as possible, without breaking or yielding it, or the parts it is joining. Which isn't much consolation, really, but it's something to think about, and to shoot for. And don't get me started on stripping the head, or galling, or fastener corrosion.

Bracket solution: tight as practical, and used blue Loctite

On the other hand, in practice, based on experience, the combination of tolerances, slop, uncertainties, and "safety factors" engineered into bolts and parts, along with "feel" and experience, most often seem to work in your favor, and most bolts would appear to do their jobs without much complaint as long as you get them tight enough.  It would be helpful, however statistically or mechanically insignificant, for providers of torque specifications for bicycle parts to indicate dry or lubricated, and if lubricated, what type of lubricant, if only to ease the mind of the mechanic a bit about one of the many factors of uncertainty.

Steel is pretty forgiving of real world factors, which is one of the main reasons we use it for so much, including fastening part A to part B on a bicycle. However, when the torque wrench clicks, it's a good idea to stop.

Other useful references I found:

http://www.brightonbest.com/catalog/UK/Socket_Screws_Metric.pdf
http://www.parktool.com/blog/repair-help/torque-specifications-and-concepts
http://www.parktool.com/blog/repair-help/basic-thread-concepts
http://en.wikipedia.org/wiki/Thread-locking_fluid
http://arp-bolts.com/pages/arpultratorque.shtml

http://www.machinetoolhelp.com/Repairing/bolt-torque-chart.html

also see my blog disclaimer

9 comments:

  1. Famous last words: just another quarter turn.

    ReplyDelete
    Replies
    1. Keeping my eyes open for surplus NASA torque wrenches for sale, now that the shuttle program has ended.

      Delete
  2. Email me if you need wisdom from the Rocketdyne fastener book. Yes, they torque fasteners up to 80% of yield. And the data says there is scatter of 4X for a given torque and thread lube. The magic is that the 80% number occurs only when you get max load for the torque applied.

    ReplyDelete
    Replies
    1. I imagine some of the tools specified in the Rocketdyne fastener book would get the job done. Some of the references here also discuss the lack of repeatability for the same fastener through several insertion/removal cycles. I am leaning toward Darin's method (see below).

      Delete
  3. My method is to close my eyes real tight and just give a little grunt as I turn. That seems to work at least 6 out of 10 times.

    ReplyDelete
  4. Steve A, how much does the torque vary according to lube? Does viscosity make a difference?
    In laymans terms please.

    ReplyDelete
    Replies
    1. limom, check out the lubricant data on the ARP site, it's a sobering validation of the variability involved in this.

      ARP lubricant info

      Delete
    2. Also, the inimitable Pocket Ref includes a table called "Effect of Lubrication on Torque" which shows a "torque reduction rating" for fine and coarse threads, from no lube, through SAE20 oil, thru white grease, graphite+oil, and dry Moly film. The torque reduction goes from 38% to 66%. In other words, if your torque wrench is set at 80% of yield dry, lubricant more than has the ability to enable yield or even failure at that torque.

      Delete
  5. Also, and I will try to make this my final addendum, since the stretch acquired through preloading the fastener which is required for the fastening job to be done is as small as a few thousandths of an inch, a bolt which is too long for its hole by as little as that can prevent the fastening job from being done by bottoming out, no matter how hard you crank on it. This is true for stem face plate bolts, which may turn out to be just annoying or cause you to ruin the stem, to a ring gear in the rear differential of a 4WD full-sized Blazer being driven hard up a rocky stream bed. A final quote from the Fastenal reference is relevant: "The X indicates that the threads are excluded from the shear plane."

    ReplyDelete

Please feel free to comment here, almost anything goes, except for obvious spam or blatantly illegal or objectionable material. Spammers may be subject to public ridicule, scorn, or outright shaming, and the companies represented in spam shall earn disrepute and ire for each occurrence.