Monday, April 4, 2005

On Rod Bolts, By Mike Gifford

My name is Mike Gifford; I'm "Outrider" in the 4Cycle.com forums, the guy that commented on how happy I was to see a fastener tightening procedure that made allowances for the differences in friction coefficient (and hence applied torque required) when different thread lubes are used on threaded fasteners. At the time, I promised you that I would eventually send you some comments on fasteners and tightening procedures.
By way of background, I started my engineering career with a BS degree in Engineering, an Ensign's commission in the Naval reserve, and an Unlimited Horsepower US Coast Guard 3rd Engineer's License for both Steam and Diesel ships. After several years as a sea-going Marine Engineer, I came ashore in 1970 and eventually went to work for the Navy. I ended up as an engineer in the Submarine Fluid Systems Division, and spent the better part of the next thirty years in various Engineering and Naval Architect billets in the Naval Sea Systems Command, all in the submarine community. Got to ride/play with things most people only see in National Geographic specials and on the History Channel, and got paid for it. And got to retire when I was 55.
Submarines have thousands of bolted joints, many in critical applications, which, in the end, accidentally caused me to build a small empire within the Navy's engineering community as a fastener expert. Amazing how a collateral duty (that's all it ever was) gets to be a big deal.

In the early 1970s, the Navy's submarine community realized that they had a bunch of bolting problems; both with fastener selection and joint assembly, and by 1978 published a Submarine Fastener Manual. In 1981 I switched jobs and, as part of my new job's duties, inherited responsibility for that manual; after that, wherever I went in the submarine community, the job description for my new position would be rewritten to include the Submarine Fastener Manual and all submarine fastener problems as a collateral duty. As one who had been assembling motorcycle and automobile engines since he was 12, and new how to use a torque wrench, I thought I had this one wired. Within 3 days of inheriting the Fastener Manual I discovered that I about had the tip of the iceberg wired, and embarked on a very steep learning curve for the next two months. The learning hasn't stopped to this day, but the curve is rarely that steep any more. My specialty was "in service" engineering - how to select a fastener that would live for a specific application, installation procedures, including procedures for tightening with a torque wrench, use of angular turn, and use of ultrasonic direct stress measurement; review of failed fastener analysis reports and developing solutions for preventing those failures in the future. The less publicly visible (but probably more important) section handled fastener materials, manufacturing processes and QA/QC testing of materials and finished products, and contributed to my ongoing education on a regular basis. So that's how I got to be a fastener geek/goon.

Random comments, in no particular order of importance:

1. While there are many materials used for bolts, cap screws, machine screws and nuts, where there are not problems with corrosive environments or other special considerations, selecting from among the various steel alloys used for fasteners is a hard act to top.

2. In the drive to reduce space and weight, ever-higher strength alloys are considered. This is a good thing if not carried to extremes, but people need to remember that high strength is no good without toughness. Just because a fastener is strong doesn't mean it is tough - VERY high strength fasteners (250ksi - 270ksi yield, as found in some heat treats of some of the Nickel/Cobalt alloys) tend to be brittle and not respond well when subjected to high shock loads. Since they don't stretch much, they break, where a tougher (though less high strength) alloy will just stretch. Of course, the tougher alloy would usually have to be a larger diameter or use more fasteners in the joint; everything is a compromise in the fastener world.

3. Concerning toughness and resistance to high shock, there is an all to often ignored quality buried in fastener material chems and physicals called "Percent Elongation" which makes screening for toughness relatively easy. The Navy's big hurdle is passing tests for resistance to "Hi Shock and Undex" ("Undex" is short for UNDerwater EXplosion). The Navy flatly refuses to allow components (including the fasteners holding things together) in critical applications to use materials with a percent elongation less than 10%, and will only VERY rarely allow materials with a percent elongation less than 10% in a non-critical application by approval of an exemption specific to that component for a specific service on a specific class of ships (we're speaking combatants here; noncombatant ships are allowed a little more leeway in their design). In general, to meet hi shock and Undex requirements, materials with a percent elongation of 15% or more should be chosen, over 18% is better, and, as heat treating processes for metallic alloys in production quantities has improved over the last 20 -30 years, many fastener alloys that were in the 15% - 18% range can now be reliably produced in the 20% - 25% elongation range, which is great for toughness. Fortunately, by the nature of the beast and the general conservatism of mechanical joint designs used in shipbuilding, any fastener material with a percent elongation of over 10% will generally pass hi-shock/Undex in a properly designed bolted joint. To protect its flanks, the Department of Defense has some specifications that it piggybacks onto commercial specs (MIL-DTL-1222, for instance) to get what they need in this respect, so an ASTM A574 4340 socket head cap screw for a critical application would be ordered with a minimum percent elongation over 10%, which is higher than the minimum demanded for 4340 socket head cap screws in A574. And many engine building applications don't have the hi-shock loads present in Undex testing and can benefit from really high strength fasteners, even if their percent elongation is less than 10%, but you do have to match the alloy chosen to its shock environment carefully.

4. Once the user the user settles on the basic fastener configuration (hex head cap screw, socket head, 12 point, etc.) and the correct material, the most important things (more important than material and basic configuration, unless grave errors are made in selecting those two items) are the geometry of the fillet radius where the unthreaded shank joins the head and the transition from the threaded portion to the unthreaded shank of the fastener. Errors in design or execution in either of these two areas can result in fastener failure where it would not otherwise occur. To eliminate problems in the threaded to unthreaded shank portion, use of rolled threads is usually sufficient. For the shank to head transition, the proper fillet radius needs to be specified, and adherence to that specification needs to be checked religiously as part of the QA program.

5. When doing qualification testing or random sample receipt inspection of fastener lots, the most effective single physical test (after a good visual inspection and dimensional checks) to verify the quality of a finished fastener (including the items in 4. above) is a wedge tensile test per ASTM F606. The beauty of the wedge tensile test is that it is done on a finished fastener, not a machined tensile test specimen, so it picks up both material problems and manufacturing defects like an inadequate fillet radius. Generally, if you purchase fasteners from an outside source, they will state what specifications are used for manufacture and random sample visual inspection and dimensional checks are sufficient. Cash flow permitting, it's nice to send a few to a lab once in awhile for a wedge tensile test by an independent lab - I was spoiled; any Naval Shipyard had a lab with lots of neat machines, including a tensile test machine.

6. With regard specifically to rod bearing cap screws for ARC connecting rods, my inspection of a limited number of these fasteners (two rods from my engine builder's stock) left me impressed. Their high quality was obvious, as was correct design and execution of the fillet radius and the thread transition (to the extent that this can be determined by a visual inspection, but I'm willing to bet that they would pass a wedge tensile test without breathing too hard), but more interesting was an extra little design feature, a reduced diameter section in the unthreaded shank of each cap screw. That little feature actually improves resistance to shock and greatly improves performance in hi-shock situations. Basically, it makes the fastener a better spring, mitigating shock damage by increasing fastener toughness with a mechanical trick, rather than exotic metallurgy. As an example of what this feature can do, the Navy has subjected a group of fasteners (studs in this case) to a shock load that would make them fail; the average stretch was 1/16" before they broke. Repeating the test with studs identical except for a slightly reduced diameter in the unthreaded section, the fasteners actually stretched an average of 3/16" prior to failure. When not carried to extremes, that little feature is an excellent way to improve high shock performance, and it doesn't take much reduction to collect that benefit; obviously, too great a reduction in diameter in the unthreaded shank will reduce ultimate strength more than it benefits shock resistance, but it is usually possible to strike an effective balance without causing problems if the fastener design isn't right on the edge of failure due to the in service load profile to begin with.

7. Excerpts from a Navy training lecture - "What your mother and the professors didn't teach you"

A. Preload range of bolted joints assembled with a torque wrench:

Most people see a torque specification for the threaded fasteners in a joint assembly and think that (1), the desired preload is achieved with great precision, and (2) that the fastener-to-fastener preload variation within the joint is small. Neither of these impressions is correct. On the best of days, the fastener-to-fastener preload variation is about 35%, and often more. It is not unusual to see the largest preload measured in the bolt circle twice that of the smallest in joints assembled with a torque wrench by an inexperienced mechanic without benefit of a proper installation process. Although other standards have been (and when there is a specific reason, still are) used, most bolted joints found on submarines (that require use of a torque wrench for assembly) have a preload established at 2/3 of yield of the weakest element of the joint (150% of yield in the case of bearing stress), and, ideally, the limit will be reached as tensile stress in the bolt or stud.

So you have a torque from a drawing, Maintenance Standard, tech manual or whatever, based on, say, 2/3 of yield. When the mechanic is done assembling the joint, all the bolts in the bolt circle would have (theoretically) a tensile load of 2/3 of yield, because the torque was chosen to give that result. Unfortunately, there is a significant fastener-to-fastener variation in friction coefficient, AND a significant fastener-to-fastener variation in short term preload loss (the relaxation that occurs in the first 2 to 10 minutes after the wrench is removed from each fastener in the bolt circle for the last time). And a few other things (all told, about 76 different things, according to the Air Force, which did an excellent study on the subject, like prying loads and cross talk. The result is that all we know for sure is that each fastener in the bolt circle, having been tightened to a mean torque mathematically equivalent to a mean preload of 67% of yield, has an actual preload of somewhere between 40% and 90% of yield. It sounds crude, but it's close enough, even for hull integrity/high shock/UNDEX/hazardous fluid, etc, services. The distribution of this preload variation is more or less a bell shaped curve in a statistically valid sample.

B. The value of process instructions:

The real case is not quite as bad as the above makes it look, as the methods incorporated into Navy/Navy approved process instructions are designed to reduce this preload spread. The reality is that the multiple passes, check passes, etc, of Navy process instructions skew the curve. The top value (90% of yield) doesn't change, but the number of fasteners below 67% is significantly reduced, and the amount by which they are under 67% of yield is also reduced, while the number between 67% and 90% is increased. Since the best defense against long term preload loss is high initial preloads and minimization of fastener to fastener preload variation, the value of good processes and training in those processes is once again proven.

C. Good shop practice and precision:

As far as fastener to fastener preload variation is concerned, the variation for fasteners in a joint tightened without a torque wrench, but in stages and with check passes, is 5% to 10% more than the variation with a torque wrench, on the average. In reality, other than being cheap to use, the torque wrench only offers 3 advantages:

1. It assures adherence to a specified mean torque, which in many joints is not a particularly significant item, except for record purposes, or where minimizing the chance of exceeding the yield strength of the material or the threshold stress level of an H2 embrittlement prone material is important.

2. It results in slightly less preload variation than the use of "good shop practice" and a box end, open end or socket wrench, in the hands of an experienced technician.

3. It offers visual proof to witnesses that each fastener in the bolt circle has been properly tightened, needed for certification records for "critical joints".

For what it's worth, the most accurate method of tightening the fasteners in a bolted joint without resort to expensive ultrasonic measuring devices, strain gauge equipped bolts or other cost increasing approaches is angular turn of the nut. The fastener to fastener preload variation runs about 15%, much better than a torque wrench can ever hope for.

When fasteners are overtorqued severely during initial installation but survive the procedure without failure, the joint is generally OK for service without further action. Various embedment phenomena and other short-term preload losses will reduce the stresses to a high but acceptable level. The exception is where the fasteners are made of materials prone to hydrogen embrittlement. They may settle out, after short-term preload loss, at a level in excess of their threshold stress level, leaving a high probability of brittle failure in the future. Resolution where H2 embrittlement prone materials are involved should always favor loosening and re-torquing, one fastener at a time, to the correct value (Note to Tom: Many bolted joints on Navy ships require hydrostatic testing for verification. If the joint integrity is violated by loosening the fasteners, an expensive retest is required. By common sense and resulting executive fiat, the Navy does NOT regard the integrity of the joint as violated if the fasteners are loosened and re-tightened one at a time, saving the expense of retesting, hence the importance of "...one fastener at a time"). If discovery is after the joint is buried by interferences and cost is too great, use of Level I certs may allow acceptance based on stresses adjusted for the actual yield rather than the min spec numbers usually used in design calculations and by PC Bolts. Fasteners of materials such as Grade 5 steel and other materials not prone to H2 embrittlement may be left as is and torqued correctly at the next disassembly and re-assembly of the joint unless the activity is having a lot of such errors, in which case remedial retorquing is necessary to get production's attention.

G. Notes on choosing thread lubricants:

2. Once in awhile the subject of the range of friction coefficients that will be exhibited by a thread lubricant will come up. When discussing this subject, insist that the range be tied to a specific material (nut and bolt or stud) combination. The reason for this is that the same lubricant often has both a different mean friction coefficient and different extremes (and can have the same mean friction coefficient, but different extremes, the high and low values that establish the range) when you compare the mean and extremes for different combinations; alloy steel and alloy steel, CRES and CRES, Monel and Monel and KMonel with a monel nut, for instance. The extremes for each material combination establish the range FOR THAT COMBINATION. Taking the high for the combination that has the highest extreme and the low for the combination that has the lowest extreme, a method that will quite often be attempted by your Nuclear counterparts if you don't call them on it, doesn't give the range, it gives a number useless in calculations and of little interest in intelligent discussions.

8. Notes on tightening procedures (the following is a slightly edited excerpt from the Navy's Submarine Fastener Manual, from the sections that are the basis for development of local activity's process instructions and instruction packages for individual work packages, where such detail is necessary for a specific work package. As you can see, the basic approach is relatively simple and grounded in common sense. It also would pass as the generic basis for ARC's specific procedures):

Before tightening any fasteners, the following should be performed:
a. Examine fasteners for compliance with marking requirements.
b. Examine the internal and external threads for burrs, nicks, metallic slivers, etc., that could cause jamming or excessive resistance to tightening. Remove or correct as necessary.
c. Ensure the threads and mating bearing surfaces are clean and free of rust, chips, or other foreign matter.
d. Ensure the nut (or cap screw) seating surface is flat and contacts the mating surface all around.
e. Lightly lubricate the threads and bearing surface with the specified lubricant and remove excess lubricant to permit air to escape from under the nut (or head of the cap screw). Flange spot facing should also be lubricated.
f. If using torque measurement method, ensure the torque wrench has a current calibration sticker. Select a torque wrench such that the required torque is between 20% and 90% of the full-scale range of the torque wrench selected.

FASTENER TIGHTENING PROCEDURES. The following procedures are applicable to nuts, through bolts, studs, cap screws, and set-studs used on flat-face and raised-face flanges:
a. Prior to applying final torque, perform the prerequisites described in steps a through f above.
b. Assure proper alignment of the mating components.
c. Where the application requires O-rings or gaskets, ensure the O-ring or gasket is in its proper position. Make up the joint evenly by tightening diametrically opposite fasteners until the mating components contact each other. This will normally be accompanied by a noticeable increase in torque when metal-to-metal contact is made. Check all fasteners to ensure that no fasteners are loose. Continue to tighten fasteners sequentially. Apply approximately ten percent of the specified torque to ensure solid part contact. Finish torquing the joint in 25 percent increments of the specified torque.
d. For determining torque values used in this procedure, refer to paragraph 5-5, use Appendix E (PC Bolts), or seek guidance from your activity's Design Division.
e. When tightening nuts in set stud and nut type joints, check stud rotation by marking with a felt-tip marker on the nut end of each stud in a direction toward the center of the flange. Check the mark on each stud after tightening to ensure the stud did not rotate.
f. After completion of the last tightening pass, wait a minimum of 2 minutes, and execute a check pass or passes until the joint holds the specified torque setting. This minimizes the effects of short-term preload loss and helps minimize fastener-to-fastener preload variation.

9. Where hydrogen embrittlement is mentioned above, it shouldn't be a problem in the environment of a rod bearing cap screw, since you need high stress (above the material's threshold stress level) in the fastener, a source of free hydrogen and an electrolyte, and an electrical potential, like screwing a Kmonel stud into an HY-80 steel pressure hull, or a steel cap screw into an aluminum rod - the whole world is a battery waiting to happen once you introduce dissimilar metals. High strength steels (140 ksi yield and above for purposes of H2 embrittlement discussions) are subject to H2 embrittlement and can be assigned a threshold stress level of 80% of yield (threshold stress level is a bit of a moving target, but 80% of yield is a good working value for high strength steels), but for embrittlement to be a problem, you have to have ALL the factors, not unlikely in the marine environment, but HIGHLY unlikely inside an engine, even if the target mean preload is 90% - 100% of yield for the rod bearing cap screws.

10. Where PC Bolts is referred to above, it is a relatively simple to use computer program developed by the Navy for calculating fastener torques for bolted joints. I can send you a copy if you're interested in playing with it - it covers through bolted joints, cap screws threaded into blind holes, and set stud and nut type joints. The present version is a pleasant little calculating machine in 16 bit Dos code, so old it lacks mouse support, but it's lean, mean and gives you torque, preload and a large collection of joint component stresses. Or you can input a torque and get preload and the stresses, or input a desired preload and get the necessary mean torque and stresses. The Navy hands it out for free to anyone that wants it. My one mule consulting outfit heads the Beta test program for the new 32 bit Windows version, which we will have out soon and which also will be given away free to anyone that wants it, once it's ready to release into the wild. As a Navy employee I headed Beta testing for the three previous versions, so PC bolts is sort of the crown jewel of the contributions I've been able to make to solving fastener problems; that, and a whole bunch of friction coefficient testing that went into the lubricant library of PC Bolts, and which led me to take note of your rod installation instruction.

11. Oh yeah, one other little item; all of the above discussions assume that when the designer chose the fastener size and alloy and the preload we are going to apply to the fastener, he/she had a reasonably good idea of what the worst case in service load will be and chose the preload to be well above that load after short term preload loss and a reasonable profile for long term preload loss. If we blew any portion of that little set of choices, we have the potential for cyclic load reversal and good old fatigue failure. Short term preload loss can run anywhere from 5% to 40%, though proper tightening procedures will make it unlikely that you will ever see the latter figure. Our preferred approach is to take the max in service load and select a preload 4 times that amount, so that you can lose 50% of your initial assembly preload to any combination of short and long term preload loss and still have twice the preload needed to do the job - crude, but it works. Most bolted joints work well because their design is conservative (if nobody goofed), and hence they are very forgiving of minor errors, abuse and neglect. Of course, the more refined your design assumptions are from a standpoint of test results or other prior experience with a specific application on which to base your decisions, the closer you can cut it if space or weight considerations intrude, without putting the joint's integrity in danger. In combatant aircraft design, there are bolted joints that are subject to fatigue failure in order to reduce size to fit in the space available. With fatigue life to failure in a cyclic range that corresponds to 12 to 18 months of service, the aircraft maintenance plan calls for automatic replacement every 6 months. That's a valid solution for a fighter plane, but probably won't hack it for a passenger car.

12. I have repeatedly used the terms hi-shock and Undex in the discussions above. To give you an idea of the magnitude of the forces we're dealing with when Hi shock and explosion testing are invoked, civil engineering design for earthquake involves designing for accelerations primarily in the 9G to 11G range. Hi shock and Undex deal with G forces in the 450G range, give or take, and greater.