Is there a rough general guideline to the endurance limit for fasteners, e.g. grade 8 bolts, as a percentage of proof load or ultimate tensile strength? Thanks!
short answer ... no
longer answer ... endurance limit would be similar to the bulk material, factoring in Kt (at the thread, think of a large number) and preload (and this is were it gets complicated).
it's fine to have an endurance limit, but what load are you going to relate it to ? the tension load applied to the bolt would be Way Overly conservative ... the "correct" load is the cyclic load portion that the bolt reacts, simplisticly applied load-preload ... which begs the question what preload to apply ? ... maybe 70% fty
When I asked a bolt manufacturer about this I was told that bolts are not subject to fatigue if they are tightened down properly.
If you are unable or unwilling to tighten your bolts down hard, or your joint is too compressible, you will have to look into bolts designed to survive fatigue. There is an example of this shown in my machine design textbook (V.M Faires). Perhaps there is something in yours!
In general, for steels the fatigue limit of the material is 1/2 to 1/3 of the UTS if you're at less than 35 Rockwell C. I think that grade 8 fasteners are somewhere between RC33 and RC39, so that method should give you a ballpark figure.
You may want to consider tightening your bolt more if it's experiencing significant cyclic loading. In most bolted joints, the larger cyclic load is experienced by the flange.
"When I asked a bolt manufacturer about this I was told that bolts are not subject to fatigue if they are tightened down properly."
Whoever said that didn't know much about the behavior of bolted joints. External loads applied to the joint will cause an alternating stress in the bolt but the amount of load going into the bolt is depends on the interaction between the bolt and the clamped material. For our cases, we have loads due to installation torque and temperature (both internal loads) and pressure and vibration. The load in the bolt will be
Ptotal = Ptorque + P(Delta T) + 1/(1+R)*[P(pressure) +/- P(vibration)] where R is the ratio - KJ/KB. Using +/- for vibe loads is conservative but we make airplane parts so conservative is good.
I recommend that you get a copy of just about any book on bolted joints written or edited by John Bickford. Google "joint stiffness ratio" and one of his books is the third link. I think his latest was called Handbook of Bolted Joints which includes a chapter on methods developed in Germany. VDI 2230 may be the best approach to computing bolt loads available today.
Important point that Overall joint design must be considered, but......
Every car driven past 50,000 miles has probably accumulated over 100 million "cycles" on the con rod bolts. Likely not very much time spent at max revs and highest inertia loads, but I'd still give most of the credit to "proper" installation torque reducing the range stress to a tiny fraction of the applied alternating stress, not the rolled threads or nice parabolic under head fillets. I think it is no coincidence I have not been able to convince any Non-believers that if their convictions are indeed courageous they should have no qualms whatsoever loosening all their short's (*) con rod and main bearing bolts/nuts, re-snugging to 3.5 lb-ft, then hopping on US I-90 and heading for the opposite coast with a shoebox half full of gas money, but without credit card, cell phone or AAA card.
You correctly point out that overall joint design is important for maximizing the fatigue life of a fastened joint. But I would disagree with your assertion that the underhead and thread root fillets are not of much benefit. With your example of conrod bolts, the underhead and thread root fillets represent points of stress concentration, commonly referred to as Kt, which can be as high as a factor of 3 or 4 in an analysis.
The conrod bolts, preloaded or not, only experience varying tensile loads for each fatigue cycle. The tension loads due to piston inertias simply relieve the bolt preload strain and don't add to it. This equates to an R value between 1 and 0. The R value is the ratio between the maximum and average fatigue load. A properly preloaded conrod bolt would have not experience an R value above about 0.5. Much less of an impact than a typical Kt for a bolt or screw thread.
So the most beneficial approach to optimizing your conrod bolt fatigue life is to minimize Kt (stress concentrations), by using large bolt underhead and thread root fillets, by using mechanically rolled underhead fillets and threads, and by using conrod bolts with a long and reduced diameter shank body.
The improvements in fatigue life provided by mechanically cold-worked surfaces (thread rolling, shot peening, fillet rolling, etc.) has been shown to provide an increase in fatigue life of at least 300 or 400 percent. Just take a look at the specs for any high performance aircraft bolt.
the first two are "internal" loads and the other two are "external," what I think you called "applied." I can't recall ever seeing an R > 4 and I wouldn't categorize 20% as being a tiny fraction particularly when the mean stress is held at such a high level; installation torques often bring you up to a high fraction of Fty. A small alternating load with a high Kt is a killer.
The stress concentrations mentioned by Terry are critical. In many of the bolted joints we analyze, the largest applied load in the whole pile is due to thermal expansion differences between the bolt (usually 17-4 or A286) and the aluminum lug. 100% of that applied load goes straight to the bolt. The head and thread forming processes are critical.
Having calculated the actual cyclic load range the bolt is undergoing (using one of the above methods), you can refer to BS7608 (Fatigue of welded structures) for experimentally-derived S-N curves for typical bolt threads so you can calculate a safe life or allowable stress range. Bolt threads under axial load are designated as Class X - this avoids having to calculate a Kt value.
Yes, proper installation torque (or rather axial force) is critical. So the better you can control your installation load of the bolt, the better you can control the fatigue life. A high installation load will in general reduce fluctuating axial loads.
A local one time yielding in the threads will redistribute the load to more of the threads and thus reduce the overall thread load.
I disagree but it may be a matter of interpretation. As long as the joint remains tightly clamped, higher installation torque doesn't change the alternating load on the bolt, see equation from 9/20/10 above. The purpose of the higher installation torque is to increase the joint separation load, i.e., to maintain the clamping load. The slope in the joint diagram increases significantly after joint separation from 1/[1+R] to 1.
i think what izax1 was referring to was the assebly techique used, which affects the tolerance of the preload torque.
if you control with a torque wrench the torque is +-33%, this limits the maximum torque you can apply (considering 133% preload) and affects the fatigue life (assuming 67% preload).
if you control with PLI washers your tolerance comes down to +-10%; clearly you can specify a higher torque and you'll get a better fatigue life (as the minimum torque is so much higher).
