Does fatigue occur in non-preloaded bolts under load range which is compressive only? 1

[siberiankhatru]

Bolts are seldom under compressive load, but what if we had an situation, where a non-preloaded bolt was under varying compressive load (load ratio R>1, i.e the loading is fully compressive).

For example, the situation could be a column base connection without grout where the anchor bolts and nuts (and baseplate) carry the compressive load from the column.

Details and FAT classes are presented in the design standards Eurocode 1993-1-9 and DNVGL RP C203 but the details are meant for bolts which are mainly or somewhat under tension. How much better FAT class would a non-preloaded bolt have if the load range was fully compressive?

 

[Agent666]

"atomic dislocations" are not what causes fatigue failure

Not sure I follow/agree though on this particular point, atomic level dislocations are the cause of the initial crack initiation. Wikipedia

Surface of material dislocates (intrusions/extrusions) on a slip plane, results in stress concentrations, material strain hardens, material eventually fractures on a microscopic level initiating a crack, crack propagates under further loading, eventually leads to fatigue failure.

Don't get me wrong, intuitively tension makes sense, but its not backed by some of the research that's out there.

I'm guessing at an explanation here but given you'll get the same dislocations and stress concentrations under compression, and same strain hardening. If you consider some steel compressed and yield it locally in areas of these stress concentrations, if you release it there might be some sort of 'effective' tension state created as the material around it that is still elastic springs back, cycle the load enough and the crack might propagate.

 

[Agent666]

Out of interest I note Eurocode has rules for dealing with fatigue and compression loading for non-welded and stress relieved details. If I'm interpreting it correctly, it allows you to effectively reduces the part of the stress range in compression to presumably allow for the improved fatigue life when all or part of the total stress range is in compression.

 

[BridgeSmith]

I'm not familiar with the Eurocode, so I won't comment on that, beyond pointing out that the figure posted shows a stress range that includes tension (stress reversal), not one strictly in compression, so I don't find that in itself to be convincing. What I quoted was from the 8th Edition of the AASHTO LRFD code published last year. AASHTO usually keeps up to date with and incorporates recent research findings, so either what you were reading is very recent, or the eggheads who propose changes to the code provisions didn't find it relevant or didn't find it convincing.

I'm not saying it isn't possible for some cracking to occur due to strain hardening in compression followed by stress relief in the component. I'm just having a difficult time envisioning how the crack would continue to propagate to failure. I also don't see how any portion of a component limited to the yield stress for design reaches strain hardening for cyclic loading.

 

[Agent666]

Just to clarify the point you are reading into the Eurocode provision, the reduction for compressive stresses applies specifically if part or all of the stress range is compressive, its not reserved to stress reversal cases that also go into tension.

I also don't see how any portion of a component limited to the yield stress for design reaches strain hardening for cyclic loading.

You could say exactly the same for tension loading though, its minute stress raisers at dislocations initiating the cracks, often at much lower average stresses than yield. Depending on the detail that is why there are all these different detailing categories for fatigue, some details are worst at concentrating the stresses to a point where they are an issue and you are only evaluating against average stress in many cases (no idea if AASHTO is based on a similar method or not as Eurocode or NZS3404 as I'm simply 100% unfamiliar with it). But just giving an opposing stance from several other standards in how compressive loads are treated for actual design for fatigue.

For all I know because these standards are older than AASHTO, the current best practice knowledge may be reflected in the latest version of AASHTO (i.e. ignore compressive loads). But I would imagine it's always been that way possibly?

 

[BridgeSmith]

AASHTO has not included any fatigue stress limits for components not subjected to tension from before I started in bridge design, so for at least the last 20 years.

 

[siberiankhatru]

A closer look at the recent DNV-RP-C203 -publication shows that fully compressed stress range with no residual stresses does not lead to fatigue:

If the thread of the bolt is rolled, there exists residual stresses and the DNV reduction factor is not zero. If the compression load is big enough, there exists local yielding in the thread, which lead to residual (tensile?) stresses and the DNV reduction factor is not zero. Thus, the compressive load range should be 60% effective.

So, according to the Eurocode and DNV, a fully compressed bolt has an effective stress range which is only 60% of a fully tensile case. I didn't find any fatigue test data of this kind of case by Googling, so I think, I will settle with this result.

 

[TomLee777]

Let's get to the basics. No matter the bolt is axial tension or axial compression, the material will eventually fail by shear due to fatigue. Steel is a ductile material. Under repeated cyclic loading, the crystals will slip in the shear plane in a microscope scale, and eventually fail on the shear plane. When the bolt is in "pure" axial compression, it has a maximum shear stress in the plane 45 degree to the axis of the bolt. Therefore fatigue failure can still take place at the plane. However, the presence of compressive stress tends to squeeze the slip plane, increasing friction of the slippage, effective increasing fatigue life as opposed to tensile stress. It is hard to explain. I found the website fatigue-life.com is helpful in understanding the basics. But in summary, bolt in "pure" compression does fail in fatigue, but not as quickly as bolt in tension. The codes and standards address this difference by "mean stress factor" as pointed out in the posts above.