Screw fatigue in bending. 2

[Biggadike]

I have a group of M8x30 High Tensile (12.9) cap head socket bolts which I am using in a vibratory system (in a vehicle). I have accelerometer data and want to calculate the fatigue life of the screw(s). I am missing two things:

1/ The S-N curve or similar for the material.

2/ The stress already existing within the screw before the oscillating bending forces are applied. - I can have a reasonable stab at calculating these but proper data would be good.

I seem to remember rolled threads giving a 1.4 stress raiser which also needs confirming.

Please don't point me in the direction of a bit of internet software, I'll never get it past our security and I've never had much success with them in the past anyway.

Anybody got any real data?

 

[nashjp]

Biggadike,

Best source I have seen lately for axial fatigue strength in threads is the German spec VDI2230. The major consideration is if your threads are rolled before or after heat treat. If rolled after heat treat you get a higher fatigue strength.

For rolled before heat treat bolts 8.8-12.9 use:
Stress Amplitude = 0.85 x (150/(bolt diameter + 45)

For rolled after heat treat bolt threads use:
Stress Amplitude = 0.85 x (150/(bolt diameter + 45) x (2 - (mean stress / yield strength))

So, Axial Stress Amplitude = 54MPa for RBHT and ~67MPa for RAHT at a mean stress near 70% of yield.

For a conservative analysis, I recommend using the axial stress numbers. Bending fatigue strength for 12.9 high strength bolts is about 1.1x the axial fatigue strength. For 8.8 and 10.9 bolts I have seen 1.2-1.5x. (I hate to guess wrong, thats why I usually stick with the axial)

Now you need to figure out what stress range is in your bolt. Any of your buddies do FEA? They will want the loading. If you dont want to wait, try VDI2230, John Bickford's books, or search Eng-Tips for some good formulas and considerations for estimating the cyclic stress in the bolt.

-Its late - all numbers provided by me after 10pm should be checked, verified and rechecked... ;-)

Jeff

 

[Metalguy]

>"I think anything which is over-engineered and therefore sees little changes in stress will deal well with fatigue."<

Good design (rigid bolted connections) and correctly installed fasteners are certainly not "over-engineered". If you have vibrating/shaking conditions, it is most important to use the above or use parts that can withstand whatever fatigue conditions you have, etc.

While little is known here about your particular conditions, so far it appears to me that "ordinary" good eng. practices would prevent failures.

 

[Biggadike]

Thanks Jeff,

That looks useful. I do know the stress profile across the core of the screw. I don't know whether the thread is rolled before or after heat treat but I can always assume worst case. The design is such to ensure the screw only sees axial stresses so no problems there. When I have calculated the stress amplitude as per your equations, is that the maximum the screw can see before fatigue becomes a problem?

Metalguy,
Whilst you are almost certainly right, I need to quantify the life that this design will give. I could just say that I've followed good design practices to avoid fatigue problems so all should be well. The thing is, this is our last shot at this so putting figures to the design is called for.

 

[nashjp]

Biggadike,

The stress amplitude calculated is for the endurance limit. Below the calculated stress amplitude, no failures should occur. You can run at higher stresses as long as you keep the number of cylces down. You can draw a SN curve on a log-log plot with the calculated endurance limit at 1,000,000 cycles and the ultimate tensile strength at 1,000 cycles and check your cycles to failure at a known stress.

Have fun with the Dynamics ;-) Jeff

 

[Biggadike]

Thanks again, I'll check that now.

Chris

 

[unclesyd]

If you will checkout the Website of my last post they have already done this for you.
It's a database in the middle of the page.

 

[Biggadike]

Unclesyd,

I can't just whizz off and check web sites out because of security restrictions on access to e-mail sites. I can only see pre-approved sites which means I have to get new sites approved first....

In this case it was quicker just to do the calculations although I will have a look at all the sites reccomended in this thread when I get a chance to process the red tape.

Following Jeff's data, I'm at 1/13.5 the stress amplitude needed to reach the fatigue limit. This corresponds nicely with my earlier life calcs of 1.5x10^11 years.

Most of these results are down to the use of good design principals which have reduced the loads the screw sees.

Looks good.

 

[unclesyd]

Heaven forbid the site is the ANSI Organization website.

 

[jetmaker]

nashjp,

Do you have a link or copy of German spec VDI2230?

Thanks.

jetmaker

 

[CoryPad]

[blue]jetmaker[/blue],

VDI 2230 is available from the German publishing company Beuth at:

http://www.beuth.de

Regards,

Cory

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips Fora.

 

[Biggadike]

From our source VDI2230-1 was £269.77. (Index house - Berks).

Unclesyd, - aren't the ANSI Organisation part of the axis of evil?

 

[bjpil]

Just a thought, if you have access to a lab could you not test several your assembly on a shaker table at both a low cycle fatigue until failure and then at a high cycle fatigue. From this you could then create your own SN curve for the assembly using some statistical equations.

BJP

 

[Biggadike]

BJP,

We have our own 'shaker' (I think of it as a giant loud-speaker) but as is always the way, only have time and resource to test the final complete assembly, hence the need for figures.

Given how many times people have generated the S-N curves for HT bolts, it suprising how elusive the figures are. There's probably a massive book somewhere we could buy for £200.

Never mind, we seem to have cobbled together enough data to be confident.