Fatigue calculations of large, non-pretensioned trapezoidal connection subject to bending moments 1

[julian89]

Hi all,

I'm about to embark on 3D, pretty large scale, finite element calculations of a large assembly for subsea operations. In this assembly, there are a couple of large threaded contacts (think pup joints and X-overs). For strength calculations, I can either do this by hand or in a finite element model. However, when it comes to fatigue calculations, explicit modelling of threads to determine the peak stresses for fatigue calculations, this becomes practically impossible due to the need for computing power. Furthermore, these threaded connections will be subject to bending moments and are non-standard in geometry.

So, do any of you know of any good resources on how to do fatigue calculations like this?

 

[rb1957]

Surely(??) we have a good understanding of the fatigue performance of threads without needing to resort to detail FEA ?

another day in paradise, or is paradise one day closer ?

 

[NRP99]

These threaded connections will be subject to bending moments and are non-standard in geometry

If these threads are special threads not covered in ANSI or ISO metric threads, still you can do analytical calculation. Check para -Fatigue Loading of Tension Joints from Shigley's Mechanical Engineering Design book.

In fact, Peterson[sup]9[/sup] observes that the distribution of typical bolt failures is about 15 percent under the head, 20 percent at the end of the thread, and 65 percent in the thread at the nut face.

Probably the shaft with special threads will be screwed in to nut-sort of hole with female threads. If that is the case, then you can start calculation at the threads engaged at the nut face.

If you want to calculate the exact peak stress range instead of applying stress concentration factor to analytical calculation, then I would suggest explicit thread model with axisymmetric approach or Sub-modelling approach to reduce computing power/time. (May be axi-symmetric approach may not apply due to non-symmetry of loads.) In sub-modelling, you can introduce local features like exact thread profile instead of of modelling those in global model.

 

[julian89]

Check para -Fatigue Loading of Tension Joints from Shigley's Mechanical Engineering Design book.

Will this work for non-pretensioned joints in bending? My concern is that the peak stresses will become very large quickly in bending since the tension side is basically a point in a cross-section, rather than a uniform distribition across the entire cross-section as with tension.

Is there any data on fatigue of bolted joints in bending? Or is the convention to simply convert the bending stresses to tension mathematically?

Thanks.

 

[NRP99]

If the threads are not tensioned, then your problem is simpler than thread fatigue. IMO you can consider the threads profile as requirement for some sort of locking and calculate the fatigue life as normally you would do for non-thread components like e.g. splined shaft.

Bolted joints are not subjected to bending directly since it will defeat the purpose of clamping. Bolt will carry major part of load in direct tension and direct shear and very minor amount of bending due to the eccentricity of loads.

 

[rb1957]

why would the bolts not be pretensioned ? Surely (??) we've learnt that loose bolts "sink ships" (ie are not a good idea). Even if you have a joint that moves, the bolt should (IMHO) be clamped against a bushing that allows the joint members to move.

another day in paradise, or is paradise one day closer ?

 

[julian89]

Note that these are not bolts. These are more like pup-joints and X-overs, i.e. joints which connect large sections of piping in the main path of load transfer. Hence, the joints are subject to both tension and bending.

The joints are held in place by shear pins, not pretension.

 

[rb1957]

ok, but still these are a common (well, common in your industry) part, so "surely" we know the fatigue behaviour and allowable loads.

if not ... fatigue test ?

another day in paradise, or is paradise one day closer ?

 

[Tmoose]

MC/I with a hefty de-rate to allowable stress due to severe but indeterminate geometrical stress concentration.

Subsea suggests some environmental salty water electrolytic factor might degrade the material and drive the allowable lower still.

I'm thinking API and MIL specs might have something useful to say.

 

[NRP99]

Taking cue from Tmoose and assuming this is threaded connection like pipe trapezoidal thread connections, API 7G, API 17TR8, API 5C1 to 5C5 may contain guideline you need. Besides you can refer "Fatigue Life Prediction of Threaded Pipe Connection" by Milad Beheshti for reference.