Steel Shaft Failure 4

[MetalGearMan]

Hi. I am a materials engineer and currently have a project that I require some help with. It is a fairly straightforward setup that I am dealing with. I have posted the setup as an attachment. The drawing was done after a modification to the system was made. The modification was moving the gear motor box further away from the main shaft with the use of a drive chain to reduce the stress on the shaft. This was done because in the original setup where the gear motor was connected directly to the main shaft, the shaft failed (almost like snapped in two). I am trying to figure out why it failed in the original setup. The shaft material used was carbon steel AISI 1080. The yield strength of this material is about 375.8 Mpa. I have done some torsional analysis on it and came up with a relatively small shear stress on the bar which is well below the yield strength. Maybe I am negelcting some stresses or loads in my calculations? It must be some other loads or stresses that caused it to fail like it did. I would really appreciate any help or ideas with this. Please do not hesistate to throw suggestions at me.

PS. The componenets resting on the main shaft and the other shafts are made of natural rubber and rotate with the shafts to move materials upwards.

Thanks

 

[CoryPad]

Did the fracture occur in a smooth cylindrical section? Or maybe at a geometrical feature (stress concentration) like a step, groove, fillet, etc? How do you know the yield strength with four significant figures?

 

[MetalGearMan]

I have attached a pic of a similar shaft failure. However there is a groove running through the top of the shaft that extends through the length of the shaft.

I looked up the yield strength from a few websites that had that figure listed. I just needed an approximate value don't know why I used four significant figures.

 

[CoryPad]

If your shaft's fracture surface appears similar to your attachment, then it failed due to fatigue. The crack arrest marks (aka clam shell marks, beach marks) are an indication of cyclic loading. If you have a groove in your shaft, that would be a likely crack initiation site.

 

[MetalGearMan]

Hmm yes the groove is a likely initiaiton site for the crack. Any ideas on how to prevent the fatigue from recurring? What exactly is cyclic loading? Do you think that the rubber components are exerted such a weight on the shaft that along with the torsion could cause it to fail?

Thanks

 

[gwolf2]

Rubber can be heavy, looks like you have lots and the shaft is thin. Which way is gravity? Do you have a cyclic bending problem here?

The shaft looks thin and there looks like plenty of non-structural mass - do you have a pipe whip problem causing bending fatigue.

Thats all I can think of atm.

gwolf.

 

[MetalGearMan]

gwolf2

Thanks for the suggestions. I do believe that it is a cyclic bending problem. We are currently using a higher carbon steel 1440. It seems to be working along with the modification to the setup. However if fatigue from cyclic loading is the problem then I would need a better solution. Any ideas?

 

[gwolf2]

Well first find your distributed load due to the rubber, work out the bending stress and see what that tells you.

Next check out the natural frequencies of that shaft and compare with driving frequencies.

I also wondered whether the stuff you are moving is occasionally jclogging and giving the shaft a kick every now and then.

 

[AllHandlesTaken]

Yes, fatigue - have a look here for a good overview:

http://www.asminternational.org/pdf/spotlights/jfap0502p011.pdf

 

[MetalGearMan]

Yes checking the frequencies of the shaft and the motor is a good idea. So the drive frequency and the natural shaft frequency must be similar in order to prevent fatigue? I am not familiar with how to do this comparision.

The material that I am moving is recyclable materials such as newspapers, pop cans etc so I doubt it is having that effect of the shaft.

 

[MikeHalloran]

Is that some kind of powered skatewheel conveyor, or a low-budget macerator? I.e., are the objects attached to the shaft prismatical or cylindrical?

Doesn't matter much; if the chain drive helped its survival, that suggests the shaft is _way_ too skinny, and is bending under the weight of all the crap lying on top of it, or beating on it.




Mike Halloran
Pembroke Pines, FL, USA

 

[BobM3]

Sure looks like a spindly shaft (bends a lot). How did you attach the motor to the shaft originally? With a flexible coupling? Where along the shaft did it break (near a bearing support, maybe?)?

 

[Ron]

As CoryPad noted, it is fatigue. Agree with BobM3 that the spindly shaft might be an issue. Looking at the beach marks, they are radial with a single axis. This indicates that the problem might be shaft whip or oscillation that occurs at the same point in each cycle.

 

[JJPellin]

For a fatigue crack like this, I would suggest looking very closely at the location of the fracture. We had a shaft in a lobe blower that lasted for over 5 years. We replaced it and it failed in this same manner within 5 days. The crack occurred at a shaft step. The original shaft had a nice radius. The new shaft had a sharp undercut. This was a machining mistake that created a large stress concentration. The fatigue life of the shaft dropped from infinity to 5 days. I used to have some doubt about the importance of a relatively minor stress concentration. Now I am a believer. If you can send the ends of the shaft on both sides of the fracture to a metallurgical lab, they can tell you the fracture mechanism as it relates to the location of the fracture, surface finish, stress concentrations and material.

Johnny Pellin

 

[Wrenchbender]

MGman,
As materials guys, we know parts can fail at less than yield by ‘fast fracture’ or by fatigue mechanisms such as this. For the theory behind your photo of the broken shaft, look up the ‘Paris Law’.

Besides the brute force solution of a bigger shaft, other fixes can include a round bottom keyway, tougher material, shot peening the surface, higher chemical purity in the steel, finer grain size, repositioning the bearing(s), or combinations.

Also was that a typo in describing the steel that you switched to? I don’t recognize 1440; maybe you meant 4140, which can be heat treated to have better fracture toughness than a hypereutectic steel like the original 1080. (All those carbides are great for strength but can nucleate cracks.) I believe lower, not higher, carbon is better in this case. Remembering the tradeoff between YS and K_1c, I would suggest selecting the lowest carbon content consistent with the yield strength needed, and with a tempered martensitic structure.

 

[gwolf2]

"So the drive frequency and the natural shaft frequency must be similar in order to prevent fatigue?"

I recommend that you find a stress specialist in your company or outsource it.

 

[GBor]

gwolf2 should probably have added a quick and resounding, no! to your drive frequency question. Aerospace uses a +/-10% difference required as a rule of thumb. Military marine uses +/-20% difference.

You can ask in the vibration forum for additional information.

 

[MetalGearMan]

Bestwrench

Yes that was a typo I meant 4140. You are right the higer the carbon content the more brittle your material gets and more chance to nucleate cracks. Your suggestion about a tempered martensitic structure is great since it blalances out the compromised strength from choosing a lower carbon steel.
It is clear to me now that the mechanism of failure was due to fatigue. I am having some difficulty in calculating the distributed load on the shaft from the rubber components since they have an eccentric shape and the shaft is driven through the middle of them. I need this load to get the amount of deflection and see if that is what caused it to fail. In regards to 'Paris Law' how do I determine the crack size?



gwolf2

I would not be on this forum if that was an option

 

[desertfox]

Hi MetalGearMan

Yes your shaft picture is a classic fatigue failure.

Have a look at this site:-

http://www.tms.org/Students/Winners/Davidson/Davidson.html

desertfox

 

[Wrenchbender]

MG man,
You asked about crack sizes to use for the Paris Law. The final crack size, of course, occurs just before the break, and is determined by the ‘plain strain fracture toughness’ of the material, K_Ic (worst case), and the max applied tensile stress.
a_f = 1/pi*(K_Ic/sig_max)^2
Initial crack size, if unknown, can be taken as the smallest detectable flaw with whatever NDT method you are using. None? Eyeball? Then how about
a_o = 0.03 inch.

Previously you had asked what cyclic loading is. You probably know by now - as the beam rotates, the deflection/stress at any fixed point (off axis) is a sine function of the rotation angle. This is the cyclic loading; this is the range for finding delta-K in the Paris eqn. (The o’clock position of the longitudinal groove, however deep it is, will also throw in a bias.)

On shaft deflection: Reasonably conservative assumptions and best engineering judgments can simplifiy long hand calculations on fatigue life. From the drawing, it looks like you can assume a beam with a uniform distributed load for a first approximation. (or x-number of point loads superimposed).

Some other thoughts on a material fix – specify “aircraft quality” steel. Consider nitriding or carburizing both of which are supposed to improve fatigue life, I think, because a thin surface layer is placed in compression. Lastly avoid corrosive environments, plating or phosphating.