Shaft fracturing 2

[RUFUS2K]

A couple of years ago I worked on a design of a small (approx .5" dia) splined shaft that had a reduced diameter section that was intended to shear in the event that the driven load increased to some unacceptable level. Through some calculation and some experimentation, I arrived on the correct diameter for the shear section. Under static tortion testing, the shafts "twist off" at a torque of approx 350 in-lb. The shaft is driving an automotive air conditioning compressor with a clutch. I set up a test that drove the shaft at the right RPM under the operating load, and cycled the clutch approximately 30K times. (about 2 minutes on, and 5 sec off). All the lab testing indicated that the shaft should perform as designed under normal operating conditions. However, when installed in the field, the shafts began failing with very few hours of operation. The system is installed on an aircraft powered by a 6 cyl piston engine and the shaft is driven by one of the rear spline drives. The shaft is 1144 steel. The splines are induction hardened to 56C. The shear section is adjacent to the spline, but not induction hardened. We have made some design changes tht have eliminated the possibility of shaft/drive missallignment. The lab testing was done with an electric motor, could the piston pulses be causing a vibration that is causing the failures. Or is this a material problem??

 

[metengr]

RUFUS2K;
Can you upload some pictures on the web site below of the failed spline shaft and the appearance of the fractured end (view of the fracture surface looking straight at it) ? We might be able to help you. The cause of failure could be several possibilities. The key is to evaluate the appearance of the fracture surface to confirm the failure mechanism.

http://flypicture.com/

 

[MikeHalloran]

Browse this site and you'll find a fair amount of material about the normal torsional vibration experienced by an engine crank.




Mike Halloran
Pembroke Pines, FL, USA

 

[Tmoose]

http://www.dsy.hu/thermo/liazid/

The right (flexible) coupling can suck the energy out of the big spikes in the torque/rpm delivery. Conversely an unfortunate coupling stiffness can amplify the variation. I picture your device being a relatively small appendage on a large prime mover, so your device's mass properties are just along for the ride, via your effective couping stiffness. Simply introducing a long thin driveshaft can serve as the spike reducing coupling.

 

[RUFUS2K]

I have uploaded some pictures. I hope that they can shed some light on the subject. The fractured area is only about 1/4" dia. The files are located at:

http://flypicture.com/bin/?id=rtr8kKzf

http://flypicture.com/bin/?id=rtr8kK3b

http://flypicture.com/bin/?id=rtr8kK3d

http://flypicture.com/bin/?id=rtr8kK3f

http://flypicture.com/bin/?id=rtr8kK3e

http://flypicture.com/bin/?id=rtr8kK3Q

http://flypicture.com/bin/?id=rtr8kK7Y

http://flypicture.com/bin/?id=rtr8kK7b

http://flypicture.com/bin/?id=rtr8kK7d

http://flypicture.com/bin/?id=rtr8kK7f

The 10 photos are of two different shafts from various angles. Since I didn't have a tripod, it was very hard to get pics that weren't too blurry.

 

[Carburize]

The "star shaped" pattern of crack growth is characteristic of torsional fatigue - reference ASM Handbook Vol 10 8th Edition page 379 figure 7

 

[TVP]

Agree that it is likely torsion overload. Two things to review include surface roughness and stresses in the area of fracture. 1144 steel contains many non-metallic inclusions to enhance machinability, but these reduce fatigue strength. If the stresses are not adequately understood, then determining the proper course of action likely will be difficult.

 

[Carburize]

RUFUS2K - can you post a side view of one of the fracture and perhaps try one from just a little further back from the fracture face to get a better focus?

 

[RUFUS2K]

CARBURIZE-
Here are some side views.

http://flypicture.com/bin/?id=rt31ka/c

http://flypicture.com/bin/?id=rt31ka/f

http://flypicture.com/bin/?id=rt31ka/e

http://flypicture.com/bin/?id=rt31ka/R

 

[Carburize]

RUFUS2K - It has all the hallmarks of a torsional fatigue failure.

 

[RUFUS2K]

After an extremely quick crask course on torsional fatigue failures, it is aparent that I have my work cut out for me. It has been many many years since I have had to dust off the types of calculations that I have seen used to shed useful light on the sources of fatigue failures. My initial (and general) question is: how do I balance the need to increase the fatigue resistance with the design requirement of the shaft, which is that it is SUPPOSED to shear when there is an overload condition? It seems that I need to increase toughness and fatigue resistance without increasing yield strength. If this is the case, what material properties am looking to change? The elimination of stress concentrations and improving surface finish I understand.

Thank you all for your help to date.

 

[Carburize]

I think you need to go in the other direction and try to find the source of the torsional exitation and eliminate it.

 

[Tmoose]

Yep, engines make varying torque, although a 6 cylinder with a big flywheel firing on all 6 is relatively smooth.

Big diameter > neck radius > spline = pretty big stress concentration due to shape. Throw in a few more de-rating factors for surface finish, naturally evil spline re-entry geometry, and if the end of the female spline happens to be the site of crack initiation. It takes a heck of a good material to overcome a pack of designed-in stress concentration factors.

http://www.pirate4x4.com/tech/billavista/PR-BV60/Materials/cs6.jpg

Gotta think turning down the shaft to a smaller-than-spline-pitch-diameter for a few inches adjacent to the spline, plus some polishing and shot peening would work wonders if you have to make these parts work.

Might be fun to stop the test on a shaft after a few hours to see what's happening.

 

[MikeHalloran]

I suspect that, if you can salvage the design at all, you will end up with a severely necked section as long, as smooth, and as small as you can possibly make it, with a spring temper, and a distinct section, much larger in diameter, with a softer temper and a circumferential groove, or shear pins in a flange. I'm sure mechanics and owners would much prefer shear pins to a complex, expensive, frangible part.







Mike Halloran
Pembroke Pines, FL, USA

 

[icrman]

350 in lb ???? And its .5 inch diameter?

So that undercut area must be a very small diameter. I could not access the photos.
I would like to know what the torque requirements are for
the compressor and "various speeds"? What is the instantanious shock torque and full governed speed?
And how long is the shaft? How is it supported?

 

[rorschach]

My suggestion, get rid of the resulfurized steel. It may make machining easier, but those sulphur inclusions are hell for fatigue cracking. Same is true for leaded steel too. Free machining steels shold NEVER be used in high fatigue applications. They will fail every time.

My two cents.

 

[SMF1964]

My two cents: There is a difference between torsional fatigue and torsional overload. Your torsional load may be below the tensile strength of the steel, which means you can withstand an applied load of some value. A torsional overload means the torsional load exceeded that tensile strength of the material. A torsional fatigue failure, on the other hand, will give you fracture even if the magnitude of the torsional load is below the tensile strength, since repeated cycling of the load gives you an incremental propagation of a crack, rather than an overload which would be a single loading event that gives you your complete fracture.

Now, if you're torsional loads are such that you are getting stresses in the shaft that are between the yield strength and tensile strength of the steel, you will, I suspect, get a torsional fatigue failure in very short order.

I agree with RORSCHACH in that stay away from free machining steels.

Am I correct in that you are looking for a material that has incredible toughness (to resist fatigue crack propagation) yet predictable tensile strength (to give you the shear failure you desire if the load is too high).

 

[RUFUS2K]

SMF1964:

I think that your last statement is accurate. I consider the environment, driver, and load to be unchangeable. I need a material that can withstand the various inputs as they exist now. I really think that the engine characteristics must have something to do with the failures. My lab testing with an electric motor driving the system wasn't able to produce any failure resembling what is occuring in the field.

Thanks to all for the input.

 

[SMF1964]

I don't think your electric motor will not produce the same torsional loading waveform as a piston engine. A rough calculation for a 6-cylinder engine puts over a million pistons firing every hour (3600 RPM, 6 pistons per revolution). If each piston firing creates an impulse torsional load, you're going to need to be operating at stresses below some variation of an endurance limit (a small fraction of yield, let alone tensile strength) under normal circumstances while still being able to shear if your loads drift too high (by some value above normal conditions that I'm guessing is less than double normal torque?

Just out of curiosity, having stuff snap into two pieces that is attached to the engine of an airplane while it's flying - isn't that something to be avoided? Could you make something like a clutch coupling that separates under excessive RPM?

 

[MikeHalloran]

He could use a belt, which has two virtues:

- When the compressor locks up, it doesn't take the belt long to burn up and fall off.

- The natural frequency of the belt can be tuned to isolate the compressor from the engine's torsional vibration.



Mike Halloran
Pembroke Pines, FL, USA