question about spline design

[PearlRock]

Hello everyone, I designed a spline in this thread:

http://www.eng-tips.com/viewthread.cfm?qid=373420

Now I'm doing an FEA torsional analysis of my shaft to check for fatigue issues. I modeled the spline along with the shaft. Peak stress is showing to be about 80MPa (12ksi) localized around the teeth roots.

Now, when I did my spline calculations, everything seemed to be within the design stress limits as specified by the machinery guide, but now I'm worrying because the fatigue limit for my steel is suggested to be about 35MPa (5200psi) with a safety factor of 2. Is this stress FEA is giving me a fatiguing stress? Do I need to be worried about this?

This has to be a fatigue stress~~~ a start/stop fatigue. This stress would not exist at zero speed. Therefore the spline will fail, yes?

I also have a very nasty torque profile. Peak torque is about 2X higher than the average torque at constant speed conditions. If this torque pulsation is damped away, then my stress would be reduced by 50%, putting me in acceptable stress ranges. The question is though......how much damping will occur. Unfortunately I can't answer that.

Have a good evening.

/PR

 

[Dougt115]

From what I have been working on.

Fatigue is an issue. The 2x is a design margin and may not be enough depending on your loading and material. You can get SN curves for most materials. The curves are not application specific so you may need to test to verify it meets the requirements.

Reverse bending is the worst condition so preloading in compression usually produces the best results. Sorry but your start stop sounds like reverse loading unless you can preload the spline some how.

Also on the SN curve look at your design cycle life. You may find a higher stress is acceptable if you have a short design life. Some steels are considered to have infinite life at 10^6 cycles so if your stress is below the corresponding value you are good. But if you only need 10^4 then your stress could be as much as 50% higher and still work.

Good Luck

 

[CoryPad]

35 MPa is ridiculously low for steel. Can you describe how you came about this value?

 

[tbuelna]

You did not mention what steel you are using for this spline, but a tensile fatigue limit of 5.2ksi sounds extremely low. The peak stresses that you see around the (external?) tooth roots in your FE model are likely tensile. Machinery's gives recommendations for tensile stress limits for various steels (table 7) based on a simplified equation (equation 9) that uses adjustment factors for things like load cycles, misalignment/load distribution and application conditions. If you are relying on the analysis approach shown in Machinery's equation 9 for tensile stress limits, then you don't need to apply the added 2.0 FoS.

Regarding your FEM results, you need to be careful in how you interpret them. Accurately modeling something like a spline joint is very tricky since even extremely small variations in the tooth flank surfaces, alignment between the matings parts, or stiffness of the associated structures can have a huge effect on load distribution and stress levels in the spline teeth. If you are analyzing for HC fatigue with your FEA, then you should be using a composite load case that is a cubic mean of all the torques/cycles the spline is subject to over its anticipated service life. Each start/stop only amounts to one load cycle, which may likely be fully reversing. But if there are any torque variations during a single rotation of the spline in constant speed operation, then these must be taken into account in the composite load case.

 

[PearlRock]

You guys beat me to it. I think I am looking at improper fatigue properties. I'll post back when I learn more about my material, hopefully in the next day or so.


@tbuelna, good points. Yeah, I wish I had more experience with FEA modeling this type of system. I've already noticed that small changes in the constraints has a fairly significant impact on the results. The nuances of this analysis has taken me a bit by surprise, because the analysis is ultimately a simple torsion study.

 

[Dougt115]

The 80 MPa is the calculated stress, the 35MPa is his stress limit for fatigue. He never said what the material is. But steel should be closer to 30Ksi and higher. These stress levels sounds more like a plastic or ceramic than a metal.

The fatigue analysis should be giving you the stress level needed to survive. Based on your design, materials, load profile, and cycles your max stress should be 35MPa. You are at 80MPa.

Good luck with the modeling.

 

[PearlRock]

Tbuelna recommended me to look at Vacuumelt Nitralloy N from Latrobe based on my spline. I've been looking at the vacuumelt; however, I called Latrobe the other day, and they seemed to imply 300M was easier to procure, so I got a quote for that. I've been following a number of shaft and vibration design guides, which state that the torsional fatigue stress limit is very low. One is an old MIL STD. Looking back, Tbuelna gave me a fatigue curve for an airmelt 4340 with ~400MPa for unlimited fully reversed axial fatigue life. One of the references recommends at safety factor of 4 or more for such an application, which would put my design in the green zone, but this is not the most conservative design recommendation I have. In truth, I don't know how these references are calculating these "stress estimations" and applying the correction factors. These guides could be very generous if they are using hand calculations to compute stress.

The more I look at this, the more I think it will be fine. I redid the simulation to improve the accuracy and reduce the stress concentrations, and the stress seems to have gone down to be more inline with machinery guide spline calculations. If these results are right, then I have meet all my design guide recommendations. These results seem to make sense, because the rest of the shaft isn't likely to see stress much greater than what exists in the spline itself, and I believe the spline itself will be robust.