Applying torque to a shaft

[flyforever85]

My question might sound very simple but I can't think of a the right answer:

I have a shaft, a very simple cylindrical shape, with a spline at the bottom. This spline is connected to another component and the torque is 32,000 lb in. Where would you apply the torque? On the side of the teeth of the spline? Because that's where the torque will be applied in real life. Any other suggestion?

 

[FEA way]

Well, if it is applied there in real life then do the same in the analysis. Or model this part attached to the spline (that I guess transmits the torque to the shaft) and apply torque on it. Of course this will require proper contact between the part and spline. Anyway you will need coupling constraints to apply torque on the boundary of solid body.

 

[flyforever85]

The shaft has the spline that will interact with splines of another component. Refer to Fig 1 of this free paper, I have exactly the same model.I'm bothering of simulating the shaft only (so I have only the grey part of Figure 1). Now, the torque should be 32,000 in lb since I have an engine of 260 HP at 256 rpm that moves two shafts.

I created an RP and this RP is coupled with the surfaces of the splines that will touch the splines of the external component. I apply the Torque on the reference point. The solution seems to make sense in terms of pure distribution, but I have way too much stress and I can't figure out why (700 ksi!!) so I'm wondering if I applied the load in the proper way. The shaft is a pretty thick piece of AISI 4140 and it is used already with that HP and rpm so I'm pretty sure I'm making a mistake somewhere.

Any tips is greatly appreciated.

 

[FEA way]

What are the boundary conditions - how did you constrain this model ? Can you attach a picture with BC and the one with results ?

Make sure that all units are correct. Almost 5 GPa stress indeed seems really off. Also it would be a good idea to do some rough approximation with analytical formulas for stepped shafts subjected to torsion. There are some equations for splines too.

 

[flyforever85]

hope the links works:
This is the part so you have an idea of the component. The far right is fixed as BC

https://files.engineering.com/getfile.aspx?folder=f2aa016d-b93b-496b-9b6f-09e681eda211&file=Part.tif

In the middles of the two splined I put the RP and it is coupled wih 1 surface of each tooth of each spline

https://files.engineering.com/getfi...09bc-433d-b115-121d75c3931e&file=coupling.tif

general results: this is a different version and stresses are not real

https://files.engineering.com/getfi...6b24-458f-97bd-fcb6bfc04471&file=results1.tif

Zoom out on the highest stresses area

https://files.engineering.com/getfi...1baf-4b05-b94e-a80cb19661cd&file=results2.tif

hope this helps!

 

[FEA way]

It would be more realistic to replace fixed constraint with supports simulating bearings. This should reduce stresses by eliminating overstiffening. However the results are so high in this case that probably there is another serious error. That’s why I suggested to check all units (common mistake) and perform simplified hand calculations. They should give you an idea of the stress values that can be expected in this model.

 

[flyforever85]

I agree, quick hand calculation with a constant section hollow tube result in 17 ksi which is very good. I can't think of anything else beside the wrong application of the load and I'm trying to find a way out. Units to me seem to be correct, I'll double check them one more time.

Thanks

 

[FEA way]

In the mechanics of materials books you can also find examples for stepped shafts as well as notch stress concentration factors. However it's clear that the results are too high. Check if the reaction moment at the support is equal to applied torque just for sure.