pic: Broken Output Shaft

This is the output shaft we broken in San Diego QF2-2.

Referenced in this post: http://www.chiefdelphi.com/forums/showpost.php?p=976105&postcount=19

Wow, looks like our arm after Champs in 2007.

What was the hole for? An encoder? Looks like that was the cause of failure.

It almost looks like a dog shaft to me from a transmission. Which would also explain why it failed where it did.

the pullback for our kicker was originally made of 3/8 aluminum, something similar happened, 340 ft lbs on a 1/8 key in aluminum, doesn’t end well, after making it outa steel we were fine though.

What did the re-design look like? Did you switch to steel?


Those grooves cut into the shaft are a HUGE no-no is shaft design. It probably didn’t help that there was a hole drilled through the middle, but those grooves are big stress risers where the shaft is experiencing the most stress.

Funny, that’s how tons of teams have held on their wheels, gears, sprockets, and everything for years. I’d probably argue that a keyway increases stress in a shaft more than a snap ring groove.

The grooves are very large stress risers in the shafts. That being said, it doesn’t mean that putting grooves in a shaft is a bad thing, its just something you should consider when properly designing a shaft.

A keyway may be more of a stress riser in a certain load application. However, for most of the applications we look at in a shaft, like handling the torque of a drive train or cantilevering a wheel, the keyway probably does not create nearly as much stress. It’s all about the orientation of the problem.


Could someone with some MechSkills post an FMEA of the stress in a design like this vs a key-way?


I’m certain the large slot for the dog shifter was more of a stress riser than the snap ring grooves.

And Chris, not that most teams use snap rings on the extremes of the shaft, where there is no load being applied; it’s less common to see them between loads like this.

It wouldn’t be a universally accurate demonstration.

The location of all such features and loads would have a large effect on the result.

The chart on this webpage should sum it pretty nicely. For some reason I can’t directly link to it, and google images doesn’t turn up any other good ones, so go to this website and search for “Graph illustrating the effect of proper fillet radii on stress concentration” (minus the quotes) and there it will be.

The factor on the left (Kf) is a multiplier to the max torsional stress (and when you exceed the max torsional stress of the material, it will fail!) you find if there was no change in diameter. So, you can see it isn’t hard at all to multiply your stress significantly for relatively small changes in shaft diameter.

The initial design of our robot included a hanger that was powered off a PTO from our drivetrain. Step hexing is really pushing the limits of our machining resources and was something we wanted to try and avoid. By having a single speed transmission we figured we could get away with using a 1/2" hex shaft instead of the 5/8" hex shaft commonly found in other 2 speed transmissions. However, we still needed a way to shift the drivetrain into neutral. The hole at the end of the shaft is there for a shifter to shift the output shaft into neutral.

The shaft failed right where the snap ring groove is. We actually continued to run the same (old) shaft design on our practice robot until just before Atlanta when both shafts failed in exactly the same spot.

In further analysis, we had a couple things working against us:

  1. The hole diameter for the shifting pin was made too big and the hole was drilled too deep (past the snap ring groove).

  2. The snap ring groove further compromised the strength of the shaft. I don’t have the numbers in front of me at the moment, but I believe the wall thickness would have been just under 1/8".

A few days before ship day, we made a decision to ditch the hanger so that we could put more effort into making our ball grabber and kicker better. When we ditched the hanger we no-longer had a need for a transmission with a PTO. We also didn’t want to remove the PTO entirely just in case we wanted to add a hanger later.

After San Diego when we sat down to begin redesigning the transmission we stripped out the entire PTO. Since now we were now running a pure one speed, we could get rid of the slot and hole for the shifting assembly.

Before we crated in our robot in San Diego we pulled a couple drive axles (which were also 7075 Aluminum) from the robot just check and make sure that they weren’t breaking or bending. Those shafts were okay so we knew that 7075AL would be fine for our redesigned output shaft.

This. They’re fine if they’re no appreciable torque is being transmitted through it, which is the proper way to utilize snap rings. Heck, two of the FSAE cars I worked on had drive shaft components retained with snap rings in no-load areas.

About an FEA model of the stress concentrations… in my experience many FEA programs don’t pick up on stress risers much of the time. I just tried to do it with Solidworks Simulation and the program didn’t pick up on the groove as a stress riser. This is a problem when a designer doesn’t use common sense and simply trusts FEA results without question.

Here is an interesting FSAE thread on a similar topic with a very good explanation about why the drive shaft failed at the groove.

Again, non-torsionally-stressed grooves are okay. Torsionally stressed grooves will very likely fail.

You have a large aspect ratio in this problem - long shaft and short, small keyway.
Make certain you Add-in SolidWorks Simulation under the Tools menu. SimulationXpress will only give you a simple approximation. You need full SolidWorks Simulation.

You will need to adjust the mesh to be finer around the keyway.


What keyway?

sorry - tried to answer this on my iphone. maybe it is a grove?

Learning how to use COMSOL 4.0 at work (an excellent FE program for those who are interested, much better than COMSOL 3.5) I decided to see if it could show the stress risers in this particular situation, and it did so quite well. I’ve attached two plots. One plot is of the surface stresses, which does indeed show the stress concentrations in the groove. However, I find the slice plot much more illuminating as to why this groove is a stress riser. If you look at how the shear stress flows through the part you can see that it likes to stay near the outside surface, but the snap ring groove abruptly forces stress towards the middle, compressing the stress (if you will) around the groove.

Pay no attention to the magnitude of the stresses, I just picked a load number out of the air that was close-ish.

PS- I hope no one is too upset about reviving an older thread.

Stress plots posted above.

No FEA is necessary. Contrary to popular belief. A lot of good analysis can be done with a book, pencil and a napkin! All good ME design books will have a reference to Stress concentrations. They are generally relatively simple to use as long as you can read a chart.

Engineers edge (www.engineersedge.com) is the 1st pass that I usually take when I’m looking and I don’t have my book handy and the internet is available. Here is the section on stress concentrations: http://www.engineersedge.com/calculators/stress-concentration/stress-concentration-shaft.htm

This won’t give you the stress unless you know the rest of the numbers, but you can at least compare Kts for the known geometry to determine what has less of an impact.

Shigley’s Mechanical Engineering Design is my text book of choice. It was required in my college ME Design class. It is still something that I use regularly and has a lot of great references.

Good luck!