Rear axle broke right inside cover

[Shawn T. W]

I calculated it. A mechanical engineer would see that the axle is 30.7% weaker compared to full material (bending forces, torsion calculation I leave to yourself), not considering notch effects etc. .

While I'm not doubting you ... May I suggest that Kubota may have taken that into consideration, and just used a stronger alloy?

Let's say they used "alloy A123" at that diameter it had a strength off "100 Lbs" and then bored the hole for whatever reason in it and now it's only good for "70 Lbs" ... So they chose a alloy "B456" that at that diameter was good for "200 Lbs"! Then bored the hole in it, so while weaker than before the boring, it still was good for "140 Lbs"

?

A few broken axles, even on a"new" tractor, which this wasn't ... Compared to the hundreds of thousands of tractors that haven't broken an axle yet, is nothing to have a "class action lawsuit" about, especially since they are different models!

 

[McMXi]

In this day and age these types of components are reliably designed using FEA and different combinations of geometry, materials, and loading conditions. Its not infallkble but it does not make basic mistakes.

Dan

Shear, bending and buckling stresses in shafts are well understood and have been for many, many, many decades. Material selection and heat treatment are also well understood. Many parts aren't designed using FEA, even today, but rather are based on empirical data collected over decades in combination with known best practices.

As for FEA, it's only as good as the inputs so mistakes can and are made, and given the cost of software and paying engineers to use it, it's typically used for more elaborate modelling of parts with complex geometry and loads. Even then, parts are tested empirically to validate the model.

 

[TheOldHokie]

Shear, bending and buckling stresses in shafts are well understood and have been for many, many, many decades. Material selection and heat treatment are also well understood. Many parts aren't designed using FEA, even today, but rather are based on empirical data collected over decades in combination with known best practices.

As for FEA, it's only as good as the inputs so mistakes can and are made, and given the cost of software and paying engineers to use it, it's typically used for more elaborate modelling of parts with complex geometry and loads. Even then, parts are tested empirically to validate the model.

My experience was seversl decades ago and involved bigger and more complicated systems where FEA was used on everything and took hours of super computer time to run the models.

I dont know what tractor designers do today but my ancient (2008) copy of Autocad Mechanical has 2D FEA modeling built in and and wiĺl model any shaft I can draw in a matter of minutes.. Why would Kubota engineers pass that up?

Dan

 

[McMXi]

My experience was seversl decade ago and involved bigger and more complicated systems wher FEA was used on everything and took hours of super cimputer time to run the models.

I dont know what tractor designers do today but my ancient (2008) copy of Autocad Mechanical has 2D FEA modeling built in and and wiĺl model any shaft I can draw in a matter of minutes.. Why would Kubota engineers pass that up?

Dan

There's no doubt that Kubota engineers use FEA, but I would guess that they use it on far more complex parts rather than relatively simple parts that haven't changed much over the last 50 years or more. Talking of Kubota engineers, I had three of them (from Japan) come to my house last winter along with three product managers out of Texas. It was a cold day but they were excited to see an MX and M in action.

 

[whitetiger]

The thread is title a Broken Axle which happens of several models, I was simply sharing what I have been seeing going on the Topic of Broken Axle...

It's not my suit, But i'm watching it.

Several machine shops verified it as a design flaw, But you do not have to be a mechanical engineer to see it.

Still waiting for "Who verified the break as a design flaw??????

 

[GreensvilleJay]

I don't care how 'good' the FEA code is,it's still being run on digital computers with inherent,known flaws. The BEST ,most acurate testing is done in the Real World.

 

[TheOldHokie]

I don't care how 'good' the FEA code is,it's still being run on digital computers with inherent,known flaws. The BEST ,most acurate testing is done in the Real World.

You think digital computing is inherentky flawed and untrustworthy.?

Dan

 

[Hugo Habicht]

I don't care how 'good' the FEA code is,it's still being run on digital computers with inherent,known flaws. The BEST ,most acurate testing is done in the Real World.

Digital computers typically have no flaws. And if you stay away from microsoft the propability of error free software is pretty good too.

And if you do not trust computers, you can calculate the strength of the axle with pen and paper, the equations are known and do not require FEA.

Testing is always required, I agree, humans do make mistakes.

 

[TheOldHokie]

Digital computers typically have no flaws. And if you stay away from microsoft the propability of error free software is pretty good too.

And if you do not trust computers, you can calculate the strength of the axle with pen and paper, the equations are known and do not require FEA.

Testing is always required, I agree, humans do make mistakes.

"Pen and paper"

That evokes memories of college physics. Formal exams were "blue book" events were your answers were recorded in little blue booklets with ruled pages.

My class was AP and given in the freshmen year as opposed to the nornal sophmore year. As a consequence most of us were takeing the prerequisite calculus courses conqurrenyly. As you might guess the class as a whole struggled with exams. The lone exception was Mr Denardo, a student from CA who was there on a full athletic schokarship (Fullback on the football team). He took exams using a ballpoint pen and finished with perfect scores in half the alloted time.

The professer assigned alphabetic grades based on a curve and by the third or fourth time the curve presentation was always prefaced by the words "first we throw out Mr Denardo"

PS> We were allowed to use a calvulator

Dan

 

[targabill]

I am going to reframe from any further comments as this is currently in discussion with some very high level people at Kubota right now. I will start a new post with all the details when some conclusions are made.
I have nothing to do with the class action, as I mentioned, except I will sign up as this happened to me also.
With my very very limited search I found 8 people that this happened to and they all broke in the exact same place.. I have no idea how to calculate % it is weakened but when you take a 2" diameter shaft, and you drill out the center leaving 1/4" around the outer edge it has to weaken it by a substantial amount.

The biggest concern is the Safety issue, when my Axle broke the wheel with the hub actually fell off the tractor as the break is right inside the outer bearing leaving nothing to hold it.

I'll be signing off for now, thanks everybody for chiming in with informative info.

-Bill

 

[TheOldHokie]

I am going to reframe from any further comments as this is currently in discussion with some very high level people at Kubota right now. I will start a new post with all the details when some conclusions are made.
I have nothing to do with the class action, as I mentioned, except I will sign up as this happened to me also.
With my very very limited search I found 8 people that this happened to and they all broke in the exact same place.. I have no idea how to calculate % it is weakened but when you take a 2" diameter shaft, and you drill out the center leaving 1/4" around the outer edge it has to weaken it by a substantial amount.

The biggest concern is the Safety issue, when my Axle broke the wheel with the hub actually fell off the tractor as the break is right inside the outer bearing leaving nothing to hold it.

I'll be signing off for now, thanks everybody for chiming in with informative info.

-Bill

Cant wait to here the details.

 

[GreensvilleJay]

You think digital computing is inherentky flawed and untrustworthy.?

Dan

I don't think , I know the 'math' is. Have 5+decades cutting code in Assembler + 9 or 10 other languages. The root problem is that all floating point 'math' IS flawed( ie NOT accurate ). Even if you use 64 bit for PI and others, it's still not good enough to, say, get you to the Moon and back.

 

[TheOldHokie]

I don't think , I know the 'math' is. Have 5+decades cutting code in Assembler + 9 or 10 other languages. The root problem is that all floating point 'math' IS flawed( ie NOT accurate ). Even if you use 64 bit for PI and others, it's still not good enough to, say, get you to the Moon and back.

That is pure nonsense.

Dan

 

[GreensvilleJay]

Dan sounds like you've never actually looked at how FP numbers are crunched inside any digital computers. There are lots of books you can read about it. Heck even banks do NOT use FP for doing math.

 

[TheOldHokie]

Dan sounds like you've never actually looked at how FP numbers are crunched inside any digital computers. There are lots of books you can read about it. Heck even banks do NOT use FP for doing math.

You would be mistaken. A brief personal history.

I started "cutting code" in 1971. I was a student trainee in the Applied Mathematics Laboratory at what today is the Naval Ship Research snd Development Center., Bethessa MD.

The primary mission of the AML was the development of advanced numerical methods (e.g. FEA) and in particular developing codes for implementing those methods on "state of the art" digital computers. At the time the AML computers were an IBM 7090 and the second of only 2 Univac LARC supercomputers ever built

UNIVAC LARC - Wikipedia

en.m.wikipedia.org

I worked for Dr.. Daniel Shanks who had just published the first ever calculation of PI to 100 decimal digits using the 7090. The 7090 had a 63 bit double precision floating point data format - the sign bit of the second double word was not used. Wikipedia has a nice bio on Dr Shanks.

Daniel Shanks - Wikipedia

en.m.wikipedia.org

During my 15 years there I programmed in muultiple assembly, snd high levek. languages. I helped evaluate, specify, procure, benchmark, and develop applications on many different general purpose and "super" computers. Thats many different architectures and many different floating point formats. I am intimately familiar with the methods used to represent and "crunch" floating point data while controlling rounding error to obtain high degrees of accuracy.

Now as regards the decimal precision and accuracy needed to get to the moon and back may I suggest you read this.

Why NASA Only Uses A Handful Of The 62.8 Trillion Known Digits Of Pi In Its Calculations

The question is how much precision do you really need?

www.iflscience.com

Dan

 

[GreensvilleJay]

well you're one out of 1,000 that understands why FP isn't as good as the other 999 that think it is !

 

[McMXi]

You can calculate the strength of the axle with pen and paper, the equations are known and do not require FEA.

Testing is always required, I agree, humans do make mistakes.

This was my point that FEA is a useful tool, and particularly for complex geometries and loads, but not needed for simple parts such as axles. Even with FEA, which is invaluable when designing a complex part that has to be of a certain mass/material while meeting a required factor of safety, testing will be required due to many variables that can't be accounted for using FEA.

Axles are simple, but if you think about off roading, one of the ways that axles fail is when a wheel/tire is airborne with significant torque applied to it. The moment it makes contact with the ground there's an instantaneous load applied to the axle with lots of variables to consider such as friction, deformation of the tire, tire/wheel slip, mass etc. FEA wouldn't be much help here.

 

[TheOldHokie]

well you're one out of 1,000 that understands why FP isn't as good as the other 999 that think it is !

Anyone that programs engineering applications professionally understands rounding error in floating point operations. More to the point that rounding is a complete non-issue in 99.999999% of the places it is used.

Dan

 

[TheOldHokie]

Axles are simple, but if you think about off roading, one of the ways that axles fail is when a wheel/tire is airborne with significant torque applied to it. The moment it makes contact with the ground there's an instantaneous load applied to the axle with lots of variables to consider such as friction, deformation of the tire, tire/wheel slip, mass etc. FEA wouldn't be much help here.

Au contraire. Thats exactly the type of thing you can model with FEA.

Dan

 

[Hugo Habicht]

I don't think , I know the 'math' is. Have 5+decades cutting code in Assembler + 9 or 10 other languages. The root problem is that all floating point 'math' IS flawed( ie NOT accurate ). Even if you use 64 bit for PI and others, it's still not good enough to, say, get you to the Moon and back.

If you are not happy with floating point use integer, my preferred choice when resolution matters and range is not important. And you get the same result on every processor.

And of course banks do not use floating point. Has to be accurate to the cent / penny / öre etc. etc.

And to the moon, appr. 300000km away, 24 bit mantissa floating point should be good enough to get you back to within a mile or so. Ever seen the Apollo computers?