Roller shaft failed 6

[sokenyou]

In a paper converting mill, one of our draw roller shaft failed at the weakest point (near the bearing and chamfer with the radius of 0.5). Rachet marks (but shiny) indicates that fatigue cracks were initated at a few locations along the groove. The eccentric pattern tells me that the load on the shaft was imbalanced BUT it's only pressing on tissue papers, and to our knowledge, there was no excessive loads. Two shafts failed within a month... it's really frustrating.

 

[mfgenggear]

sokenyou

Do a full failure analysis.
don't jump to conclusions.

what is different now than what was before.
verify verify verify, or measure three times then cut.
was it the correct bearing clearance/press fit , contact the bearing manufacture make sure what is recommended by the mfg.
Is there history that can be looked at?
what is it the correct material & heat treat.do a sample spectra analysis, run met labs of the failed shafts. does it meet print?
is the correct hardness?
was is it processed correctly during manufacturing?
was any short cuts taken during manufacturing cycle.
are the certifications required for a paper trail? verifies the correct material, heat treat, any process, NDT.
have the bearings mfg changed, is it the correct class of precision bearings, who installed them?
was there any critical operations where operators changed.
was the out side diameters ground or machined correctly?
was there non destructive test completed, on the shafts,IE mag
and so on.

Mfgenggear

 

[chicopee]

If this equipment is covered by your insurance company, have its representative do a vibration analysis.

 

[Tmoose]

Very generally, probably too generally, a material with greater ductility will also have lower yield, ultimate and endurance limits, so fail sooner if subjected to the same (alternating bending loads). It won;t bend unless the stresses exceed the yield strength.

if you really want some advance warning I'd use Zyglo or spot check or some similar dye penetrant product to look for cracks initiating. Might have to shift the bearing over to get a look at the critical area.

The new condition causing failure >>could<< be as simple as a new belt drive assembled with greater overhang distance and tightened by a muscular young millwright.
Or, maybe your roll supplier already (and unbeknownst to you) switched to some low grade badly processed steel.
or is adding a stub or creating a radius by adding material using questionable and uncontrolled weld techniques.


I'd definitely be doing a lot more inspections of broken and new parts, and the installation looking for clues, and some basic design checking of what's there.

Look at the effect corrosion has on steel's endurance limit. page 82 here -

http://www.timken.com/en-us/Knowled...s/Timken_Practical_Data_For_Metallurgists.pdf

 

[sokenyou]

Another shaft went this morning! Same failure - rotation bending.
I am out of idea except spending $30,000 for a brand new roller from the vendor

This is a bitch!

I have gone through a few directions
1. Alignment - we used self-aligned bearings
2. Heat Treatment - hot rolled steel and don't know how to proof
3. Fretting corrosion - why is it only happening to this guy and not the upper roller

But none of them seems to be the root cause.
DAMN!

 

[Cockroach]

Obviously you have a loading issue. It cannot be this " fretting corrosion"' whatever that is, or heat treating. This would happen to so many shafts in such a small period of time?

Look at the bending moment diagram and you will probably see a maximum stress just outside that bearing position. How do you know you don't have a rotating deflection? That would mean you have some sort of bend that when rotating, it goes from a compressive state to tensile condition twice per cycle. So you are old working the shaft at that location.

Show the complete shaft with all loads and positions of these bearings. Clearly you are missing a major issue in the dynamics, I can't see a word solution such as the discussions to date really helping you much.

Regards,
Cockroach

 

[Tmoose]

I'm sorry you're having trouble.

Are the broken shaft from one supplier, and was that "the vendor."

Depending on your location there may be shaft repair geniuses around who can securely attach new stubs of good quality and accurate manufacture.

What is the heat treat you were expecting?
If this is a previously successful design then the manufacturing is highly suspect. There may be some clues in the broken pieces, but you seem to have given up on the investigation.

 

[mfgenggear]

Do a full failure analysis.
don't jump to conclusions

you are not giving enough information to the senior design engineers here to help you.
I see repeated questions that has been given above by all who have posted.
take a check list & verify what each of the questions that have been asked.

if this beyond your experience, " not your capability" then seek help from an outside source.
discuss the issues with your supplier, do the vibration analysis as suggested.
eliminate what it is not.

mfgenggear

 

[sokenyou]

Thanks mfgenggear,

I want to do a full analysis but budget is an issue here. I can't just do a vibration analysis and it turns out it's a waste of money.

You said I am not giving out enough information, but what information you think it's missing that I haven't provided?. Please kindly advise. I have some experience doing failure analysis but I do admit this one is a challenge for me.

We discussed the issue with the supplier but we have tried what they suggested but the root cause is still not determined and nothing much too helpful from the local shop that refurbished our roller shafts either.

We plan to install another refurbished one and do an alignment check or if budget permits, vibration analysis.

If there is non-proprietary information that you think it's useful for the group to help me resolve the failure, I am glad to provide.

Thank you very much

 

[dwallace1971]

How much does it cost each time you have to shut down your operation and replace the part? I can't imagine it would take many iterations of that to pay for a proper analysis of the problem.

"On the human scale, the laws of Newtonian Physics are non-negotiable"

 

[vpl]

Sokenyou, after reading all the above posts, I don't think anyone is going to be able to solve your problem for you at this point. However, just replacing parts is obviously not a reasonable solution. At some point, it is likely your management is going to decide to replace the person (like they did with your predecessor) as the issue isn't getting fixed.

You earlier stated that a new roller will cost you $30,000. I suggest going to management with an engineering analysis that is something along the lines of "We've broken x number of shafts in the past y months. Each time one breaks, it costs us this much in downtime $$$ as well as this much in costs $$. Here's what I've tried so far, for a cost of this $$$. I could recommend ordering a new roller for a price of $30,000 and this much in time and labor $$$ to install, not to mention this much lost production time $$$. I don't want to keep throwing money on this. I think it would be wise to do a complete analysis, including some vibration analysis, to get an idea of where the problem is really occurring. Here's my list of what I think needs to be done and what it will cost. Yes, in the end we'll probably have to make some changes, but at least then we'll know what we're doing so it will be one-time changes not this constant drain. I think this will be money well spent in keeping us up and running."

Saying you have the money to keep replacing parts but not the money to find out what the problem is poor engineering. Your management might not want to hear that, so you need to make your case. But sometimes there's a limit to what you can do over an Internet forum.

Patricia Lougheed

******

Please see FAQ731-376: Eng-Tips.com Forum Policies for tips on how to make the best use of the Eng-Tips Forums.

 

[mfgenggear]

http://www.rexnord.com/sites/Process/Services/Renew/Documents/108-010.pdf

see the above article it gives clear concise failure analysis examples.
note page 14 shafts

Mfgenggear

 

[mfgenggear]

http://www.koyousa.com/brochures/pdfs/catb3001e.pdf

Bearing Failures,Causes & counter measures

Mfgenggear

 

[Cockroach]

I'm done with it.

The magic words, "budget is an issue here" told me everything I needed to hear. That coupled with wordy problem solving, no free body diagrams, no loading, lack of clear, concise engineering practice. I've asked for some numbers, I don't think you got a handle on what the problem really is. So we could be talking about fairies dancing on the end of a pin. You just can't do engineering like that, which makes me wonder about your practice.

Good luck with the wonder through the wilderness.

Regards,
Cockroach

 

[mfgenggear]

http://www.engineersaustralia.org.a...tees and Panels/Engineering Design/Part 5.pdf

Pay attention to page P5-22 & on

Mfgenggear

 

[mfgenggear]

Torsion of Shafts
like this

http://www.me.mtu.edu/~mavable/MEEM2150/Chap5.pdf

Mfgenggear

 

[mfgenggear]

like this too

http://www.nd.edu/~manufact/FME pdf_files/FME_Ch11.pdf

Mfgenggear

 

[sokenyou]

Thanks mfgenggear for all the useful links....greatly appreciate it.
After reading all your comments, especially vpl and cockroach's, I feel bad.... I agree I should've done the engineering analysis properly and get to the bottom of it to provide good recommendations.

Thank you very much.

 

[Tmoose]

If the broken pieces are around it's not too late to learn something.
A first cut at stress in the shaft can be done just knowing the geometry.
I'd spend $30 at the welding supply store and get a few cans of SpotChek dye penetrant chemicals to inspect the "refurbished" shaft before installing it.
Also the areas on either side of the busted shaft(s).

They don't happen to be welding up the bearing journals to refurbish the shafts, to they?

Dan T

 

[mfgenggear]

I realy agree with Tmoose,
or send it out to a third party NDT lab.
use MAG & DEMAG per ASTME1444/E1444M & MIL-STD-1907 Grade A
it would be a minimum charge.

Mfgenggear