Good Day All
I am looking for a way to monitor a slow speed shaft approx 15 RPM which is on our dryer trunions for shaft failure. We had a failure yesterday and there was not an indication from the online system that there was an issue. Funny thing was though the sensors on the opposite side trunion did pick up on an axial event at the time the failure occurred. But only minutes before complete failure vauge to say the least considering it was on the opposite side of the dryer. I tried once before to resolve this issue but as with most things that happen once in a blue moon it was put aside to move onto another issue. Happens alot I know but now we are back into finding out how we can prevent this. If the drum is full and catches fire when this failure happens then the drum can actually warp from the heat we have already had this happen and do not care to repeat history. So from the vast expanse of knowledge on this forum I am hoping to have someone suggest how this can be achieved. I thought of a couple of things but after consideration do not think I have found the answer. Any idea, concept or methodology would be appreciated and wlecomed. Thanks in advance and if info is needed I will supply what ever I can to help.

Lee,
There was a similar thread going on one of the other boards recently which may give some useful info on monitoring slow rotating equipment.
See:
http://maintenanceforums.com/eve/forums/a/tpc/f/119103451/m/3101077213

quote:

We had a failure yesterday and there was not an indication from the online system that there was an issue.

Funny you should mention an online system, what with all the talk about how good they are and how it will not be long before they replace the guy with the magic box.

You mentioned "shaft failure",What type failure, actual shaft or bearing or what?

Have you a picture or drawing of the setup of the dryer?

Hi Folks
Vibeguy I have already got good info on the bearing end of this thank you .
Ralph
I know what you are saying but we had the same issue the last time one of these failed. I had contacted one of the kiln people we deal with and was told after some discussion that in this type of event such as a shaft failure this noise does not travel well through steel. In the last occurance we did get a reading it was at the same time the shaft failed. This crash was not near as big as the last and I was suprised to see nothing on that particular bearing at either end of the shaft. What really made me take a second look was the bearing opposite did show an event at the exact time very wierd. We have two sensors per bearing one horizontal and one axial both peakvue readings with the option to obtain a regular spectrum if you wish. This is a drum aprrox 30' dia.and around 60-70'long has a two thrust wheels on the outfeed tyre. The bearing setup for the trunion which is about 4' dia and 2' thick is an SKF 23234cc in each housing. I have no issues with the online system as far as monitoring the bearings it does this job very very well. I know the online is not meant to monitor shaft breakage so we should not pick it apart for not being able to perform this function. What I would like is any information on how to monitor a shaft for breakage the breaks in this case happens right where the shaft meets the wheel. I will post pics

some more

Lee,

Judging from what you replied, the bearings are still in good condition, no wreackage in the bearing, huh, just shaft breakage?
The "cutting/worn" area in the picture with your hand showing, what caused this, continued running after the shaft broke? What was it rubbing on?
Is this the trunion wheel that drives the drum or is it driven by a chain?

This message has been edited. Last edited by: Ralph Stewart,

lee,
I did a ton of work on crack shaft detection for the nuclear industry during the 90's and have several patents on the method. We never did patent the method I'll give you here. Instrument the end of the shaft with a torsional measurement system. Perform a rotor dynamics model of the roll. Identify the torsional modes that have a high degree of torsional bending in the area where the break occurs. I guarantee that you will see a large percentage decrease in the torsional frequency of interest as the shaft starts to crack. The threshold of crack identification for the method was approx. any crack greater than 3%.
Good Luck.

Lee,

Do you have any monitoring on the Driver, at all? I'm unsure if the system is the same as the one I had worked/proposed for a client in the far east; in a Cement Plant.
Monitoring of the Motor (DC) current and temp, did help prevent similar failures, for them. Was this checked, in this instance?

Cheers...Rajan Muthukrishnan

I'm not too clear what I'm looking at.

Is the steel plate with the welded on lifting lug the end of the dryer/drum? If so, the details around the inserted shaft leave a bit to be desired. Generous radiuses, residual compressive stress (shot peening, nitriding) and protection from fretting, scarring, etc can make huge improvements in fatigue strength. Like, to never failing again.

A spotcheck/zyglow inspection a few time a year might have detected that crack when it was just starting, maybe even a long time ago

Anyhow, if I needed to monitor similar shafts I'd look at gundrilling a few blind axial holes to the the depth of the failure, tapping the ends NPT, squirt in several ounces of bright or fluorescent dye (spotcheck?) and pressurize them, maybe even leaving a gage or some of those tire pressure monitors on the end of the 15 rpm shaft. Loss of pressure or appearance of dye during the PDM route would be an indication to check further.

The appearance of the shaft makes me think one centered hole would not leak soon enough.

Might need 3 or 4 equally spaced out nearer the surface

Lee,

I was also thinking about torsional resonance as Ron suggested. You could also monitor a bending vibration mode that would have high stress at the failure point. In fact either a torsional or bending natural frequency may have contributed to the failure. In the case of bending natural frequency, it should be measurable on the bearing housings. I would use high frequency resolution, high number of averages (32), and log amplitude. A crack would cause the bending frequency to reduce. I am using this method to monitor a large pump case for cracking.

Walt
w_f_strong [at] msn [dot] com

This is from left field, but I know that audio sensors have been used to detect the sound of concrete breaking in parking garages and whatnot. Perhaps someone has a similar sensor to detect the sound of a cracking shaft.

Lee,

I see where you said that this happened at least once before and the drum warped from a fire inside.

Is this latest one in the same position as the previous one that broke?

Does a chain drive the drum or does a set of the trunions drive it?

Have you a picture of the drum and its trunion which keeps breaking. How ong is "once in a blue moon" between breakage?

There has been a lot of discussion in this topic in the nuke industry as Ron mentioned, and especially in the last few years since an RCP pump shaft cracked at a TVA plant a few years ago.

Since it's not on the motor side, I haven't paid much attention. From what little I have heard, you might possibly catch a change in vibration (perhaps an increase in 2x) but only if you are very very lucky. I think various off-line NDT inspection are emerging as the most popular method for early detection.

Classical and proven way is to monitor phase shift of 2xRPM vector, that works at least for not so symmetrical 600MW generator rotors at 3000RPM where there are real examples. Resonance change as discussed should also work. If you do knock test it for that resonance, do it in a couple of rotor angle´s like 90 deg apart so you are sure you do it with a crack opened. Olov

I'm not sure that I can contribute to any discussion on a method of detecting a crack in a shaft rotating at 15 rpm but...you could detect when the shaft had broken by comparing the rotational speed at either end of the shaft. This could be done by placing a toothed wheel at each end of the shaft and observing each of these by a magnetic pickup.
ou would then need a method of comparing the speed at both ends of the shaft and triggering a trip when a difference in speed is detected. If you have a DCS system, making each speed output available in 4-20mA format would make comparison of the speeds fairly straightforward

or....

How about using proximity probes to observe each end of the shaft i..e axial arrangement. I would have expected a well developed crack to have caused the apparent length of the shaft to grow by a couple of thou - at 200 mV/mil (or greater) you should get a detectable change.

This could be a relatively inexpensive way of doing things since, with one (or two) probes 'looking' at each end of the shaft, you would only need a multimeter to measure the dc gaps volts - bang the readings into a simple excel spreadsheet to trend and look for differences between the ends.

With proximty probes, I guess you could consider installing a monitor designed to detect abnormal growth in turbines or use a datacollector to look for waveform changes that would indicate any 'swashing' that might be associated with a crack.


Just thinking out loud really...

Just a at the time of the moment idea, if both ends are available, you could measure the variation of the shaft twist by having two edddy or tacho probes picking up 1 puls per rev each and by comparing the pulse timing getting a number on the shaft twist in degrees or in parts of that and then at normal constant load see the normal value. If the diameter of the shaft is reduced by cracking, this twist angle should change. This would require pretty constant load or when empty. Olov

We had a cracked motor shaft, the motor tripped on high vibration and never wants to restart. Upon opening the coupling cover, it was revealed a crack on the motor shaft end.

Glory be great day in the morning
Gentlemen this was like christmas or drilling for oil and striking a gusher. Was not expecting this many answers overnight!!!
First thing first This trunion set up does not drive the drum this unit is driven through a santisalo gearbox, fluid coupling on the drive side and a gear on the output the other two unit have the same drive input type just the final drive is a chain. We monitor the shaft for motion and this is what tripped. It may have spun a couplke of times due to inertia but this time the shaft did not come apart and cause the trunion to turn in the frame as it did last time. Goood thing!! There was also no fire to cause the drum to warp another good thing. All the ideas you have will be considered as you have made a few. I also made note that the nuke regulatory commission has made some comments and recommendations regarding shaft failure as well and I am in the process of reading these to see if we can apply anything to our situation.Dan this is pics from the side of the trunion wheel before they lifted it out and the end of the shaft.The drum rides on four of these units.Okay so I will take some time and digest all this and willpost the results when we arrive at a descision again thanks for the quick and great responses this place is like having more the a few brains and boy does it help

I realize you want to "monitor" for the break, but I would focus on th root cause of the break. Our job as an analyst, is to "stop" failures if at all possible. THere has got to be a reason for this.

Where I worked as a millwright for many years we had two drums or kilns or dryers (whatever one wishes to call them ) about 200 feet long and at least 20 feet diameter and in the 38 plus years I was there we never had a trunion shaft to break.

Does your drum wobble or run off center at the trunion which might cause a load change on this shaft which would cause it to flex back and forth until breaking.

I recall a 4 1/2 inch shaft coming out of a gearbox driving a paper machine dryer section that kept breaking from being misaligned and flexing back and forth and it only turned about 15 rpms. After 2 breaks in less than a month, a universal joint drive shaft solved this.

Point is::: A huge shaft will break if continued flexing is involved, just as a small piece of wire will, when bent back and forth.

Just some thoughts, that may be way out in left field.