I work at a Sawmill/Planer Mill operation, we presently employee about 150 people, me handle short logs and process them into structural lumber, about 72 million fbm per year. In both our Saw mill and Planer Mill we use high speed/high precision bearings, ex. SKF 6317 YC782 Bearings, either in a oil feed or grease feed auto lube system. The problem we are encountering is that when these bearings start to go they deteriorate very quickly, we use live time temperature sensing. This system does not give us enough warning before failure. What I need to know is there a better failure detection system that will give me enough warning to plan a scheduled shut down. The Temperature sensors only give us about a days warning until failure, I was hoping for at least 2 weeks or more.

Mitchell,
I use CSI's PeakVue technology. SPM has their Shock Pulse and the traditional method is demodulation.
They all attempt to remove the low frequency data, using a high pass filter, and emphasize the ultrasonic vibrations.
One of my co-workers put me on to a method that he uses. He gets a bearing fault from demodulation and checks "time to failure" using his ultrasonic gun. He monitors the ultrasonic audio while lubricating the bearing. When he detects the lubrication entering the bearing, he stops and listens. If the bearing noise stays reduced, he can schedule the replacement during a normal down. If it becomes noisy again, he takes an un-scheduled down and has the bearing replaced soon.

This message has been edited. Last edited by: Jon McFadden,

Mitchell,
I spent my first 7 years as a vibration analyst, in the wood products industry. I'm familiar with the equipment your describing. To answer your question: Yes, there are ways to detect bearing failures more than the day before it fails. I've tracked bearing failures for as long as 1 year. When I was in the wood products industry, I had my alarms set so I could give a 30-60 day notice of problems. Temperature readings are not the best method, because the temperature will not increase dramatically until the equipment enters the final stages of failure. There are a number of options open to you. Vibration will be your best indicator. Because of the inherent high vibration in your milling process, it will be best to employ HFD (high frequency detection). This filters out the lower frequencies that will be generated by the saws and milling heads, but will detect high frequency stress waves caused by metal to metal impacting that occurs when you have a bearing defect. Data collection can be done in a number of ways. Hand held data collectors, permanent mounted transducers that send data to a collection point or a computer station. They even have wireless systems you can set up and transfer data via your phone. I could go on all day about the different equipment, methods of data collection and type of data to collect. You may want to hire a reliability consultant to evaluate your needs and make recommendations. You've already made the right choice by posting your question on this board. You won't find a better place to get your questions answered. You may get more than you bargained for though.

I'm with Don 100%. I started out in a OSB mill, and we implemented vibration monitoring on our equipment from mill start-up.

Warning: I'm a vendor.

I agree that vibration monitoring can be just as effective in sawmills and planing operations as it can in pulp and paper, where there's a much more in-depth history of success. Running a paper mill today without vibration monitoring is almost unheard-of.

While continuous monitoring is convenient (and we certainly sell several solutions for doing it), monthly surveys with an appropriate handheld analyzer should give you substantial warning of impending failure (read: months). The right analyzer for the task would be one that gives both detection and diagnosis of emerging problems, and allows you to correct underlying conditions that cause bearings to fail before the bearing itself is damaged, such as correcting imablance conditions, as an example. Analyzers with FFT capability have been the gold standard in this respect. . .but they're no longer the only option. Our algorithm, Spintelligence, offers true root-cause diagnostics that are easier for front-line personnel to understand, freeing reliability professionals to concentrate on other aspects of the overall reliablity program.

There's more info on our instruments and approach at http://www.spintelligentlabs.com

Eric

Vendor,
Why not enjoy the full performance of FFT, Demod in a package for humans with plain English evaluation and builtin rule based evaluation system in the palm of your hand using ME42 the Intelligent Vibrometer and having backup in tricky cases by sending the data by email for human review. Enter the speed of your machine, take the readings as your are guided to do and you will have the result on the screen in your hand in clear text. Complete with PC reporting software: http://www.conditech.com/catalog/
All complete at the same price as systems w/o builtin intelligence, resellers covering US/Canada and more than 30 countries wait for your inquiry, 7 languages available.
Olov

There are several Sawmills in the Northwest using SPM to monitor super precision bearings on planners, curve saws, etc. The key thing to remember is that any super precision bearing (ABEC7 / 9) has very little internal clearance between the rolling elements and the raceways. The lubrication film thickness is very thin under this preloaded condition thus a short time span from a warning to a failure. The Shock Pulse Method will measure the film thickness upon start up and will detect installation faults without setting a baseline. Early detection is the key here. A portable instrument frequently used will offer great insight but an inexpensive on-line system will insure real time knowledge and reliability. Another bearing to watch for lubrication film thickness is the band-saw bearings with stationary shafts.

I thought the ABEC/RBEC number was just the tolerance class telling how precise the dimensions are but not the internal clearance.

The internal clearance would be something like C1, C2, C3, C4. I have a hard time decoding that suffix to determine the internal clearance. If it's a tight bearing I expect to see a C1 or C2. All I see is C7 which if it were a clearance would be loooose but I don't think the nubers go that high. Maybe these bearings have some other system outside the normal numbering system C1-C4? I know I'm confused.

The following suffix designation on a SKF Bearing goes as follows on a 6317 YC782
6317 - is a single row, deep ball bearing
Y - Press Brass Cage Ball Guide
C782 - C78 + C2 (C78 = Dimensional & Running accuracy to ABEC 5 Tolerances; C2 = Clearance)
Hope this clears things up.

The ABEC numbers are solely for the tolerances of the bearing and have nothing to do with clearances. This is a pdf of the tolerances of the various classes from the Barden website http://www.bardenbearings.com/scan%20pdf/ABEC%20Brochure.pdf

quote:

Originally posted by Stan Riddle:
I'm with Don 100%. I started out in a OSB mill, and we implemented vibration monitoring on our equipment from mill start-up.

What type of press did you have?

Mitchell,

I wonder if you have performed some kind of failure analysis on the failed bearings?
Was it fatigue type of failures in the races(pitting, spalling, etc.)?
Or may be it was a cage failure?
Or possibly improper bearing loading/installation errors?

What I am trying to say is that natural bearing wear will produce patterns even in normal ( not demodulated) vibraion data well in advance (more then two weeks) before a catastrophic failure actually occurs. But if you are experiencing unexpected failures, it is either vibration signals were missed by the analyst or rapid progression of a failure, such as due to loss of lubrication or cooling loss, had occured. In the latter case early bearing detection methods (like demodulation) won't help much.

If possible please post pictures of the failed bearings.

David_g,
I think temperature sensing is all they are doing right now.
Michell is looking for another way, besides temperature sensing to catch the bearing failures sooner.

Thanks Mitchell and Mike B for answering my question.

Mitchell - I think it is widely agreed that vibration provides a very early warning for most bearing degradation mechanisms - much earlier than temperature in most cases. If there is damage on the races or balls you can detect it with vibration analysis.

And I agree with David that if you are having repetitive failures it can be informative to cut apart a removed bearing and try to understand what went wrong.

To give a little more back ground on these bearing failures this is what has happened over the last two years (in our Planer only, I am not directly involved with the Sawmill yet). We did send out the failed bearings for analysis and found that installation was the problem, retraining started and the failure rate went from every 2 months to every 4 months. Then we took a look at the auto lub system and found that greasing some bearings was more benifical, the failure rate went to every 6 months. To extend this period we started to replace the bearings on a calendar cycle and have not had a failure since, but we are changing bearings that in most cases are perfectly good and there is always the chance of human error occuring during installation even with the best training. We know there are still other underlying root causes to our bearing failures and will eventually find them, but we still need early detection. Run longer, run faster, run cheaper, that's the name of the game. Early detection is not the cure but it will help with recovering lost down time.

Mitchell, another huge benefit to having vibration equipment, is being able to verify that installation was done properly. As you're finding out, some installed bearings last longer than others. You're heading in the right direction with your PM's, but taking your program to the next level will help reduce costs even more. As you stated, some of the bearings that are being replaced on a PM schedule, don't have anything wrong with them. We had spindles running at 7200 rpm, with ABEC 7 bearings that lasted 10+ years. The biggest problem we had, was operators over tightening belts on the spindles when they started getting "ripple". It's good that you've increased the life of your bearings from 2 months to 6 months, but to be blunt, that's not good enough. If the average life of your bearings is 6 months, then you have an underlying problem somewhere. There are some relatively inexpensive analyzers that you could get started with. If your management requires cost justification, that can easily be done. Let us know if you need some ideas on how to do that.

If you were going to start vibration monitoring, I would include HFD Pre and Post Lubrication readings. One (Pre) for condition of lubrication in bearing and when it is no longer acceptable. Second (Post) for how it affects the HFD reading and for determination of how long it remains low or acceptable. (Example come back thirty to sixty minutes after lubrication and take the Post HFD reading to record whether or not it remains down. If it does the problem was lack of lubrication, if it does not remain down then most likely you have infant bearing faults beginning or another condition that you will have to determine the origin of at that time. You must remember two things. In my experience when using HFD readings for lube condition data the further away you are from the bearing (I.E. Motor end bell non-drive) the HFD readings will not be as reliable, due to the motor's high frequency influence. The second is that if resonance is present in any form this will also render the HFD reading unreliable. Therefore, an ENV spectrum (demodulated) with a number 3 filter is used, for every horizontal position @ every bearing of concern. The overall noise floor rise will indicate the lube condition (unacceptable) in this instance. After lubrication, the risen spectrum noise floor should sub-side. These spectral readings will not be influence by the high frequency created by the motor components because of the filters contain within them. I would set a Pre and Post ENV(3) spectrum as well if you are going to be completing the lubrication during data collection. Not only will the ENV(3) spectrum allow you a look at the lubrication health in the bearing, it will allow you a very early look @ infant bearing faults thus allowing you more time to trend them. Every bearing is different dependant on load, speed, and environmental conditions so lubrication will vary upon these conditions. I use certain manufactures monitoring equipment and this is their nomenclature for this spectrum, the other companies describe this type of spectrum differently. So, dependant on what manufacture you choose the descriptive term for this spectrum I have described will differ. It has work well for me. That is not to say my statements within this reply will be widely accepted. One last comment if you will allow is that over lubrication will harm bearings more readily than to little. Therefore, if not vibration data for lube at least utilize ultrasound while introducing lubrication to your bearings to insure that over lubrication is not contributing to the premature demise of your bearings.

Regards, hope this is of some value to you.
These observations and opinions are just what they are mine and have served me well.

Mitchell,

*** vendor warning ***

Failure analysis is an excellent point to start when improving reliability. The only problem is that many bearing manufacturers blame the installation, the installers blame the bearings and lubricants, and the lube supplier ... (etc., etc.) Find an independent (unbiased) lab for failure analysis.

As far as what direction to take from here, I would affirm that vibration analysis is THE most cost effective tool for advance diagnostics (read: planning). There are two "basic" options that are commonly used:

1) spend "a bunch of money" on vibration instruments and training
2) hire an experienced consultant to monitor the vibration

The benefit of option #2 is that IF if doesn't help your situation, you haven't invested too much capital. Additionally, you gain a familiar source to train your people, should the technology prove effective and you decide to purchase instruments.

*** the vendor warning is up there because we sell vibration analysis service (but I would recommend you use someone local, and we are in Ohio)

Take Care,

To answer Sparky first I'l have to get back to you on sensor location, I'm not directly involved in that Plants Operations.

To answer Chris, as of now, we are looking for an out side contractor to do walking around testing. In the past we were doing this but lost the service do to insurance problems on the Contractors side.

I am interested in finding someone to do Bearing Failure Analysis beside a bearing Manufacture. Can anyone give any suggestions as to who? We are loacted in Chipman, New Brunswick, Canada.