Dear Chris, I will surprise you by jumping all the comments above and go to your initial question. These three spectra show an oval outer race of that poor bearing. I have seen this many times and when it has been possible to adjust the shape of the outer ring, it has disappeared.
If you can, consider checking if the surface where the pillow block stands is twisting or bending the oouter ring. You could carefully loosen the shadow bolt and see if the peak changes. Not sure but possible. In any case, no reason to swap the bearing yet. I would measure to 2500 Hz to catch the real faults building later.

Hi Arne. Always glad when you join us.

Our of curiosity, did you see the later spectra:
http://www.sixteen-eleven.com/maintenanceforums/

They show higher harmonics in the 60kcpm-120kcpm range and some impacting in the twf.

Arne,

Great to see you posting again!

Unfortunately, in this case, the bearing housing base is dowel pinned in two spots. This bearing is the fixed bearing (as in non-expansion), so adjustments are not really an option. The old millwright trick of rolling the outer race around 180 degrees to run on clean land, was also considered by the customer. It may yet be used if the bearing condition degrades.

However, you do bring to light something that has not been touched on by the previous posts: the effects of operational adjustments.

Being able to adjust or correct a machine condition (misalignment, bearing load, balance, etc.) before it causes a failure is a large factor in the cost avoidance benefits that Reliability / PdM technologies can offer companies. Would you think, perhaps, that being able to detect various conditions, should lead to "pre-emptive" maintenace? Going beyond preventive, predictive or proactive, should there be a category where machine conditions are "tweaked" according to measureable data (IR, vib, temps, etc). to achieve optimum performance? Or is this still under the existing categories of maintenance methods? Just curious ...

Take Care,

Chris - perhaps you should reread the posts on the first page of this thread and think about them in light of Arnie's comments.

Update - the customer had a shutdown from a steam tube leak and is intending to change out the bearing that I have posted data on. The planned October shutdown is now off the calendar, so they want to take the extra precaution of replacing it now. The downside will be that I won't be able to post supporting data regarding the failure rate.

Duncan - Do you see something I missed on the previous posts? I just spent the last several hours helping plan a 30 day shutdown, so maybe you could clarify what you mean.

Chris - you said it's a fan application. How is the fan driven? Is it the drive end bearing or the impeller end bearing? Exact measurement recommendations will follow after I get the answers.

Duncan,

DE or ODE ... umm, yes ... both (you are going to love this explaination)

The fan is single inlet with floating bearing on inlet side.
1200 RPM motor coupled to fixed bearing on one side.
1800 RPM motor coupled to floating bearing on other side.

The fixed bearing, running at 1200 RPM is the data posted.
In 15 years, they have only run the 1800 RPM twice.

The purpose of the two speeds, is that the boiler has three mills and needs the flow from the 1800 RPM. However, they do not have enough demand so they run two mills and the 1200 RPM.

So why the two motors, instead of a dual speed or VFD? No idea.
This configuration has been in place longer than I have.
For what it's worth I have this same configuration in another plant also.

I get to spend all my free time for the next 30 days helping to rebuild a turbine and somewhere in there, I will be helping replace the bearing. I doubt I'll be posting much during that time, but I'll have some data on the new bearing when it's all over. Perhaps we'll be able to salvage the old bearing and I can add some pics for this thread.

Take Care,

On recent spectra on 1611.com: The 2000 Hz spectra are in acceleration which highly stresses the higher multiples of BPFO as well as noise from - I guess - non perfect greasing. If you swap spectra to velocity and compare with what many hundreds of such machines show for years, it would not be possible to call all bearings for repair. But the pending damage from running (note: in my opinon) an oval bearing housing and not very perfectly clean greasing will surely reduce the life time and a spall will finally start. From what has been shown on this fan (although a more complete view on all bearings would give much more meat), had it been here, I would not consider replacing the bearing, but double check the greasing hygiene and maybe use the stop for checking the dust seals.

The area of operational adjustments: Hmmm, most of the analysts I know around let the diagnose on spot guide them to act right on if possible. If the pump is screaming, she or he walks to the control room and tries to figure out what is wrong. All too often it is a small thing to fix. And so on. Improving machines in small steps when you can is very popular from owners since anyone knows the cost of brushing your teeth is drastically cheaper than replacing them on a regular basis. All these activites range from moving resonances to keeping things clean, and so on. What we have been calling proactive maintenance for many years.

In this case I understand that the problematic bearing is pinned and hence has rings inside that fix the thrust to this one. An idea to change the situation could be to move the fixing to the other bearing. But to check that the surface below the bearing is flat and will not distort when bolts are torqued is important. It would probably be quicker to check this and eventually add a thick backing plate or a stiffening wedge across underneath during the revision, rather than swapping the bearing that could include much more handling of motor and coupling and alignment work. Well, not fair for me to try to judge these things from abroad, sooo many local aspects rule in reality. Good Luck in any case. If you swap the bearing, please make sure to measure out the outer rings shape before letting go the fixing (could take the seals out and measure through that gap with a small dial gauge when turning around with housing tight) as well as save it for a knowhow feedback.

Chris - do you have the physical dimensions of the SKF 1315?

Attached is screen from the SKF interactive engineering catalogue shows those dimensions. By the way this isavailable free on-line... doesn't even need sign-up.

skf1315.ppt (108 Kb, 15 downloads)

They also have a pdf generator (in case you can't read powerpoint...didn't want to add a jpg which would kill the screen width)

skf1315.pdf (9 Kb, 10 downloads)

pete - thanks.

Chris - I asked for the dimensions to use DREAM to calculate the approximate dB level for the various direct related spectral lines in the envelope spectrum. Unlike Ascent + Diagnostics, DREAM exposes these levels to the user as part of the diagnostic measurement point setup.

As I mentioned previously using 20 Hz shaft speed, for the bearing flaws, use a 4 KHz to 5 kHz demodulator input band pass filter with a 500 Hz maximum frequency with 400 lines and at least 12 averages.

The defect sensitivities at the threshold between medium and severe defects are:

Rotation around outer race - the symptom is 1x rotation frequency and is usually, but not always, caused by unbalance. - 15 dB

Non-uniform radial tension of the outer race this is a 2x rotation frequency symptom caused by outer race/housng issues (oval outer race). - 10 dB

Misalignment of the outer race - this is a 2x outer race fundamental frequency (fr). 15dB

Outer race wear/flaws - harmonic series for outer race fundamental frequency 15 dB

Inner race wear/flaws - harmonic series for inner race fundamental frequency 10 dB

Rolling elements harmonic series of rolling element fundamental frequency 13 dB

FTF 15 dB

Slip of race - 9 dB

The difference between wear and flaws are how the terms of the harmonic series decrease with increasing order. If there are mixed symptoms, you really need to be able to use the software diagnostics but for simpler cases, eyeball inspection of the envelope spectra can work.

Remember that bearings need a 50 to 100 hour run in period for reliable diagnostics using envelope spectra. Before the bearing wears in and depending on the surface finish quality, you may see things that look like flaws( usually low absolute amplitude) that disappear after wear in.

To also evaluate the shaft line including it's effect on the envelope specta, you also need a normal spectra with history and that's really beyond simple eyeball diagnostics. What I've done with this setup is to consider only the bearing.

Please plot the spectra with a dB vertical axis and a linear frequency axis.

Finally, you can evaluate lubrication using average level of the overall filtered bearing using the zero frequency envelope spectrum line. For this, you either need history or else a comparison of a known, good, similar bearing measured in exactly the same way. Other vendors use the same method.

Arne - I'd be interested in knowing what criteria you are using.

typos r'us - sorry!

Duncan

This message has been edited. Last edited by: Duncan Carter,

Chris - I've seen the two motor approach used on belt driven a HVAC fan mounted on spring isolator. In this case, the "engineer" who said it ignored the low spped end of the spec, used a very small, light weight motor, moving a isolator resonance up to the operating speed of the high spped motor. This, plus belt tension plus pulley runout, had the effect of causing repeated mount hardware failures of the small motor.