Introduction:
Fluid film bearings are mainly monitored with proximity probes. It is often stated that “you can’t detect early defects in fluid film bearings with normal vibration techniques (Velocity, Acceleration or a bearing condition unit)”. But in fact you can detect the effects of a fluid film bearing deteriorating with a normal accelerometer.
Under abnormal circumstances metal to metal contact might occur, leading to occasional high-frequency noise that can be detected with normal vibration equipment. The following case study is a great example of this and also using lubrication analysis as part of a maintenance program.
This case history covers a Production facility Extraction Fan which has been monitored as part of a site wide Condition Based Monitoring program. The drive of this program is to integrate condition monitoring techniques and to drive the maintenance program.
This fan unit had a motor that is a direct drive to a fan shaft, the fan shaft has a white metal fluid film bearing. The fan is a standard overhung centrifugal of about 2.5 meters diameter.
Vibration Analysis:
We flagged this reliability risk very early in the program for having a considerable 1 Order impact present in the PeakVue™ data (see Figure 1).
Figure 1 shows the initial Vibration data collected on the fan bearing with the 1 Order event present.
Vibration Trend Analysis:
Figure 2 is the PeakVue™ Trend of the bearing as it was nursed through until maintenance could be conducted.
This showed the initial level, reduction in levels after an oil flush, then a period of monitoring until another oil change and a bearing inspection.
Lubricant Analysis:
As the initial fan data had a dominant 1 order event and was being monitored on a monthly basis, we needed to determine whether the event was consistent or deteriorating, and what possible causes were.
Oil samples were then taken (see image 1), on visual inspection the oil was in a very poor contaminated condition. The lab report (see image 2) stated a serious concern with particulate matter contamination and high levels of Sodium, Iron, and Tin.
The indications of Tin suggested probable sleeve (Babbitt) wear of the bearing.
Image 1 is the Fan DE bearing oil sample:
Image 2 shows the initial oil sample report and diagnosis with emphasis on the high Tin levels:
Maintenance Actions:
This fan was a critical component of the facilities production process, however with a plant upgrade planned for the very near future, the decision was made to closely monitor the deterioration of the assembly rather than to rectify this potentially expensive piece of equipment.
The temporary measure of an oil flush and change was conducted immediately, with a visual inspection planned for the next shutdown.
Maintenance Inspection:
During the shutdown the bearing housing was split and the bearing shells separated. The damage to the bearing was very extensive with ‘scalloping’ of the Babbitt material evident in the direction of rotation forming a build-up at the end of the lower shell. The cause of this is most likely the failure of the lubricants ability to sustain an adequate oil wedge between the shaft and the bearing.
Image 3 shows the Fan DE white metal bearing upon inspection:
Image 4 shows the Lower half shell of the white metal bearing with a piece of ‘free floating’ Babbitt that was found in the sump:
Conclusion:
In conclusion, with the company’s utilisation of all the available Condition Monitoring technologies and tools, they were able to monitor and be consistently and accurately informed of the state of deterioration of the bearing. This allowed them to implement a rolling program of temporary measures to stave off what was essentially an unserviceable critical machine until the Factory upgrade was conducted.
.
A reliable plant is a safe plant
…..an environmentally sound plant
….. a profitable plant
……a cost-effective plant
_________________________________________________________________________________________
7th August 2018: Additional data as requested by JUAN CARLOS URQUIOLA
Below is the Velocity data from September 2017 and June 2018, there is a difference in the number of an amplitude of the 1 order as associated harmonics.
Below is the Acceleration high frequency spectrum, this shows two mounds of activity that has side bands of 1 order and 2 orders.
As noted in other comments, this is a good example as to how you can not only forecast but predictively and with minimal risk, provide a scheduled shutdown. Looking at the final phase of the failure, I am rather surprised that it did not have a catastrophic spell you as noted in other comments, this is a good example as to how you can not only forecast but predictively and with minimal risk, provide a scheduled shutdown. Looking at the final phase of the failure, I am rather surprised that it did not have a catastrophic Failure, spinning both have’s in its bore for example.If the author of this review is reading this, it would be good to see what your maintenance frequency was for this piece of equipment prior to the visual observation seem in the oil analysis.
Hello Denis,
We are contractor’s supporting this manufacturing production business. We try to help guide their maintenance and reliability, but change in a factory +30 years old is slow.
We are not sure of their maintenace practice for this asset, they tend to leave things alone.
Regarding the oil analysis we could visually see the oil was in a very poor state with floating particles.
This fan is heavily engineered and of an old design. The bearing housings are water cooled jackets and the two bearing halfs are secured well.
Kind regards
The Seasoned Analyst
Congratulation. Good detection Thanks for sharing you case. Only I have a question. Did you see a high-frequency noise or any kind of anomaly there? Could you share the normal data please?. On sep17 levels looks similar than last data… but data collected subsequently shows a significant decrease.I supossed Load conditions are identical.. so I am wondering what happend.. Did you do something? Thanks in advanced for the information.. and congratulation.. really good detection.
Hi Juan,
Thanks for your comment, I will add some additional data at the end of the original post.
The only change that reduced the levels was site performed an oil change, took out a lot of the contaminates and put new oil in.
Kind Regards
This is why regular oil sample analysis is so valuable in the early detection and prevention of catastrophic failures in capital rotating assets.
This is actually a huge failure of the proponents preventative maintenance sytem.
Once repaired, 6 monthly oil sample analysis should be carried out. Once good lubriation is established the sample period can be extended.
Hello Will,
Totally agree with your comments and great advise, thanks.
Hopfully this company now sees the benefit of a regular oil analysis CBM program.
Kind regards
To correct my original statement/typo. It should read 6, 1 monthly oil samples should be taken taking care not to remove too much of the oil inventory. To add, proximity probes are really only useful in this case for detecting an out of balance of the rotor. With correct contamination free lubrication these bearings should last indefinitely.
The bearings in the photograph look like they have been damaged by very recent and catastrophic oil contamination. The fact there was no reported oil contamination does not make sense.
If it were my call, I’d say they ran it dry of oil over months or years of neglect or an oil supply failure, panicked, then put some heavily contaminated oil in when they realised their mistake.
In my humble opinion anyway.
I was involved in a study of which the conclusion of this study was that eddy current probes are unsuitable for large fans. The main reasons for this were;
1) There is no direct relationship between the eddy current probe readings and the shaft movement inside the journal.
2) The quality of the eddy current readings would deteriorate with time.
3) Eddy current probes are not good at detecting problems with the support structure.
Hi Remy, thanks for sharing ypur knowledge, comments and input. Appreciated.
Useful case study…
Please send me such case studies
Hello Veeresh, just subscribe to my blog though the web site and you will get an automatic email every time I put a new case study up. Regards
Just curious to know, doesn’t this bearing have a bearing metal temperature measurement thermocouple(s). Obviously, this bearing showed a severe wipe and it would be surprising to know if there was no indication of high metal temperature and/or vibration during or post the initial wipe event.
Hello Kiran, unfortunately there are no temperature monitoring or measurements on these fan bearings and the vibration data was initiated post bearing damage. Regards
Looking at your spectrum, the large number of harmonics in the vertical direction was likely due to “looseness” within the bearing due to increasing clearance. The dominant peak at 4 kHz was likely an end-stage “screech”. I have found using headphones (Bluetooth earbuds with a 2140) very valuable for this sort of thing, especially where noise is intermittent and a snapshot spectrum may miss it.
Hello, Do you still have the oil sample with the metallic wear debris. I would like to evaluate detection with our MetalSCAN sensor (ODM) technology.
Hello, unfortunately the sample remained with the client and has since been disposed of.
Can you also post the velocity spectrum of the bearing monitored?