Low-speed bearing
monitoring
Challenges and Solutions

Presented by: Oscar van Dijk
Presented to: BRCE conference
Date: 23 March 2016

Agenda
1. Low-speed bearing monitoring
● Challenges
● Solutions
2. SKF's commitment in this field (History, techniques, patents)
3. SKF Multi parameter measurements
4. Case studies
5. New approaches
● HAL
● AEE

© SKF Group

Slide 3

22 March, 2016

1
Challenges

Challenging monitoring applications
● Large size bearings are usually more difficult to monitor due to:
● Low speed applications
– Steady speed
– Non-steady speed

● High “noise”, low bearing defect energy
● Contamination and insufficient lubrication

The most common problem is lack of understanding of:
• The application conditions
• The appropriate instrumentation, techniques, and
configurations

© SKF Group

Slide 5

22 March, 2016

this originates from one or more common mistakes in setting up the vibration measurement in the first place. ● Some common errors include: – – – – – – © SKF Group Low-frequency sensor use Incorrect mounting method Incorrect Enveloping filter selection Inadequate measurement time length Not accounting for non-steady speed Not accounting for load zone Slide 6 22 March.Low speed bearing monitoring – Issues ● “Undetected” failures on slow speed machines leads to the assumption that finding them with vibration is not possible. ● In most cases. 2016 .

© SKF Group Slide 8 22 March.well below the upper boundary of the demodulation BPF.Low speed – low frequency accelerometers? ● There is a common misconception that a low frequency response accelerometer needs to be used on low speed machines because of the lower energy levels/bearing frequencies. ● However.2 to 2 300 Hz Impact energy . the upper 3 dB point is only 2.3 kHz . ● A typical 500 mV/g low frequency model has a lower 3 dB point of only 0.4 to 1 000 Hz ±3 dB: 0.2 Hz. 2016 Frequency range: ±5%: 0. ● The use of a standard 100 mV/g sensor is better – these typically have an upper 3 dB point around 14 kHz. fully encompassing the BPF range.6 to 700 Hz ±10%: 0.

For low-speed – standard accelerometers No signal components from within this range are used in the band 3 enveloped acceleration CMSS 2100 Frequency range: ±5%: 3.0 to 5 000 Hz ±10%: 1. one should still use the standard sensors! . 2016 For Bearing Analysis on machines with a low rotational speed and envelope band 3/4.0 to 5 000 Hz ±3 dB: 0.5 to 14 000 Hz CMSS 2200 / CMSS 2200-M8 band data capture range Useful range of sensor No signal components from within this range are used in the band 3 enveloped acceleration band data capture range Frequency range: ±10%: 1.0 to 9 000 Hz ±3 dB: 0.7 to 10 000 Hz Useful range of sensor © SKF Group Slide 9 22 March.

Mounting is important – use stud mounting ISO standard 5348 Non-linear response Amplification by the sensor Harmonics and energy from ‘over-rolling’ of a defect are still present in this range Reference Linear response Near zero amplification by the sensor © SKF Group Slide 10 22 March. defect frequencies are gearmesh can be in this area in this range . 2016 Fundamental bearing Noise from e.g.

© SKF Group Slide 11 22 March.Sensor mounting makes a big difference ● Here is an comparison between data recorded from two identical 100mV/g ring-mode sensors: Magnet – One mounted by drill & tap/stud – One mounted by magnet. 2016 Stud Mount . but particularly so for slow speed machines. ● This is true for most measurements.

– Slow variations in the speed can mask bearing frequencies © SKF Group Slide 12 22 March.Constant speed? ● With low speed machines – Measurement times are longer – Typically we would like 10 – 15 events during the measurement time ● Constant speed – With long measurement times. 2016 .

8 RPM Delta 9.Speed variation over 16 minutes 11.2 RPM 0.4 RPM Speed changed: • 16% during the interval • About 8% within a typical measurement window © SKF Group Slide 13 22 March. 2016 .

Sensor Location ● Small signals – Mounting the sensor will make some difference © SKF Group Slide 14 22 March. 2016 .

Sensors – Load Zone ● The next consideration is location of the sensor – as close to the load zone of the bearing as possible. Outside Load Zone © SKF Group Slide 15 22 March. 2016 . Load Zone ● Again. here is an illustration.

2 SKF Innovation in monitoring and tribology measurements .

2016 . in order to use frequency analysis for diagnostics (1990s) ● Developed and patented HAL and CTA (2000s) ● Developed an patented RFT (Repetitive Fourier Transform) ● Developed Acoustic Emission Enveloping. – conducted extensive testing in relation to lubrication and bearing life © SKF Group Slide 17 22 March.SKF bearing monitoring highlights ● Developed High-frequency detection technique (1990s) ● Pioneered the use of multi-filter acceleration enveloping in data collection devices (1990s) ● Patented (1985) and developed the use of practical acoustic emission measurements using data collectors.

3 SKF Multi parameter measurements .

2016 .Multi-parameter measurements © SKF Group Slide 19 22 March.

a rolling element rolling over a pit (defect) ● Impact signals have a high frequency content ● These impacts can excite a resonance © SKF Group Slide 20 22 March.Impact signals Where do those bursts of energy come from? ● Impacts – for instance. 2016 .

40 kHz Felt Band 1 © SKF Group Slide 21 22 March.Hz FFT Spectrum Gears and bearings Band 4 . 2016 Rolls Bearings Band 2 Band 3 Frequency .gE Filter Selection The acceleration enveloping process groups energy related to repetitive impulsive signals Bandpass filter 5 Hz .10 kHz 5 kHz .100 Hz 50 Hz .1 kHz Accelerometer Enveloping Detector 500 Hz .

4 Case history / examples .

steady speed This Lime Kiln Trunnion uses roller bearings that turn between 3 to 6 RPM For years the predictive team at this facility tried to predict failures with no success. 2016 . © SKF Group Slide 23 22 March.Case 1: Low.

2016 . We watched it due to production schedule.Early Detection of Very Slow Speed Defects SKF Enveloped Acceleration detected the problem nearly 6 months before the bearing was removed from service. © SKF Group Slide 24 22 March.

Sept 5. 2016 .5 months after initial detection © SKF Group Slide 25 22 March.

Oct 4.Failure is imminent The FFT has no discernable peaks. but the overall is much higher now. 2016 . © SKF Group Slide 26 22 March.

Appropriate choice of accelerometer and measurement location © SKF Group Slide 27 22 March.Trendable results aid in predicting time of failure Signal processing filters selected properly. 2016 .30K-600Kcpm for bearings. Database setup for several rotations of the shaft.

Case 2 . 2016 .Twin Roll Wash Press © SKF Group Slide 28 22 March.

What is a Twin Roll Wash Press? © SKF Group Slide 29 22 March. 2016 .

Wash Press vibration obscured by nuisance frequencies ● Hydraulic noise dominates the frequency range of the Env3 filter. 2016 . ● We will need to apply IMx Order Tracking. © SKF Group Slide 30 22 March. ● The speed varies constantly while collecting data – indicated by wide skirts around the hydraulic pulsation energy.

Full range spectrum shows areas of interest Typical vibration analysis looks in this range for bearing defects. 2016 . but this energy is due to intense hydraulic noise. © SKF Group Slide 31 22 March. which will mask most bearing defects.

2016 .Hydraulic noise dominates lower frequencies This frequency range includes Envelope Filter 3 range 30k to 600k CPM © SKF Group Slide 32 22 March.

2016 .Point setup based on frequency distribution These frequencies are less likely to be caused by pure hydraulic noise. © SKF Group Slide 33 22 March. Envelope Filter 4 is good choice to remove the hydraulic noise but leaves energy from bearing defects.

2016 . The machine speed varies during the Measurement time © SKF Group Slide 34 22 March.Vibration peaks obscured Envelope Filter 4 has been applied but still no bearing defects appear.

Outer Race defect identified Outer Race defect Clearly separated from Hydraulic noise This clean data was collected by using envelope band 4 filtering. 2016 . Additionally.Drive Side Bearing. © SKF Group Slide 37 22 March. SKF’s IMx Real Time Order tracking was applied.

Inner Race defect identified © SKF Group Slide 38 22 March.Drive End Bearing. 2016 .

After removal.Outer Race defects Deep pits in both roller paths through out load zone © SKF Group Slide 39 22 March. 2016 .

Press Side Roller Path Press Side Motor Side Inner Race etching at roller interfaces Outboard edge corrosion pitting Press Side © SKF Group Slide 40 22 March. 2016 Motor Side .Inner Race.

5 New approaches .

HAL/HAI – Harmonic Activity Index Harmonics Non-Harmonics Harmonic Content / Non-Harmonic Content Dimensionless.5 – Alert ● ~5 and above – Danger Set up as a derived POINT SKF Patented Technology © SKF Group Slide 42 22 March. 2016 . more “universal” alarm set-up for example ● 1 – Clear ● ~2.

Frequency spectrum © SKF Group 22 March. 2016 .

2016 .02.2009 04.2009 19.03.11.02.01.2009 © SKF Group 22 March.09.Example 3 : Dryer bearing Vibration trend 06.2009 HAL calculation of the outer ring frequency HAL calculation of the inner ring frequency 09.2010 22.06.01.2010 02.2010 15.2009 11.

Example: Conti-press Conti-press Speed main bearing: Bearing type: bearings © SKF Group Slide 45 22 March. 2016 3 – 7 rpm Spherical roller Bearing .

Trend with overall evaluation Bearing replaced © SKF Group Slide 46 22 March. 2016 .

2016 .Trend with HAL evaluation Bearing replaced © SKF Group Slide 47 22 March.

2016 .Comparison between overall and HAL evaluation Bearing replaced (no change in over-all) Bearing replaced (large HAL drop) © SKF Group Slide 48 22 March.

Bearing replaced (January 2010) © SKF Group Slide 49 22 March. 2016 .

If corrected. 2016 Poor Lubrication and Contamination results in ~50% of premature bearing failures! Using a technique like AEE can help to detect poor lubrication and contamination. it can help to prevent further bearing damage. .Early bearing failure detection Typical bearing failure modes include: ● ● ● ● Fatigue ~34% Poor Lubrication ~36% Contamination ~14% Poor fitting ~16% © SKF Group Slide 50 22 March.

Test Rig used for Acoustic Emission Envelope © SKF Group .

2016 .Trend in AEE and Lubrication Film AEE Pk Trend of AEE Pk-Pk Increase of about 100x © SKF Group Slide 52 22 March.

Questions? Questions? 22 March. Slide 53 2016 .