Here is the third of the five case studies brought to you with the Reliability Training Institute. In this case study we show how vibration analysis detected a hidden failure in a variable frequency drive. Hopefully, you are finding them of use and helpful.
As promised and to follow up on my previous post on bearing failures here is the first of five technical detailed case studies presentation. This is my first voice over presentation so please excuse the English Farmer accent mixed in with some Australian twang.
A profitable plant is reliable, safe and a cost-effectively maintained plant.
I often am lucky enough to use different vibration technologies and this post is a great example of how a vibration analysis program can protect the business using the SPM HD Enveloping technique. This post is with thanks to assistance from Dean Whiteside.
As part of a routine vibration data collection program a change in condition was noted at the fan motor vibration levels. Vibration monitoring frequencies were increased to daily as the defect deteriorated. This enabled planning and a controlled change out of the motor.
On inspection of the vibration data, bearing outer raceway (BPFO) damage was diagnosed at the drive end motor bearing. This was clearly evident in the SPM HD Enveloping.
Figure 1 shows the Acceleration RMS trend from the motor drive end (DE) bearing location. This shows the steep increase in the impacting levels with an exponential increase in the final days of monitoring.
SPM HD Enveloping Data:
Figure 2 shows the SPM HD Envelope spectrum from the motor drive end bearing. This technique shows a clear impact at 3.09 Orders that matches the defect frequency for the bearing fitted. There are many harmonics indicating a very impactive signal.
Summary of vibration:
There is a clear distinct defect in bearing outer raceway, at these levels this would confirm a spalling to the raceway.
Due to the risk of failure, a new motor was sourced and placed on-site encase of instant catastrophic failure. The risks was discussed with production and was deemed too high to the process and a plan was put in place for a controlled stop. But prior to this date there was an unexpected line stop, and as the motor was all prepared on site, the motor was changed during this downtime.
Vibration data after controlled change-out:
Figure 3 are the SPM HD Enveloping spectra from before and after motor change out. The top plot showed the clear bearing damage and now with the replacement motor there are no bearing defect signals present.
Figure 4 is the Acceleration trend from the motor drive end bearing location. This trend shows the increasing and then the lowest record level with the new motor installed.
Bearing Inspection: After sectioning and cleaning
inspection, it was found as expected, a large visible defect in the loadzone of
the bearing outer raceway. Motor Drive End Bearing FAG 6316-C3
Image 1 is the drive end bearing sectioned.
Image 2 is the defect located in the loadzone of the outer raceway.
Notice the flat bottom of the spalled area and the “neat” cracks around it. These are cracks that have come to the surface and in time, more material will break away.
Image 3 is the defect located in the loadzone of the outer raceway.
Particle over roll as the bearing comes out of the load zone.
ISO 15243:2004: 5.1.2 Subsurface initiated fatigue. Primary causes of Subsurface initiated fatigue are repeated stress changes and material structural changes. This leads to microcracks under the surface, crack propagation and then spalling.
The bearing is damaged as soon as spalling occurs. Spalling gradually increases and gives rise to noise and vibration levels in the machine. This machine was stopped and repaired before the bearing collapsed. The period from initial spalling to failure depends on the type of machine and its operating conditions.
What is sub surface fatigue? In a rotating bearing, cyclic stress changes occur beneath the contact surfaces of the raceways and rolling elements. Consider the rotating inner ring of a radial bearing with a radial load acting on it. As the ring rotates, one particular point on the raceway enters the load zone and continues through an area to reach a maximum load (stress) before it exits the load zone.
During each revolution, as that one point on the raceway enters and exits the load zone, compressive and shear stresses occur. Depending on the load, temperature and the number of stress cycles over a period of time, there is a build-up of residual stresses that cause the material to change from a randomly oriented grain structure to fracture planes.
In these planes, so-called subsurface microcracks develop beneath the surface at the weakest location, around the zone of maximum shear stress, typically at a depth of 0,1 to 0,5 mm. The depth depends on the load, material, cleanliness, temperature and the microstructure of the steel. The crack finally propagates to the surface and spalling occurs.
This is another example of how vibration technology
and knowing system health and risk of failure enables data driven decisions to
benefit the business. The motor was replaced when the line was down due to an
unplanned shutdown, with no additional downtime occurred.
If this motor had failed without any planning this would have lost product and reduced profit. In addition these actions have protected the customers, supply chain and brand our reputation.
(Case Study Electrical Defect detected thought CBM)
This is a case history brought to you with data from James Pearce – another great find! This shows how utilising multiple CBM technologies, with a certified and experience technician, can help prevent unplanned failure to assets.
Using vibration analysis and thermal imaging condition based monitoring techniques a change in condition was found and a diagnosis of electrical issue with the VFD was given. From this the variable speed drive history parameters were interrogated. This confirmed it was indeed an electrical issue. Further analysis carried out by the site electrical supervisor pinpointed the IVI card as the issue. The IVI card controls a lot of optic connections controlling the IGBT’s. This was replaced and the vibration, temperature and current reverted back to normal.
We have been monitoring assets at the production facility utilising vibration analysis and infrared thermography. On a routine survey a change in condition was noted and investigated.
The motor in this case study is a 4 Pole 50Hz AC motor on a Siemens Variable Speed Drive. This asset has 2 of the same motors both driving a roller each to crush and grind product.
On-Site CBM Recommendations:
Motor: It was reported on the day that the windings temperature has been higher in the warm weather and is 10oC warmer than the comparable motor. This survey there has been an increase in the electrical activity across the motor. Please note we can only detect indications of an electrical anomaly. Recommended actions to investigate the electrical drive.
Vibration Analysis Data:
The dominant change in condition in the vibration data was an appearance of running speed electrical frequency in the PeakVue data and the increase in the high frequency electrical data.
Figure 1 compares the last four PeakVue acceleration spectra taken from the motor non-drive end. This displays the normal 2xLF activity and then the appearance LF activity this survey.
Figure 2 compares the last two Velocity spectra’s. This shows the increase in the high frequency electrical activity. The top plot is the normal activity and the bottom plot is with the defect.
Data with electrical defect
Thermal Imaging Data:
The thermal data below compares the suspect motor and the comparison motor. These motors are on the same asset performing the same duty at the same time.
This data confirms that the windings are indeed warmer on the suspect motor.
Electrical Supervisors Investigation:
Below trace shows the current varying.
The below trace is the Phase 1 Current under load conditions, only reading positive part of cycle.
This compares Phases 1 and 3 motor current under load conditions. Phase 1 only reading positive part of cycle.
On start-up temperatures all came back to normal.The IVI card in the inverter was replaced. The below plot is Phases 1 and 3 motor current equal after changing IVI card, under no load conditions.
An Insulated Gate Bipolar Transistor (IGBT) is a key component in what makes up a VFD (Variable Frequency Drive). An IGBT is the inverter element in a VFD, pulsing voltage.
IGBTs have become highly reliable devices that can handle high voltage devices and are able to switch in less than a nanosecond.
The IGBT acts as the switch used to create Pulse-Width Modulation (PWM). An IGBT will switch the current on and off so rapidly that less voltage will be channelled to the motor, helping to create the PWM wave. This PWM wave is key to a VFDs operation because it is the variable voltage and frequency created by the PWM wave that will allow a VFD to control the speed of the motor. Therefore, without the IGBT switching the current on and off so rapidly a PWM wave—and the speed control that comes with it— could not be created.
The IVI card in the drive controls a lot of optic connections controlling the IGBT’s
Failure mode ISO 15243: 5.4.2 Subsurface initiated fatigue
Is this normal Fatigue Failure, how many of you get to see a bearing actually fail from normal fatigue? Usually we come across bearing failures/damage due to secondary factors such as misalignment, over or under lubrication, imbalance, resonance and poor installation……
This is also a great example of how important knowing the asset you are monitoring is as to know when to remove the asset from service, ensuring that the client has got the maximum life out of the asset for the associated risks.
# You can’t analyse what you don’t know or understand #
This Case Study Application:
This is a DC motor that is direct coupled driving a gearbox.
We have been monitoring this motor since 2006 and in May 2017 a subtle change in the PeakVue level was noticed, closer monitoring was initiated and a bearing inner raceway frequency was found. Next in June 2018 there was a further step change that prompted the decision to remove from service as we felt the risk of failure was too high.The motor was overhauled at the next opportunity, this was in July 2018.
Figure 1 is the Velocity spectrum, there are no indications of any defect in this data.
Figure 2 is the Peak Acceleration 10 KHz FMax trend from October 2017 until change out in July 2018, this displays an increasing trend.
Figure 3 is the PeakVue Max Peak Acceleration trend from October 2017 until change out, this also shows the increasing trend.
Figure 4 is the PeakVue spectrum. This shows the running speed activity and a beautiful text book bearing ball pass inner raceway defect frequency with harmonics and sidebands at 1 Order.
Figure 5 is the PeakVue time waveform, this shows a distinct periodic impactive activity.
Figure 6 is the Auto correlation of the PeakVue time waveform. Auto correlation is great tool for distinguishing periodic activity within a time signal. This data shows us that there is a defect that is modulating by 1 Order. Therefore a component on the motor shaft, rotating with the motor shaft has a defect.
Figure 7 is a zoom in on the Auto correlation of the PeakVue time waveform. From this we can see that the 1 Order activity is side banded by the inner raceway defect frequency.
Images of the bearing defect
Image 1 is of the bearing inner raceway. This shows the track of the rolling element in the race way, due to the DC drive, also within the arrows there is the defect.
Image 2 is a microscopic image of the defect. Has anyone else pulled a bearing with this type of defect?
Suspected failure mode is ISO 15243: 5.4.2 Subsurface initiated fatigue,
The images show that this bearing had reached its end of life, the cyclic stress changes occurring beneath the contact surfaces had initiated subsurface micro cracks, and this would have been in part of the bearing at the maximum shear stress. We are at the point where the crack has propagated to the surface and spalling has started to occur.
A special thanks to James Pearce for the data and working with me on the analysis.
Electrical defect found with Velocity data – Case Study
Has anyone found many electrical defects though vibration analysis? We know that VA will show the indications of electrical activity but not necessary the severity. This case study shows that the Velocity vibration data can indicate what the cause of the vibration problem is, this will enable the engineer to target the investigation.
Thanks to James Pearce for the data. linkedin.com/in/james-pearcevibrationanalysis
A routine client called after the operators noticed an increase in noise and vibration from a main plant drive motor. This is a DC motor and usually operates around 400-500 RPM. This is a rather old motor and drive system.
Initial Vibration Survey:
On attending site vibration data was collected, analysed and before leaving site recommendations were given.
Figure 1 is the Velocity Spectrum collected from the motor. This showed a 1 Order amplitude of 0.07mm/s RMS, with a dominant peak at 49.95Hz with an amplitude of 3.2 mm/s RMS with many harmonics. The motor was operating at 384 RPM during data collection.
Figure 2 is the PeakVue Spectrum. This displayed a dominant peak at 149.86Hz, 3xLf. This was also sidebanded by running speed.
The recommendations was to check all supply cable connections and inspect the variable speed drive components for condition.
The site electrical engineer was dispatched to inspect the drive for this variable speed motor. Upon inspection 2 Thyristors were replaced and all electrical connections checked for security.
The operator then reported that the vibration magically disappeared.
Post Maintenance Vibration Survey:
Vibration data was then collected after maintenance. The motor was running at a higher speed of 456 RPM on the follow up survey.
Figure 3 is the Velocity overall trend from the initial survey and post maintenance survey. This trend shows the reduction on motion from 4.301 mm/s RMS to 1.162 mm/s RMS.
Figure 4 compares the before and after maintenance Velocity Spectra’s. From this you can see the dominant 49.55Hz and harmonics have disappeared. The only activity left is a peak at 299.74Hz again sidebanded by 1 Order.
This again shows the benefits of sending a certified, experienced and correctly mentored Vibration Engineer and not a data dog to investigate vibration issues. James quickly pinpointed the cause of the excess vibration that enabled the client to efficiently target the area of concern and quickly rectify the issue saving time and money.
This month’s blog is to promote the thinking that when drive trains are aligned they should be aligned to the bearing tolerances and not the coupling tolerances. In addition how many people receive an alignment report with a soft foot check? We have found that some companies allocate their employees a laser alignment kit tell them what buttons to press and send them in the field. Without proper training and mentoring how will these employees learn correct Precision Alignment? Without correct training they will not know how to fix problems if they don’t understand fully what they are doing.
This month’s blog shows the importance of Precision Alignment including soft foot check and that the users of laser alignment equipment should be properly trained and mentored in Precision Alignment.
We were called to investigate an apparent increase in vibration levels after a high pressure hot water pump was replaced with a new pump end and a reconditioned drive motor. The operator felt that it was not running as smooth as the old pump set.
For this survey James used the CSI 2140 Dual channel Machinery Health Analyser. Data analysis was carried out using the CSI AMS Machinery Health manager software V5.61.
Vibration data including Velocity, Acceleration and bearing condition unit PeakVue was collected from each bearing location as close as possible to the source. Where applicable additional data including high resolution vibration data was collected.
There are elevated directional Velocity vibration levels when running at 2680 RPM (Low speed). This is due to a coincidence of a system natural frequency being excited by a motor Soft Foot condition.
Check/inspect condition of the foundation, looking for looseness and any deterioration in the base plate.
Perform precision alignment that must start with a soft food check and soft foot elimination. Followed by precision laser alignment.
If these actions do not resolve the issue then stiffening of the base may allow for improved precision alignment and may move the system resonance out of the running speed range.
There are increased directional Velocity vibration levels at the motor when running at 2680 RPM.
After reviewing the vibration data it was decided to perform further checks and the motor holding down bolts was loosened one time when in operation, this is to check for distortion of the motor rotor to stator air gap. During this test it was and found that the Velocity amplitude reduced. The amplitude reduced to its lowest level when the motor non-drive end foot bolt (bolt closest to pump #1) was loosened (see figure 1&2). This indicates there is a soft foot issue.
In addition an overall vibration coast down test & resonance bump test was performed. This data confirmed a natural frequency at 2X 2680RPM (see figure 3).
HPHW Pump #2 Motor Non-Drive End
The motor has elevated directional Velocity vibration levels. By loosening one motor fixing foot bolt at a time, the Velocity amplitude reduced. The amplitude reduced to its lowest level when the motor non-drive end foot bolt (bolt closest to pump #1) was loosened.
Figure 1 compares the Velocity spectra when running at 2680RPM, for the one order levels, as found state (4.332mm/sec RMS) & where the amplitude decreased the most after the motor foot bolt was loosened (2.651mm/sec RMS).
HPHW Pump #2 Motor Non-Drive End
Figure 2 is a photo of the motor indicating which foot bolt was loosened which resulted in the best decrease in amplitude.
HPHW Pump #2 Motor Drive End
Figure 3 is the data from a resonance bump test & overall vibration coast down test, performed at the motor drive end (DE).
The top plot bump test result indicates a system natural frequency that will coincide with twice the running speed (when running at the low speed) and amplify the vibration levels.
The bottom plot amplitude peak from the coast down test also confirms this condition with a peak at 5336 RPM, twice the running speed at the low speed setting.
This is one I recently finished and thought it would be a great one to share so people know what can be achieved.
We had three pump sets suffering from elevated vibration levels when operated in different combinations. Conventional vibration analysis was performed and this indicated a structural resonant condition.
The pump motors are mounted on a false floor:
and the pump barrels are below the floor:
The pump with the worst motion was on pump 3, the one far away from the edge of the drop. Also this pump has the least structural support under the floor. When ran in certain combinations pump 3 would be excited very badly.
The cost effective solution.
I designed a vibration dynamic absorber.
Dynamic Absorbers are often overlooked and not used, they can be seen as a band aid or a last option for some vibration problems. Whereas in some cases they can be the only cost efficient option, and they are very effective.
The Vibration Dynamic Absorber is a unique bespoke item, maintenance free, that is designed to absorb unwanted energy. It is tuned to have the same resonant frequency as the structure to set up an out of phase signal reducing the signal generated by the structure.
How did I design these?
For this one it was more of a ‘gut feel’. I looked at the motor and then drew out a design that wouldn’t look out of place when mounted, and that had some adjustment to it when fitted as theory doesn’t always pan out in real life. Then from this I worked backwards to get the correct material dimensions/configuration so it was resonant at the target frequency. I also made some weight configurations so I could cover my target range.
I will be going back in 6 months to see how it fairs. I did consider a round bar and weight but thought that with the rectangular bar you have more control on what way it will be resonant. As once you have performed phase analysis on the motor you then know what way it is moving and can mount the absorber accordingly.
Image of Pump 2 Vibration Dynamic Absorber:
Image of Pump 3 Vibration Dynamic Absorber:
Pump 2 Live Motion Video
Pump 3 Live Motion Video
Pump 3 Slow Motion Video
What am I covering?
On pump 3 I am covering the one problem frequency, 1 Order, but the two arms are of different lengths in terms of the length from the point of pivot (clamping) to the mass. Also the arms are of different dimensions with different mass at the end so they could be tuned to the same frequency.
I also did find that the sweet spot was not necessary the point of higher deflection of the absorber and that the three motors all reacted differently.
Final Review of actual vs theory:
I have had time to review the final theoretical tuning of the three pumps to actual results. They are all different and no one motor is the same, they all have their own personalities dynamically wise.
Pump 3 had the highest overall vibration, one dominant frequency at 1 Order on pump 3 and this was successfully reduced.
Pump 1 and pump 2 had two frequencies in the data. And both of the vibration dynamic absorbers were tuned to the lower frequency not the one order.
Table of final overall levels:
Motor NDE (Top)
Motor DE (Coupling end)
Motor NDE (Top)
Motor DE (Coupling end)
Motor NDE (Top)
Motor DE (Coupling end)
Pump 1 actually showed the text book results. The theoretical calculations for the tuned damper was for the lower frequency not the running speed (1520 CPM yes they are on soft start VFD). It actually split the frequency – text book……….beauty!!
I have more questions and theories now, this is pretty exciting stuff. Hopefully I can keep this going on other pumps.