The World’s Leading Information Resource For Maintenance & Engineering Professionals.

Social Media

Inaccessible Fan – Bearing Diagnosis By Fernando Cabrera and Ernesto Wiedenbrug, Ph.D.

Whitelegg Figure 1Situation

One of sixteen new 25kW 2-pole motors had an insulation failure just a couple months after being installed. The motor application is line-operated, direct across the line-started fans.
A closer look at the failed motor showed that the root cause of the insulation failure was mechanical. Parts of the failing drive-end bearing were flung into the end-turns of the windings, causing a phase-phase short circuit, as seen in Fig. 1.

Deeper Analysis

Proper Predictive Maintenance practice requires looking into every fault to see whether it was a random fault, or whether there is a systemic problem. Vibration analysis is the preferred technology to investigate bearing problems, but just like many other applications in this major copper mine in
Peru, these fans were physically very difficult to reach, making vibration analysis unfeasible. This is why the decision was made to use Torque Spectrum Analysis on the remaining 15 identical motors, looking for possible developing bearing problems.

Torque Spectrum Analysis
Previous uses of MCSA for Bearing Faults

whitelegg2As stated earlier, vibration is the preferred technology for finding bearing problems, but in some occasions the use of technologies requiring physical proximity is unfeasible. This is why much research has been done in finding feasible alternatives. Even though MCSA hasn't shown success in
100% of the cases, it remains the most promising alternative.

Lubrication problems: Harry Smith applied Torque Spectrum Analysis successfully in Excelon's Limerick nuclear power plant. He was able to identify remotely which out of a group of identical motors were most in need of being pulled for maintenance [1].

Single Point of Failure: Precursor to this effort, Don Doan investigated whether MCSA could be used effectively for diagnosing the mechanical condition of vane-axial fans [2- 3]. This research was performed at TXU's nuclear power station Comanche Peak.

Don's work included comparing the current and torque signatures obtained at the MCC with the vibration signatures on the outside of the fan's duct; using the vibration signatures on the motor as the reference. His work concluded that MCSA, in particular Torque Spectrum Analysis, was effective to find mechanical unbalance issues, and could also be used to find deteriorating bearings with a single point of failure. Figure 2 shows the SKF 6502 bearing with a scratch in the inner race which he used for this study.

Single Point of Failure in Torque Spectrum Analysis

Vibration Associates' chart [4] shows that a BPFI signature will result in vibration spectrum signatures in accordance with (1), which means that peaks will show at multiples of the BPFI, with sidebands of multiples of 2x RPM.

The torque spectra of BPFOtrq, BPFItrq and BSFtrq have the identical format, which is very similar to the BPFIvib case (2): Comparing the torque signatures for BPFOtrq, BPFItrq and BSFtrq with the BPFIvib, the only difference is that the torque signatures have sidebands at 2∙felec, while the sidebands of BPFIvib are at a distance of 2∙RPM.The EXP software includes a bearing database and a tool designed for the analysis performed in this case study. Figure 3 depicts the torque spectrum on logarithmic scale, and the vertical lines show a group of signatures in accordance with equation (2). The particulars of this
technique are taught in the SKF Dynamic Monitoring Level II class [5].


whitelegg3Performing the analysis over the remaining 15 motors showed that 3 additional motors were developing bearing failures. Proving to be systemic, the problem had to be dealt with immediately. All bearings were upgraded from the 62XX series to the sturdier 63XX series and the motors were put on the watch list using Torque Spectrum Analysis to ensure that this upgrade took care of the problem. After one year of successful operation without development of damaged bearing signatures

in any of the 16 motors it was concluded that the systemic problem has been resolved.


Even though vibration is the preferred method to find developing bearing problems, it is sometimes impossible to perform. Done properly, MCSA using Torque Spectrum Analysis can sometimes lead to the right answers for such challenging cases, since it is performed remotely from the MCC.

Having found the systemic problem allowed to exchange the motors and avoid unplanned downtime. The savings of having found this issue during the new motors' warranty period was U$48,000 - barely a fraction of the total savings if one included the avoidance of unplanned downtime costs.

References / Further Reading

[1] "Rotating Element Bearing Diagnostics in a Nuclear Power Plant: Comparing Vibration and Torque Techniques", H.R.
Smith, E. Wiedenbrug, M. Lind, IEEE SDEMPED 2007
[2] "Using Motor Electrical Signature Analysis to Determine the Mechanical Condition of Vane-Axial Fans", Donald Scott
Doan, Thesis, University of North Texas, May 2002.
[3] "Comparison of Duct-Mounted Vibration and Instantaneous Airgap Torque Signals for Predictive Maintenance of
Vane Axial Fans", D. Doan, E. Wiedenbrug, CMD 2005.