Detecting Core Loss Problems with Power Factor

1 Star2 Stars3 Stars4 Stars5 Stars (1 votes, average: 5.00 out of 5)

The ALL-TEST Pro OL was used to win a court case resulting from core damage due to stripping processes in 2014 (the first ever application of ESA in a successful court case). The key is to use power factor and identify motor loading. A damaged core requires more energy to generate the magnetic field, known as kVAr at the same load, kW, although the measured voltage and current will be larger (kVA). power factor

Figure 1: kVAr, kW, kVA and Power Factor (Cos Θ)

In the court case, the power factor was 0.044 when the investigation was started with an expected value over 0.6 at no load conditions. When the motor was evaluated it was found that the core was badly damaged and big chunks of stator core were missing and epoxy used in place. When the stator core was replaced in the same operating conditions the power factor increased to over 0.6.


Figure 2: Power Factor 0.044 with 164 HP feeding the core and only 7.3 HP at the shaft at no load


Figure 3: Condition of Core before Replacement


Figure 4: Power Factor 0.6 with the motor overloaded (157%) on this 100 HP motor after core replacement

When comparing multiple 4160 Volt, 2500 HP, 3600 RPM electric motors at 84% load, a recently wound motor was more than 0.10 less than the other machines. This would indicate that the core was in poor shape confirmed by reports that the stator was running hotter than it was prior to the repair or compared to the other machines.

Motor3 power factor

Figure 5: Good 2500 HP 3600 RPM motor at 84.8% load (Power Factor 0.866)

Motor4 Data

Figure 6: Power Factor 0.795 at 84.1% Load on twin 2500 HP 3600 RPM

The energy used to overcome the poor power factor results in increased heat as the increased power is increased eddy current and hysteresis losses as the insulation between laminations breaks down.

Bearing Frequency iPhone App

1 Star2 Stars3 Stars4 Stars5 Stars (1 votes, average: 5.00 out of 5)

For simple frequency calculations there are apps including Motionics REBvibe Rolling Element Bearing Vibration Fault Frequency Calculator.  The frequencies can be determined by either knowing the bearing dimensions or size.  The associated frequencies can be converted to ESA signatures by adding and subtracting the line frequency (ESA is amplitude modulated versus vibration).

2015-08-23 20.08.01

2015-08-23 20.08.09

2015-08-23 20.08.45





Case Study: Whoops – An Unexpected Casting Void

1 Star2 Stars3 Stars4 Stars5 Stars (1 votes, average: 5.00 out of 5)

During the class a 1250 horsepower, 690 Volt motor was randomly tested with the ALL-TEST Pro 5 using the ALL-TEST Pro 33 methodology (see ALL-TEST Pro 5 vs ALL-TEST Pro 33 video at and in less than a minute the students and instruments had identified that there was a problem with the rotor bars. The following day the motor was connected to a Variable Frequency Drive and brought up to 515 Vac to allow a little slip with the motor unloaded.

PPF of unloaded 690 volt motor

Above: Electrical Signature Showing Rotor Bar Faults At No Load on VFD

The GTI Predictive VibeRMS was taken to evaluate the vibration related to the rotor. At no load there were no frequencies in velocity or acceleration related to the bars. It is noted that the ~130,000 CPM frequency associated with the drive did appear (the drive was not tuned as part of this evaluation). However, pole pass frequency sidebands most definitely appeared.

June2010 068

June2010 071

Above: Examples of Casting Voids in Cast Rotors

It was concluded that the cause was most likely a casting void.

Casting voids appear in all cast rotors and appear as the sectioned rotor end rings below.

Case Study: Milled Rotor Bars and Fault Detection

1 Star2 Stars3 Stars4 Stars5 Stars (1 votes, average: 5.00 out of 5)

A common method of ‘testing’ diagnostic technologies is the deliberate damage to rotor bars. This is done either by attempting to crack the rotor bar or by drilling/milling it out. Over the years I’ve noted both of these last and know that it is extremely lucky to catch the problem. Why? Well, when you drill or mill the rotor bars, in addition to a few extra forces that are brought into play, laminations can be shorted giving the impression that there is just a casting void because there is continuity along the sides of the slot.

In addition to this, there are forces that work on the two remaining walls in the rotor that generate a ‘pushing’ and ‘pulling’ force similar to the operation against the rotor bar, itself, with the result that the generated fault area works much like a switched reluctance rotor. It would be expected at low to no load that this would be virtually indetectable and at higher loads would vary depending upon the current involved (the greater the current, the greater the magnetic forces working on the surfaces).

We had a chance to work this theory during the EMD class at Dreisilker. A 30 horsepower 1765 RPM motor had a large section of rotor removed by milling. The original intent some 6-7 years ago was to test offline and online tests to see if the rotor bar could be detected. When I came across this I saw the large gap and suggested, instead, that it would have the effect of acting as more than one rotor bar and may cause some unusual readings due to that spacing.

General 180General 181

Above Figures – Top: Stator with milled rotor; Bottom: Milled Rotor

When performing a rotor test using the ALL-TEST Pro 5 dynamic test it was noted that the instrument determined that the rotor was bad. There were 2 identical motors and I sent the students off to test both without identifying which one had a problem. Motor #2 was the one with the milled rotor bar and that was the one where the fault was detected.

The motor was put on a dynamometer and tested at no load, half load and full load. The results were as seen in Figures 1, 2 and 3. What was interesting in the loaded conditions was that the pole pass frequency sidebands had harmonics. Normally you see harmonics when you have multiple groupings of broken bars meaning that the size of the slot was signficant enough to prove the fault. Finally, the fault should be fairly constant regardless of load. That did not appear to be the case here.

PPF 30 hp no load

Motor at No Load PPF close to Line Frequency

 PPF 30 hp 100 Load

Motor at 45% Load Harmonics of PPF

PPF 30 hp 50 Load

Motor at 95% Load Harmonics of PPF

Also, note that the PPF sidebands at no load were most likely mixed in with the line frequency as the motor was operating with virtually no slip. It was noted, as well, that the motor was run up to full speed and load resulting in a steady current at nameplate, which would not have happened with a real broken rotor bar (current would have fluctuated, at the least). An extremely high vibration was found at half load which was noted by the observers and as a 1X RPM frequency in ESA.

Conclusion: as expected, milled and drilled rotors do not equate to real conditions when looking for broken rotor bars. On the other hand, the faults were still detected.

ALL-TEST Pro and GTI Compatibility with Windows 10

1 Star2 Stars3 Stars4 Stars5 Stars (1 votes, average: 5.00 out of 5)

I have not received anything official from ALL-TEST Pro or GTI, but made the leap to Windows 10 due to a series of horribly embarrassing issues with Windows 8.1 updates. I waited until after the EMD class, just in case as I didn’t want to have to scramble to install and test software on another computer at the last minute.

The upgrade from Windows 8.1 on my Lenovo laptop (16GB RAM and !TB drive) took about 3 hours after a 45 minute download of the upgrade. So, overall, it took about 4 hours. Following the upgrade I immediately tested the ability to use the ALL-TEST Pro 5 MCA Pro software including uploading. A feature that was not functioning in Windows 8.1 was the erasing of the ALL-TEST Pro 5. Not only did all functions work after the upgrade, but I was finally able to erase the ATPro5.

I tested the PowerSight Manager software and ALL-TEST Pro OL 6.31 software, including uploading from the data collector, and was able to perform all functions with Windows 10. In fact, the Powersight Manager popped up an update notice (first time I ever received one) and gave access to a Beta version that kicked butt.

GTI Predictive’s cloud-based systems actually performed better than Internet Explorer when using Microsoft Edge, which is the replacement for Explorer. I was very happy with this!

Of note with the upgrade to Windows 10, the random crashing I was dealing with on several levels has not happened and the colors seem richer and there is more contrast, something that was lacking in Windows 8.1 (and 7). It has more of the feel of Windows 7, which is a nice improvement.

Overall, I was very worried about this upgrade and the issues did not appear. In fact, all of my saved passwords still existed in the system. So, following a 4 hour pause, I am up and running and possibly more effective. Only the future will tell.