Abstract

Spike Energy™ has long been regarded as a valuable tool used for detecting bearing and gear defects, cavitation, and rubbing. But, it has been neglected when considering its ability to assist with electrical fault diagnosis. Because it has the capability to demodulate the sidebands associated with high frequency carriers, faults associated with the rotor bar assembly in ac induction motors can be detected. A case study supporting this idea is presented at the end of the article.

Background

A simplified flow chart of the Spike Energy™ signal processing is shown in Figure 1. The vibration signal from an accelerometer is passed through a high frequency bandpass filter. The purpose of filtering is to reject normal rotational vibration components, such as unbalance and misalignment, while allowing the vibration generated by impacts to remain. There are six choices for the lower corner frequency, fc, of the band pass filter: 100, 200, 500, 1000, 2000, 5000 Hertz. The higher corner frequency, fd, is 65 kHz. The filtered vibration signal passes through a peak to peak detector with a proper selected output time constant. The detector detects and holds the peak to peak values with a pulse duration t . Then, it decays at the rate of the time constant until the next pulse occurs. The Spike Energy™ detector repeats this process. The gSE overall readings are determined by the intensity and levels of the peak to peak values. The fast Fourier transform (FFT) of the signal from the Spike Energy detector yields the Spike Energy™ Spectrum.

Spike Energy™ is a special case of an acceleration measurement that uses a high frequency band pass filter, demodulation techniques, and true peak to peak amplitude detection to accurately describe energy caused by early bearing and gear defects. Each of these types of faults produce a high frequency carrier and modulating sidebands. The carrier is either the bearing component’s natural frequency or the gear mesh frequency respectively. The modulating sidebands are caused by load and speed changes.

But these mechanical faults are not the only defects with a high frequency carrier and modulating sidebands. An electrical defect, particularly with regard to the rotor bar assembly, works similarly. The rotor bar pass frequency (RBPF) is calculated in the following equation:

RBPF = N x rpm, where N = number of rotor bars

FIGURE 1 - SPIKE ENERGY SIGNAL PROCESSING

In general, the higher the number of poles in a motor, the fewer rotor bars it has. A 2 pole motor may have as many as 60 rotor bars, while a 4 pole may only have 45. These examples would generate frequencies in a range from 81,000 to 216,000 cpm. The RBPF will normally have sidebands associated with it. The most dominant would occur at the RBPF plus or minus twice the ac line frequency. In the absence of a variable frequency drive, the ac line frequency is 3,600 cpm in the United States . Since each cycle of power provides a positive and negative power pulse, vibration is generated at twice its frequency, or 7,200 cpm. This power frequency will modulate the amplitude of the RBPF since the very existence of the RBPF depends on the supply of power. Occasionally, running speed sidebands exist with the RBPF due to load and speed variances. Typical spectral data depicting the RBPF with its sidebands is illustrated in Figure 2.

The presence of a low amplitude 1X RBPF with its 2X AC line frequency sidebands is normal in the velocity spectrum. A good rule of thumb is less than 0.05 in/s. When the rotor bar assembly becomes loose, it is the harmonics of this frequency that give the fault away. The 2X RBPF amplitude can easily be ten times the amplitude of the 1X RBPF peak when defects appear. In addition, a close inspection of data associated with this type of fault will show the presence of 3X RBPF and 4X RBPF peaks. These would give frequencies easily above the 300,000 cpm Spike Energy™ high pass filter range. Since all multiples of the RBPF are usually accompanied by the 2X AC line frequency sidebands, a Spike Energy™ spectrum, with its demodulation techniques, can be applied. The 2X AC line frequency sidebands associated with the high frequency carrier (the multiples of the RBPF) will be presented in the Spike Energy™ spectrum when the rotor bar assembly becomes loose.

FIGURE 2 - TYPICAL ROTOR BAR PASS SIGNATURE

In summary, it is important to realize that the Spike Energy™ spectrum has no prejudices with respect to forcing frequency sources. Any fault producing a high frequency carrier along with modulating sidebands can be detected and presented clearly with the Spike Energy™ spectrum. The typically dominating high amplitude, low frequency fundamentals, such as unbalance, misalignment, etc., will be ignored because of the high frequency filter, and the less dominating, but very significant high frequency faults can be readily seen.

Case Study

Vibration data is collected monthly on a 50 hp chilled water pump motor installed at a resin producing facility. The data is collected with an IRD Model 943 accelerometer, the IRD Model dataPac 1500 data Collector/Analyzer, and a magnet mount. The data is stored and presented with the IRD Model IQ2000 software.

The outboard bearing of the motor is labeled position 1, and the inboard bearing is position 2. The pump is mounted horizontally, so the direction labels are as expected. The velocity spectrum collected on the outboard bearing uses a maximum frequency of 360,000 cpm, 8 averages, and 1600 lines. This gives a resolution of 225 cpm per line, and an accuracy of ± 337.5 cpm for the Hanning window (since it requires 3 lines to describe a peak). The velocity spectrum collected on the inboard bearing has a maximum frequency of 12,000 cpm, 2 averages, and 3200 lines. This setup gives a resolution of 3.75 cpm per line, and an accuracy of ± 5.625 cpm. The Spike Energy™ spectrum collected on both the inboard and outboard bearings uses a gSE detection range from 300,000 cpm to 3,900,000 cpm. After the demodulation, the data is displayed at a maximum frequency of 60,000 cpm, 4 averages, and 400 lines of resolution. This configuration results in a resolution of 150 cpm per line and an accuracy of ± 225 cpm.

Figure 3 depicts the overall Spike Energy™ trend which might normally call the analyst’s attention to a bearing problem or cavitation due to the rise in levels in February and March.

FIGURE 3 - SPIKE ENERGY TREND

Investigation of the outboard position velocity spectrum waterfall in Figure 4 shows a dramatic increase in the 2X RBPF frequency during February and March. This would indicate the loose rotor bar assembly condition, but does not, on the surface, explain the Spike Energy™ increase.

A similar presentation of the Spike Energy™ spectrum in Figure 5 gives the presence of 2X AC line frequency peaks and several harmonics during the same months as the RBPF harmonics increase.

FIGURE 4 - WATERFALL VELOCITY SPECTRUM PLOT

FIGURE 5 - WATERFALL SPIKE ENERGY PLOT

At first glance, the presentation of the velocity spectrum at the outboard bearing in Figure 6 does not show any multiples higher than 2X RBPF. But, by converting the linear peak velocity amplitude scale to a log dB scale, additional information can be seen in Figure 7. The 3X RBPF and 4X RBPF peaks and their 2X AC line frequency sidebands are easily seen. The 4X RBPF peak is high enough to exist within the high pass filter used by Spike Energy™. A frequency magnification shown in Figure 8 makes the 7200 cpm sidebands clear.

FIGURE 6 - VELOCITY SPECTRUM

FIGURE 7 - LOG SCALE SPECTRUM

FIGURE 8 - MAGNIFIED LOG SCALE PLOT

These sidebands present along with their high frequency carrier explain the existence of the 7200 cpm peaks shown in both the outboard and inboard bearings’ Spike Energy™ spectrums illustrated in Figures 9 and 10 respectively.

FIGURE 9 - SPIKE ENERGY SPECTRUM

FIGURE 10 - SPIKE ENERGY SPECTRUM

Figure 11 has velocity data from the inboard bearing. The lack of significant amplitude at 7200 cpm (less than 0.02 in/s), and the absence of pole frequency sidebands around either running speed harmonics or the 7200 cpm peak indicates that there is no other significant electrical problem found in the vibration data.  

FIGURE 11 - HIGH RESOLUTION VELOCITY SPECTRUM

In March of 1996, the recommendation was made to replace the motor. The motor was changed immediately and the April 1996 data is presented in Figures 12 and 13. The outboard bearing velocity spectrum show a drastic reduction in levels at the RBPF harmonic. The Spike Energy™ spectrum shows no distinct peaks as a result of the repair. The old motor was disassembled to find significant charring of the rotor laminations due to the looseness.

FIGURE 12 - VELOCITY SPECTRUM

FIGURE 13 - SPIKE ENERGY SPECTRUM

Conclusion

The Spike Energy™ spectrum is unique in its ability to demodulate those telltale sidebands which accompany a variety of problems. It is important not to limit one’s thinking to only mechanical problems of this nature. Any fault producing data with a high frequency carrier and modulating sidebands can have its presentation enhanced by the Spike Energy™ spectrum.

References

1. Xu, Ming, “Spike Energy™ and Its Applications”, The Shock and Vibration Digest, May/June 1995 (Volume 27, Number 3).
2. Berry , James E., “Concentrated Vibration Signature Analysis and Related Condition Monitoring Techniques”, Technical Associates of Charlotte, Inc., 1994.