A study of rolling element bearing defect analysis

A Study of Rolling Element Bearing Defect Analysis
Extensive literature is available on diagnosing rolling element bearing defects using vibration analysis. However, answers to many questions remain elusive, such as the effects of rotor weight (load), speed, and damage severity on the vibration signature. SpectraQuest plans to publish a series of articles addressing these issues. This article,first in the series, will cover the effect of rotor weight, speed, and defect severity on outer race fault spectra. SpectraQuest’s Machinery Fault Simulator (MFS) provides a platform to study bearing faults. The following tests are performed on MB ER-10K bearings with lightly and moderately faulted outer race. To acquire and analyze data, VibraQuest data acquisition and analysis software,SpectraPad portable data acquisition device, and six PCB accelerometers were used.

Test Setup

Test Setup – with one loader installed The whole test setup is demonstrated in the picture above. The belt drive is not connected to prevent interference in spectrum. The faulted bearing is installed in the inboard bearing housing. Accelerometers are installed in vertical and horizontal directions on themotor and two bearing housings.

Faulted bearing

Two loaders are installed The above picture shows the configuration with two loaders installed.

SpectraPad Portable Data Acquisition Device All the sensors are connected to the portable SpectraPad, shown above. The SpectraPad communicates with a laptop via PCMCIA.

Test Procedure
1. Mount the lightly faulted bearing in the inboard bearinghousing. 2. Mount all the sensors and connect them to the SpectraPad unit. 3. Start VibraQuest software data acquisition panel, and run the MFS with 3 different rotor weights (normal configuration of 2 pound weight with 2 disks, 1 loader of 14 pounds, and 2 loaders totaling of 26 pounds) at 4 different rotating speeds (1000, 1500, 2000, and 4000 RPM). Collect the data while machine runs in a steadystate condition. Data is collected for each configuration at a max frequency of 5kHz with 3,200 spectral lines and 32 blocks. 4. Repeat the above steps for moderately faulted bearing.

Test Data Analysis
The data files are analyzed using the Rotating Machinery Analysis (RMA) module in VibraQuest. MB ER-10K bearing parameters: Number of rolling elements: 8 Rolling element diameter: 0.3125 inchPitch diameter: 1.319 inch Contact angle: 0 degree To calculate the BPFO multiplier, the following formula is used: Nb Bd BPFO = * (1 − * cos θ ) Pd 2 Where, BPFO = Ball pass frequency multiplier of the outer race Nb = Number of rolling elements. Bd = Rolling element diameter, inches Pd = Pitch diameter, inches θ = Contact angle, degrees Fitting the parameters into the formula, we get BPFOmultiplier = 3.052. This multiplier is used to obtain the fault frequencies for each RPM. VibraQuest enables the user to resize the block sizes to obtain different spectral resolution. The frequency spectral size can be up to 102,400 lines. This feature was helpful in finding the fault frequencies needed to diagnose the bearing faults in this experiment. Often a high resolution is needed to detect faultfrequency, especially when the fault frequencies are very close to a multiple of the running speed. The following tables illustrate the resolution needed to detect bearing faults when fault frequencies are close to a multiple of the rotational speed for a Hanning Window function. If a different window is used, the resolution may have to be even higher.

Typical Spectra

Above is anillustration of typical spectra showing the outer race fault defect for 1000 RPM running speed. Note that there are several harmonics of running speed but the amplitudes are very small and the bearing fault amplitude is even smaller. A very high resolution is needed to detect outer race defects in the presence of third harmonic of running speed. The harmonics can happen due to many reasons, such as small...