| INTRODUCTION |
| A vibration monitoring program at the Cincinnati
Bulk Mail Center (BMQ of the US Postal Service was started in October,
1996. The B-14 conveyor drive was classified as a critical drive for the
movement of mail through the BMC. The vibration program identified a serious
intermediate shaft looseness or impacting problem with the drive unit on
B-14. The intermediate shaft vibration levels were monitored closely and
trended for increased vibration. Through the Christmas season, the unit
showed increased vibration levels indicating that the problem was getting
worse. Once parts were available, a replacement to the gearbox was made
during a scheduled maintenance period before catastrophic failure. |
| This case history displays some of the vibration
data used in the diagnosis of the problem. A basic explanation of the internal
configuration of the gearbox is included to help understand the machine.
The driver is a 4 pole AC induc-tion motor operating at a speed of about
1785 cpm. The motor /gearbox is close coupled and the double reduction
gearbox has two gear meshes that are identified as high speed gear mesh
(HG) and low speed gear mesh (1,G). The shaft rate frequencies and gear
mesh frequencies were calculated using the gear teeth ratios. See Figure
1. |
|
|
Figure 1. Schematic of the B-14 conveyor drive and the primary
forcing frequencies. The motor housing is bolted directly to the gearbox
so that the prime mover is a close coupled unit. All shafts are in the
horizontal plane.
|
| VIBRATION SETUP |
| The BMC maintenance staff used a triaxial
accelerometer sensor to collect vibration data in three axes (axial, radial,
and tangential). The digital data collector was configured for 800 line
vibration data using both the Fast Fourier Transform (FFT) and amplitude
demodulation signal processing techniques. The FFT spectra were collected
using two frequency ranges. The low range data covers a range from 0 to
150 Hz, and the high range data is from 0 to 1500 Hz. The demodulated spectra
were collected using a frequency range from 0 to 300 Hz. |
|
|
Figure 2. Low range vibration data from October 96 (an the left)
and just before replacement in March 97 (right). Note harmonic series at
1x1 spacing.
|
| VIBRATION SPECTRAL
DATA |
The first vibration test was conducted on
October 4, 1996.
The prime mover was replaced in early March of 1997. Figure 2
shows the first and last set of data taken on the faulty machine. The tangential
spectrum shows a strong 1x1 harmonic series. Note that all prominent peaks
in the spectra are harmonics of either IxM or W. |
| The spectra above are displayed in triaxial
format. The upper spectrum is axial vibration, the middle spectrum is radial
(vertical) vibration, and the lower spectrum is tangential (horizontal)
vibration. The frequency scale is shown in units of Orders. The order 1.00
corresponds to the motor rotational rate (IxM). The order 0,296xM correspond
to the intermediate shaft rate (lxl). The amplitude scale is displayed
in units of Velocity decibels (VdB) using a velocity of 10-8 meter/sec
as a reference. |
| The replacement unit, while also a double
reduction gearbox, came from a different manufacturer. The nameplate data
was slightly different and had different gear tooth counts. Following the
replacement of the unit, another vibration test was conducted. The spectra
from this test are shown in Figure 3 to allow comparison of a fault free
machine with the faulty machine. Note that the low speed gear mesh (LG)
is lower frequency than the previous gearbox due to the differences in
gear tooth count. |
| VIBRATION DEMOD
DATA |
| The demodulated spectra are shown in triaxial
format in Figure 4a for the October 1996 data set. The February
1997 data is shown in Fiqure 4b. The harmonic spacing is clearly
at a frequency that corresponds to lxI. Note that the noise floor is higher
for the February data while the amplitudes in the harmonic series are slightly
lower. |
|
Figure 3. Low range vibration data after replacement of
the drive unit. Note that lxI harmonic series is not present in
the fault free machine.
|
| Again for comparison purposes, the post repair
demodulated spectra are shown in Figure 5. The amplitude scaling
is slightly different from those spectra in Figure 4 due to a much
lower noise floor. |
| VIBRATION ANALYSIS |
| The first vibration test showed a significant
intermediate shaft rate (lxI) harmonic series in the FFT spectra (Figure2). |
|
|
Figure 4a. Demodulated spectra, October 96.
|
|
|
Figure 4b. Demodulated spectra, February 97. The amplitude scaling
is slightly different from Figure 4a.
|
|
|
Figure 5. Demolulated spectra, March 97. Note the amplitude
scale is different from the data shown in Figure 4.
|
| The strong lxI harmonic series in the demodulated
spectra (Figure 4) confirms the intermediate shaft problem and actually
shows the fault much more clearly. The shaft rate harmonic series with
high amplitudes indicates a looseness or impacting problem with the intermediate
shaft or one of the components on the shaft. The BMC maintenance technicians
continued to monitor the machine on a monthly basis until a replacement
was available and the conveyor could be taken out of service. Each monthly
survey showed the strong lxI harmonic series up to the point of replacement. |
| When the unit was replaced, the BMC maintenance
technicians disassembled the defective gearbox and found a broken tooth
on the low speed pinion. The broken tooth was causing an impact each time
it meshed with the gear on the output shaft. The tooth next to the broken
tooth also had a severe crack and came off when handled by the analyst.
The photographs show the intermediate shaft on the left and the broken
teeth on the right. |
| BENEFITS |
| Represented in this case history are two obvious
benefits to any facility with critical machinery in an industry where down
time creates a hardship. First, the BMC technicians were able to identify
a significant problem on a critical unit and continue to evaluate the risk
of continued operation over the five months they needed to run it. |
| Second, they were able to anticipate the parts
required for repair, order them, and schedule the repair before failure
occurred. |
| Third, if left to run until complete failure,
the low speed pinion would have continued its self- destruction and cata-strophic
failure would have probably been the result. Cata-strophic failure usually
causes collateral damage (which results in significantly more down time
and repair costs) and typically happens at the worst possible time. |
| Michael Johnson is a Senior Mechanical
Engineer with the DLI Engineering Division of PredictDLI. He has six years
experience as a Naval Nuclear Propulsion Officer and six years as a senior
application engineer for the DLI Expert Automated Diagnostic System. He
is a registered Professional Engineer with experience in marine, semiconductor,
manufacturing, power education, and facilities engineering. |
| Presented at the 1998 Machinery
Reliability Conference in Charlotte, NC. |
