|
Case
study
Slow
Speed Diagnostics
Many
industrial machines operate at such slow rotational speeds that vibration
generated by common machine malfunctions, such as imbalance, misalignment,
and bearing faults, falls below the sensor frequency response range. The
impulses created by these phenomena are diminished both by their
structural path through the machine as well as by the low frequency sensor
cutoff. For these reasons, major concerns in selecting condition-based
maintenance programs for slow speed machines have focused primarily on the
service life of the rolling element bearings being used.
Typical
slow speed machines include ball mill mixers, antenna drives, hydraulic
electric generators, and rail car turntables where the rotational speed is
often less than several hundred rpm and can be as low as a fraction of an
rpm. Even though defect impulses occur at extremely low frequencies, the
nature of the impulses is such that high frequency components can be
analyzed for bearing faults by using acceleration enveloping methods.
Case
Study - Rail Car Turntable
In
a steel mill, during one phase of operation, scrap is added to a furnace
feed hopper from a rail car that is positioned by a turntable under a
traversing crane. Recently there were problems associated with the
turntable thrust bearing.
The
rail car turntable is a critical operation in the steel process, and
unscheduled outages incur high production costs. The thrust bearing must
sustain the combined rail car and scrap weight as it completes one
revolution, taking about 258 seconds. It is therefore critical that the
bearing performs reliably. Proactive condition-based monitoring can ensure
such reliability. If failure modes are avoided and scheduled maintenance
intervals are extended, then large savings can be realized in maintenance
costs and process downtime.
An
accelerometer was magnetically attached as close to the lower bearing race
as possible. It was immediately recognized that conventional vibration
measurements would not be fruitful in evaluating the bearing condition, so
a Microlog data collector/analyser was used to measure in the acceleration
enveloping mode. To eliminate noise and unwanted signals, the bandpass
filter selection was set to a high pass corner frequency greater than 10
times the running speed. The use of a filter separates the defect impulses
from the low order vibration components. Higher frequency filters are
sometimes preferred when structural resonances amplify the higher harmonic
components of vibration. In this case, the 5 to 100 Hz and 50 to 1000 Hz
settings were used to make alternate measurements.
The
measurement was more difficult to make since the table rotates one
revolution in approximately 258 seconds. In order to capture the bearing
defect signals, a wide data sampling time window was required. A 96 second
data sampling window was obtained by setting the FFT line resolution to
6400, and the maximum frequency range to 4000 pm
Time
window = # of lines 96400 lines) divided by f (4000/60) or, T = 6400 x 60
= 96 seconds 4000
Figure
1. Time domain in
AE measurement P1 - P2 = 28.5 secs. P3 -
P2 = 27.4 secs
Figure
1 shows a time domain plot using the 5 to 100 Hz filter in the enveloping
mode. Repetitive impulse peaks appear at approximately a 28 second
interval.
Figure
2 shows an expanded display of the first peak which indicates an impact
with its structural response lasting for 0.5 seconds. A similar expansion
of the other peaks showed the same rise and fall response curve. The 28
second interval bewteen peaks corresponds to the outer race bearing defect
frequency of the model 29452 spherical roller bearing.
Figure
2. Time
domain 3 sec expansion about the first peak
The
Frequency Analysis Module database from SKF (Figure 3) identifies the
outer ring defect frequency as 9.12 Hz for an inner ring speed of 1 Hz.
Since the table rotational interval is 258/9.12=28.2 seconds, the
diagnosis of an outer race defect was confirmed.
Figure
3. 29452
Thrust bearing specification
The
bearing was replaced and examined for structural faults. A defect was
discovered in the outer race which provided further confirmation that
acceleration enveloping methods can be successfully applied to very slow
speed bearing diagnostics.
Vibration
analysis of very slow speed machines is difficult to perform because of
sensor low frequency limitations. In addition, the low frequency related
vibration responses are characterised by very low amplitude signals, and
are usually buried in noise.
However,
impulse forces caused by bearing defects and gear faults can often be
identified by enveloping techniques. In this case study, the spectrum of
the enveloped signal did not provide sufficient information to permit a
positive identification; the time domain data however, did provide the
diagnostic clues.
The
data collectors/analysers and on-line monitoring systems used and supplied
by Apt Technology include menu selectable acceleration enveloping and
allow selection of the optimum filter for a particular measurement
circumstance. In this case study, the lower frequency bandpass filter
provided the best results for positive identification of the bearing flow.
Top
of page |
Back
to Case Studies
|
