Vibration Analysis Reveals
Bearing Defect
On July 26, vibration readings on an extruder drive motor
exceeded the fault alarm on three parameters: vHFD, pk-pk waveform, and
max-pk waveform.
The waveform revealed spikes showing up once per every revolution of shaft
speed.
These three alarms are all good indications of bearing problems.
The spikes were occurring just under the running speed pointing to inner
race
defects. Confirming this in a vibration spectrum was next. In
Peakvue there were clear peaks at running speed and ball pass frequencies
inner
race
(BPFI).
In a normal velocity spectrum, the fundamental bearing frequency
does not normally show up. To match bearing frequencies, the bearing
sometimes rings at its natural frequency and we look for sidebands
of defect frequencies.
How long will this bearing last? Looking at the trend pk-pk
waveform, there was a big jump from
6 g's to 26 g's in a 69-day period. I knew from experience that it was
not an emergency, so I wrote a work request on July 26.
The Planner scheduled a time convenient
with production to allow for the bearings to be changed out. On July
29, 2004 the motor was taken out of service for repairs. One important
thing I noticed was that by turning the shaft by hand, it turned easily
and I felt no defects.

I
requested that the old bearings be saved for visible inspection.
I took photos of the inner and outer
race which do have clear defects.
Their were no visible defects on the balls.
Now, how long do you think
this bearing
would have lasted? Another week or several more weeks? I had been
tracking this bearing defect since March (4 months). The visible
defect looks relatively
minor but was impacting in the load zone. Trying to predict how long
a defected bearing will last is just a guess. Knowing the defect is
there is the best reason for
corrective action. Bearing defects will not go away; they can only
get worse until the machine fails.
There are many different ways to determine how long a bearing
defect will run to failure. Load and speed are the two most common factors
to consider. We can also try to track a defect
through the four stages of bearing failure. But trying to track a bearing
to near failure can be risky at best. Once you determine that the
bearing is more than a minor defect, change it out. And as demonstrated
in this case, good alert and fault levels can also help.
Submitted by Carlos Hernandez,
Reliability Technician, Engelhard
|