I´ve been monitoring since last year an EP200 Ingersoll Rand Screw Compressor. This unit has run for 76,000 hours with no internal maintenance. IR recommends overhaul at 40,000 hours (motor and compressor bearings replacement).

Four months ago bearing wear was detected in AC motor, I recommended bearings replacement and maintenance people did the job recently. Motor bearings had signs of severe spalling. The idea was replace compressor bearings at the same time, however, some parts were not in place at time for logistic problems. Maintenance people assembled the motor with new bearings and the compressor with old bearings.

Compressor data show some signs of wear (high frequency peaks and some enveloping patterns), but overall levels are about 5 mm/seg and 12 gE (not abnormal). Lobe mesh frequency is dominant but looks at normal levels.

My question is: the idea is to extend components estimated life using monitoring technologies, but how long should I extend this process in a safe way in this machine ? Some kind of failures are not progressive. Do you have some real cases or recommendations ?

I´m thinking in no matter what, to recommend compressor bearings replacement at 80,000 hours, which will be twice the manufacturer bearing life estimation.

Comments welcome.

Hey guys some comments. Considering the scenario, do you consider a better idea:

1. To continue monitoring the compressor ?
2. To call for a preventive maintenance ?
3. Other one.

I need to sent a recommendation soon for this machine. "To replace or not replace ... that is the question"

Walter
Quote
Compressor data show some signs of wear (high frequency peaks and some enveloping patterns), but overall levels are about 5 mm/seg and 12 gE (not abnormal). Lobe mesh frequency is dominant but looks at normal levels.

Overall levels and numbers for gE can IMO be very deceptive if fault frequencies and I quote “some signs of wear high frequency peaks and some enveloping patterns are present” then determine what these frequencies are
If found to be fault frequencies then you have a problem; there is nothing in the market at this time that can predict a catastrophic failure.
The decision is simple inform the customer that there are fault frequencies present and recommend replacing defective components
Unfortunately we live in a blame culture society if it fails unexpectedly be prepared to get the blame

Mike