Below is a case study from the paper, “Babbitted bearing health assessment,” by John Whalen of John Crane, Thomas Hess of Dupont Engineering, Jim Allen of Nova Chemicals, and Jack Craighton of Dupont Engineering at the 41st Turbomachinery Symposium in Houston, Texas.
The case study involves a large centrifugal compressor in charge gas service in an ethylene plant. The unit was running fine when a power outage shut the plant down. The train came down fine but when brought back up it was noted that one bearing exhibited high levels of vibration.
This is the low pressure “A” case compressor located between the steam turbine driver and a gearbox which drives the medium pressure “B” case compressor. Alarm and trip levels were set to 3 and 5 mils respectively, and the compressor normally runs around 4750 rpm. The bearing in question is a 6” 5 pad LBP bearing with an L/D ratio of 0.5 loaded to 225 psi.
The low pressure compressor outboard bearing vibration had increased from under 1 mil of vibration pre-outage to over 4 mils vibration after coming back up; this vibration was substantially synchronous, had a forward precession, a round orbit and was speed dependant, there was also a slight phase shift. All of this (and other data, history and condition analysis) pointed to a balance condition change on that end of the machine.
There was a lot of speculation on how the rotor became unbalanced however rotor dynamic analysis indicated only a coupling unbalance (or other similar synchronous forcing function) could cause the vibration levels observed.
The rotor model the rotor deflection shape from an unbalance at the first wheel are presented. Note that this is a drive though machine so there is a coupling on each end. Also note that the analysis predicts that for this unbalance the vibration at each bearing should be about the same. However the actual machine had 5 mils vibration on one bearing and 1 mil on the other. It was determined that an unbalance between bearings could not duplicate the vibration spread observed.
When applying an unbalance to the coupling we find the roughly 5 to 1 variation in vibration as observed from one end of the rotor to the other. It was determined that somehow the coupling became unbalanced – even though the vibration on the other side of the coupling remained low. This analysis also replicated the speed dependency found in the actual machine.
Now the focus shifted to determine if continued operation would lead to an unplanned shutdown due to bearing failure. The dynamic force imparted on the bearing due to a coupling unbalance sufficient to raise the vibration to 5 mils is. Note the significant dynamic load (which is zero to peak). The bearing had (roughly) a static load (journal load) of 4,000 pounds and a dynamic load of +/- 20,000 lbs.
Analysis of the steady state condition of the machine predicted a maximum film pressure of 750 psi so one could estimate that the dynamic pressure was 3750 psi. Analyzing this as a compressive stress of 750 psi and an alternating stress of 3750 psi we can then look at the babbitt compressive strength at 180 °F of 4000 psi and see that we are very near failure.
It was decided to reduce speed to maintain vibration levels at or below 4.5 mils and continue to run. This of course resulted in reduced plant output put did allow the machine to continue to run. Financial analysis dictated that they run for another 3 years under this constraint. The bearing health was monitored by observing shaft centerline data, bearing temperature readings and performing frequent oil analyses.
An alignment consultant was brought in and determined that there was an offset misalignment that would have presented itself as a synchronous force on the rotor. When the machine came down they found that the coupling had spun and friction-welded itself to the rotor – this caused the misalignment and change in balance condition. It was theorized that there may have been a significant liquid ingestion when coming back up after the power outage.
What did this bearing have going for it that allowed continued operation for 3 years under this extreme dynamic loading condition? The OEM bearing was replaced a long time ago with a bronze and babbitt bearing with ball & socket (B&S) pivots. The B&S pivots deteriorated very little over this run thanks to the significantly lower stresses compared to non ball & socket designs. The user is convinced they would not have been able to run this long under these conditions with the originally supplied bearings. The fact that the bearing was running at only 180°F was also a major factor as this meant the babbitt strength was still relatively high.
The bearings were removed and inspected and it was found that the lower two pads had opened up about 2.0 mils due to some pivot wear and babbitt loss. The bearing on the other end had the same bore as when it shipped 11 years earlier. The rotor, bearings and coupling were replaced and the machine came back up with vibration levels on both ends under 1 mil.