In my previous column, I presented a case study that portrayed how having specific knowledge about a given situation can result not only in finding a viable solution but also in changing people’s minds about the abilities of oil analysis.
In keeping with the theme of “knowledge is power”, I’d like to present a case study in which a client received great oil analysis data; however, the data was not utilized properly to indicate the true condition of the equipment until failure was simply unavoidable.
Oil analysis was regularly performed on a ship loading conveyor. While the wear metal results were higher than what would be expected from this type of gearbox, they were relatively consistent over time. When applying alarms and limits based on actual historical data from this specific unit, one could consider this unit to have been running in its normal range (Figure 1).
In March 2004, conditions took a turn for the worst. Levels of iron, chrome, nickel (elemental components of steel) and lead increased dramatically. These metals all indicate bearing wear in a gearbox. The increases in metals were noted on the laboratory report; however, no direct recommendations were made other than to change the lubricant.
In June 2004, the oil sample for this gearbox showed yet another significant increase in wear debris. Again, the only recommendation was to change the lubricant. As custom alarm levels were placed on this sample data, it was obvious that this gearbox was in an imminent failure mode (Figure 2).
Prior to making a formal recommendation, data from other PdM technologies was considered. Vibration analysis indicated an abnormality on the input bearing confirming oil analysis data.
After understanding and applying correct alarm levels, comparing data with on-site technologies, and discussing the results with site personnel, the recommendation was made to replace the gearbox. No further testing was required on this sample to understand the severity of the current condition.
This client did not have a spare gearbox on-site. The gearbox was ordered and scheduled for replacement based upon the lead time for the new gearbox. Catastrophic failure occurred on this gearbox one day after receiving the replacement unit and exactly two weeks after receiving the recommendation for replacement.
An inspection was performed to determine the cause of failure. During the investigation, there was a concern about the viscosity of the lubricant. The lubricant used prior to failure was an ISO VG 150. The viscosity result of the March 2004 sample indicated a significantly different lubricant.
It was later realized that the manufacturer-recommended viscosity grade for this application was an ISO VG 680 - the grade of lubricant specified to the laboratory as the grade used. The lab continuously mentioned that the grade of the samples was not consistent with what was identified for use. When the client changed to the correct lubricant, the gearbox experienced the failure (Figure 3).
During the teardown investigation, the channels designed as the lubricant’s path to the bearings were found plugged by sludge. Due to the lower viscosity, the ISO VG 150 lubricant was able to work past the sludge to lubricate the bearings just enough to keep from excessive wear.
When the lubricant was changed to the manufacturer-recommended ISO 680 lubricant, the sludge was sufficient to block the flow of lube to the bearings. This quickly resulted in the catastrophic wear condition.
The technician drawing the oil sample mentioned that when the previous gearbox was in place, little oil flow was noted during sampling. On the other hand, the new gearbox which began its service with the proper lubricant showed a high level of oil flow during the sampling process.
Understanding the oil sample data and realizing how to apply proper alarms significantly reduced downtime with this failure. It is easy to fall into the routine of using lubricant changes to solve problems. In this case, it is apparent the problem started well before oil analysis was performed.
If closer attention had been paid to the data in the beginning stages of the oil program, an overall failure may have been prevented.
See the article “A Hands-on Approach to Limit Selection and Use.” Practicing Oil Analysis magazine, September-October 2005.