Skip to main content

Read our primer articles on Oil Analysis and Tribology

We had a failure of a sleeve bearing on a 400HP motor last week. The inspection revealed a small concentrated wear area with significant groving in the bearing material. Oil analyis was performed in the middle of August and showed a slight increase in wear metals. Iron went from 1PPM to 2PPM, Tin from 1PPM to 6PPM and copper from 1 to 2 PPM. We have 5 of these motors and the only difference in the analysis on the failed bearing was the 2 PPM of both iron and copper. This seems to be a very low threshold for calling a bearing problem so my question is what is the general criteria for sleeve bearing oil analysis? I wouldn't want to pull a motor and send it for a $15KV bearing rebuild based on a 4 or 5 PPM increase in tin. Vibration and acoustics showed indicated normal prior to the failure.
Original Post

Replies sorted oldest to newest

Elemental spectroscopy, whether done by ICP or arc flash, suffers from size limitation, meaning, all metal particles greater than 10 micron in size will not be detected. Analytical ferrography and/or acid digested ICP results would give you more adequate results. Of course, if you do an acid digest elemental spectroscopy then you also need to establish adequate baseline for this lubricant.
Rotrode spectroscopy can get all of the metals. Good luck finding an economical lab for that.

I have Herguth run ferrogams, Patch and or magnetic as I request. Perhaps getting that done on some "good" and "bad sampples may give some insights.

If I'm really concened, I have them do an Automated Feature Analysis (AFA) in the SEM.

AFA looks at EVERY particle on a patch and gives me size and composition. It has allowed me to determine which bearing in a multi bearing reservoir was misbehaving.

As to vibration, we have some 1200 HP 150 RPM vertical machines, we use Peakvue (One of CSI's high frequency techniques) on them. And have found issues in the past. (See above)

Did you do any analysis of the failed bearing? (Got pics?)

Sample methods? Are you sampling running? Using a vampire pump, etc? If you shut it down to sample it you can be missing evidence due to settling. Plus sample tubes are sometimes difficlut to control. So consistency become an issue

Unles you are at a Nuclear plant, call up Trico and get some sample ports. Slap them in and then you can get repeatable samples while running. Build a new trend history.

If you are at a Nuke, let me know, we'll talk about how's, why's and how some of us are managing to get the paperwork done.

As to pulling a motor for a $15K rebuild based on a few PPM, Don't get emotionally attached to the idea that that is a bad ting. If it is reality when all else is done, it is reality and much better than grinding a machine to a halt.
You havn't mentioned particle count and PQ, if you current analysis does not include particle or PQ faliures can be as you experienced, without warning, with Particle Count and PQ once these two commence to elevate then check by microscope, wear debris analysis to "see" what the increases are to cause particle count PQ to elevate, as the ohter three replies have pointed out, the size limitation with spectrographic analysis reders it of limited use for examples of WDA
Thanks Rob S
attend Reliable Plant 2024
Link copied to your clipboard.