We have used a CSI 5100 system for quite a few years as a quick first check of oil quality for our plant. I routinely run 40-60 samples per month. I recently added a hot plate to do a crackle test, just as a backup or confirmation for water content testing.
I have understood for a long time that the tolerance for the 5100 in reporting water can be quite high. We have for years used 1000ppm and visually checking for water content under a microscope as a general standard. I recognize this is not “Karl Fischer” quality or accuracy, but I’m not running an ASTM lab either. I can say that we have not had a water related failure in nine years.
I was testing my January sample of Mobil DTE Extra Heavy from a Cooling Tower gearbox when I got an unusual result. I tested the oil using the crackle test and had no indication of water.
Then, using the standard procedure for the 5100, I cut the sample 50/50 with kerosene and the resultant test reported in excess of 15,000 ppm. I examined the sample under the microscope, and water was present in the sample. I tested the mixture on the hot plate, and it did crackle.
Next, I tested the kerosene and it did not crackle. So I have concluded that the kerosene did not introduce any water into the oil.
I retested the uncut oil again, and it did not crackle. So now I have a crackle test on the uncut oil that indicates there is no water present, and a 50/50 mixture tested on the 5100 that indicates an excessive amount of water is present, along with a visual confirmation and a crackle test of the mixture – all positive for water.
I’ve come to the conclusion that the emulsified/free water present in the second test was present in the original oil, but was being held in solution. When the kerosene was mixed with the original oil, the new mixture could no longer hold that much water in solution, and it dropped out as emulsified/free water. The 5100 reported what it saw: water droplets on its grid and properly reported water contamination.
I suspect the problem in not with the 5100, but rather the test procedure. I have come to believe that mixing the kerosene with the oil and then testing for water gives an inaccurate result. This has happened more than once with other samples.
For the time being, I am relying on the crackle test as a pass/fail indicator for water content. Testing the oil twice with the 5100 would be too time consuming for our program – I’m probably like you – I’ve got other things to do in addition to handling the oil program.
Any thoughts?
Original Post