Skip to main content

Read our primer articles on Oil Analysis and Tribology

I have 5 oil sample locations from identical equipment and for the last year one of the locations has had a low viscosity. When calculating the average and standard deviation should I include the low viscosity readings or exclude them from my calcs? I want to base an oil change on the out-lier equipment on the 2x standard deviation.
Original Post

Replies sorted oldest to newest

quote:
calculating


Dear Rgf12,

From my side, normally if these sampling point is integrated in one forced and cirlulated lubrication system, the oil viscosity shall equal

no need to calculate standard deviation (STDEV) to get viscosity based line. normally we just put limit +/-10% from its original viscosity (Reff: ASTM D 6224)

but for establishing metal wear based limit, STDEV is required
My first cut would to use all the data, unless there is a clearly evident outlier (specific data point, or machine). Calculate the average and standard deviation. I would then throw out the data that exceeds the average +/- 2 standard deviations and re-calculate the new average and new standard deviation. This is an interative process, so you may do it a couple of times. One thing to be concerned about it do you have enough data to be confident that the calculated values represent reality, especially if you are throwing data out of the analysis.

Good luck!
Post
×
×
×
×
Link copied to your clipboard.
×