Experiment proposal on oil analysis

[455 Bravo Uniform]

On a separate thread there was discussion about oil analysis where some posters, including I, commented on repeatability of results, detection limits, and other quality assurance/control questions. What I am proposing has nothing directly to do with the usefulness of oil analysis, that’s a different matter. I do oil analyses because I think it’s useful. You do you.

Here is my proposal- I can take multiple samples during an oil change and send them in for analysis. This would be a “blind” study; not entirely a study in the academic publication research sense, but a test which we could use to answer questions for use independently. We could then share the results with the lab and others if appropriate. I want to be careful here - I don’t want to damage anyone’s livelihood.

We use one engine and one oil dump. Take 3 samples - one at the start, middle, and end of a draining. This is in order to get 3 different concentrations. Then we split each sample into 2 different “aliquots”. We’d have a total of 6 samples. We would hope that each pair of samples have nearly the same results.

We would label them differently (that’s the “blind” part). We could use the automotive oil analysis, instead of airplanes, just so we don’t get names, tail-numbers, or email addresses messed up.

So this might be in the neighborhood of about $200. If anyone is interested, chime in here, then PM me. I will pay for one sample analysis (about $35). Looking for 5 others who might be interested in funding this and playing.

It might take me a while to start this, just looking for interest at the moment; maybe after the holidays.

 

[Pinecone]

FYI quality labs do internal QC similar to this.

And also, for many types of analysis, there is a commercial validation system where monthly a sample is sent for analysis. The results are sent back and compared to the known sample and with all other participating labs.

 

[455 Bravo Uniform]

FYI quality labs do internal QC similar to this.

Indeed they do. I spend millions of dollars annually on environmental lab analysis for work. That’s where my curiosity comes from.

 

[Lon Stratton]

Would be interesting to see the differences in the samples. I have always taken a small amount from those three points every time I change oil thinking I would get the most accurate analysis. But all three in the same bottle.

 

[chemgeek]

A properly functioning analytical laboratory should be running validation standards with each set of sample runs. However, this can be a false sense of security depending on how one is actually making sample measurements. Validation standards have controlled matrices, unlike samples. Each wear metal measurement technique has its own foibles as well. For example, liquid sampling methods that depend on sample nebulization (AAS/FES,ICP-AES) can be very sensitive to sample viscosity. TXRF is relatively insensitive to sample matrix effects, but can be more limited in which elements it can detect. No one method is optimal at detecting trace amounts of all wear metals/elements. And all methods are subject to method-specific detection limit issues for each element.

Oil analysis should be treated as a general, long-term trend monitoring approach. The absolute values may not be as accurate as you might suspect. General trend monitoring MAY help identify a developing issue if accompanied by other, matching symptoms. But a short term change in a wear metal (unless it is an order of magnitude or more) is unlikely to be informative. A one-time change of a wear metal concentration from 3 to 6 ppm, for example, is unlikely to be significant in and of itself. Users of oil analysis would be better informed if the labs provided both DLs and confidence limits for their measurements. (On the other hand, this information might point out that some measurement values may be of very limited usefulness due to DL or uncertainty issues.)