Unfortunately you cannot "measure wear" via a $30 spectrographic analysis. Wear is measured in a much more sophisticated and complicated way using standardized ASTM tests such as D6709, D5966, D8279 and D6984. If you could instead use a cheap and quick method it would be utilized.
And comparative wear between oils is even worse since Blackstone themselves have stated that there is no statistically significant difference in UOA results between any oils. If you had a bad test then that is due to your engine design or condition or due to your specific operating conditions. It is impossible to discriminate wear results between oils using a UOA. Not that people aren't constantly trying to do so as you can read on here of course. But all that means is that irrelevant and erroneous conclusions are being drawn from a data source that is improper and inadequate for such conclusions.
I will disagree with most of what you say, but only on technicalities beause of your generalizations ...
I've worked with Blackstone a few times to get bulk data (no personal data involved; just raw numbers). The folks at BS are not statistical process people; they are oil sample people and honestly, most of the lab "techs" are just Joe Average guys who've been taught to run equipment at a basic level. They are honest and hard working, but they don't do statistical analysis for a living (like I do). The reason they state they don't see any significant difference is because they don't see it; it's that simple. And ... this is a BIG point not to miss ... it's a VERY FALSE presumption to assume that they cannot see a difference, because that would infer that there's a difference to be found. The absence of significant differences does not imply it cannot be found as the ONLY answer; the other answer could be that the difference cannot be found because it does not exist! If you want to prove that Bigfoot (Sasquatch) exists, you go looking for it. If you find Bigfoot, then you have imperical proof. But the absence of finding him does not prove he does not exist; it only proves you didn't locate him. That is a super-important distinction.
You most certainly can discern wear metals which would be far out of "normal" ranges. Normal being defined as outside of a statistical limit prescribed by a multiple of standard deviation.
You can used macro data to understand how the general population of your equipment (engine, trans, gearbox, diff, whatever ...) will and does respond to a multitude of inputs, and then compare/contrast your equipment to the overall population.
You can use micro data to understand how your one piece of equipment responds to a variable under controlled conditions.
HOWEVER ... both these techniques (macro and micro) require gobs of data and time; something most home-based UOA enthusiasts don't have the wallet or patience for.
UOAs are indeed a cheap means of seeing a portion of wear. They do not see all wear (nothing above 5um generally). But that view of the smaller particles does imply some sense of an overall wear trend development. In healthy equipment, the trends of spectral analysis have shown to be very reliable and predictive. At times, even catastrophic events can be predicted if caught at the right time (mostly a game of luck).
UOAs also are FAR, FAR cheaper than some of the alternatives to measuring wear in other means. For example, you could tear an engine down and measure the bearing clearances and visually inspect the parts. Three massive problems arise with this option ...
- the cost of labor and time is SUPER EXPENSIVE as contrasted to a $30 UOA
- the R&R of the gages used is typically very poor
- presuming your engine would be in good shape overall, you would put it back together. Now that you've done that, you introduce all manner of variation in terms of various torques when reassembling the engine; this introduces such a massive variable that the next time you'd want to tear the engine down, you really don't have a "baseline" with any credibility.
There have been some SAE studies which showed good correlation between UOA spectral wear analysis and other forms of wear measurements (electron bombardment, component weight analysis, etc). The mantra of the studies was not correlation as the main intent, but it was discovered that wear analysis methods did show correlation as a side benefit. There have also been SAE studies which show UOA wear data also has good correlation to some particle count analysis methods as well.
Are UOAs perfect? No they are not.
Are UOAs the cheapest, quick, reasonably reliable means to establish some manner of wear trending? Yes; most certainly they are.
The problem with almost any lab data study is that the sample sets are just too darn small and the variation is a huge topic quietly ignored. Most any company does not want to sample dozens upon dozens of it's product. They want to do things like computer modeling based on one or two tests, or they like to do HALT (highly accelerated lifecycle tests). These two methods introduce massive amounts of R&R error, and a wildly unrealistic understanding of standard deviation.
I do agree with you that most folks can't use a UOA properly, and that causes them to misunderstand what can and cannot be gleaned from UOAs.
The greatest issue with UOAs for most people is they don't have the training to understand what to collect, what to control, what to allow as a variable, how to process the data and how to interpret the data. They also lack the time and money to PROPERLY study baselines and variables. That group of challenges does NOT mean UOAs aren't a good tool to use. It simply means most folks don't know how to use the tool properly, and what the limitations of that tool are. Frankly, most BITOGers use UOAs as toys because they don't understand how they should be used as tools. If that's the generalization we can agree on, so be it. Otherwise I'm doing to have to say we'll agree to disagree.