Because there are too many uncontrolled variables. The ASTM test for measuring wear due to the oil is complicated and rightly so. It isolates the one single variable of the oil.
Measuring something is often the easy part. Ascribing the observation to one isolated singular variable is the difficult part.
I think I understand our diversion of thoughts. This is a matter of how one views analysis and math models.
I look at things from a view of statistical analysis, and all the detailed math modeling.
The term "uncontrolled variables" as you use it would imply that because things are "uncontrolled", there is no validity to test results.
That's a laymans interpretation and very common. But it's also not true. In EVERY set of test data, there is variation. Some of that variation is controlled, some is not. The desire is to control as much as possible, but there is not a requirement to do so. The concept of "control" is often misinterpreted.
You do not need to "control" all inputs. You split them into two groups; controlled and uncontrolled.
- With the controlled group, you have the ability to manipulate the input(s), and if they correlate to some output, you can then work to assign causation, if possible.
- With the uncontrolled group, there may be correlation or not, and be causation or not, but you have no ability to manipulate the input.
Having an "uncontrolled variable" does not, in any manner, invalidate a test. It only needs to be accounted for in the data processing, and acknowledged in the analysis. Having an uncontrolled variable only limits you to further refining any result; it does not in any manner invalidate a result.
What we need to recognize is that there are many ways to measure "wear". Some methods are very precise, but require time and money consuming efforts and not at all usefull to the average person. Other methods are more general and very easy to apply.
The issue with very specific ASTM testing is that it focuses on only one aspect; that is both a good and bad thing. It can prove/disprove something in a unique way, but that method may also overlook other factors; it's sort of the "can't see the forest for the trees" concept. UOAs are just the oppostive; they see a large amount of information all at once, but they cannot focus on any one thing.
Using electron bombardment (just one example) is a very accurate way to measure wear on a cam journal, but to do so, you have to control the test in a lab, and it means tearing down an engine. You can find very specific info regarding some element of the lube as a manipulated input, but it ignores all other aspects of wear elsewhere. You can also only rely on that one example, as any effort to reassemble the engine again introduces a slew of other variables and would multiple the factors on orders of magnitude. Teardown analysis is great, but it's a one-and-done method.
UOAs cannot distinguish Fe of a piston ring from the Fe of a bearing journal, etc. But when the UOAs shows "low" wear numbers and low wear rates in succesive reports, there is good cause to believe that all is well. The technology of ICP itself is certainly proven and sound; that's not in question. UOAs also cannot see wear particles above 5um. As discussed before, using a UOA is a means of seeing some wear directly, and inferring other wear indirectly. UOAs also require no disassembly, obviously. They don't disturb the continued relationship of all parts relative to one another.
I believe it s grossly inaccurate to claim that UOAs don't track wear. They most certainly do. It's just a matter of how one wants to track that wear, and then acknowledge the benefits and limitations of that info; how it's collected, analyzed and reported.
Once the UOA data is derrived, it's then a matter of proper processing the data to reveal the rates and trends using statistical analysis. If one understand how to do this, and what the results can and cannot reveal, then you have a good, cheap way of knowing what's going on. Further, UOAs have been shown many times to be able to reveal some problem as it's developing; we've seen many examples in Blackstone news letters. It is wrong to say that every UOA will accurately predict all problems, because they are not that reliable in terms of a 100% guarantee. But they are FAR, FAR better than not doing anything. It's far better to have a false positve and seek a second opinion, then to never get a test in the first place and be blind-sided by an unforseen event. UOAs are very accurate, but not 100% accureate.
UOAs are not a means of singular determination for comparing/contrasting two lubes; that's a fools errand. But they are a good way to determine overall equipment health. They cannot ascribe a specific failure if something is wrong; they can only sound a general alarm. But that does not, in any way, make them useless. Any UOA which shows an undesirable value in wear or trend, is a notification to go look deeper into something. It's wrong to say UOAs cannot measure wear; they most certainly can. It's just that they cannot tell you what's wrong; you have to go figure that our for yourself. They can also imply when things are ok; they can show you if your data is "normal" (good averages with typical variance).
If we can agree on these priciples, then there's room to move forward.