I'm not aware of any statistical study done on a UOA's accuracy. I would LOVE to design/participate in a study that got down to the nitty-gritty of UOA accuracy. Do a real R&R analysis; that would be awesome. Alas - I doubt any typical UOA service wants to find out how good or bad they really are; may hurt sales. And the cash outlay belies my level of interest; I cannot afford $1000 or more for UOAs in the interest of science solely to established their performance level.
Just like this thread started out many pages ago, there are some things that are not impossible, but difficult to prove. So we exist in a world of theory in some manner here ...
Does a BP system affect UOAs? Several plausible answers, based upon your position of interest:
- Most BP filters commonly available for automotive apps are reasonably effective (essentially absolute) at 3um and above. But they are not as effective below 2um. However, as they load up, they will become absolute towards ever smaller particle sizes, because unlike a FF filter, they don't "have to" flow. They can be allowed to blind off. Of course as they load up ever more, they become more efficient. But the effectiveness drops off because their flow rate ever dwindles, leaving more and more in the FF stream. It's a catch-22 situation. This is very reasonable common sense. It could easily be proven with enough PCs collected with a high frequency.
- most ICP spectral analysis can see from sub-micro up to 3um very well, and up to 5um reasonably well. Past 5um the ability of ICP drops dramatically. Each system and machine is going to be a bit different, but these are reasonably accepted as valid.
- Since BP filters can remove particles 3um and larger, and ICP can see particles 3um and larger, then it's very reasonably to believe that BP filters will remove some portion of wear particle evidence, and do so with even greater resolve as they mature. The issue is that it's very hard to know just how much this has effect. As I mentioned, it will have a great deal to do with the effective RATE of particulate removal as that filter media matures; it's not a static process - it's ever evolving throughout it's life cycle.
There are plenty of studies that show UOAs can reasonably detect "normal" wear rates (that which can be seen within it's spectrum). UOAs will never see anything larger than 5um, so they "sample" the overall wear. But that's OK, because we still see "normal" in terms of a ratio of overall wear. Unless there is a catastrophic acute event, the wear rate seen in a UOA will accurately reflect the overall wear rate shifts in trending. That is what the studies establish. When they compare/contrast the wear rates of UOA evidence to that of other means (rad bombardment; IRRAS; mass weight; etc) they see good correlation. They may not come up with the same values, but they do show the same relative shifts in wear trends. Hence, a UOA is reasonably accurate at predicting wear rates and trends.
PC analysis does a fantastic job of viewing the size and quantity of particulate. Unfortunately it does nothing to tell us about composition of the particulate.
UOAs can tell us about the composition of the wear metals, but only in context of % in the viable range, and speaks nothing to size.
My "normalcy" article clearly has data that shows undeniable proof that wear rates drop as the sump matures. Does not matter how much particulate is present in terms of filtration. Wear metals speak to the relative wear shift. In a FF-only system (a traditional filter system), there is no ability of the FF filter to have any reasonable effect on a UOA; they simply cannot filter down into that range (below 5um) with any ability to shift the data whatsoever. A BP system does have the ability to shift the data, but we'll never know by how much because no one has done such a study to my knowledge. I would love to design the DoE and preside over one, but I don't have the time/cash to do so.
UOAs most certainly have value. But they are not perfect. Same can be said for PCs, etc. But, used together, along with other tools like visual observations, we can get a fairly decent idea of the overall wear trends of the equipment. UOAs absolutely must be used with the understanding of their benefits and limitations. And the data they provide is nearly worthless unless one understands and has access to the "normal" distribution of the data, it's range and trends and variance. If you don't know this, a UOA is really handicapped. UOAs are a direct view of lubricant health, but only an indirect view of equipment health. But as I stated elsewhere in this thread, they are a nice low-cost tool that is proven reasonably accurate. Some of the alternatives may be a tad more reliable, but are grossly over-expensive and out of typical range for most folks in time and money.
If someone is looking for a perfect answer, they're not going to find it here at BITOG, because of the nature of the tools we have at hand.
On a side note, I desperately want to do this in the scope of tire alignment as well. My personal "feeling" (admittedly an emotional position) is that tire service centers that do alignments have incredibly poor R&R stats. But I cannot seem to find one that wants to participate, and I surely don't have the money to pay for 30 to 50 alignments over a weekend.
.