Forgot: Which studies show the connection between wear and UOA results?
- Thread starter bulwnkl
- Start date
- Status
This isn't gas engines, but in 2004 the air force did a study of their UOA program dealing with jet engines, and found of the 2 million UOAs that were done, only 200 or so were false positives or false negatives. All the rest showed either normal wear metal results (and no engine damage) or abnormal wear metal results on the UOA, correlated with engine problems. That is a pretty good connection, if you ask me.
https://www.spectrex.com/html_files2/technical-notes/oilanalysis-particle-analysis-engine-oils.pdf
https://www.spectrex.com/html_files2/technical-notes/oilanalysis-particle-analysis-engine-oils.pdf
Originally Posted by bulwnkl
I thought I had a couple things on point, but I don't. Point me to something?
Seems like visual inspection of motor oil & cutting open oil filters can be a better gauge of impending catastrophic engine failure than UOA, the particles seen in a UOA are microscopic, too tiny to see with the naked eye. I've always consider3d a UOA as more of an OIL tool than an engine tool-is the oil no longer usable, or can it keep going type of test.
I thought I had a couple things on point, but I don't. Point me to something?
Seems like visual inspection of motor oil & cutting open oil filters can be a better gauge of impending catastrophic engine failure than UOA, the particles seen in a UOA are microscopic, too tiny to see with the naked eye. I've always consider3d a UOA as more of an OIL tool than an engine tool-is the oil no longer usable, or can it keep going type of test.
Not sure why we need a study when iron or aluminum or other metals PPM must come from the bearing surfaces. We know the source because its the only possible source.
Some may be due to acidic corrosion on non-bearing surfaces, but that is a form of wear or certainly a problem.
Some may be due to acidic corrosion on non-bearing surfaces, but that is a form of wear or certainly a problem.
I was afraid the answer to my question would be that no one knows, and clearly it is. Thanks anyway.
The AF paper is interesting, but not quite on point and underscores the need for additional testing in order for a UOA to be predictively useful.
The AF paper is interesting, but not quite on point and underscores the need for additional testing in order for a UOA to be predictively useful.
OVERKILL
$100 Site Donor 2021
Doug Hillary's write-up on the main page of this site delves into the usefulness of UOA's and their respective limits in terms of resolution:
https://www.bobistheoilguy.com/used-oil-analysis/
https://www.bobistheoilguy.com/used-oil-analysis/
At one time I had been referred to work done in _maybe_ the ‘80s from a major auto manufacturer that supposedly correlated UOA results to wear measured precisely via other mechanisms. However, I was never able to find such. I have a reasonably good understanding of what UOA is not, and of at least a good deal of what it can't do.
What I don't have, or don't have any longer, is anything that shows how tight the correlation is between wear and UOA results. Other info suggests the correlation isn't very good, or in any event is unreliable.
What I don't have, or don't have any longer, is anything that shows how tight the correlation is between wear and UOA results. Other info suggests the correlation isn't very good, or in any event is unreliable.
dnewton3
Staff member
There is some decent info on connection between UOA data and other forms of wear measurement.
There are some SAE studies that have closely related info in their data; the inference is that UOA data can be correlated to other means of measuring wear. Most of these used UOA data and some other means such as electro-bombardment, or elemental weight analysis. I caution all to consider these topics where reading studies:
1) HALTs have not shown to be good predictors of normal operating conditions
2) You need to buy/read the studies; don't just peruse the synopsis
https://www.sae.org/publications/technical-papers/content/780184/ wear relative to filtration
https://www.sae.org/publications/technical-papers/content/881825/ wear relative to filtration
https://www.sae.org/publications/technical-papers/content/2007-01-4133/ wear relative to TCB
https://www.sae.org/publications/technical-papers/content/902238/ wear relative to filtration
This study, in particular, shows excellent correlation between PC counts and UOA element analysis. The inference being that particles in the oil stream are comprised of not only things like soot and other combustion byproducts, but also elements of metal wear. The "cleaner" the oil stream, the lower the UOA wear data. The correlation was very accurate in this study. In short, the wear data seen in UOAs (5um and smaller) is reasonably reflected in PC data (3um and higher), where % concentrations show good correlation.
but OTOH ... This shows that when it comes to filtration, you cannot discern filter efficiency to wear data in a UOA. Filter efficiency is different from wear, relative to UOA data. So this becomes a two-pronged comment.
From one POV, you could argue that UOAs are not a great way to measure wear.
But from the other POV, you could argue that filtration is not the only controlling entity of wear, and that until you isolate the individual wear contributors, this study only proves that filtration has negligible effect on wear. (OCI duration and TBC being the other key inputs that were not studied in this Fleetfilter effort).
https://www.cumminsfiltration.com/s...ct_lit/americas_brochures/SB_LT15105.pdf
For example, the infamous GM filter study from 1988 discussed the use of percent weight loss analysis as the main study effort. They also did PC analysis. They speak to the specific elements of Fe, Tn, Pb and Cu; these elements can ONLY be discerned with spectral analysis (you cannot tell a the type element of a particle in a PC count). So they did show that elemental analysis in UOAs shows correlation in % concentration to the other forms of wear analysis like the PC and weight loss methods. They also casually mention that you'll never see real differences in UOAs because the wear rates they induced in the HALT simply are never seen in the real world. They admit that UOAs cannot show filter differences simply because the delta wear is so flipping small that you cannot discern a difference. And so to the point I always make; if the wear is really low, the filter efficiency is moot.
This one is one of my favorites:
https://www.sae.org/publications/technical-papers/content/952557/
It's the Donaldson "total filtration" study.
Here, it's clear that the best oil filter is a good air filter. There's no talk of UOA data; it's all mass loss derived. This study looks at wear from various sources; air, fuel, combustion, OCI duration, etc. They do not address the TBC as a controlling factor; I wish that were included.
No one should be foolish enough to think that UOA data is perfect; it's not. It's a compromise of efforts, costs, data and accuracy. But that can be said true about many other forms of wear discernment; it's not like elemental weight analysis or electro-bombardment are not without challenges and concerns as well. Even tear-down analysis has it's issues with gauge R&R; it's highly suspect to repeatability concerns. UOAs have been shown to have good (not perfect, but very reliable) accuracy in "normal" conditions. They show you a portion of wear; that at or below 5um, generally in ICP analysis. What wear exists above 5um is unknown. But "normal" wear isn't really about huge chunks of flying asteroids in your oil, anyway; it's about the small stuff. UOAs are by far and away the easiest, cheapest way to discern most wear. Many of the studies I linked did show reasonable correlation between UOA data and other forms of measurements.
Other forms of wear measurement have issues, as I already stated. Electo-bombardment is sensitive in some elements, but not all. Percent mass loss is only accurate if the component measured is 100% of one element (say a pure aluminum sleeve bushing), but if you have a multi-element component (babbit bearing), you have to visually interpret the % loss of each as the wear layers reveal themselves. Teardown analysis is super sensitive to the previous set-up efforts, as well as human inaccuracy in measurement repeatability. THERE IS NO PERFECT WAY TO MEASURE WEAR; ALL HAVE THEIR PROS AND CONS.
Anecdotally, I'd also point to many of Blackstone's articles where they have UOA data that points to engine issues, and then they save an engine from impending doom. This is admittedly not always accurate, but if you're the person that saved an engine from destruction because you caught a timing chain guide going out before you crash hard parts, you're a happy person.
https://www.blackstone-labs.com/wp-content/uploads/2019/08/Mustang-report.pdf
https://www.blackstone-labs.com/wp-content/uploads/2018/09/Eng-Feb-2018-1.pdf
https://www.blackstone-labs.com/wp-content/uploads/2018/09/ENG-Report-1.pdf
https://www.blackstone-labs.com/wp-content/uploads/2018/09/ENG-report-July-16.pdf
https://www.blackstone-labs.com/wp-content/uploads/2018/09/7-13-ENG.pdf
etc
I'm sure I've missed a few studies; these are just the ones I've bought or gleaned from various sources.
There are some SAE studies that have closely related info in their data; the inference is that UOA data can be correlated to other means of measuring wear. Most of these used UOA data and some other means such as electro-bombardment, or elemental weight analysis. I caution all to consider these topics where reading studies:
1) HALTs have not shown to be good predictors of normal operating conditions
2) You need to buy/read the studies; don't just peruse the synopsis
https://www.sae.org/publications/technical-papers/content/780184/ wear relative to filtration
https://www.sae.org/publications/technical-papers/content/881825/ wear relative to filtration
https://www.sae.org/publications/technical-papers/content/2007-01-4133/ wear relative to TCB
https://www.sae.org/publications/technical-papers/content/902238/ wear relative to filtration
This study, in particular, shows excellent correlation between PC counts and UOA element analysis. The inference being that particles in the oil stream are comprised of not only things like soot and other combustion byproducts, but also elements of metal wear. The "cleaner" the oil stream, the lower the UOA wear data. The correlation was very accurate in this study. In short, the wear data seen in UOAs (5um and smaller) is reasonably reflected in PC data (3um and higher), where % concentrations show good correlation.
but OTOH ... This shows that when it comes to filtration, you cannot discern filter efficiency to wear data in a UOA. Filter efficiency is different from wear, relative to UOA data. So this becomes a two-pronged comment.
From one POV, you could argue that UOAs are not a great way to measure wear.
But from the other POV, you could argue that filtration is not the only controlling entity of wear, and that until you isolate the individual wear contributors, this study only proves that filtration has negligible effect on wear. (OCI duration and TBC being the other key inputs that were not studied in this Fleetfilter effort).
https://www.cumminsfiltration.com/s...ct_lit/americas_brochures/SB_LT15105.pdf
For example, the infamous GM filter study from 1988 discussed the use of percent weight loss analysis as the main study effort. They also did PC analysis. They speak to the specific elements of Fe, Tn, Pb and Cu; these elements can ONLY be discerned with spectral analysis (you cannot tell a the type element of a particle in a PC count). So they did show that elemental analysis in UOAs shows correlation in % concentration to the other forms of wear analysis like the PC and weight loss methods. They also casually mention that you'll never see real differences in UOAs because the wear rates they induced in the HALT simply are never seen in the real world. They admit that UOAs cannot show filter differences simply because the delta wear is so flipping small that you cannot discern a difference. And so to the point I always make; if the wear is really low, the filter efficiency is moot.
This one is one of my favorites:
https://www.sae.org/publications/technical-papers/content/952557/
It's the Donaldson "total filtration" study.
Here, it's clear that the best oil filter is a good air filter. There's no talk of UOA data; it's all mass loss derived. This study looks at wear from various sources; air, fuel, combustion, OCI duration, etc. They do not address the TBC as a controlling factor; I wish that were included.
No one should be foolish enough to think that UOA data is perfect; it's not. It's a compromise of efforts, costs, data and accuracy. But that can be said true about many other forms of wear discernment; it's not like elemental weight analysis or electro-bombardment are not without challenges and concerns as well. Even tear-down analysis has it's issues with gauge R&R; it's highly suspect to repeatability concerns. UOAs have been shown to have good (not perfect, but very reliable) accuracy in "normal" conditions. They show you a portion of wear; that at or below 5um, generally in ICP analysis. What wear exists above 5um is unknown. But "normal" wear isn't really about huge chunks of flying asteroids in your oil, anyway; it's about the small stuff. UOAs are by far and away the easiest, cheapest way to discern most wear. Many of the studies I linked did show reasonable correlation between UOA data and other forms of measurements.
Other forms of wear measurement have issues, as I already stated. Electo-bombardment is sensitive in some elements, but not all. Percent mass loss is only accurate if the component measured is 100% of one element (say a pure aluminum sleeve bushing), but if you have a multi-element component (babbit bearing), you have to visually interpret the % loss of each as the wear layers reveal themselves. Teardown analysis is super sensitive to the previous set-up efforts, as well as human inaccuracy in measurement repeatability. THERE IS NO PERFECT WAY TO MEASURE WEAR; ALL HAVE THEIR PROS AND CONS.
Anecdotally, I'd also point to many of Blackstone's articles where they have UOA data that points to engine issues, and then they save an engine from impending doom. This is admittedly not always accurate, but if you're the person that saved an engine from destruction because you caught a timing chain guide going out before you crash hard parts, you're a happy person.
https://www.blackstone-labs.com/wp-content/uploads/2019/08/Mustang-report.pdf
https://www.blackstone-labs.com/wp-content/uploads/2018/09/Eng-Feb-2018-1.pdf
https://www.blackstone-labs.com/wp-content/uploads/2018/09/ENG-Report-1.pdf
https://www.blackstone-labs.com/wp-content/uploads/2018/09/ENG-report-July-16.pdf
https://www.blackstone-labs.com/wp-content/uploads/2018/09/7-13-ENG.pdf
etc
I'm sure I've missed a few studies; these are just the ones I've bought or gleaned from various sources.
Last edited:
Originally Posted by dnewton3
There is some decent info on connection between UOA data and other forms of wear measurement.
...
but OTOH ... This shows that when it comes to filtration, you cannot discern filter efficiency to wear data in a UOA. Filter efficiency is different from wear, relative to UOA data. So this becomes a two-pronged comment.
From one POV, you could argue that UOAs are not a great way to measure wear.
But from the other POV, you could argue that filtration is not the only controlling entity of wear, and that until you isolate the individual wear contributors, this study only proves that filtration has negligible effect on wear. (OCI duration and TBC being the other key inputs that were not studied in this Fleetfilter effort).
https://www.cumminsfiltration.com/s...ct_lit/americas_brochures/SB_LT15105.pdf
For example, the infamous GM filter study from 1988 discussed the use of percent weight loss analysis as the main study effort. They also did PC analysis. They speak to the specific elements of Fe, Tn, Pb and Cu; these elements can ONLY be discerned with spectral analysis (you cannot tell a the type element of a particle in a PC count). So they did show that elemental analysis in UOAs shows correlation in % concentration to the other forms of wear analysis like the PC and weight loss methods. They also casually mention that you'll never see real differences in UOAs because the wear rates they induced in the HALT simply are never seen in the real world. They admit that UOAs cannot show filter differences simply because the delta wear is so flipping small that you cannot discern a difference. And so to the point I always make; if the wear is really low, the filter efficiency is moot.
This one is one of my favorites:
https://www.sae.org/publications/technical-papers/content/952557/
It's the Donaldson "total filtration" study.
Here, it's clear that the best oil filter is a good air filter. There's no talk of UOA data; it's all mass loss derived. This study looks at wear from various sources; air, fuel, combustion, OCI duration, etc. They do not address the TBC as a controlling factor; I wish that were included.
Hello Dnewton, yes, very true indeed
Is there any study that links "washing" paper air filters to any significant increase in wear patters in on-road engines ?
Specifically, if we were to wash, dry and reuse stock air filter elements, would it definitively result in decreased air-filtration efficiency ( say for example, after the 1st wash/dry and reuse cycle, will the air filter allow more dirt ingress into the engine (that is, into the combustion chamber as well as the engine oil) ?
Thanks in advance for your very detailed posts here.
There is some decent info on connection between UOA data and other forms of wear measurement.
...
but OTOH ... This shows that when it comes to filtration, you cannot discern filter efficiency to wear data in a UOA. Filter efficiency is different from wear, relative to UOA data. So this becomes a two-pronged comment.
From one POV, you could argue that UOAs are not a great way to measure wear.
But from the other POV, you could argue that filtration is not the only controlling entity of wear, and that until you isolate the individual wear contributors, this study only proves that filtration has negligible effect on wear. (OCI duration and TBC being the other key inputs that were not studied in this Fleetfilter effort).
https://www.cumminsfiltration.com/s...ct_lit/americas_brochures/SB_LT15105.pdf
For example, the infamous GM filter study from 1988 discussed the use of percent weight loss analysis as the main study effort. They also did PC analysis. They speak to the specific elements of Fe, Tn, Pb and Cu; these elements can ONLY be discerned with spectral analysis (you cannot tell a the type element of a particle in a PC count). So they did show that elemental analysis in UOAs shows correlation in % concentration to the other forms of wear analysis like the PC and weight loss methods. They also casually mention that you'll never see real differences in UOAs because the wear rates they induced in the HALT simply are never seen in the real world. They admit that UOAs cannot show filter differences simply because the delta wear is so flipping small that you cannot discern a difference. And so to the point I always make; if the wear is really low, the filter efficiency is moot.
This one is one of my favorites:
https://www.sae.org/publications/technical-papers/content/952557/
It's the Donaldson "total filtration" study.
Here, it's clear that the best oil filter is a good air filter. There's no talk of UOA data; it's all mass loss derived. This study looks at wear from various sources; air, fuel, combustion, OCI duration, etc. They do not address the TBC as a controlling factor; I wish that were included.
Hello Dnewton, yes, very true indeed
Is there any study that links "washing" paper air filters to any significant increase in wear patters in on-road engines ?
Specifically, if we were to wash, dry and reuse stock air filter elements, would it definitively result in decreased air-filtration efficiency ( say for example, after the 1st wash/dry and reuse cycle, will the air filter allow more dirt ingress into the engine (that is, into the combustion chamber as well as the engine oil) ?
Thanks in advance for your very detailed posts here.
dnewton3
Staff member
Originally Posted by fpracha
... Is there any study that links "washing" paper air filters to any significant increase in wear patters in on-road engines ?
Not that I'm aware of, but I'm certainly not the end-all/be-all librarian of study data by any stretch.
My concern surrounding washing common paper (cellulose) air filters would be the issue of fiber break-down. Cellulose does a great job of absorbing moisture, and it can easily be "dried out", but most all cellulose filters have treatment added to them to give the fibers both functional form and also resilience against shedding ancillary sloughing of tiny fibers. If you were to "wash" these air filters, I have no idea what would happen to the treatment over time, and the subsequent shedding and breakdown of the fibers.
Could be an interesting study, but not one that I'm aware of being done so far, because the typical dry filter is made for a "one and done" use cycle.
... Is there any study that links "washing" paper air filters to any significant increase in wear patters in on-road engines ?
Not that I'm aware of, but I'm certainly not the end-all/be-all librarian of study data by any stretch.
My concern surrounding washing common paper (cellulose) air filters would be the issue of fiber break-down. Cellulose does a great job of absorbing moisture, and it can easily be "dried out", but most all cellulose filters have treatment added to them to give the fibers both functional form and also resilience against shedding ancillary sloughing of tiny fibers. If you were to "wash" these air filters, I have no idea what would happen to the treatment over time, and the subsequent shedding and breakdown of the fibers.
Could be an interesting study, but not one that I'm aware of being done so far, because the typical dry filter is made for a "one and done" use cycle.
I think I'm out of luck (maybe more like out of desire to keep searching?). It seems that the best we can get from a UOA is: So long as the engine is broken in, and so long as it's running well and there isn't anything amiss in any way, a UOA gives a sorta-kinda-probably correlation to wear.
The trouble with that is, under those conditions one needn't bother running UOA. If one already knows everything is fine, there's no point. It's when something is _not_ right, and the correlation goes away, that we would need the data from UOA. I've had engines in the fleet fail outright and just out of the blue, yet ‘forensic' UOA levels are either not flagged or barely-flag-able (depending on which lab's criteria one uses). Immediately previous UOAs were exactly ‘on trend.'
I am unable to reconcile my own thoughts on this topic, though, because it's only automotive (on-road) things where I can't find utility in UOA. Not even during the time I sank a lot of money into it to try to develop a better maintenance system (that's when one of the engine failures occurred). I have better confidence in what I've received back when looking at non-automotive machinery, stationary engines, and the like, but I can't quite provide the level of data I'd like to support my thoughts on that topic.
The trouble with that is, under those conditions one needn't bother running UOA. If one already knows everything is fine, there's no point. It's when something is _not_ right, and the correlation goes away, that we would need the data from UOA. I've had engines in the fleet fail outright and just out of the blue, yet ‘forensic' UOA levels are either not flagged or barely-flag-able (depending on which lab's criteria one uses). Immediately previous UOAs were exactly ‘on trend.'
I am unable to reconcile my own thoughts on this topic, though, because it's only automotive (on-road) things where I can't find utility in UOA. Not even during the time I sank a lot of money into it to try to develop a better maintenance system (that's when one of the engine failures occurred). I have better confidence in what I've received back when looking at non-automotive machinery, stationary engines, and the like, but I can't quite provide the level of data I'd like to support my thoughts on that topic.
That's an interesting observation. People associated with the UOA field admit that a UOA will quite possibly not be able to indicate/predict an oncoming catastrophic engine failure (presumably, the chunks will in this scenario be too big to be detected). I remember reading one employee of a UOA company (I couldn't find the story after a quick search), who talked about a family member who did UOAs religiously on a truck, and IIRC they looked good, until the truck was unable to drive up the ramps. Needless to say, that guy wasn't too happy with this whole UOA process.
Just out of curiosity, could you tell me more about these engines with good or decent UOAs before failure? Are we talking about one or two, or a bunch? I'd be interested in hearing more about the circumstances of these.
Originally Posted by bulwnkl
The trouble with that is, under those conditions one needn't bother running UOA. If one already knows everything is fine, there's no point. It's when something is _not_ right, and the correlation goes away, that we would need the data from UOA. I've had engines in the fleet fail outright and just out of the blue, yet ‘forensic' UOA levels are either not flagged or barely-flag-able (depending on which lab's criteria one uses). Immediately previous UOAs were exactly ‘on trend.'
Just out of curiosity, could you tell me more about these engines with good or decent UOAs before failure? Are we talking about one or two, or a bunch? I'd be interested in hearing more about the circumstances of these.
Originally Posted by bulwnkl
The trouble with that is, under those conditions one needn't bother running UOA. If one already knows everything is fine, there's no point. It's when something is _not_ right, and the correlation goes away, that we would need the data from UOA. I've had engines in the fleet fail outright and just out of the blue, yet ‘forensic' UOA levels are either not flagged or barely-flag-able (depending on which lab's criteria one uses). Immediately previous UOAs were exactly ‘on trend.'
A couple come to mind immediately. One was a 5.3L Chevy. Driving down the freeway it locked up tight (I was driving). Miles were >180k. Engine was replaced. Another was a Dodge hemi. I wasn't driving it, and I don't recall what my guy said exactly, but that engine also needed replacement. <100k miles.
dnewton3
Staff member
UOAs have multiple uses, and unfortunately here on BITOG, it seems folks don't use them to their full spectrum.
Doing a UOA not only allows one to track the equipment health, but also it's a ROI tool. As your knowledge matures with the application, so can the OCI mature in duration; that increases the ROI.
A fully mature, well managed maintenance plan is not only about the equipment but also the costs.
Doing a UOA not only allows one to track the equipment health, but also it's a ROI tool. As your knowledge matures with the application, so can the OCI mature in duration; that increases the ROI.
A fully mature, well managed maintenance plan is not only about the equipment but also the costs.
- Status
