Originally Posted by DGXR
The following graphic is pulled from this article:
How to Choose the Right Oil Filter for Your Car
Obviously it is an oversimplification -- many owners are getting 300,000+ miles using standard economy filters -- but the point is still valid: yes efficiency matters. How much does it matter? That depends on the individual owner. There is a balance between efficiency, economy, and reasonable service life of the engine. I am not trying to get 800,000 miles out of my engines because it would be more micromanagement and more stress and more work than it's worth, to me.
Buy a well made vehicle and stay on top of maintenance, and it will last "a long time." That is good enough for me.
All -
That article is based on the infamous GM filter study (SAE 881825).
I cannot believe that we have to have this discussion over and over and over and over and ....
That GM filter study is the most over-used, misunderstood, useless and misleading study that has graced the pages of magazines and websites for decades now. It is, essentially, worthless to the common man, and has no use in the real world. And I mean NONE, ZIP, NADA.
Allow me to quote myself from years back ...
If, after reading my assessment, you want to have a meaningful discussion about the facts of how/why/what that GM filter study does and does not prove, then I'll listen. But I encourage you to buy/read the study BEFORE you post up another word ...
Originally Posted by dnewton3
First of all, we should agree that there are two filters on the typical engine; an air filter and an oil filter. The air filter generally deals with silicate ingestion; the oil filter deals with soot generated from the engine, as well as anything ingested that would pass into the lube system, and wear particles themselves.
We need to understand how a "normal" engine ingests contamination via air filtration. I offer Jim Allen's excellent explanation here:
http://www.bobistheoilguy.com/forums/ubbthreads.php/topics/3229015/5
Depending upon how often you change air filters, you can significantly alter the ingestion rate. Just as with any filter, frequent changes actually REDUCE the efficiency. So I'm going to make some assumptions upon average folks and not anal-retentive BITOG over-achievers ...
Using Jim's data, I'll estimate that approximately .75 oz of dust ingested over perhaps 30k mile air filter change intervals. That is equivalent to about 21 grams of dust.
In regard to the GM filter study, I call into question not the validity of the study itself, as I understand the premise of its intent, but rather the application of the study to real world use of filters in everyday lives of millions of pieces of equipment. I'll go over my contentions one at a time:
1) Contamination loading:
In the GM study, they dumped 50 grams of fine AC dust into the sump every hour, for 8 hours. (Page 2, first paragraph). That is 400 grams of contamination over the 8 hours of testing. They did this to "accelerate" the wear attributed to differing filtration levels. For us to understand how much this relates to the "real world", we have to understand how much dust would enter an engine during normal use, and then figure an estimated mileage duration that would infer, presuming average air filter changes and loading. Using Jim's data, we can use the average of 21 grams of dust every 30k miles. Considering GM induced 400 grams of dust in the entire test that would be roughly equivalent to 19 air filter changes. Multiply that FCI quantity by the miles per change and you can see that the contamination loading was equivalent to 570k miles of typical road use dirt ingestion. Yes - you read that right; the sump in the GM study suffered Five-Hundred-Seventy Thousand miles of contamination loading based upon typical air filter changes. As I already stated, this is somewhat dependent upon your air filter change interval, etc. While we could debate this exposure duration, let us just agree it's a LOT of contamination represented by a LOT of miles. Whether you think it's 400k miles, 500k miles, or 600k miles is of no real consequence to me. Most folks NEVER own and operate a vehicle or other piece of equipment this long. This represents a HUGE amount of dust ingestion; more than a lifetime for most folks.
2) Oil sump changes:
In the GM study, they never changed oil for the duration of the test. While they did filter it, they never changed it, relative to each filter used. Each sump lasted 8 hours for each filter trial. Given that the sump endured an approximation of 570k miles of contamination ingestion, the OCI duration equivalent in terms of ingestion loading was also 570k miles. That does NOT mean the other contributors to contamination were equal; there is no reason to believe that soot loading was very high as only 8 hours were run per test. Soot loading is a factor of incomplete combustion byproducts; that is not an issue here because the engine simply didn't run long relative to the real world OCI. In other words, the engine did not burn 570k miles worth of fuel; it only burned 8 hours of fuel, so the soot loading would have been very low relative to the ingestion of the fine dust. But the "age" of oil in terms of the variable manipulated (fine AC dust loading to affect wear) was prolifically long to say the least. A "typical" person would perhaps OCI every 5k miles, and would have seen 114 oil changes relative to the contaminant loading. To put that in perspective for 8 hours duration, they would have changed oil every 4.2 minutes to represent "normal" OCIs in terms of contamination. But they never changed oil at all. And so the sump loading of contamination was allowed to become extremely prominent to say the least. The overall presence of particulate was WAY more than a typical sump would ever see even in a worst case scenario. Why do we want to understand this? Because, while the filtration was manipulated ABOVE 15um, the net result was that a huge amount of small particulate stayed in the sump for the entire 8 hours! Any particle that was 5um, 7um on up to 10um was able to continually circulate repeatedly with no capture at all! Those particles (and there were certainly a LOT of them according to the data) just floated along indiscriminately and did damage while no filter was able to remove them. Therefore, because they didn't change oil, they never got rid of the small particles (5-10um) that do a lot of damage. They dumped in 500k miles of dust, and then never addressed particles that are capable of damage below 10um. That 10um size is important and will be discussed further down; see the * … In short, because they never did an OCI for the equivalent of 500k+ miles of dust ingestion, the UOA wear data represents a LOT of metals due to smaller particulate never leaving the system; never at all.
3) Add-pack condition:
In the GM study, because they heavily dosed with dust, thereby creating artificial wear rates over one R-E-A-L-L-Y L-O-N-G OCI, the additive package was greatly overwhelmed. The anti-agglomerates and detergents were so hopelessly over-run that I cannot really find a way to describe or define how it could be measured. Let it suffice to say there was no hope that the additives would have been able to handle the loading. Referring to the OCI duration in point #2 above, a 570k mile OCI with only oil filter changes isn't representative of real world add-pack health. Admittedly, silica is not directly controlled by dispersants, but they can alter the ability of the add-pack to function when their concentration is so grossly high. I don't know of any SAE study or ASTM test that can show us a definitive cut-off point or direct correlation, but I highly suspect the 570k miles of equivalent silica is "over the top" to say the least. I'll note this as well; because the test was only run for 8 hours, we can exclude soot contribution to the loading of particulate; the engines simply did not run long enough to really produce a significant amount of soot. Eight hours is only one full day's drive, after all. Overall, this topic is moot in terms of wear contribution. And so, the VAST majority of wear is only attributed to the equivalent of ingestion wear and not hydrocarbon byproducts.
4) Filter efficiency and the state of claimed improvements:
In the GM study, all filters were rated at 98% efficiency (a fairly good rate overall) at the desired particulate range as the starting point. They tested eight (8) filters total; four for a diesel engine and four for a gasoline engine. The four diesel filters were rated at 40um, 15um, 8.5um and 7um; all rated at 98% first pass. The gasoline engine filters were rated at 40um, 30um, 25um and 15um, again 98% first pass. They used the 40um filter as a "baseline" for performance. Now, we need to understand that today's "typical" filter is nowhere nearly that bad in terms of performance. Many filters are available that can be 98% at 25um or maybe even 20 um, some are even 99% at 20um. Therefore, the "baseline" of the "improvement" in wear reduction really isn't based upon a realistic starting point. We can easily get a decent filter that is 95-99% efficient at 20um from any manner of brands. The use of a 40um filter for a starting point may or may not have been reasonable back in 1988 when the study was posted, but it's not anywhere reasonable today as most filters are much more efficient than that. So the claim by GM that filtration can reduce wear by "70%" is biased in that they started from such a poor state to begin with. That "70%" wear reduction rate was based upon contrasting the 40um filter to a 15um filter in the gas engine application. They showed a 70% reduction of wear going from the worst to best filter at 98% efficiency. But in today's world, it would be easy to start at 20um as "baseline". And frankly, you'd struggle to find a filter that would be so efficient at a significantly smaller size anyway in terms of full-flow performance; I'm not aware of a filter that is commercially widely available that would be 98% at 15um off the shelf for the typical automotive application.
* Also, they noted that while single pass filtration efficiencies can predict relative wear data shifts, multi-pass filtration can also narrow performance disparity when averaged over the life of the test. And I quote:
"
Even though filter (A) was rated at 40 micron, it effectively removed particles down to 10 micron. To do this, recirculation of the oil through the filter was required." In other words, use your filter and the efficiency increases! Just as Jim Allen's data shows in air filtration, that same concept applies to oil filtration. The longer they used the 40um filter, the better job it did, and to a point where at 10um, there was a convergence of filter efficiency between all filters tested!!! To quote the study:
Note that concentrations converged above 10 micron for all filters. (page 4, fourth paragraph).
In essence, if you use a 40um filter long enough, it will perform as if it were a 15um filter as the pores close down. And any particulate smaller than the typical pore size after multi-pass, will pass ANY filter media anyway. This is why I state that once a filter is appropriately defined in terms of efficiency and pore size, using a "better" filter really does not show any real-world tangible wear reduction. Here is why this happened, so read VERY CLOSELY and UNDERSTAND the cycle of the test protocol.
- They dump in 50 grams of dust (equivalent to 70k miles of ingestion all at once!), and this is done once every hour
- Wear escalates because the FIRST SEVERAL PASSES of the oil allows a lot of garbage to continue around in circulation and generate wear in the engine
- As the media loads up, REGARDLESS of the starting pore size rating, the filter essentially loads to a point where ALL filters tested see performance converge above 10um
Why is this important to understand? Because the filters with larger pore sizes allow a lot more stuff to circulate in the first few passes, causing a LOT of wear in the first few minutes of each hour's "ingestion". But after those first several passes, the filters will all settle to a reasonably similar pore size with good efficiency. The wear spikes at the front end of the contamination load in the test, and then it falls dramatically after a few minutes because the media of ALL filters becomes loaded to a point where 10um pores are about the only thing remaining! The filter was ONLY changed once the dP would approach bypass. Until then, the filter just continued to load up and all filters loaded equally well after the first few minutes.
This is why I state that using a "better" filter really does not reduce wear in a tangible manner for the average garage engine in a typical application. While the first pass efficiency may result in a tiny fractional difference, the multi-pass effect over 5k-15k miles is moot because all the filters essentially load up equally. And because we don't "spike" dirt into the engine (the air filter stays in place and the soot production is a low constant), there will never be a cause for wear to escalate arbitrarily.
(NOTE: The ONLY time we typically see wear escalate is at the front end of an OCI, and that is because of the removal of the tribochemical barriers by the add-pack, as established and proven in the Ford/Conoco study 2007-01-4133. It has nothing to do with filtration in this regard.)
How would all this relate to the real world? Well - if you're inclined to "spike" your engine with dirt by arbitrarily removing the air filter for a several weeks and driving through a dusty bean field all day long, then this would roughly be a reasonable equivalent. Your wear would escalate dramatically until your oil filter would capture what your missing air filter did not. And don't forget to not change oil for while you're at it!
Here's what I do like about the GM study: they did show a reasonable correlation between wear data in UOA analysis and wear data as measure by % weight loss concentration. This is actually one part of the study I like and believe has merit, although it is only mentioned in passing. They did both methods, as well as relied upon former studies also linking wear data tracking methods to show that UOAs can be reasonably used to track relative wear conditions. They also noted that physical measurement methods are prone to errors; you cannot disassemble an engine multiple times during a test and expect repeat-ability as thing like bearings and such will be altered by the removal and re-installation. However, changes in weight of components had a reasonable correlation to percent shift in UOA spectral analysis; I agree with this!
And so, I contend that the GM filter study was a lab test that did prove what it set out to prove. It showed a reasonable correlation of wear reduction to filtration pore size at a stated efficiency relative to the first few passes. But that entire test was heavily biased towards accelerated wear to a point where no "normal" equipment would ever be allowed to run. The test bordered on, in my opinion, absurd. I would liken such treatment to abuse or neglect. Some would contend that they did this to "accelerate" the wear to simulate 500k+ miles of use. OK - I might agree with that. But again, they did not also do the things in that simulated 500k miles which ALSO go along with wear control. They didn't change oil at a reasonable frequency; they didn't change oil at all! Therefore the wear they induced was ONLY applicable to someone who runs a 500k mile OCI, and only manages the oil filter to a point where the component is changed only when the dP across the media is at 10-20psi (a point at which most any normal filter would already be in constant bypass due to complete media blinding anyway) ….
Here is a quote I agree with, but only because they confine their statement well:
By comparing filter bench test performance with the engine wear data, it becomes apparent that a filter's single pass efficiency correlates very well with its ability to control abrasive engine wear." (page 5, paragraph 1)
Why do I agree? Because they state it related in SINGLE PASS scenarios. And this is proven true when you never change oil and also dump a slug of garage into the sump!
But if you change oil with normal frequency, and maintain a reasonable air filter situation, and you allow the oil filter to control contamination via MULTI-pass, you'll NEVER see this kind of disparity between filters.
Do you see the difference between what they did in their "test" and what the real world does in the garage?
And so I disagree with anyone who says that study has merit in the real world. No one I know of, nor any maintenance program I'm aware of, uses such parameters to run their equipment.
And GM even acknowledged this on page 2, in the last paragraph …
Used oil analysis from engines in the field will not typically show such a clear correlation since wear metals generated between oil changes will be a much lower concentrations.
In other words, they know that because OCIs were negated AND contamination was grossly overdone in their test, simple routine maintenance will not ever result in such wear rates, therefore the filter disparity will never materialize. So GM went to great effort to correlate UOA wear data with weighed component wear data, and then clearly states that real world usage wear data will never show filter performance differences because wear is just never, ever that bad in normal circumstances.
In short, I agree that the test proved what it set out to prove. What I disagree with is that the study has any valid application to real world situations. And anyone who states such will have to prove to me just how they think 500k mile OCIs with single-pass base-rated at ol-skool 40um filters is applicable to today's equipment management.
![[Linked Image]](https://www.bobistheoilguy.com/forums/attachments/usergals/2019/02/full-39486-30658-gm_study.jpg "[Linked Image]")
