Comparing oil analysis results in a “fleet” of same vehicles running 2 different oils?

[2.7ecoboostFordBronco]

I’ve read that using consumer grade used oil analysis to compare the relative suitability of different oils doesn’t produce reliable information, but sometimes I have to learn things the hard way.

It gets boring where I live and I’m always looking for entertainment. My spouse and I drive the same vehicle with the same engine, with about the same mileage.

Beyond my own amusement, would it be a useful exercise to run two different oils in the vehicles, do Horizon UOA’s and oil changes at the same intervals, then compare wear results after a few years?

We could use the same gas, same filters, and we could alternate who drives which vehicle from week to week.

I don’t know which oils I should try yet. But I have a few months before I would start this and am open to recommendations, even if it’s just to get a life and leave the testing to the professionals.

Edited to add: since user CarbonSteel is running M1 ESP 0W-30 in the same vehicle, publishing his horizon UOAs, and seems to have no intention of switching, maybe I could use his results as an experimental control if that doesn’t sound too off base.

 

[PontiacHO]

That would be an interesting comparison although my 25 year career in fleet management found that used oil analysis were flawed. Pulling from the same vehicle, sending two separate specimens, both sent to the same lab yielded two entirely different results. Same thing when using a new lab. We were curious if lab(s) UOAs were worthwhile. In my case - no. We found what worked. The oil, filters, and intervals based on vehicle type-use and were adhered to strictly. If a driver-vehicle missed a preventative maintenance, we shut off access to fuel. That gets their attention.

 

[kschachn]

Comparative oil quality measurement is difficult and expensive. No $30 spectrographic analysis is able to isolate that one variable. Day after day, week after week the method is laden with uncontrolled variables which are far more significant than the oil. In your case the longer you run this test the more unpredictable it will get. Not all experiments improve with age.

If that were a usable tool then there would be no need for tests such as ASTM D6709 or others in the Sequence tests. It would save everyone a lot of time and money.

 

[dnewton3]

I disagree; UOAs most certainly can be used to compare/contrast the performance results of lubricants. However, most folks don't understand how to use the UOA as a tool properly. The reliability and quantity of data is paramount and goes unheeded by most.

Often people want to strike down the UOAs because they believe that some inputs are uncontrollable. True; some are. But that does not automatically make the tool unusable. Rather than claiming the data is of no value, it is acceptable to consider those inputs as part of the natural variation of the process. Things like ambient temps, fuel quantity, operational patterns, etc all play into this. These things are the very inputs that the vehicle sees every single day. You don't ignore them, you account for them in the stdev calculations.

Sometimes it is better to talk outside the realm of BITOG, for the examples to make sense ...
Consider your favorite basketball team. They don't always play in the same arena each night. They don't even play the same night every week. They don't have the same ambient temps, nor same noise levels, nor same lighting, or even the same opponent each week. There are dozens of inputs that change every single game. But you can still track each player's individual performance, and the team's performance, and get reliable analytical data about their averages and trends (and near the end of season, the stdevs). It's foolish to just say "Well, we can't control 100% of the surrounding environment, and so there's no reason whatsoever to make any effort to track performance or make any conclusions." That's just absurd. Think of repetitive tasks in sports; things that happen multiple times every single game such as taking shots (2pt; 3pt), FTs, rebounds, turnovers, etc. Are we to conclude that because we don't "control" the entirety of the conditions around the game, that we can't track the data accurately?

Same goes for equipment UOAs. The normality of stdev's takes into account the inputs which are uncontrolled. To not do so would be ridiculous; no one operates their ICE in a lab. The resounding claim against using UOAs has always been "too many variables; you can't control everything in the field" ... to which I say "since when have you been able to control everything in your life?" And I counterthrust with this statement ... Since when has any HALT ever produced data which represents real world conditions?

There have even been studies which looked at multiple methods of tracking wear, and found that UOAs do show good correlation between wear metals in the UOA and wear tracked by other means such as radioactive traces, component weight analysis, etc. Further, there have been filter PC studies done which show good correlation between PC counts and wear metal counts.

The greatest violation of statistical analysis comes from the typical BITOG mentality in which some believe that a few UOAs (or less) are sufficient to make an accurate conclusion; that is patently false - it completely ignores the topic of "normality". So I do agree that a few UOAs are not a good tool in the manner which many use them here on BITOG. But that does NOT make UOA data, in and of itself, useless. It only means the user is inept at understanding the benefits and limitations of the tool; they don't understand the proper techniques and methodologies of statistical data analysis.

*********************

For the OPs quest here, he's going to have to be willing to put a LOT of time and money into this. Minimum 30 samples for each subject vehicle. It's going to take a lot more time than a "couple years". Consider that if one averages 15k miles a year, it would take 10 years of data collection, if 5k mile samples were taken. (5k miles x 30 samples is 150k miles ... divided by annual avg 15k miles ... 10 years!). By then he either lost interest in the experiment, or the vehicles were traded/totaled, or the oil formation specs have changed, etc ...

 

[2.7ecoboostFordBronco]

Would borescope images to go with the UOA’s be of any benefit? I kind of suspect even at 50k miles, which is probably where they’ll be in 2 years from now there wouldn’t be enough build up to be noticeable, but I could be wrong. Null hypothesis?

 

[dnewton3]

You can use a borescope to see stuff; often can see x-hatching if it's still in good shape. But the problem with visual checks on something like that is that it's a subjective criteria, not objective. You may or may not see a difference, but it is very difficult to quantify a difference, if any is detected at all. Generally borescopes are good for confirming suspected damage or other maladies; not really useful for quantitative assessments.

 

[Lilliston]

I agree that UOA can be useful for what you propose, OP. I also agree that it probably won't be.
(-:

A borescope would be handy to look at intake valve stems and back faces, if the engines are direct-injected. You could even push past the valves and look for scuffs in the cylinders. If one engine has noticeable or significant scuffing and the other doesn't, as a kinda-big example, I'd like to hope the UOAs would have been significantly different leading up to that visual discovery. I wouldn't hold my breath, but I'd use that information to help me decide whether or how much I wanted to rely on UOA in the future to estimate wear.

I think UOAs can be entertaining. I think there's opportunity for them to be useful, too. I came across a thread here about a 'test' where a fellow was trying something over the course of much more time and miles than (I believe) he was accustomed to doing. He was watching UOAs along with other things, and was very pleased. OTOH, during the quick thread and data review, it was crystal clear that a key wear metal was moving higher exponentially over time. The other things he was watching revealed probable cause(s) for that. Others in the thread waved it off as nothing, not significant, no big deal, or just said maybe next check would be time to stop and change. Would I have caught the initial ramp-up in wear at the time? Probably not. It was a very useful series, though, to show at the end what had been happening. THAT is the kind of thing I would use to inform me of preferred tactics in the future. The person did not collect 30 samples, and it didn't matter. The trend was clear and consistent across the entire period of the test. He wasn't comparing fluids or vehicles or filters.

I took all that to be indicative of what I see at work and in other places just every day. People sometimes 'interpret' things to mean what they'd like them to mean. Other times they use a scattering of nonsense to try to distract from something clearly happening right in front of them.

 

[2.7ecoboostFordBronco]

OK, thanks, I think will still do the UOA's. Even if I don't make it 5+ years plus to get useful data, I will still get some entertainment out of it. I think I will set up both vehicles on a 5000 mile OCI. My initial idea was to run one on the OEM brand/manual recommended 5W-30 and run the other an HPL/Amsoil type "premium" 0W-30 or 5W-30. I'm not sure I want to subject one of the engines to the OEM branded oil, even if it would be fine on 5000 mile OCI, so I was thinking of using VRP 5W-30 instead. Again, just for entertainment.

 

[fdcg27]

UOAs are entertaining and we may just learn something from what you propose to do.

 

[jetman]

@2.7ecobootFordBronco
Over the yrs I've pulled sample from vehicles of mine, sent to different labs, and results were always near identical.

Oil analysis results one time in 2019 helped me to save a Kubota diesel engine of mine. Like most diesels, at half-life rod bearings needed to be changed.
https://bobistheoilguy.com/forums/threads/mobil-delvac1300-super-15w40-kubota-v2203-107-hrs.317850/