Mobil 1 and 3.6 Pentastar

[goingplacesanddoingstuff]

UOA's lack the resolution to allow you to discern meaningful differences in wear performance between different oils, that's not their intended purpose, but unfortunately, believing they can results in many people chasing their tail.

Engines are complex mechanical devices with various systems in them that can indeed respond differently to different base oil blends, additive combinations, FM formulations...etc. This can result in subtle differences that may, or may not, be apparent to the end user. This could be something as simple as increased/reduced volume of consumption due to the PCV setup, or an apparent reduction/increase in NVH due to the FM interaction with something like cylinder deactivation. However, UOA's aren't going to give you any insight into this, nor does it mean wear performance is going to be any better/worse.

Personal anecdote: Years ago, I had very high levels of consumption with AMSOIL AZO 0W-30 in my Expedition 5.4L. M1 0W-30 didn't consume at all in the same vehicle, nor did Motul 0W-30. I did several OCI's to see if the consumption tapered off (which it can with a chemistry change), but it never did, so I had to stop using that oil in that vehicle. Do I know why that oil consumed? No. Do I know HOW that oil consumed? Also no. Had I done UOA's on that oil, would it have given me valuable insight into this phenomenon? absolutely not.

I agree. My point was not that UOAs can discern details on any differences, but rather that differences unto themselves are evidence of something (it doesn’t matter what). Given how complex engines are, it is very unlikely that any two UOAs will always be the same when varying only oil type.

Had you done UOAs, while it may not have given you insight, it could have shown a difference in fuel dilution, rate of degradation, etc. Any difference there is evidence that the complex system that is the engine has variables within it that respond to the input variable. It’s like “black box” testing.

 

[OVERKILL]

I agree. My point was not that UOAs can discern details on any differences, but rather that differences unto themselves are evidence of something (it doesn’t matter what). Given how complex engines are, it is very unlikely that any two UOAs will always be the same when varying only oil type.

The error bars on UOA's are such that, even if they were the same, they still aren't, or even if they aren't, they still could be. Re-tests are always interesting on here because many things change.

Had you done UOAs, while it may not have given you insight, it could have shown a difference in fuel dilution, rate of degradation, etc. Any difference there is evidence that the complex system that is the engine has variables within it that respond to the input variable. It’s like “black box” testing.

Doubtful, the engine was port injected and not one known for fuel dilution. I expect it would have shown strong TBN and generally just cast a huge question mark on why it was disappearing.

Using spectrography to trying and discern the difference in wear between approved oils is like trying to judge the size of a mountain by rubbing the sand that's at the base between your fingers. There is meaningful data to be gleaned from the sand, and its rate of accumulation, but that isn't one of them.

 

[goingplacesanddoingstuff]

The error bars on UOA's are such that, even if they were the same, they still aren't, or even if they aren't, they still could be. Re-tests are always interesting on here because many things change.

Doubtful, the engine was port injected and not one known for fuel dilution. I expect it would have shown strong TBN and generally just cast a huge question mark on why it was disappearing.

Using spectrography to trying and discern the difference in wear between approved oils is like trying to judge the size of a mountain by rubbing the sand that's at the base between your fingers. There is meaningful data to be gleaned from the sand, and its rate of accumulation, but that isn't one of them.

I should not have provided examples. There are many ways any two UOAs could show any kind of differences. Again, any differences could be cited as evidence. It really doesn’t matter what they are.

You have a complex system with data in and data out. If the data out changes given different data in, you know there’s a difference in how the system responds to the inputs. That’s all my point is trying to make.

It is a fair point that if two of the same data ins give different results greater than the magnitude of any of the other data ins then it’s not a statistically meaningful exercise. This would be an interesting experiment.

 

[OVERKILL]

I should not have provided examples. There are many ways any two UOAs could show any kind of differences. Again, any differences could be cited as evidence. It really doesn’t matter what they are.

Any differences could be cited as being within the error bars for the test, ergo, they cannot be reliably cited as evidence. This is why we trend UOA's, not just do samples of one and try and draw wild, baseless conclusions from them. The purpose is to have a trend so that significant deviations from that trend can be used to potentially identify developing issues with the equipment, ideally before they cause expensive/extensive damage.

 

[goingplacesanddoingstuff]

Any differences could be cited as being within the error bars for the test, ergo, they cannot be reliably cited as evidence.

I understand. This is why we do statistical significance and students t-tests in science. See my edit above.

I don’t necessarily believe that in all cases all inputs would always fall within a statistically insignificant margin of all outputs in this case. We may need to let this crux of the debate rest here. I’d be very interested if there were experimental data for this scenario.

We are quickly delving into the philosophy of scientific inquiry. Hats off to you.

 

[kschachn]

I understand. This is why we do statistical significance and students t-tests in science. See my edit above.

I don’t necessarily believe that in all cases all inputs would always fall within a statistically insignificant margin of all outputs in this case. We may need to let this crux of the debate rest here. I’d be very interested if there were experimental data for this scenario.

We are quickly delving into the philosophy of scientific inquiry. Hats off to you.

Even when they do you're still left with attributing an observation to an isolated variable or variables. This is the biggest problem with UOA on here, there is no control on the multitude of variables. Much like observations of fuel economy.

 

[OVERKILL]

I understand. This is why we do statistical significance and students t-tests in science. See my edit above.

I don’t necessarily believe that in all cases all inputs would always fall within a statistically insignificant margin of all outputs in this case. We may need to let this crux of the debate rest here. I’d be very interested if there were experimental data for this scenario.

We are quickly delving into the philosophy of scientific inquiry. Hats off to you.

Yes, that's exactly where this leads

Not the best evidence, but it's what we have: If you look at some of the UOA re-tests and even tests of the same oil from different labs, it becomes quite apparent that the level of variation is significant.

Then, as @kschachn notes, you are trying to correlate that with something else. Then you have to consider that in terms of what you are are observing with respect to the metals component, is within an extremely narrow range of particle sizes able to be seen by the equipment.

One example:
https://bobistheoilguy.com/forums/t...3-labs-1-with-pc-tan-tbn.365649/#post-6403094

Here are VOA's done at three different labs:
https://bobistheoilguy.com/forums/t...ferent-test-laboratories.344857/#post-5860305

 

[goingplacesanddoingstuff]

Yes, that's exactly where this leads

Not the best evidence, but it's what we have: If you look at some of the UOA re-tests and even tests of the same oil from different labs, it becomes quite apparent that the level of variation is significant.

Then, as @kschachn notes, you are trying to correlate that with something else. Then you have to consider that in terms of what you are are observing with respect to the metals component, is within an extremely narrow range of particle sizes able to be seen by the equipment.

One example:
https://bobistheoilguy.com/forums/t...3-labs-1-with-pc-tan-tbn.365649/#post-6403094

Here are VOA's done at three different labs:
https://bobistheoilguy.com/forums/t...ferent-test-laboratories.344857/#post-5860305

Woof, yeah, that’s bad data. Even with imperfect and limited measurements it is still possible, using some clever techniques, to ascertain statistically significant information. That’s one of the craziest things I find about empiricism.

The null hypothesis would be that there is no variation in UOA when varying only oil type:

https://www.britannica.com/science/Students-t-test

Theoretically this technique could be used.

 

[buster]

You could have oil A show an average of 20 ppm of Fe and oil B show an average of 10ppm of Fe, but oil B could be more prone to allowing deposits to form, or have higher volatility etc. I think years ago many of the oils that were touting how well they did on the IVA wear test (Valvoline in 2008'ish) were mediocre at best at preventing high temperature deposits. So, it's obvious you need to have a very well-balanced oil that can do everything at a pretty high level and not just excel in one particular area.

 

[OVERKILL]

Woof, yeah, that’s bad data. Even with imperfect and limited measurements it is still possible, using some clever techniques, to ascertain statistically significant information. That’s one of the craziest things I find about empiricism.

The null hypothesis would be that there is no variation in UOA when varying only oil type:

https://www.britannica.com/science/Students-t-test

Theoretically this technique could be used.

An additional issue here is that even if there is a difference identified using this technique, that is, that we reject the null hypothesis, there is still the problem of having to infer what that value actually means, because something like wear is inferred, and larger particles aren't captured by the sampling method.

If we are looking at something like TBN, which is directly measured, or fuel dilution (ignoring Blackstone, which doesn't measure it directly via GC), or the virgin element concentrations of the product, that's one thing.

However, if we are looking at iron ppm or some other metal where the assumption is that it comes from wear (but can also come from chelation and corrosion), the potential source is a variable with its own unknown range, as is the range for the particle size distribution, which can fall outside what's visible to the measuring device. This complicates things. And then you have the ever varying operating conditions of the equipment compounding this.

So, ignoring that two identical engines won't produce identical wear profiles (though they should be more similar), if you ran a 10,000km OCI interval experiment for 200,000 miles with the only difference being the lubricant (so identical operating conditions somehow, including climate, driving profile...etc) and had the following results:
- Oil A: 2.5ppm/1,000 miles of iron
- Oil B: 2.2ppm/1,000 miles of iron

Without doing an actual tear-down with measurements, you still can't conclude that Oil B provided less wear than Oil A. Oil A could have more polar base oil chemistry that naturally chelates some metal and could actually provide a lower wear rate, despite yielding higher ppm/1,000 miles. Oil B could be allowing more deposits to form, which capture some of the iron particles as they don't stay in suspension, artificially reducing the apparent wear rate.

 

[BornSlippy]

I have noticed UOA's on oils that have esters in them tend to have elevated iron levels. Is that because the engine is wearing faster on the oils with esters, or are the oils with esters liberating the iron laden deposits and now ending up in the oil and the UOA shows elevated iron? The knee-jerk reaction would be to condemn that oil and swear off of it for good.

 

[buster]

I have noticed UOA's on oils that have esters in them tend to have elevated iron levels. Is that because the engine is wearing faster on the oils with esters, or are the oils with esters liberating the iron laden deposits and now ending up in the oil and the UOA shows elevated iron? The knee-jerk reaction would be to condemn that oil and swear off of it for good.

I have not noticed that. It shouldn't matter if the oil is formulated correctly.

Red Line uses substantial esters (guestimates range from 20-40% in some grades) and it was hypothesized that the esters were "scavenging" and or interacting with Pb in some bearings, which showed up as high Pb or elevated wear metals. I do not know what type of esters Red Line uses and whether they interact in this manner. Amsoil and HPL also use esters and they do not show elevated wear metals.

Esters are known to compete with antiwear additives and other additives for surface space. I suppose if not properly formulated, wear would be higher among such oils.