Using UOAs to find the best viscosity for your engine?

[cammyfive79]

I read a long thread here recently where @dnewton3 said something along the lines of "as thick as necessary but no more" (I'm sure that's not exact but the same point stands) in regards to selecting the proper viscosity for your engine. Would the methodology to test for that look something like this:

1) After 2 OCIs using, let's just say 0W-20 from Brand X, run a UOA noting the value of wear metals.
2) Switch to higher grade (0W-30) from Brand X and run another UAO after 2 OCIs and compare wear metals from prior UOA.

I know there would be a lot of variables, but if, for example, the wear metals increased or statistically stayed the same going up to a 0W-30 would it be safe to say that you should back track to the 0W-20? Or vice versa if the wear metals dropped a bit with the 0W-30 that you should continue on that path? Assume you're using the 0W-20 and 0W-30 from the same series so the additive packs are the same and you're primarily dealing with a matter of only different viscosities.

This thought was ringing in my head, but I also know that things like HTHS are important in the longevity of an engine and a 0W-30 is going to have a higher HTHS than a 0W-20. I realize that plenty of cars using any off the shelf 0W-20 can go for longer than most folks would care to own one but if one was curious, what data/tests would you use to determine what viscosity is ideal for your engine; how would know when they've hit the ideal (loose term here, yes) viscosity without going to the far end of a spectrum where there is no further useful return?

 

[OVERKILL]

You can't do that. Go back and read more of what he has said. You'd need hundreds of UOA's to even begin to see statistically useful information and even then, due to the nature of UOA's, the source of the material is inferred, you don't actually know where it came from and what the mechanism was, be it mechanical wear, corrosion, chelation...etc.

 

[cammyfive79]

You can't do that. Go back and read more of what he has said. You'd need hundreds of UOA's to even begin to see statistically useful information and even then, due to the nature of UOA's, the source of the material is inferred, you don't actually know where it came from and what the mechanism was, be it mechanical wear, corrosion, chelation...etc.

Then how does one choose viscosity with an eye for evidence? I know with your Ram you decided to stick with 0W-20, but in a truly great blend. What led you do that vs others that just decided to go 30 or 40wt? Much of what I've read seems to indicate that 0W-20 (and lower) are primarily/only for fuel economy and that the average driver would only see a fraction of a percent increase maybe.

Honestly, your use case is what I was thinking about when this came into my head. Supercar 0W30 (or 0W40 even) would have a higher HTHS than Supercar 0W20, would the former(s) not intrinsically protect better/show less wear over the long haul than the latter? After you and others explained it to me, I do get your desire to run full-SAPS VW502 but why would you not run that same blend in the higher grades with the higher specs? When does going up the viscosity scale turn into a counter productive situation?

 

[OilUzer]

Keep it simple and when in doubt, use a thicker oil.

 

[facility1]

when in question use the engine oil that has the most published information (transparency) as in product data sheets etc , then go from there and learn and study proper info (not most obsolete U-tube junk) what it means pertaining to your viscosity and brand use. rely less on marketing. or what others use or even it's what I always used in past. Engine oils are seeming in the constate state of re-formulation, seek only current up to date information.

 

[Glenda W.]

There was a time when every owners manual had a viscosity chart based on air temperature. IMO oil temperature is an even better way to determine a viscosity. But I do realize that may not always be possible. I’m not convinced a UOA could pin point a viscosity based on wear numbers.

 

[dnewton3]

OP -
Please read this: https://bobistheoilguy.com/used-oil-analysis-how-to-decide-what-is-normal/


I am not a chemist nor a tribologist. Rather, I went to college and became a mechanical engineering technologist. I spent more than a decade as a Statistical Process Quality Control Engineer for a multi-national HVAC manufacturing company, (concurrent with my LEO job; had to make ends meet, etc). My experience I can contribute to BITOG is that I understand how to collect, process, analyze and summarize data. I know how to write FMEAs and DOEs, etc. I've been trained in Six-Sigma and Shannin Red-X methodologies, etc.

The most important thing which most folks don't understand about data analysis is that is it critical to get a good understanding of the variation (and there is ALWAYS variation) of any studied component , element or process. You have to have enough samples to determine a reliable standard deviation value. Thirty samples are an absolute minimum; fifty would be better. And that concept applies to each input studied.

- Want to know which viscosity is "better" for one specific application? You'd have to use the same lube brand and formulation, and then do 30 UOAs of one vis and then 30 UOAs of another vis. Pretty much no one here has the time/money to do so. Micro analysis is very expensive and consumes lots of effort and requires a lot of patience.
- Want to know which viscosity is "better" as an overall choice across a broad application, you have to apply macro data analysis. You need lots more data to crunch. You need hundreds of UOAs and all the time and money that has to go with them.

Oh, and when you're done reading that "normalcy" article linked above, please go back and read it again. You probably missed something the first time through. I get those comments all the time. When you're done reading it for a second time, if you have questions, feel free to ask.

 

[Hohn]

Two facts defeat the OP's efforts before they even start:
1) There is no "perfect" viscosity
2) Even if it existed (and it doesn't), we don't have the measurement tools and resolution to determine such.

Why is there no "ideal" viscosity? This IMO comes down to:

• The boundary conditions under which that viscosity is "perfect" only exist for fleeting moments of steady state conditions which never exist in typical usage. Change load, temperature, RPM, oil age, or any of several other factors and the "perfect" viscosity is no longer perfect. Even with a warmed engine on the highway with the cruise set at 70, you still have changes in load due to wind, hills, etc.
• Different parts of the engine each have a different preference as to what is too thick or too thin. For some parts of the engine, there's nearly no such thing as too thick (yes, journal bearings in theory have an upper viscosity limit but in practical terms in a warm engine, you could run SAE60 without issue at all). Chains probably prefer the highest viscosity you can muster. Same thing for piston pins. Other parts of the engine probably want a very thin oil to enhance flow where direct oiling isn't possible-- ring packs come to mind. Definitely piston cooling nozzles or any parts where oil cools more than lubricates.

It's quite likely that as you change viscosities, you're simply slightly altering the mix of sources of wear metals within the engine. You might not see much if any net change as a result. There's always a tradeoff and there's a reason we don't see big difference in wear across viscosity ranges.


I looked into many oil viscosity guidelines within internal engineering standards and supplier documentation and they aren't much help because the "requirement" is to use the viscosity you tested and validated. In other words, because 15w40 is the market-available diesel oil, we (and our suppliers) have engineering requirements to make sure the design works with 15w40. The "viscosity requirements" from Bosch for high pressure fuel pumps that are oil lubed basically come down to "use an 40 grade." They also have a requirement that oil not drop more than 2 cSt in service. So if you start with a 16cSt oil and it drops t o 14cSt, it is both still in grade and no longer meeting Bosch requirements. It's it's essentially a backdoor fuel dilution limit, obviously.

Is it rather interesting that on the fuel-lubricated surfaces of the pump, there is only a minimum viscosity requirement from Bosch and no upper limit....Hmm.

 

[OVERKILL]

Then how does one choose viscosity with an eye for evidence? I know with your Ram you decided to stick with 0W-20, but in a truly great blend. What led you do that vs others that just decided to go 30 or 40wt? Much of what I've read seems to indicate that 0W-20 (and lower) are primarily/only for fuel economy and that the average driver would only see a fraction of a percent increase maybe.

Honestly, your use case is what I was thinking about when this came into my head. Supercar 0W30 (or 0W40 even) would have a higher HTHS than Supercar 0W20, would the former(s) not intrinsically protect better/show less wear over the long haul than the latter? After you and others explained it to me, I do get your desire to run full-SAPS VW502 but why would you not run that same blend in the higher grades with the higher specs? When does going up the viscosity scale turn into a counter productive situation?

The HEMI isn't known for bottom end problems, so sticking with the spec visc wasn't a concern in my applications. It technically calls for a 5W-20, but the following year they switched that to 0W-20. The SRT calls for 0W-40, so that's what it gets. My concern was more the neutered API additive package, hence the full-SAPS oils, which the SRT originally spec'd.

If this was an engine family known for bearing wear problems, I would have stepped up probably to the same oil I'm running in the SRT.

 

[thescreensavers]

You don't actually know where it came from and what the mechanism was, be it mechanical wear, corrosion, chelation...etc.

This! Big part I failed to realize for too long.

 

[cammyfive79]

OP -
Please read this: https://bobistheoilguy.com/used-oil-analysis-how-to-decide-what-is-normal/


I am not a chemist nor a tribologist. Rather, I went to college and became a mechanical engineering technologist. I spent more than a decade as a Statistical Process Quality Control Engineer for a multi-national HVAC manufacturing company, (concurrent with my LEO job; had to make ends meet, etc). My experience I can contribute to BITOG is that I understand how to collect, process, analyze and summarize data. I know how to write FMEAs and DOEs, etc. I've been trained in Six-Sigma and Shannin Red-X methodologies, etc.

The most important thing which most folks don't understand about data analysis is that is it critical to get a good understanding of the variation (and there is ALWAYS variation) of any studied component , element or process. You have to have enough samples to determine a reliable standard deviation value. Thirty samples are an absolute minimum; fifty would be better. And that concept applies to each input studied.

- Want to know which viscosity is "better" for one specific application? You'd have to use the same lube brand and formulation, and then do 30 UOAs of one vis and then 30 UOAs of another vis. Pretty much no one here has the time/money to do so. Micro analysis is very expensive and consumes lots of effort and requires a lot of patience.
- Want to know which viscosity is "better" as an overall choice across a broad application, you have to apply macro data analysis. You need lots more data to crunch. You need hundreds of UOAs and all the time and money that has to go with them.

Oh, and when you're done reading that "normalcy" article linked above, please go back and read it again. You probably missed something the first time through. I get those comments all the time. When you're done reading it for a second time, if you have questions, feel free to ask.

Thanks for taking the time to supply the free education; something I never turn down. That being said, I feel you (and others) have answered my question - doing a simple A-B test between two grades isn't enough to make any sort of definitive conclusions.

These are just musings in my mind as my car "recommends" 0W-20 but recommended 5W-30 in the years prior so I'll be fine with that. I, like many here I assume, get wrapped up in the 'what if' with various discussions and how that would relate to me.