3RZ 280k - What I've learned from 9 oil reports

[JZiggy]

Oil analysis reports from the engine's life 230k - 280k

I've switched back and forth between 5W-40 diesel oils and RGT 5W-30

The way I am summarizing the data is to normalize the wear metals by 1000 miles of use on the oil.

I then averaged the Fe and Al between RGT and the diesel oils... assuming this is statistically significant (which may be a stretch) I am seeing that Fe wear is identical but Al wear is higher with the RGT. In no particular case has any RGT run done better than any diesel run.

Once I run out of my stash of M1 TDT I think we'll stick with T6

(Note the one Fe "excursion" value of 14 was found after hundreds of miles of desert offroad driving, so I'm OK with it)

 

[BMWTurboDzl]

IIRc the margin for error is around +/- 5ppm so basically there's no difference.

 

[JZiggy]

IIRc the margin for error is around +/- 5ppm so basically there's no difference.

That's a pretty crappy margin of error and it would surprise me if that was the case. I'll email them and see what they say about it.

Also -- "Hi!" from another Atlanta resident

 

[kschachn]

Agreed. Without passing the data through a PhD mathematician I’d say it’s easily within normal variation. Besides there’s no way to conclusively assign that observed difference to one isolated variable (the oil) since that’s not in any way the only variable in everyday driving. People are constantly trying to do so here but it’s an impossible task and unwarranted.

 

[JZiggy]

Agreed. Without passing the data through a PhD mathematician I’d say it’s easily within normal variation. Besides there’s no way to conclusively assign that observed difference to one isolated variable (the oil) since that’s not in any way the only variable in everyday driving. People are constantly trying to do so here but it’s an impossible task and unwarranted.

Not to be contentious but we don't need a PhD to do some meaningful statistical work. If we know the 99% confidence intervals in the measurements I think we can get somewhere. Agreed that there may be other variables not considered. But provided no major changes are made (ie, significant mechanical work) then these variables become less important as the sample size increases.

 

[Kuato]

And the 3RZ is in what year and vehicle?

Bet the oil specification is 0w-20....and no harm.

 

[JZiggy]

1999 4Runner

I believe 5W-30 is spec'ed for US, up to 20W-50 for other markets...

 

[wpod]

We can see that all the oils used kept the parts adequately separated. Great engine.

 

[BMWTurboDzl]

That's a pretty crappy margin of error and it would surprise me if that was the case. I'll email them and see what they say about it.

Also -- "Hi!" from another Atlanta resident

This may be of interest https://bobistheoilguy.com/used-oil-analysis-how-to-decide-what-is-normal/

 

[JZiggy]

We can see that all the oils used kept the parts adequately separated. Great engine.

True statement!

 

[bbhero]

You need at least 30.... Yes minimum 30 data points to be considered statistically significant.

 

[Danh]

Plus, even virgin oil will have some contaminants. Not at all unusual to see 1-2ppm of aluminum in a VOA sample.

 

[JZiggy]

Blackstone states accuracy of +/-5% for wear metal readings, which translates to small fraction of a ppm.

 

[dnewton3]

So, in terms of statistical accuracy, there's far too little data to claim either oil did better or worse than another. To be able to see something statistically significant, you first have to understand how much "normal" variation exists in the typical data set (macro or micro). Read the "normalcy" article in the sticky section.

That being said, the wear looks fine with either lube.