BMW N57 - 30 grade v 40 grade

[Vulusmaximus]

Pulls the pin, throws the subject out there and quickly steps back...

Gentlemen, I've been collecting BMW N57 used oil analysis...

I've have now collected 36 oil samples results on 27 engines. Excluding five because:
a) the oil is not known,
b) it's a run in oil, or
c) there is a known problem with the engine or something odd going
provides a sample size of 31 (which is just over the statistical minimum required for meaningful analysis). 16 samples of 40 grade and 15 of 30 grade. Applying Trimmean principles the highest and lowest results have been excluded from each grade so leaving 27 samples.
.
The resultant data when put into a chart form shows that the 40 grade suffers a 6.6 ppm wear rate per 1,000 miles, where as the 30 grade is 10.6. This means that on average a N57 using 30 grade is suffering 60.6% more wear than one using a 40 grade. If I expand the data set to include used oil analysis for the earlier M57 (which is a very similar), a final total of 42 samples being considered, the result is pretty much the same, with the resultant difference actually widening slightly.

Food for thought for the N57 owner...

 

[BISCUT]

Not knocking your efforts, but would like to know if there was any consistency in oils. I'm running Mobil 1 Euro 0w40 in a 14 N20 and while it's a 40wt it is not what would be considered a thick for grade oil. The N20 shears the 40 out of grade at 6k OCI.

 

[fantastic]

Thanks for piecing that graph together. Interesting to ponder.

 

[edyvw]

N57=absolutely a 40 grade. Timing chain likes that better.

 

[Vulusmaximus]

Not knocking your efforts, but would like to know if there was any consistency in oils. I'm running Mobil 1 Euro 0w40 in a 14 N20 and while it's a 40wt it is not what would be considered a thick for grade oil. The N20 shears the 40 out of grade at 6k OCI.

Here' the all wear chart I have

These are the corresponding oils:
1) Shell Rotella T6
2) Shell Rotella T6
3 )Millers CFS NT+
4) Millers CFS NT+
5) Liquid Moly
6) Ravenol RUP
7) Millers
8) Pentrite Enviro+
9) Pentrite Enviro+
10) Liqui Moly
11) Castrol Edge M
12) Castrol Edge M
13) Fuchs Titan GT1 Flex 23
14) Titan Supersyn Long Life.
15) Liqui Moly Top Tec 4100
16) Pentrite
17) Castrol Edge
18) Acea C3
19) Quantum
20) Castrol Edge
21) Mobil 1 ESP
22) Motul 8100 X-Clean Gen2
23) Motul 8100 X-Clean Gen2
24) n/k
25) Tripple Qx Synplus PD
26) Millers EE Performance C3
27) Castrol Edge Professional
28) Quantum
29) Quantum Long Life III
30) n/k
31) n/k
32) BMW
33) BMW

 

[BMWTurboDzl]

Are you adjusting for 30 grades which are LL04 or LL12fe+?

 

[BMWTurboDzl]

Pulls the pin, throws the subject out there and quickly steps back...

Gentlemen, I've been collecting BMW N57 used oil analysis...

I've have now collected 36 oil samples results on 27 engines. Excluding five because:
a) the oil is not known,
b) it's a run in oil, or
c) there is a known problem with the engine or something odd going
provides a sample size of 31 (which is just over the statistical minimum required for meaningful analysis). 16 samples of 40 grade and 15 of 30 grade. Applying Trimmean principles the highest and lowest results have been excluded from each grade so leaving 27 samples.
.
The resultant data when put into a chart form shows that the 40 grade suffers a 6.6 ppm wear rate per 1,000 miles, where as the 30 grade is 10.6. This means that on average a N57 using 30 grade is suffering 60.6% more wear than one using a 40 grade. If I expand the data set to include used oil analysis for the earlier M57 (which is a very similar), a final total of 42 samples being considered, the result is pretty much the same, with the resultant difference actually widening slightly.

Food for thought for the N57 owner...

View attachment 299293

Are you adjusting for 30 grades which are LL04 or LL12fe+?

 

[Vulusmaximus]

No, I haven't done so thus far.

 

[kschachn]

How are you eliminating all the numerous other variables that influence a simple spectrographic analysis? The relative effect of the oil is pretty far down into the noise.

How are you accounting for wear that doesn’t show up in that analysis?

And how many samples have you determined constitute a statistically significant size?

 

[wavinwayne]

Interesting data. Thanks for putting it together.

 

[Vulusmaximus]

How are you eliminating all the numerous other variables that influence a simple spectrographic analysis? The relative effect of the oil is pretty far down into the noise.

How are you accounting for wear that doesn’t show up in that analysis?

And how many samples have you determined constitute a statistically significant size?

To answer your questions:
1) I'm not.
2) I'm not.
3) It needs to be a minimum of 30. Quoting from the web on this topic; 'The number 30 is often used as a rule of thumb for a minimum sample size in statistics because it is the point at which the central limit theorem begins to apply. The central limit theorem states that the distribution of sample means will be approximately normal, regardless of the distribution of the population from which the samples are drawn, as long as the sample size is large enough.'
Obviously the larger the better which is why I mentioned I can boost the sample size including the very similar M57, but it doesn't change the resultant message. My data collection is still continuing, so I will just continue to add to the sample size over time.

I'm just sharing this info for interest, primarily because I haven't seen anything else out there focussed on the N57. Other engines may see different results of course. This 30 v 40 analysis is a spin off from looking at the data set to try and determine what a normal wear result looks like for the N57 and then what oil is giving the best result. Some would appear to be better for instance at keeping the Iron rate low, but but at the expense of the Aluminium rate. Obviously this data set is (way) too small to draw any concrete conclusions from at this stage. What we do know about the N57 is that it has a nasty habit of spinning a bearing without any warning. Keeping the Ali wear rate down has be a priority IMHO.

Anyway, like I say, makes you think...

 

[kschachn]

Yeah wear testing between motor oils is complicated and expensive. Spectrographic analyses are not the right tool for that. Your data presentation is colorful but meaningless in that regard.

 

[dnewton3]

Ah, yes .... The dangers of improper statistical analysis applied.

Allow me to interject some thoughts here. The greatest error you're making is that you are confusing macro and micro statistical analysis methodologies; you're trying to get a micro answer from macro data.

- while I agree that 30 samples is a minimum, that should be per variable
- these you have collected would be run in several different cars (engines) meaning there's a lot of variation from driver inputs, environmental inputs, etc

If you are intending to only study the effect of viscosity (say 5w-30 versus 5w-40), then you would need to study the same brand and type of lube, in the same application. For example, only study Mobil 1 in those two grades, of the same "type" (eg: Mobil 1 FS European Car Formula) because the additive packages would be similar, if not the same. You would need 30 samples of each, from the same singular application (same driver and car). Meaning you'd need 60 samples total from that source. And, at 5k mile OCIs, that would be 300k miles of testing, without incurring major engine issues or internal work (as those also would induce variables).

Please, when you have time, go read this article on "normalcy" of UOA data processing:
https://bobistheoilguy.com/used-oil-analysis-how-to-decide-what-is-normal/

Bottom line: If you want to draw conclusions from macro data, it's perfectly acceptable to do so, but you're going to need a LOT more data, AND the understanding of how to process it.