List of oils by base-oil-quality index: 0W-20 oils

Status
Not open for further replies.
Originally Posted By: Gokhan
In the formula for 0W-xx oils, I chose 3,500,000. This is because according to the ASTM D5293 - 15 paper, the value for DV-35 is typically 1.75 times the value for DV-30. This way we can at least roughly compare the 0W-xx oils to 5W-xx oils using the BOQI.
LOL, not in my reading of the standard. I take it that you took the calibration fluids in table 1 as the source of the "fact"...how did you come up with 1.75 ? Just plucked it so that your scale would fit/overlap with your preconceptions of quality ?
 
Originally Posted By: Shannow
Originally Posted By: Gokhan
In the formula for 0W-xx oils, I chose 3,500,000. This is because according to the ASTM D5293 - 15 paper, the value for DV-35 is typically 1.75 times the value for DV-30. This way we can at least roughly compare the 0W-xx oils to 5W-xx oils using the BOQI.
LOL, not in my reading of the standard. I take it that you took the calibration fluids in table 1 as the source of the "fact"...how did you come up with 1.75 ? Just plucked it so that your scale would fit/overlap with your preconceptions of quality ?
OK, I'll put my money where my mouth is while we wait for the OP to justify his "typically 1.75 times" statement. The Standard quoted is ASTM D95293 "Standard Test Method for Apparent Viscosity of Engine Oils and Base Stocks Between –5 and –35°C Using Cold-Cranking Simulator" This, as per the standard, is the DV of the calibration oils for calibrating your Cold Crank Simulator rig Here's an excerpt of table 1 from the Standard....in the range of oils useful for the 0W, 5W, 10W comparison. Colour coding is mine, bearing in mind that for a multigrade, the label MUST indicate the lowest W grade that the oil passes, so 2,000 is likely to meet the next grade lower...So I've calculated the ratio of DVs for one grade tested at the next lower temperature, along with the DV-35/DV-30 for all of them to explore the veracity of the "states 1.75" claim. I stand by my statement that the OP has chosen 1.75, NOT because the paper that he quotes states it (clearly it doesn't) but to suit the order in which he wants his comparison "tool" (a term I'll use loosely) to "prove". If anything, the paper more supports OVERKILL's "5 degrees doubles/halves the DV" depending on which way you are taking an oil. (Not also, the finished oils aren't the test oils, they have waxes and other interactions that make the "5W at -40C" pour point tests on youtube farcical as well.)
 
First, no, I didn't doctor the DV-35/DV-30 etc. ratios. Yes, I used the values from ASTM D5203-15. Calibration oils are "Newtonian oils with low cloud points (low wax)." In other words they are quality oils but they don't contain VII. When I was researching for typical/average DV ratios, I tried hard so that the oils averaged are representative of either 0W-xx or 5W-xx oils. For that reason I chose oils satisfying both DV-30 < 7000 and DV-35 > 2500 (or perhaps 3000). With that restriction you get 1.746 - 1.757 (depending on 2500 or 3000, respectively); therefore, I estimated that DV-35/DV-30 = 1.75. I did a similar range restriction for 5W-30 vs. 10W-30 comparison for things to make sense, which resulted in DV-30/DV-25 = 1.16. You can contest these values. However, the BOQI values in the tables are all consistent with what you would expect. For example you get similar BOQI for GTL-based Pennzoil Platinum, Quaker State, etc. for 0W-20 vs. 5W-20. There is no surprise there. In fact, you even get similar BOQIs for GTL-based batches of Pennzoil Yellow Bottle 10W-30 and 5W-30, while BOQI is miserably low for a typical conventional 10W-30. To me these are very encouraging checks. And then PAO-based Mobil 1 0W-20 and Amsoil Signature Series have higher BOQIs, which has nothing to do with the scaling factor chosen, as you can compare them within the same viscosity.
 
Originally Posted By: Gokhan
I did a similar range restriction for 5W-30 vs. 10W-30 comparison for things to make sense, which resulted in DV-30/DV-25 = 1.16.
Correction: which resulted in DV-30/DV-25 = 1.73. 1.16 is the scaling factor in the BOQI equation using DV-25.
 
I don't think my error analysis question has been answered. I will state that this index, useful or not, is, like most of these things do, expressing too many significant figures. The error in CCS isn't bad, but is still far above what's being shown in the index. The error in Noack is significantly worse. You'll get one significant figure out of this.
 
Originally Posted By: Gokhan
First, no, I didn't doctor the DV-35/DV-30 etc. ratios. Yes, I used the values from ASTM D5203-15. Calibration oils are "Newtonian oils with low cloud points (low wax)." In other words they are quality oils but they don't contain VII.
My chart was the 2008 version...so you are suggesting that the Newtonian calibration oils have changed substantially between 2008 and 2015. Can you share the basics per my chart ? Name/designation and DV-25, DV-30, and DV-35 ?
 
Originally Posted By: Gokhan
You can contest these values. However, the BOQI values in the tables are all consistent with what you would expect. For example you get similar BOQI for GTL-based Pennzoil Platinum, Quaker State, etc. for 0W-20 vs. 5W-20.
Ahhhh....a method premised around cognitive bias...I expect that I'll get these results, and using the manipulation that I develop get results that meet my expectations, therefore both my expectation and technique are correct.
 
Originally Posted By: Shannow
Originally Posted By: Gokhan
You can contest these values. However, the BOQI values in the tables are all consistent with what you would expect. For example you get similar BOQI for GTL-based Pennzoil Platinum, Quaker State, etc. for 0W-20 vs. 5W-20.
Ahhhh....a method premised around cognitive bias...I expect that I'll get these results, and using the manipulation that I develop get results that meet my expectations, therefore both my expectation and technique are correct.
That pretty much says it all in a nutshell. The final touch is to give this "finding" an official sounding name and post it as a lubrication standard used in the industry.
 
Well, if you use 2.00 instead of 1.75, thinner oils will fare better with BOQI. I don't think 2.00 is correct. You're making the mistake of considering a too large range or a too small range. You need to narrow it to the relevant viscosity range as I explained for the comparison to make any sense. Regarding your 2008-version table, it doesn't cover a good range for DV-35. Despite that you still get 1.9, which is only 9% off my 1.75. Error analysis: Of course, it's about two significant figures. I include the third figure for the calculation's sake. I disagree that it's only one significant figure. NOACK definitely has two significant figures. You simply don't claim that 5.0% and 14% NOACK are the same because they both round off to 10%. In fact they are as different as day and night.
 
Originally Posted By: Gokhan
Well, if you use 2.00 instead of 1.75, thinner oils will fare better with BOQI. I don't think 2.00 is correct. You're making the mistake of considering a too large range or a too small range. You need to narrow it to the relevant viscosity range as I explained for the comparison to make any sense. Regarding your 2008-version table, it doesn't cover a good range for DV-35. Despite that you still get 1.9, which is only 9% off my 1.75.
I've both offered you to include the "good range" from the 2015 paper...and you didn't. I've included data from a supplier of calibration fluids which supports the "around 2" number in general.
Quote:
Despite that you still get 1.9
...OK, 1.96 now "rounds down" to 1.9. When I was at school, 1.96 rounded up, but I guess things have changed, yet again.
 
Originally Posted By: Gokhan
NOACK definitely has two significant figures. You simply don't claim that 5.0% and 14% NOACK are the same because they both round off to 10%. In fact they are as different as day and night.
LOL, you and the rounding errors...14% is clearly 20% Re NOACK... If the repeatability is 0.81%, and the reproducability is 1.62% (both in terms of actual total evaporation), then 9% could be either 8%, or 10%...there's no "two significant figure" accuracy in that band.
 
Originally Posted By: Shannow
If anything, the paper more supports OVERKILL's "5 degrees doubles/halves the DV" depending on which way you are taking an oil.
That's because the doubling/halving rule came from folks working in the industry and is generally supported by all of the data we've seen. Of course that said, it is a rough estimate to be used for the sake of comparison and not in any way meant to be used to obtain exact figures.
 
Originally Posted By: Gokhan
Well, if you use 2.00 instead of 1.75, thinner oils will fare better with BOQI. I don't think 2.00 is correct. You're making the mistake of considering a too large range or a too small range. You need to narrow it to the relevant viscosity range as I explained for the comparison to make any sense.
Doubling/halving is a rough guide that came from the oil industry and lacks the precision to be expanded to two decimal places. This is because of the inherent variability of what's being discussed. Your questioning of its legitimacy hinges on its ill-fitment to the comparison portion of this theory you've come up with in creating your "index" so you are "fixing it" to fit, even though, as Shannow has aptly demonstrated, that ~2 is the appropriate figure when calculated. Perhaps you should re-evaluate your process here and see if you can find a way of making it work properly with 2 as the value shrug
 
Originally Posted By: OVERKILL
Of course that said, it is a rough estimate to be used for the sake of comparison and not in any way meant to be used to obtain exact figures.
Thanks! Some people will take things too literally. He took your ballpark estimate of double/half as 2.0/0.50.
Originally Posted By: Shannow
Appears also to be 2015, published in the 4th month according to the link.
Nope. If you look at the lubes there, they are the same lubes as in D5293-08, not D5293-15. Therefore, they didn't get a chance to update to the latest standard at the time of their publication. The lubes in D5293-15 are entirely different. The problem with the numbers in their catalog is that they are all over the place, between 1.7 and 2.3. Taking the average with such random numbers doesn't help. However, if you look at the numbers in D5293-15, DV-35/DV-30 = 1.677, 1.711, 1.724, 1.742, 1.760, 1.791, and 1.813 for CL130 - CL200, respectively, and they are also uniformly increasing. Clearly these lubes are systematic and they have fixed a problem in D5293-08. Moreover, as I stated before, using 2.00 instead of 1.75 would favor thinner oils over thicker oils when the base-oil-quality index (BOQI) is calculated. Therefore, I'm not trying to bias the calculations in the direction of thinner oils as you may be thinking. What's the point of doing all this analysis if you're not even going to be honest to yourself? It would be a great waste of time doing so and where is the fun of doing that when you aren't learning anything from your results because you doctored them and can't trust them?
Originally Posted By: Shannow
When I was at school, 1.96 rounded up, but I guess things have changed, yet again.
Ha? I wasn't even using a calculator and I just estimated the average to be around 1.9 when looking at your data. I don't think we were even looking at the same numbers. Why bring this up in this manner?
Originally Posted By: Shannow
LOL, you and the rounding errors...14% is clearly 20% Re NOACK... If the repeatability is 0.81%, and the reproducability is 1.62% (both in terms of actual total evaporation), then 9% could be either 8%, or 10%...there's no "two significant figure" accuracy in that band.
Are you kidding? Let's use the scientific notation: 14 = 1.4 x 10^1 (two significant figures) = 1 x 10^1 (one significant figure) = 10 (one significant figure) So, you need at least two significant figures in this case. "Significant figure" does not mean that the figure has 100% accuracy. You can have a +/- 7% error in 14, which makes it in the range 13 - 15. This doesn't mean that you should reduce it to one significant figure and express it as 10. That would make no sense and represent a +/- 50% error instead of the actual +/- 7% error. (Also note that when you do calculations, you never round off until the final answer.)
 
Originally Posted By: Gokhan
The problem with the numbers in their catalog is that they are all over the place, between 1.7 and 2.3. Taking the average with such random numbers doesn't help. However, if you look at the numbers in D5293-15, DV-35/DV-30 = 1.677, 1.711, 1.724, 1.742, 1.760, 1.791, and 1.813 for CL130 - CL200, respectively, and they are also uniformly increasing. Clearly these lubes are systematic and they have fixed a problem in D5293-08.
That much of a change ? That's why I asked you for the numbers. But these aren't "lubes" per se...they are reference fluids, and you are ascribing the behaviour of the reference fluids to the finished oil. Given that they've clearly (apparently) changed the reference fluids for whatever reason, is that a reasonable assumption that finished oils behave the same ? Regardless, your ranking tool due to the aforementioned errors is placing merit on lubes where they have no right to be placed...the NOACK repeatabilty/reproducability means that there's a range of 8-15 where you cannot differentiate between oils in that band. Best you could say (assuming that the metric had any validity, and that Rubicon hasn't been crossed yet is 7-10 oils in "superior", 15-20 in "pretty good", another bunch in "fair to middling", and another group in "not so good". You would get the same looking grouping sorting on NOACK, or "synthetic/dino". And of course, the disclaimer that every one of the oils has met the industry minimum standards, which is still pretty good.
 
Why does any of this matter when the oil companies are constantly reformulating their oils to meet new standard and certifications? It seems to be common knowledge here at BITOG that a two year old VOA may not apply any longer due to this. In order for this "Index" to have any merit, it would have to be constantly updated. I suspect some of the figures Gokhan is using are outdated themselves. I have no problem with him trying to come up with some formula but rounding of numbers and other factors put a shadow over the efforts.
 
Originally Posted By: Gokhan
Error analysis: Of course, it's about two significant figures. I include the third figure for the calculation's sake. I disagree that it's only one significant figure. NOACK definitely has two significant figures. You simply don't claim that 5.0% and 14% NOACK are the same because they both round off to 10%. In fact they are as different as day and night.
"Of course" isn't a good enough reason as to why it's two significant figures, nor is including the third figure for calculation's sake an acceptable reason. As for Shannow's mention of Noack's spread, I am gathering he means (X +/- 1.62) %, which makes it a little easier to determine, given that Noack is a percentage. And yes, you can use whatever the heck you want for calculations, as many digits as you like, but you have to correct in the end, especially when making some list like this. As you're well aware and have seen posted here several times before, just because Noack is printed to two or three significant figures by oil companies doesn't mean it's true. Valvoline, for instance, was very quick to point out to the PQIA the error bar of Noack when they got caught on the wrong side of it. It's great to point out two or three significant figures if your number is better than someone else; Valvoline sure abandoned that number after the decimal place in one heck of a hurry when it didn't look rosy, now didn't they? Your only going to get two significant figures on Noack with a rather high Noack. Now, you make a point to Shannow with respect to significant figures. Relying solely on significant figures is the lazy way to do this, particularly since we don't have a real error analysis to fall back on here, and we know that some of these numbers are simply using too many significant figures. Why not simply include an error bar, then? One physicist told me some years back that if you can't get the error analysis on track before you analyze the data, don't even bother analyzing the data. I say right off the bat, the first 3 or 4 places are tied for first, if the index has any meaning. 7 through 11 are much the same. 11 through 20 aren't much different. You catch my drift here. Now, what have you done to ensure this actually correlates to the real world, such as taking, oh, I don't know, some technical synthetics with known low VII content, and see what we have? As it is, with the 5w-XX table, our first two products include one that no longer exists and one that only exists when and if Shell has extra synthetic base stocks to dump into conventional. I would gather that various Red Line and Amsoil products, for instance, should do very well by your calculations.
Originally Posted By: Gokhan
Regarding your 2008-version table, it doesn't cover a good range for DV-35. Despite that you still get 1.9, which is only 9% off my 1.75.
9% difference, whether you consider it "only" or a "big deal" has to be accounted for.
 
Status
Not open for further replies.

Similar threads

GSCJR
What's The Difference? Approvals vs. Specs vs. Requirements
2
Replies
34
Views
3K
Tom NJ
Saikou
Seeking for help to find a good oil for my application (2003 Hyundai Accent)
Replies
1
Views
282
Propflux01
silversport
Wouldja Still Use...?...Old Motor Oil
2
Replies
21
Views
2K
silversport
T
Oil Recommendations for Highest Wear Protection
2 3
Replies
46
Views
9K
Blighty
D
Using Liqui-Moly 6200 0W-20 in a 2019 Subaru Impreza?
Replies
10
Views
1K
02zx9r rev2
Top