Thanks Doug...wasn't wanting to pick on them but I have noticed what i thought were some very "off' numbers. Some oils I have a pretty good idea on where the add packs should be and this time they were hugely different....Hope all is well Down Under!!
I had cause to ask B-S to retest a sample just the other day. The VOA was done by Bruce at either his in house or preferred lab in California (apparently a major lab in such testing). Blackstone, to their credit, had no problem retesting an assuring that calibrations were intact. The results were marginally different. I have an identical sample being sent to Bruce to see what he gets. I've also sent a VOA to Bruce to see if his resources get a match on what he sent me.
There was a study (I've lost the link), which oddly didn't include any USA labs, that showed that in some blind testing, labs were incapable of repeatable results with the same samples. Even after telling them that they had been in error from controls, most failed to achieve within some standard deviation of true (now I have to wonder how they established controls
).
I imagine most can be viewed in managed/controlled errors
Edit: Let me add that this was in terms of additive levels. The elemental levels fell along the lines of expectations.
Hi,
Gaery - That is my experience too - replication/repeatability is at best "within limits" and at worst almost "unbelievable"
Even with in-house (Oil Company) Labs the results can be challenged unless all formulation data is kept up to date and the "condemnation limits" adjusted accordingly. I have had many issues with both Mobil and Castrol over many years. I must add though that 98% of the time they were/are great!!
As for individual engines (or engine families) and uptake/degradation rates then personal trending is really the best path to follow
As oliver88 knows, the Porsche 911 (air-cooled family) has so many variances in model structure over time and many have added engine/technology enhancements as well - that "universal" averages mostly become quite meaningless in this case