I've been building a dataset of particle counts (PCs) from used engine oil tests, with the intent of comparing how different models of oil filter perform. All of the data is from PCs that have been posted to this site over the past 20 years or so. They're all from gasoline engines with model years ranging from the late 80's to present day. To date, there are 75 particle counts included.
Certain test results were considered to be outliers and were excluded from the dataset. Most of these outliers had high particle counts combined with high wear metals on the UOA, often with high silicon or coolant contamination as well. One was excluded due to suspected filter clogging.
Only results for the third ISO code (particles >14 micron) are shown. The reason for this is that Blackstone changed their testing methods around 10 years ago. They used to use extrapolation for determining the counts of the smaller particles, instead of actually measuring them, which basically makes those particle counts meaningless. It's also impossible to compare the PCs that were measured using the two different methods, since the old method produced results that were 5 to 6 ISO codes lower for small particles. In the >14 um size, there is no obvious difference in results between the two methods, and they're assumed to be comparable.
Wherever possible, actual particle counts were used to calculate ISO codes to one decimal place, for better precision.
Table 1 shows the average particle count results by model of oil filter, as well as the standard deviation, and the number of test reports the average is based on. The averages for almost every filter include PCs from multiple different engines. Only tests done by Blackstone are included.
The relative ranking of the filters is more or less as you would expect, with known efficient filters like the Amsoil/Royal Purple greatly outperforming known low-efficiency filters like the Purolator BOSS and OEM Toyota. The standard deviations are small enough to conclude that many of the differences between filters are statistically significant.
It isn't obvious from the table, but the particle counts have a much more obvious correlation to the oil filter model than to any other factors I could identify. The model of engine doesn't seem to be too important. Oil or filter change intervals don't seem to matter much either, though it seems there are a couple of exceptions with long OCIs. Any factors aside from the oil filter that affect the particle counts should average out, at least for those filter models that have a large enough sample size.
Table 2 shows the particle counts for filters that I only have a single test result for. It's hard to draw strong conclusions about oil filter efficiency from this due to the limited data, but here it is.
If you'd like to help contribute to this dataset, please post links to any new particle count tests to this thread. I've scoured most of BITOG for any particle count tests posted here, but if you have any PCs that you suspect I've missed, or if you're aware of PCs posted to other forums, please post links to the original source here.
If you're going to get a particle count test done, having it done by Blackstone would be most useful for expanding this dataset, and you'll have a large dataset to compare your results to as well. WearCheck also seems to do a good job with particle count tests. I would avoid SpeeDiagnostix, since all four PCs I've seen from that lab have had the exact same ISO code for every particle size, and with particle counts much lower than what is typical from Blackstone, so I question their accuracy.
Certain test results were considered to be outliers and were excluded from the dataset. Most of these outliers had high particle counts combined with high wear metals on the UOA, often with high silicon or coolant contamination as well. One was excluded due to suspected filter clogging.
Only results for the third ISO code (particles >14 micron) are shown. The reason for this is that Blackstone changed their testing methods around 10 years ago. They used to use extrapolation for determining the counts of the smaller particles, instead of actually measuring them, which basically makes those particle counts meaningless. It's also impossible to compare the PCs that were measured using the two different methods, since the old method produced results that were 5 to 6 ISO codes lower for small particles. In the >14 um size, there is no obvious difference in results between the two methods, and they're assumed to be comparable.
Wherever possible, actual particle counts were used to calculate ISO codes to one decimal place, for better precision.
Table 1 shows the average particle count results by model of oil filter, as well as the standard deviation, and the number of test reports the average is based on. The averages for almost every filter include PCs from multiple different engines. Only tests done by Blackstone are included.
The relative ranking of the filters is more or less as you would expect, with known efficient filters like the Amsoil/Royal Purple greatly outperforming known low-efficiency filters like the Purolator BOSS and OEM Toyota. The standard deviations are small enough to conclude that many of the differences between filters are statistically significant.
It isn't obvious from the table, but the particle counts have a much more obvious correlation to the oil filter model than to any other factors I could identify. The model of engine doesn't seem to be too important. Oil or filter change intervals don't seem to matter much either, though it seems there are a couple of exceptions with long OCIs. Any factors aside from the oil filter that affect the particle counts should average out, at least for those filter models that have a large enough sample size.
Table 2 shows the particle counts for filters that I only have a single test result for. It's hard to draw strong conclusions about oil filter efficiency from this due to the limited data, but here it is.
If you'd like to help contribute to this dataset, please post links to any new particle count tests to this thread. I've scoured most of BITOG for any particle count tests posted here, but if you have any PCs that you suspect I've missed, or if you're aware of PCs posted to other forums, please post links to the original source here.
If you're going to get a particle count test done, having it done by Blackstone would be most useful for expanding this dataset, and you'll have a large dataset to compare your results to as well. WearCheck also seems to do a good job with particle count tests. I would avoid SpeeDiagnostix, since all four PCs I've seen from that lab have had the exact same ISO code for every particle size, and with particle counts much lower than what is typical from Blackstone, so I question their accuracy.
