Originally Posted By: 2010_FX4
Originally Posted By: Phishin
Perhaps, to really get down to brass tacks, we should do a study like this one:
http://www.bobistheoilguy.com/forums/ubb...177#Post1250177
It is clearly apparent here, that Mobil 1 filters are far superior to Amsoil's EAO's filtering efficiency with real data in a real engine. Each filter being tested twice, giving nearly identical numbers between the two runs.
This type of test would be hard to argue against.
I will volunteer to run a particle test on the next two samples of XG2 or XG10575 filters. A few questions:
1. What filter should be used as a comparison (
Purolator is out!)
2. What mileage should the sample be pulled? 5K?
3. What lab should be used for the particle test? Is Blackstone OK?
Let me give you guys some help here for your experiment because I do a lot of this for a living and have certified various products and testing to ASTM, ISO and others as the 3rd party validation. (And actually done this type of filter testing for industrial lubrication applications several times working in conjunction with several top tier filter manufacturers) I have to be deliberately vague on some of this because I do have some sensitive product information and a legally enforceable secrecy agreement.
I have seen some good ones and some not quite as good and based on the commentary on this thread I see an inherent issue that will skew the results.
This is informal but for what I think you guys want to know, it will deliver accurate results that you can hang your hat on.
When you are designing a test you have to start at the end. You need to define exactly what the results are in term to the parameter sought. Many people accidentally render their results invalid because they either fail to control all the affecting variables (to eliminate false data) and they put too many questions in the test and then they cannot separate the data for proper analysis.
Let’s establish the question the test is designed to address
I want to know whether filter A, B or C has the best filtering efficiency in a real world scenario.
If that’s what you want to know then you have to define the criteria for that test AND you must make sure the controls are in place to not alter your question or the results in relation to the question. Here are some examples.
If you define filter efficiency in just its particulate count (which is a very valid thing) then you have to eliminate any possible affecting agents such as fuel dilution, glycol contamination and progressive wear. This is why the engine test is unreliable to address the question from a scientific perspective because every machine (even serial number 1 &2) are going to run and wear differently. All those elements (and others) affect the filter media and skew the results. Also an engine wears progressively will fluctuate. Granted the engine is a good test for the marketing department but would be thrown out as unreliable in any official test.
Another factor is the beta ratio. Let’s say you have filter 1&2 with the exact same media but one is 20% smaller than the other then theoretically they would have the exact efficiency until the filter reached the saturation point then the bigger filter would “win” in the home stretch if the time was not factored in and this will skew the data relative to the question because we didn’t say “…. Over a period of X times, cycles, miles, hours and what not” in our test question. It’s common to use miles or hours on rolling stock but if you want a measure of true scientific accuracy then use cycles or passes because operating hours or miles can be an influencing factor and an influencing variable due to actual operating conditions unless you were measuring that efficiency weighed in a specific vehicle or operating envelope.
(The experiment exceeded the service life of one brand and invalidated the test because service life was not part of the test)
The last really big factor is controlled delivery. Every filter made has either an internal bypass built in or an engineered failure point to where it will allow flow when fully stopped. (To be differentiated from varying degrees of restriction which is from 0-99% and clogged is total flow stoppage). Sadly nothings perfect and saturation/swelling from chemicals like water and a cow patty of sludge hitting it like a freight train or a severe pressure or thermal spike can and does overcome these but that’s the exception, not the norm. Also, all filters bypass some under varying loads in normal operation. But if your test is not a run to fail destructive test, this is beyond the scope.
So the test “mud” needs to be delivered at the expected operating temperature and flow/pressure that does not trigger those set points. Its normal to run sole parameter efficiency tests lower than operational parameters for a static test for efficiency alone but if you are doing it for a specific application such as a screw compressor then you would test over system parameters but you would have to change the test question to something like “I want to know whether filter A, B or C has the best filtering efficiency in an Ingersoll Rand 500HP compressor” because you now added a measure of bypass that you really cannot capture and differentiate from the efficiency data alone. (Which is what I normally do because the plant guys generally don’t care about bench tests- they are interested in their production equipment, maintenance costs and downtime)
So to build your test to give you pretty solid data in relation to the question, here is a simple way to do it.
Save about 10 gallons of oil changes, heat it to an engine temp, agitate it and pull your baseline sample to know what the counts are. (You could add media also if you wish if you really want to punish the filter)
Pour it equally into 2 clean containers
Use a diaphragm pump or gravity feed or even compressed air (using caution with the air) to move the test oil through 2 filter set ups in a low force method. (I recommend the diaphragm because you don’t want any possible wear or other pump related factor to skew the test)
Decide the number of passes or time (just keep it the same) and pour the contents from the exit container back into the supply container.
Sample the end results
For the grand finale, cut the filter to verify there was no internal malfunction or membrane saturation that might affect the integrity of your test. (And even count the contamination in the filter too if you really want to get down and dirty.)
That would be a very accurate test to answer the test question in terms that would be hard to refute by anyone. Not 100% to the scientific method but well beyond the legal standard of beyond a reasonable doubt. (I can attest that the above type test WILL win in a court of law on a warranty claim because it has in a failure analysis due to using an offshore cloned filter that claimed all kinds of stuff but did not meet the OEM requirements for a given airend and destroyed it- BE VERY CAREFUL of things made overseas when shopping aftermarket)
As far as lab selection, shouldn’t be an issue with any credible one but if you want to account for that too (just to be sure) then pick any ISO certified lab that primarily does industrial sampling because they won’t have any bias whatsoever and if you don’t tell them the type machine they wouldn’t know anyway.