Check out my magnets on my oil filter

[twX]

The oil filter on my Subaru bypasses some oil at cold idle rpms with the oil at 30 degrees C. In really cold weather, it will be bypassing over 80% of the oil flow.

How do you know that?

I calculated an estimate for oil flow at the bypass pressure, based on the pressure differential test results for the FRAM Ultra in the Ascent Filtration oil filter test. (I use a FRAM Ultra).

The XG3593A filter I use is smaller than the tested filter, and I so I estimated that it hits the bypass pressure at 85% of the flow shown in the test. As far as I can tell, the filters use the same media and number of pleats, but the tested filter is taller and has more media area.

Based on this, my oil filter would hit the 12 psi bypass pressure at 47 L/min oil flow, at the same 13.5 cST viscosity used in the test, equivalent to the flow from my oil pump at 4500 rpm. In other words, even with warm ~85 degree C, 5W30 oil, some oil will be bypassed at high rpm with this filter.

I found a source that says that pressure drop across a filter goes up linearly with viscosity. I calculated the flow at which the filter will bypass at different viscosities based on this.

Of course, a Subaru filter with its 23 psi bypass probably does not bypass as easily. However, many filters that other manufacturers recommended for this engine will bypass even more easily than the filter I'm using, especially since the XG3593A has around 15% more media area than the filter FRAM recommends for this engine.

Now, Subaru engines are atypical since they use high flow oil pumps, but I suspect that if you did the math for a more typical vehicle, the oil would start bypassing at cold idle rpms at an oil temperature somewhere between 0 and 20 degrees C.

 

[ZeeOSix]

I calculated an estimate for oil flow at the bypass pressure, based on the pressure differential test results for the FRAM Ultra in the Ascent Filtration oil filter test. (I use a FRAM Ultra).

The XG3593A filter I use is smaller than the tested filter, and I so I estimated that it hits the bypass pressure at 85% of the flow shown in the test. As far as I can tell, the filters use the same media and number of pleats, but the tested filter is taller and has more media area.

Based on this, my oil filter would hit the 12 psi bypass pressure at 47 L/min oil flow, at the same 13.5 cST viscosity used in the test, equivalent to the flow from my oil pump at 4500 rpm. In other words, even with warm ~85 degree C, 5W30 oil, some oil will be bypassed at high rpm with this filter.

I doubt your oil pump is putting out 47 L/min (~12.5 GPM) through the filter and engine at only 4500 RPM. You have a flow vs RPM vs output pressure graph of your specific oil pump? An unregulated oil pump (not connected to an engine) might flow that, but when it's connected to a flow restrictive oiling system, the pressure relief on the pump would probably cut that flow back - more so the colder/thicker the oil is. The pump will regulate a max output pressure, and the output flow has to decrease as the oil viscosity increased if the pump is in pressure relief. You would never know what the actual oil flow rate is through the engine without measuring it. Also, 13.5 cSt viscosity would be close to a xW-40 at ~100C (212F).

I found a source that says that pressure drop across a filter goes up linearly with viscosity. I calculated the flow at which the filter will bypass at different viscosities based on this.

The dP doesn't go up linearly with viscosity because viscosity vs temperature isn't a linear function. There is a correction factor in that reference source if the viscosity is different than the referenced oil used in their calculations. And the dP isn't a linear function either with flow while the viscosity is held constant - the graph on page 23 shows that.

Of course, a Subaru filter with its 23 psi bypass probably does not bypass as easily. However, many filters that other manufacturers recommended for this engine will bypass even more easily than the filter I'm using, especially since the XG3593A has around 15% more media area than the filter FRAM recommends for this engine.

Now, Subaru engines are atypical since they use high flow oil pumps, but I suspect that if you did the math for a more typical vehicle, the oil would start bypassing at cold idle rpms at an oil temperature somewhere between 0 and 20 degrees C.

Hard to say what a dP vs flow curve on an OEM Subaru oil filter looks like. Could be they have a significant;y higher dP vs flow curve than some aftermarket oil filters - only a bench test like what Ascent did would show that. I'm sure the higher oil pump flow output on some Suabru engines also have a factor, but it's not the only one. If OEM Subaru filters are relatively restrictive, then that combined with the higher flow pump would require the filter to have a higher bypass setting. But if the filter has a lower dP vs flow curve, then the bypass setting doesn't have to be as high.

The filter bypass setting is not only dependent on what the max expected dP causes by the oil viscosity and flow will be (with a loading factor too of course), but also based on the flow performance and total area of the filter media. So the filter bypass setting is dependent on many things - filter flow performance (dP vs flow curve), max expected viscosity & flow factor, and the expected debris loading factor during normal/recommended use.

As a reference, here is real measured dP vs flow data on a new PureOne filter of medium size, with an oil viscosity that equates to 5W-30 at around 200F. Even at 14 GPM this filter only had 6 PSI of dP. So if this filter had a bypass setting of 15 PSI, the max dP would only be about halfway to bypass.

 

[twX]

You have a flow vs RPM vs output pressure graph of your specific oil pump?

Here is the spec from the service manual, 60.2 L / minute at 6700 rpm. It's a Subaru thing. The spec is with 120 degree C oil, but thicker oil won't change the flow much in a positive displacement pump.

The dP doesn't go up linearly with viscosity because viscosity vs temperature isn't a linear function.

I don't really see the logic here. The dP-viscosity relationship is independent from the viscosity-temperature relationship.

There is a correction factor in that reference source if the viscosity is different than the referenced oil used in their calculations. And the dP isn't a linear function either with flow while the viscosity is held constant - the graph on page 23 shows that.

The correction factor they refer to seems to be designed only to take into account the differences in filter restriction for different models of filter. I know that it states that the correction factor needs to be taken into account for a different viscosity, but it also says that pressure drop is proportional to viscosity, and in two other places it states that the correction factor is dependent on the filter design, with no mention of viscosity. The correction factor tables they provide also only give correction factors for different filters, not viscosities. This online pressure drop calculator (albiet for pressure drop in pipes) also shows a linear relationship between pressure drop and viscosity if you input two different viscosities with the other parameters unchanged.

I'm aware that pressure drop is not linear with flow. I've taken into account, as the Ascent Filtration test tests the dP at many flows, and I've just interpolated the data points on the dP-flow curve from that test to find the pressure drop at the flows I'm interested in.

Hard to say what a dP vs flow curve on an OEM Subaru oil filter looks like. Could be they have a significant;y higher dP vs flow curve than some aftermarket oil filters - only a bench test like what Ascent did would show that.

Agreed. I have no idea what the characteristics of the OEM filter are like, aside from the fact that they are paper filters, quite small but with a healthy number of pleats, and they have been made by FRAM as well as other manufacturers.

What is the source of the dp-flow curve you posted? Is it typical that high-efficiency filters like those tested in the Ascent test are several times more restrictive than a standard-efficiency paper filter at a similar viscosity? I'm skeptical that the high-efficiency filters would sacrifice that much flow performance.

 

[ZeeOSix]

Here is the spec from the service manual, 60.2 L / minute at 6700 rpm. It's a Subaru thing. The spec is with 120 degree C oil, but thicker oil won't change the flow much in a positive displacement pump.

Notice the oil temperature (248F) and the discharge pressure at those flow rates. Like mentioned before, I doubt with colder/thicker oil that the pump would stay out of pressure relief at high RPM which means the max flow would be cut back and less than 63.6 qt/min (15.9 GPM). Obviously, some Subaru engines have a pretty crazy oil pump and put out lots of flow. As mentioned, the high flow pump is part of why Subaru OEM filters have a high bypass valve setting. The other part of the dP vs flow curve of the OEM filters is the other part, and that's unknown.

Guess if anyone is worried about the oil filter going into bypass on a high flow pump Subaru, then they should use only oil filters with the real high bypass setting regardless if it's an OEM filter or aftermarket. Some aftermarket filters have a unique filter with the higher bypass setting for Subaru applications. IMO, the higher bypass setting is more for cold start-up and warm-up periods, but once the oil is at full operating temperature there is less danger of it going into bypass. In any vehicle, it's always wise to keep the RPM down until the oil gets near full operating temperature.

I don't really see the logic here. The dP-viscosity relationship is independent from the viscosity-temperature relationship.

After looking around, looks like if the flow rate is held constant and just the viscosity is changed, then the pressure drop vs viscosity curve would essentially be linear. But of course, with the viscosity held constant, the pressure drop vs flow curve isn't linear, but exponential. On the same page now.

What is the source of the dp-flow curve you posted? Is it typical that high-efficiency filters like those tested in the Ascent test are several times more restrictive than a standard-efficiency paper filter at a similar viscosity? I'm skeptical that the high-efficiency filters would sacrifice that much flow performance.

Purolator is the source - they ran the dP vs flow test in their facility. The PureOne at the time was shown to be 99% @ 20u, so it's high efficiency. It also uses a lot of media area, which is a way to reduce the dP across the filter. Just because an oil filter is high efficiency doesn't mean it can't be made to flow well with low dP is it's designed correctly.

 

[ZeeOSix]

^^^ BTW ... looking at that dP calculation source showing how they calculate dP on their filters. There are two correction factors: 1) One is for the filter model/configuration out of the model table, and 2) One is for the viscosity correction factor if the viscosity is different than the baseline of 30 cSt. Their calculation example shows how they used both correction factors. CF was 2.00 (out of the table) for the filter they chose, and the other correction factor was (V2/V1) which was 46/30 in their example. They use a colon ( : ) to mean division. Kind of confusing unless you run the numbers to see what the are doing.

Y is the filter model correction factor out of the table, and (V2:V1) is the viscosity correction factor where V1 = 30 cSt.

 

[twX]

Like mentioned before, I doubt with colder/thicker oil that the pump would stay out of pressure relief at high RPM which means the max flow would be cut back and less than 63.6 qt/min (15.9 GPM).

This is what I was missing, and what led me to believe that all engines would bypass cold, thick oil regularly. I now think that for a filter with sufficiently low restriction or high bypass pressure setting, the bypass might not ever operate unless until the filter gets clogged up, as the pump bypass will always operate before the filter bypass can operate. Assuming the viscosity-dP relationship is the same for the filter as for the rest of the engine's oiling system, the pump bypass may reduce flow enough to prevent the filter from bypassing regardless of how thick the oil is. However it could be that if the oil is very thick, the dp-viscosity relationship is different for the filter compared to the rest of the engine.

I did some updated calculations for filter dP vs flow and viscosity for my engine, with the FRAM Ultra filter, based on the Ascent Filtration test data:

For the oil pump specification conditions, the filter dP only reaches 9.6 psi (Scenario 1). With the oil temperature at 107 C, the filter starts to bypass at 6700 rpm (Scenario 3). For a more typical oil operating temperature of 90 C, the filter bypasses around 20% of the oil flow at 6700 rpm, while the oil pump is not yet bypassing any flow (Scenario 4).

Scenario 2 shows that the viscosity required for the oil pump to start bypassing at 6700 rpm is 14.8 cST. The filter would have a dP of 21 psi if it did not have a bypass. The OEM bypass spec of 23 psi would be perfectly suitable for a filter with this level of restriction, but the FRAM filter would be bypassing around 28% of the oil flow. I would expect the percentage of flow bypassed to always be around this level while the oil pump is in bypass, regardless of what combination of high rpm ond high viscosity is causing the oil to bypass. If the OEM filter has the same restriction as the FRAM, I would expect the bypass to never open, as the oil pump would always start bypassing first. This seems like the way to properly design a filter for an application.

I was under the impression that oil filter bypassing was quite normal with cold oil and not a big deal. Now I don't believe this is true. The FRAM Ultra will filter sub-22-micron particles better than the OEM filter, regardless of whether the FRAM filter is bypassing oil. Above 30 micron, its efficiency may only be around 70% while it is bypassing, vs 90 to 99.9% for the OEM filter. I believe that most engine wear is caused by sub-30 micron particles, but is this only because oil filters are very efficient at capturing anything larger? What size particles cause most wear on engines that do not run oil filters, or that only run bypass filters? Given that most engine wear occurs during engine warm-up, the oil filter's performance in this regime seems the most important.

In any case, I found a PurolatorOne with a 20-30 psi bypass rating that is recommended for a different model of Subaru. It's quite small and I can't be certain it won't bypass even with the higher bypass pressure setting, but it may be a better bet than the FRAM Ultra in this regard. I could always go back to OEM, but after spending a few dozen hours researching oil filtration, that would feel like a personal failure

 

[ZeeOSix]

A few comments on the Subaru oil pump specs. Do you or anyone you've seen have an oil temperature and oil pressure gauge on that engine ... preferably digital gauges that have at least a 1 degree and 1 PSI resolution to monitor oil temps vs RPM to see if it correlates to the specs Subaru shows in the service manual?

At 6700 RPM with 120C (248F) 5W-30 oil it's saying the filter and engine is getting 60.2 L/min and the corresponding oil pressure at that flow and oil temp/viscosity should be 46.6 PSI. BTW, your table shows "L/s", but should be L/min.

So 60.2 L/min is 15.9 GPM. What's the sump capacity on the engine? If it's assumed to be 5 quarts, then a 5 qt sump is basically fully cycled 12.7 times every minute ... or once every 4.7 seconds. That seems pretty crazy to me. Under those constant flow conditions, can the sump level even stay at a safe level and not allow the pump to start sucking air? Can 5 qts of oil fully drain back to the sump in less than 5 seconds?

BTW, I dug into more info on the PureOne dP vs flow curve I posted earlier and the test oil viscosity was 11.3 cSt. So shift the whole curve up ~20% to compare as if the viscosity was 13.5 cSt like Ascent used.

In all these Subaru high flow oil pump discussions I always end up suggesting that if someone doesn't trust the filter maker's called out application for their Subaru, then either use the OEM filter or find an aftermarket filter maker that shows a unique filter application for that engine that shows a higher bypass valve setting than their similarly sized filter for other non-Subaru engines.

 

[Zee09]

Very magnanimous of you!

 

[twX]

So 60.2 L/min is 15.9 GPM. What's the sump capacity on the engine? If it's assumed to be 5 quarts, then a 5 qt sump is basically fully cycled 12.7 times every minute ... or once every 4.7 seconds. That seems pretty crazy to me. Under those constant flow conditions, can the sump level even stay at a safe level and not allow the pump to start sucking air? Can 5 qts of oil fully drain back to the sump in less than 5 seconds?

It's 5.4 quarts with oil and filter change. Incidentally, the Subaru WRX STI has an oil pump rated even higher at 63 L/min, a sump capacity of only 4.5 quarts, and is known to have issues with oil starvation at high cornering loads. It's common for owners of those cars to install larger, baffled aftermarket oil pans.

 

[ZeeOSix]

I think Subaru engineers went a bit off the deep end with the oil pump volume. My Z06 stock oil pump only puts out around 7 to 7.5 GPM at 6500 RPM. High performance after market oil pumps for that engine put out around 9 to 9.5 GPM.

It would still be interesting if someone could actually measure the oil temp and oil pressure on one of those Subaru engines to confirm is it matched the service manual specs. My Z06 had oil temp and pressure digital gauges, and I plotted the oil pressure vs oil temp vs engine RPM, and when the oil was hot (5W-30 Mobil 1 at 200F) the pressure would not get high enough to hit oil pump relief. And I'm sure the flow would not have put the filter in bypass either.

 

[DGXR]

If you need that many magnets, you have bigger problems than oil and oil filtration.

I don't think this is a need, it's not a band-aid to keep the engine alive for another year before it grenades.
This is just another layer of wear protection above & beyond the oil filter and won't hurt anything. We have seen pictures of cut open oil filters that were run with magnets on the outside -- it's obvious what the magnets catch on the inside of the metal canister.

 

[sk_pete]

80c will kill neodymium.
Go alnico, samarium or ferrites

 

[Furious]

It will keep the fine metal filings stuck to the inside of the filter instead of circulating through the engine. A magnetic oil plug would probably be just as good?? IMO But this can't hurt......Good on you.

My S10 which has a 4.3 comes factory with a magnetic drain plug, every oil change the magnet was clean, and because the drain plug was kind of chewed up from 20 years of use replaced it with a non-magnetic one.

Vehicles been running 200,000, 300,000 upwards of a million miles without magnets with just regular oil changes, in my opinion a magnet while certainly not hurting anything has very minimal to zero benefit on a regular daily driver.

I could see it being more useful in heavy duty applications and race cars.

 

[daman]

Looks good to me just the right amount

 

[garageman402]

@cman, did you change that filter yet? Did you cut it open to see how much you caught?

 

[MrQuackers]

This is no good! It could potentially create a magnetic flux within the can. And if by some mishap the can becomes ungrounded it could hold a magnetic charge. Which creates capacitance. Meaning you have just unintentionally created a flux capacitor. Then you get up to 88mph, next thing you know - boom! You’re back in 1985. Tread lightly, sir.

Hmm, that was a good year

 

[SubieRubyRoo]

You'll get nearly the same effect with a bunch of store bought neodymiums.

Id argue the button the stream is probably the best method - and way cheaper than a filtermag.

Although if one tracks the car this may not be the best idea… it’s been awhile since I checked the actual temp but somewhere below 300*F IIRC neodymium’s lose their magnetism. Then, not only is the button not collecting ferrous material, it’s also free to bounce around in the space between the return tube and the gasket.

 

[UncleDave]

Although if one tracks the car this may not be the best idea… it’s been awhile since I checked the actual temp but somewhere below 300*F IIRC neodymium’s lose their magnetism. Then, not only is the button not collecting ferrous material, it’s also free to bounce around in the space between the return tube and the gasket.

With the setup I saw I don't think it could make its way into the tube.

80c will kill neodymium.
Go alnico, samarium or ferrites

An n35AH will go to about 425, but I doubt you'll find those at home depot.

Oddly the shape helps with temp resistance.

 

[dumkid]

Nice magnets

 

[Al]

Here are mine on my '18 and 22 Forester