Oil Filter Testing

The FAA’s annual survey indicates piston aircraft fly around 17 million hours per year in the U.S. An assumed fleet average of 45 hours between oil changes suggests we buy more than 350,000 filters annually. The world numbers are far larger.

So what exactly are we buying?

Foremost, an oil filter should remove abrasive particles, most of which are too small to see with the naked eye. Because of their invisibility, particles less than 100 microns tend to be treated as an abstract threat, like bacteria. We know they do damage, but we may not give them the respect they deserve.

The second purpose revolves around debris we can see. Obvious metal bits provide a tangible indication of something amiss. This threat we recognize and so we diligently cut filter cans and examine the folds within like a Roman haruspex examining the liver of a sheep. Sadly, our ability to predict future events is often no better, but it does provide for airport conversation.

Less recognized is a filter’s capacity for storage. A good filter will lock away 20 to 25 grams of dirt, carbon and metals, in bits sized from 2 microns (µm) to chunks. Capacity buys time. A healthy engine supplied with clean intake air has little need for capacity. In truth, it hardly needs a filter…as long as things are going well. However, consider a tappet beginning to spall. Perhaps an oil change shows a little bit of ferrous material in the filter, but not enough for concern. Capacity allows us to reach the next 50-hour filter cut. Should things progress to rapid self-disassembly, it gets us to the next airport.

So how do we know the filter will do the job? Unfortunately, we often don’t. You might be surprised to learn there is nothing in the Federal Aviation Regulations to require or define filtration for piston engine design. Anyone is welcome to stuff something in a can, call it a filter and apply for an STC, where the sole criteria for approval is likely to be container integrity under pressure and a bypass valve. Only turbine design has a specific requirement related to filtration, stated in FAR 33.71: “The type and degree of filtering necessary for protection of the engine oil system against foreign particles in the oil must be specified.”

Thankfully, SAE ARP 1400B exists and spells out specific requirements. It is voluntary, not regulatory, an industry agreement first published 50 years ago. You won’t find it in any of the service publications, but in response to questions for this article, representatives from Lycoming, Continental and Superior unanimously stated filters for their products were expected to meet the requirements of ARP 1400B.

So what does it require? Structural compliance requires verifying proof pressure and burst pressure (400 and 500 psig), pressure drop across the element, element collapse pressure and a variety of other points, right down to the torque capacity of the big nut on the end of the can. Users may assume it won’t leak, blow up or collapse internally.

Performance compliance means meeting minimum filtration efficiency and contaminant capacity values. Efficiency is expressed as a cumulative average, in percent (%) or beta ratio (ß). Cumulative percentage is easy enough. For any given size, it’s the percentage of all particles that size or larger captured with each pass through the filter. Beta ratio is simply the number of upstream particles divided by the number of downstream particles. For math geeks, Efficiency Percentage = 1- (particles out/particles in), or ((ß – 1)/ß)*100. The rest of us need only remember larger values are better. ARP 1400B’s prescribed minimum values are a cumulative average of 75% at 25µm (ß25 = 4) and 90% at 40µm (ß40 = 10), measured per ISO 4548-12.

Cumulative percentage and beta are the common language of the filter engineering world, with their meanings carefully defined in standards. Unfortunately, buyers rarely get to see complete technical reports. Instead, the marketing departments tend toward vague terms, typically claiming a “micron rating” that may be nominal or absolute. A nominal efficiency rating is often understood to be the particle size above which the filter stops 75%, but there is no standard definition. Absolute is assumed to be the largest single particle that can pass through the element. The number on the box could be nominal, absolute or complete fantasy, so caveat emptor applies.

Capacity is straightforward. As a filter media becomes saturated with captured material, its resistance to flow rises rapidly. Capacity establishes the total mass held by the filter when differential pressure (deltaP) across the element reaches a predetermined value. The minimum per ARP 1400B is 15 grams at 8 psi for a filter with a built-in bypass valve, typically identified as xx108 and xx109 series in the GA world. The spec is 15 grams at 20 psi for a filter without a bypass, but the higher pressure has no practical value, because those filters (xx110 and xx011) are installed on an adapter with its own 11 to 13 psi bypass.

Testing Filters

First, strange as it may seem, we need standard dirt. The world’s largest supplier of ISO 12103-1 dust is Powder Technology Inc. They mill and size material found in the Salt River area of Arizona into four closely controlled grades of test dust, largely silicon dioxide with a few other compounds. ISO 4548-12 specifies Grade A3 Medium, with a particle size distribution as seen in Figure 1.

The sizes highlighted in green approximate the typical 4 to 40µm range of a filter test conducted per ISO 4548-12. A system can be calibrated to report additional efficiencies for the large particles, but there’s not much point, as all the decent oil filters are above 95% cumulative efficiency (ß40=20) at 40µm. Particles at the small end of the range can be detected with ICP spectrographic analysis.

Roughly 90% of the world’s filter test installations are built by Bonavista Technologies, in Tulsa, Oklahoma. They’re about the size of a small van and when fully optioned can cost as much as $1 million. The system requires a skilled operator, but like CNC machining centers, automation results in repeatability with high accuracy. How high? I asked an experienced project engineer at Bonavista, who said any user should be able to generate efficiency values with less than 10% variation as compared to any other lab. The best labs will vary by little more than 1%. Those estimates are matched by the round-robin exercises outlined in ISO 4548-12 itself. Although it’s possible to introduce variations with different batches of test dust, dilution ratios and counter calibrations, they are remarkably small.

The ISO 4548-12 multi-pass method is conceptually simple. A dilution system continuously injects contaminant into a fluid loop at a carefully metered rate. The fluid in the loop (MIL-H-5606, i.e., Aeroshell 4) is dosed with an anti-static and held at 100° F to control viscosity. A constant displacement pump moves it around the loop at 6 gallons per minute, similar to our small flat engines at cruise rpm. A laser-based device (most commonly a Pamas 4132) records particle count and size before the flow enters the filter. Downstream, an identical device counts and sizes the particles that passed through the filter. The flow continues around the loop, where it picks up more contaminant.

As the test progresses, the filter accumulates particles, which hinder flow. The differential pressure, measured across the element, slowly rises but as the element nears capacity, the rate of rise steepens significantly. The test is terminated at some specified deltaP, typically selected to remain well below the opening pressure of the filter’s bypass valve. The total capacity of the filter is recorded in grams; see Figure 8.

Five filters were selected as representative of current Experimental use. They were a Champion 48108-1, a Tempest 48108-2, a Wix 51515 auto filter, a K&P S-15 reusable screen filter and a Challenger 48108C screen filter. The Champion and Tempest filters were obtained from Aircraft Spruce stock in mid-January of 2024. The Wix was purchased at a local auto parts store. The K&P and Challenger were new, uninstalled units donated by Jon Friedemann and Phil Sprang. All were run using the parameters spelled out in ISO 4548-12. Each run was terminated when the filter collected enough contaminant to push the element pressure drop to 10 psig above clean.

Looking at the Data

The summary table in ISO 4548-12 format can be seen in Figure 3A. We’ll use the Champion as an example. Efficiency was recorded at 10-minute intervals until termination at 110 minutes, then averaged. Recall “cumulative” means “this size and larger,” so the filter captured 4% of all particles 5µm and larger, 61% of all particles 25µm and larger and 99% of all particles 40µm and larger. Figure 9 presents the same efficiency results in graphical format.

Sharp readers will note neither Champion nor Tempest met the 75% at 25µm standard of ARP 1400B. To be fair, 1400B specifies 8 psig termination, not 10, and efficiency tends downward as the filter collects more trash. An average based on 8 psig termination would look a little better, but we rarely collect even that much dirt. A new, clean Champion meets the standard and a new Tempest gets close (Fig 3B). If you like ’em, use ’em.

Retained capacity is more than adequate with the Champion, Tempest or Wix. None rise above 8 psig until well beyond the 15 gram minimum of ARP 1400B.

The Wix demonstrated better efficiency than the Champion or Tempest. However, be aware 1400B’s structural standards don’t apply to auto filters. For example, Wix lists the 51515’s burst pressure at 290 psi, well below the 500 psi of the aircraft brands. Although unlikely to be an issue for a user running multigrade aviation oil in a temperate climate, use of 50-weight oil in low temperatures could change the picture. Minor caveats include the lack of an installation nut, no safety wire provision and users must add a nipple if installing on a xx110 application.

As noted, this particular Wix was selected merely because it was representative of a respected auto filter. Quite a few Experimental/Amateur-Built owners already fly it as a replacement for a 48108. However, it’s certainly not the only choice and readers should not consider its appearance here as a recommendation. Readers who wish to go full auto might also consider offerings from Baldwin, Donaldson and Purolator, which like Wix is a Mann+Hummel brand. Read those specification sheets!

The test values recorded for the two screen filters are not misprints. A woven screen offers one opening size across its entire face, defined by the wire size and weave pattern. As such, they tend to act as a sieve or separator, allowing essentially all contaminants smaller than the openings to pass while blocking all contaminants larger than the openings. This is distinctly different from a depth media, a comparatively thick fiber mat with a limitless variety of pore sizes and flow paths. While a screen tends to sort particles in an absolute pass/no-pass manner, a depth media stops some percentage of all particle sizes.

At a glance, the report raised two obvious questions. On their website, K&P states, “The medical grade stainless steel cloth that we use is consistent across the entire media surface and is rated at 35 microns,” while Challenger, on their own website, states, “The medical grade, stainless steel mesh used in our filter is rated at 22 microns.” So why did neither screen register efficiency in the 20 to 40µm test range? And why the same results for both?

Let’s look at the images. Figures 4 and 5 show photos taken at 10x magnification. These and others identify both the screens as 50×250 plain Dutch weave, a widely used industrial cloth.

In addition to warp and weft counts, the cloth can be further identified by measuring the wire diameters (0.0055-inch and 0.0045-inch, respectively) and a woven thickness of 0.012 inches. A key visible characteristic is the widely spaced warp wires and the tightly compressed weft wires. When viewed perpendicular to the surface, there are no visible apertures. Flow passes through the weave at an angle, as seen in Figure 6.

To ensure accuracy, the samples were sent to a professional wire mesh calibration lab for a second opinion. The lab confirmed near identical counts for the S-15 (51×252) and Challenger (51×252), as well as the fabric type and typical performance.

So what does it mean? The widely recognized absolute micron rating for this fabric is approximately 60µm, with a reported range of 56µm to 64µm absolute, depending on source. The reported nominal rating appears to be around 40µm. These values align with the demonstrated ISO 4548-12 performance. The screens are too coarse to catch much of anything within a specified test range ending at 40µm. Instead, they captured only the largest particles in the A3 size distribution. Continuous contaminant injection combined with limited removal eventually raises the circulating concentration to a level beyond the capability of the particle counters.

Note the retained capacity and test time. The screens clogged at 3 grams, well below the 15 gram requirement of ARP 1400B. As they clogged, the deltaP ramped up sharply, rather than exhibiting the more gradual rise of a depth media. It’s a characteristic of screen filtration called blinding—the obstruction of the openings in the filter, which are uniform and limited in number. When operating in a constant-pressure system, blinding generally results in the formation of a filter cake and a drop in flow. However, when operating downstream of a constant displacement pump (i.e., an engine oil pump), the pressure must spike.

The screens did offer less clean pressure drop, a detail often quoted in their favor. However, here it is limited and doesn’t last long. There simply isn’t enough media area. Figure 7 is a visual comparison of a Champion element and a Challenger element, hung together on the shop door. Figure 8 is a summary of pressure drop across the filter elements vs. circulated contaminant.

So What Is Good Enough?

Although reference literature clearly shows high-efficiency oil filtration significantly reduces wear, the practical benefit depends on the operating environment. Dirty air, sand on the ramp, maybe a little rust? Run the best filter you can find. However, given minimal airborne dust and limited corrosion, our engines do fine with limited filtration. Many make TBO with nothing more than a strainer screen. It catches the big stuff and it buys time when things go wrong. The 25-hour oil change removes the small particles.

Increasing the change interval requires better fine filtration, as oil alone can’t carry the dirt load. In 1975, when ARP 1400B was first adopted, spin-on filters were only 20 years into general acceptance on new cars. The 1400B standard of 75% @ 25µm and 90% at 40µm was state of the art. It allowed aircraft oil change intervals to extend to the automotive standard of the time, about 2500 miles for “severe use,” which is 50 hours at 50 mph.

Our next culture shift is on the way. Lycoming Service Letter L270 already approves 100-hour oil changes (after the initial 50-hour change) with continuous use of an approved unleaded fuel. The interesting detail is the filter requirement, which remains at 50 hours. That too will change when the aircraft industry adopts synthetic filter media. Eventually.