Filtration in live animal production inherited its ratings, its test velocities and its efficiency bands from commercial HVAC. Most of that transfer works. These are the places where we keep running into questions.
Standard practice in live animal filtration was built on commercial HVAC research and adapted as the industry grew. That has served producers well. None of what follows is a criticism of it. These are five places where the numbers stop short of the question we were actually asking, and where we would be glad to be pointed at work we have missed.
ANSI/ASHRAE 52.2-2017 sets 492 feet per minute as the default test face velocity where no other velocity is specified. That is appropriate for the commercial HVAC systems the standard was written for. A filter in an attic housing over a ceiling inlet, or seated in a wall at the end of a barn, is not that system.
Working the barn figure from published numbers: a nominal 24 by 24 filter presents four square feet of face, so the standard test moves roughly 1,970 CFM through it. Reicks, presenting at the 2009 Banff Pork Seminar, put it plainly: a 24 by 24 filter supplies about two adult swine at maximum summer ventilation. Extension hot-weather rates are 500 CFM per head for a sow and litter, 300 for a breeding sow or boar, and 180 for a gestating sow. Two head at those rates is 1,000, 600, or 360 CFM across four square feet, which is 250, 150, or 90 feet per minute. A barn filter runs somewhere between one-fifth and one-half of the velocity it was rated at.
Pressure drop falls with velocity, somewhere between linearly and quadratically depending on where the media and the pleat geometry sit on the curve. Across this velocity range that is a reduction of roughly half to ninety percent, which is a wide band but a real one: the airflow penalty of a high-MERV filter in a barn is materially smaller than the published resistance figure implies.
Efficiency is less simple, because the capture mechanisms do not move together. Diffusion capture strengthens as velocity falls, which favours the fine end. Inertial impaction weakens, which works against the coarse end. Interception is unaffected. Which effect dominates in a given band, on a given media, across this velocity range is not something we have found measured. The inlet velocity figure quoted in ventilation guidance, 600 to 1,000 feet per minute with a year-round goal near 900, is a different measurement at a different plane and does not answer it.
The open questionWe have not found fractional efficiency published for the same filter at test velocity and at barn velocity. The pressure-drop side of the change is predictable from the curve. The efficiency side is not, because diffusion strengthens and impaction weakens as velocity falls and nothing we can find establishes which one wins in which band.
What that means nowThe resistance figure on a datasheet overstates what a high-MERV filter actually costs a barn in airflow. The efficiency figure at that same lower velocity is neither confirmed nor contradicted, so it stays a rating rather than a measurement of what the filter does in service.
Sources ANSI/ASHRAE 52.2-2017 §8.1.2 · Reicks, Banff Pork Seminar 2009 · NCSU Swine Extension hot-weather ventilation rates. The two-head-per-filter figure is a practitioner rule of thumb, not a measurement, and everything derived from it inherits that.
ASHRAE 52.2 permits seven test face velocities: 118, 246, 295, 374, 492, 630 and 748 feet per minute. The standard also requires the rating to be reported with the velocity it was measured at, so a filter tested at the default reads as MERV 8 at 492 FPM, not simply MERV 8. In practice the velocity travels with the number on the paperwork and drops off in conversation.
246 feet per minute sits at the top of the range a barn filter appears to operate in, and it is already one of the seven. Running the same filter at both velocities produces two complete reports on the same media by the same method: fractional efficiency across E1, E2 and E3, plus initial resistance, at each condition. The comparison would be per band, since that is where the mechanisms pull in different directions.
The most direct evidence the industry has is the University of Minnesota PRRSV challenge work, which established biological endpoints: whether a given system prevented infection in susceptible animals. A bench test does not substitute for that. It does measure outcomes rather than mechanisms, so it cannot separate how much protection came from filter efficiency, how much from system integrity, and how much from operating conditions.
A paired-velocity test would speak to the first of those, and not to the other two. We have not found one published for a filter sold into livestock production. The velocity is already permitted and the method does not change, which is what makes it worth asking about.
The open questionWhether any filter sold into livestock production has been reported at both 492 and 246 feet per minute. Both velocities are already permitted under the same standard, the method is unchanged, and the output would be two complete fractional-efficiency reports on identical media.
What that means nowA MERV number is a result at one test condition, not a property of the filter. Ask which velocity produced it, and treat the answer as the starting point for a barn calculation rather than the end of one.
Airborne pathogens travel attached to something: a droplet nucleus, a skin cell, a fragment of feed or bedding dust. That carrier particle, not the bare organism, is what reaches a filter. So the band a carrier occupies is the number that governs whether a given filter protects against a given disease.
Cascade-impactor measurement with infectivity resolved by size fraction exists for PRRSV and swine influenza A, from the Alonso work at Minnesota, plus avian pathogenic E. coli in a BSL-2 chamber rather than a barn. PEDV was measured by size and shown to be infectious, but no size threshold for isolation was reported, so it sits half in and half out. For the other pathogens we have looked at in swine and poultry production, the band is assigned by inference from carrier material and route, with no direct measurement we have found behind it. African swine fever, foot-and-mouth, Mycoplasma hyopneumoniae, Newcastle disease, infectious bronchitis and Salmonella sit in that group.
Our own carrier table publishes those inferences with the basis marked on every row, because the alternative is presenting a guess as a finding. Marking them does not turn them into measurements.
The cascade-impaction design that produced the PRRSV numbers is more than a decade old, and as far as we can tell it has not been extended to the organisms in the second group. For a producer weighing filter class against African swine fever, the band in question is inferred rather than measured, and it is worth knowing which one you are looking at.
The open questionCascade-impactor work with infectivity resolved by size fraction exists for PRRSV, swine influenza A and avian pathogenic E. coli. For African swine fever, foot-and-mouth, Mycoplasma hyopneumoniae, Newcastle disease, infectious bronchitis and Salmonella, the band is inferred from carrier type, with no direct measurement we have been able to find behind it.
What that means nowBefore a filter class is chosen against a specific disease, check whether the band driving that choice was measured or inferred. Our carrier table marks the basis on every row for exactly this reason.
A pre-filter on the upstream face of a main filter captures coarse dust and extends the life of the expensive element. The mechanism is not in dispute and the field record supports it. The part that stays hard to pin down is how much, for a specific pairing, at a specific dust load.
Life extension is not a property of the pre-filter. It is a property of the combination, the barn, and the site. Some manufacturers do publish airflow for a main filter with a pre-filter installed, at stated static pressures, so for those products the resistance side of the pairing is available before anything is ordered. The part that decides the economics is the other side: how much longer the main lasts behind that pre-filter, under that barn's dust, on that site.
The published work we can find on pairing is bench-scale. A 2025 study in Processes matched five pre-filter grades against two ultra-high-efficiency main media and reported life extensions of 4.65 and 5.25 times, with service lifetimes measured in minutes rather than years. Useful, and a long way from a V-bank in a barn attic. Its own authors note that pre-filter and main-filter matching lacks systematic study.
What would speak to the barn case is a paired comparison at production conditions: mains run to end of life covered and bare at a documented dust load, resistance and efficiency measured on both.
The open questionWe have not found a paired service-life comparison published at production conditions: mains run to end of life covered and bare, at a documented dust load, with resistance and efficiency measured on both. Bench work on pairing exists, at service lifetimes measured in minutes.
What that means nowHow much life a pre-filter adds on a given site is not a figure a supplier can hand over, so asking for it tends to lead nowhere. The answerable version is the inverse: given what the mains and the pre-filters cost to run, how much life would a pre-filter have to add before the arithmetic favours it. That threshold can be calculated exactly.
Air sampling is the obvious way to ask whether pathogen is present around a site, and the results get read the way a lab result usually gets read: negative means absent. The published record we have found does not support that reading, in either direction.
During the 2015 avian influenza outbreak, a research group sampled downwind of three infected farms and collected 104 samples across three sampler types. None yielded viable virus. Their own modelling of the same outbreak put several of those farms at medium to high probability of airborne exposure. The authors concluded that caution is needed when interpreting negative air samples, because they may not indicate a genuinely safe aerial environment.
The reverse error is just as common. A separate study from the same institution found infectious virus in only 29 percent of the samples that returned viral RNA. Roughly seven positive PCR results in ten did not yield recoverable live virus. A detection is evidence that material arrived, not that what arrived could still infect an animal.
Both errors point the same way for a producer. Sampling answers a narrower question than it appears to, and the number that comes back carries the sampler that produced it: in that same comparison, high flow samplers collected more total material while low flow samplers reported higher and probably more accurate concentrations.Zhao et al., Scientific Reports 2019;9:11755 · Raynor et al., PLOS ONE 2021;16(1):e0244977
The open questionSampler type, flow rate, and the gap between detectable RNA and recoverable infectious virus all move the result. We have not found a protocol that establishes what a negative sample around a livestock site is entitled to conclude.
What that means nowA negative air sample is one input, not an all-clear, and a positive PCR result is evidence that material arrived rather than proof it could still infect. Design decisions hold up better anchored to the building than to a sample.
Whether it is a retrofit, a new build, or a system that has been in for years and is not performing as expected, the conversation starts with what you have.