Subscribe

Summary

We think that pathogens which can replicate outside human hosts and infect people through airborne particles, food, water, or contact with surfaces—environment-to-human or ‘E2H’ threats—are one of the most concerning forms of severe biological risk. This post summarizes a hypothesis, which we and others are still testing, that even in a worst-case scenario, humanity may be able to survive such a threat by quickly making low-cost, scalable bioshelters and PPE.

In another blog post, I describe what E2H threats are, why they’re an especially concerning category of extreme biological risk, and why reducing people’s exposure to outdoor-originating aerosols is probably the hardest part of surviving the release of an E2H pathogen. Here, I focus on why we think it may be possible to robustly protect against outdoor aerosol exposure, making humanity resilient to even the most extreme biological threats.

Keeping aerosols out, removing them if they get in

Depending on their location, people breathe in tens of thousands to millions of bacteria every hour spent outside, and being indoors generally does not reduce this exposure much[1]. If a lethal pathogen was replicating in the environment, it could occupy a share of the background microbial ecosystem, making outdoor air a dangerous exposure. To survive, people would need respiratory PPE when outside and somewhere with low particle concentrations where they could remove the PPE to eat, drink, and sleep. We can call this space a bioshelter, and it would offer protection by keeping outdoor aerosols out, and quickly removing them if they get in.

In any building, there are multiple sources of air. Some sources, like air that infiltrates through cracks in the walls or comes in through open doors and windows, has effectively the same concentration or particles as outside air. Other sources of air have much lower particle concentrations, such as air that is recirculated through the building HVAC system and passes through a filter. The ratio of “contaminated” to “clean” airflows sets the exposure of someone in a bioshelter relative to their exposure if they were just outside. If 10% of the airflow in a bioshelter is effectively unfiltered and 90% is fully filtered, then an occupant’s exposure to an E2H pathogen is cut by 90% compared to being outdoors. We think that consistently reducing exposure to airborne E2H pathogens by 99.9% (a 3-log reduction) is a robust target for most scenarios[2]. Our best guess is that airborne pathogen exposure can in practice be roughly approximated as a fixed fraction of PM10: the mass concentration of sub-10-micron-diameter particles in the air.

Given this, we want the ratio of unfiltered airflow to filtered airflow to be <0.001 (whereas a ratio of ~0.1-0.7 is common in most buildings[3]). This can be done by increasing filtered airflow and/or reducing unfiltered airflow.

Assuming obvious things are done like closing doors and windows (more on opening doors below), the main source of unfiltered airflow is infiltration (air leaking in through building cracks[4]). Another source of particles is air brought into a building through a filter (mechanical ventilation), since no filter is perfect and some particles will get through or around a filter. If the airflow from mechanical ventilation is 1 air change per hour (ACH)—meaning a volume of air equal to the volume of the building is brought in every hour—and the filter has an efficiency of 90%, we can say that the effective rate of unfiltered airflow is 0.1 ACH.

The main sources of clean air are air that moves through a filter, either air that’s recirculated or air that’s brought in from outside. A bioshelter could also include air cleaning tech, like UVC, that kills pathogens in the air. The rate at which these technologies kill pathogens can be described in terms of how much clean air you’d need to bring in to achieve the same level of pathogen removal/dilution (more on this below).

If we define

  • M → mechanical ventilation, air brought in through a filter
  • E → efficiency of the filter used for mechanical ventilation
  • I → infiltration, air that leaks into a building
  • C → (equivalent) clean air provided by recirculating air through a filter or air cleaning technologies

Then we can get the ratio of unclean to clean airflow as:

\[\frac{I + M(1-E)}{I + M + C}\]

To get the log reduction (with the target of 3), we can just take the negative log of this ratio.

So we want low infiltration, high clean airflow, and air brought into a building to pass through a high efficiency filter.

How to turn a house into a bioshelter

While there are indoor spaces with low outdoor-origin particle concentrations in defense, manufacturing, and healthcare settings, we want robust bioshelters that are much more widely accessible for society to survive a severe E2H bioattack. People could shelter in retrofitted public buildings, much like people might shelter in schools, office buildings, or community buildings for natural disasters, and we expect such public shelters may be an important part of an effective response. However, public shelters also come with several drawbacks like requiring advanced coordination and the risk that one person could contract an infection and pass it to others[5]. If it was possible for people to turn their homes into bioshelters using only materials around them, this policy could be extremely robust to many scenarios, possible even if broader coordination, manufacturing supply chains, or other components of a food response broke down.

Turning a house into a reliable bioshelter, especially to the specification given above (3 log reduction in exposure to outdoor aerosols) is a tall order. We are not confident this is technically or practically possible. However, preliminary analysis suggests it may be doable, and we’re excited to be actively working to test these ideas more. Mostly because of what is available to us in terms of data and physical spaces to experiment with, most of our work so far, and the analysis summarized below, is focused on the US, but we think it will be important to extend this analysis to other countries that have different building stocks and material availabilities.

Pressurization

An important step to making any bioshelter is keeping infiltration to a minimum. The most practical way to do this is to “pressurize” a building, actively pumping in air (through a filter) so that air only flows outward (and not inward) through cracks. The amount of air one needs to pump into a building is determined by how much of a pressure gradient you aim for, and how leaky the building is.

The pressure gradient target depends on the wind conditions. If the outside air was perfectly still, then any positive pressure gradient between the house and the outside would mean that all air flows out, and there’d be no infiltration. However, this gradient can be reversed if wind blows against the house, generating pressure on the side the wind hits. The table below shows the pressure generated by wind, as a function of wind speed (in both m/s and mph).

Wind speed (m/s) Wind speed (mph) Pressure (Pa)
5 11 15
10 22 61
15 34 138
20 45 245
25 56 383
30 67 551
40 90 980

So a building with a pressure of +75 Pa (when there’s no wind) would generally not experience infiltration as long as the wind speed was below ~11 m/s (~25 mph), but would if the wind got stronger than this[6]. Wind conditions will of course depend on a shelter's geography, how much it is surrounded by structures or plants that block wind, etc. We are unsure what pressure level is needed, but something between 25-75 Pa is probably reasonable, and similar targets are used for military shelters designed to protect against chem/bio attacks.

Whether a space can be pressurized to a given pressure target depends on its leakiness (which sets the airflow requirement) and whether it has a fan that can provide that much airflow (though a high-grade filter). Many US houses have pretty powerful HVAC fans, but they are not rated to provide high airflow through a high-grade filter, and many houses are sufficiently leaky that reaching 25-75 Pa would require a lot more airflow than we expect is possible for normal HVAC (supply) fans to deliver[7]. Importantly, it’s also quite rare for US houses to have any mechanical ventilation (any air actively brought into a house), but it may be possible to create a makeshift intake duct and use the existing system to ventilate a house for pressurization.

The solution in many cases is probably to focus on pressurizing just one part of a house. A single room, especially if it doesn’t have exterior walls, would experience much less wind than the house as a whole, requiring less pressure to keep out infiltration. As a smaller space, a single room would also probably require much less airflow to achieve a given pressure level than a house as a whole. As a simple design, a house’s HVAC system could be set to max fan and all vents inside the house except those in the bioshelter room could be closed, pumping a lot of air into the single room. That air could go through a makeshift filter, which we believe can be made to be extremely efficient (filtering out >99.9% of small particles) while having a low enough resistance to flow to allow a typical house HVAC supply fan to deliver a large volume of air[8]. This also avoids the need to improvise an outdoor intake vent.

Pressurizing a room in a house to avoid outside contamination is actually not a new idea. A 1980s study by the Oak Ridge National Laboratory (ORNL) found that a vacuum cleaner could pressurize a living room (in the study’s case, for about 8 hours), achieving a ~800 fold reduction in small particles meant to represent radioactive fallout from a nuclear meltdown or attack. This reduction is close to the target we think would be robust to most E2H bioattacks (~1000 fold), and while their method was more temporary than we may need to survive a severe bioattack (vacuum motors can burn out if used continuously), it is a good proof of concept.

Internal shelters

Per above, we expect it may be easiest to achieve a high level of protection in individual rooms in a house, rather than robustly protect the whole house. Another advantage of internal shelters is that the air they bring in, and the air that infiltrates in, may already have a lower concentration of outdoor particles. This effectively “stacks” the protection levels that two shelters, one nested in the other, can provide. If the air in a house already has a concentration of outdoor particles that is an order of magnitude lower than outside, and a shelter inside the house draws on this air when pumping in air for pressurization, then if it brings this air through a 99% efficient filter, the air that comes out will have 99.9% fewer particles compared to outside. Likewise, even the air that leaks into this interior shelter (say, if there’s a draft in the house or the shelter door is opened) will only contain 10% as many particles as air that leaked in directly from the outside. This “double bubble” effect argues for layered defense, where some effort is made to reduce particle concentrations in the whole house (say, by running some portable air cleaners), then one room is pressurized into a robust shelter, and perhaps occupants even spend time in a tent or some other smaller shelter within this room.

The feasibility of different nested shelter designs will vary a lot from house to house, but the general idea makes us more optimistic about achieving high protection levels: a 1000-fold reduction doesn’t have to be achieved in a single shot, where a strong wind, opening a door, or some other temporary lapse in protection could massively increase exposure, and all filters don’t have to provide >99.9% filter efficiency (though we think some DIY filters can), but a 99% efficient filter drawing in already-clean air may be sufficient for an internal shelter.

Extra air changes

Even with pressurization and internal shelters keeping incoming particles to a minimum in occupied spaces, it may still be useful to further reduce equilibrium particle concentrations by providing higher clean air flow (or equivalently, removing infectious particles faster). This added equivalent clean airflow is the C in our expression from earlier (produced again below) that shows the ratio of outdoor aerosol exposure inside a shelter vs outdoors. Since the other dominant term in the denominator is mechanical ventilation (M, the airflow pumped into a shelter to pressurize it and keep the air breathable), the impact of added clean airflow C can be assessed by comparing it to M. If C ≈ M, then the clean air roughly halves the particle concentration (by doubling the denominator). If C is ~10x M, then it cuts the concentration by about one order of magnitude.

\[\frac{I + M(1-E)}{I + M + C}\]

There are a lot of ways to add (equivalent) clean airflow to a bioshelter, including chemical- and radiation-based air cleaning technologies that can accelerate infectious particle decay while being safe for human occupants, like UVC or glycols. However, while some of these may be able to scale up quickly in an emergency, we expect that basic portable air cleaners will be the most quickly accessible option. This is both because many US households—about ~30-40%, according to some surveys—already have portable air cleaners (PACs), and also because they are in theory fairly straightforward to make.

If (DIY) PACs are used to clean the air of a whole house, these PACs may be able to reduce particle concentrations by about half. This is valuable, but not an order-of-magnitude difference. In smaller shelters such as internal rooms, where the airflow delivered by PACs is a larger share of the total volume, they may be able to reduce concentrations more, perhaps as much as 1 OOM. PACs may also be useful in establishing effective airlocks (more below).

Making a portable air cleaner mostly requires a fan and filter material, and makeshift PACs using HVAC filters and box fans were popularized during the COVID-19 pandemic (e.g., Corsi-Rosenthal boxes). About half of US households report having a window or floor fan, which often deliver high airflows and can usually be moved around, making them decent candidates for DIY PACs. Typical airflows might be about 500 to 4000 m3/h, though when paired with filters to make a PAC, the clean airflow will be lower, perhaps about 100-1200 m3/h[9]. Normal household PACs similarly provide about 50 to 1000 m3/h clean airflow.

An added clean airflow of about 100-1200 m3/h is about 0.25-3 ACH for an average US house (which is about 400-500 m3). This is probably a similar or slightly smaller amount of air as might be used to pressurize a house to make it a bioshelter, so PACs would at most roughly halve outdoor particle concentrations. A single room in a house, however, might only be about 15 to 50 m3, so PACs could provide many, perhaps dozens, of ACH in these shelters. However, their proportional contribution might still not be that large if more air is used to pressurize a small shelter.

Air locks

Even if a shelter is usually able to effectively exclude or quickly remove outside particles with pressurization and added air changes from PACs, opening doors may cause a large lapse in protection[10]. A good practice to mitigate this is to have an airlock- a space with doors to both the main shelter and the outside that aren’t opened at the same time, so that a shelter is never exposed directly to the outdoors. Airlocks can keep outdoor particles from leaking in both by having an intermediate pressure between the main shelter and the outdoors, so that air tends to flow out from the main shelter into the airlock and from the airlock to the outdoors[11], and by allowing air in the airlock to be “purged” of outdoor contaminants before potential exposure to the shelter. The latter function can be achieved with portable air cleaners (or other air cleaning tech).

If the main shelter is a single room in a house, the airlock could be an adjacent room. If the whole house is used as a shelter, the airlock might be a foyer or other entryway. The volume might be pretty small, meaning that a portable air cleaner could deliver many ACH of clean air, potentially allowing the airlock to quickly get to a low particle concentration, even if it briefly had a lot of outdoor airflow. During this time, someone entering a shelter could keep PPE on, only opening the door to the shelter after airlock particle concentrations had gone down a lot[12].

Beyond air cleaning, airlocks might also serve as a place for people to disinfect clothing or equipment as they enter a shelter. As with household surfaces, this might be done with a disinfectant like HOCl, which conveniently is safe for human skin.

DIY PPE

Respiratory PPE would be another key part of a successful response to a severe E2H pathogen release. PPE would be most important for people who have to work outside, such as construction or maintenance workers, but would also be necessary for people who need to work onside if they can’t sleep at work and have to commute. Even for those who can largely shelter in place, PPE would be important so they could get packages from outside or maintain a high survive temporary lapses in the protection provided by their shelter caused by wind or opening doors.

For the kind of environmental threat we are considering here, the most promising DIY PPE is probably a loose-fitting powered air purifying respirator, or PAPR. A PAPR uses a fan to pull air through a filter and push that filtered air into a hood or facepiece. If enough air is supplied, the hood stays slightly pressurized and air leaks out rather than in. This means the headpiece does not need to form a perfect seal around the face, making these respirators easier to DIY.

The airflow requirements for PAPRs is not large, and could probably be provided by several types of household fans. Humans breathe about 6 L/min at rest, roughly 16 L/min during normal activity, and up to ~140 L/min during very heavy exertion. Standard loose-fitting PAPRs are designed to supply at least ~170 L/min, and DIY PAPRs may aim for a bit of a safety margin by providing something like 200-400 L/min. Importantly, this airflow has to be provided through a high-grade filter, which likely has substantial resistance to flow.

Car cabin HVAC fans look like a particularly promising option. There are roughly 280 million cars in the US, almost all of which have cabin HVAC fans. These fans often move several thousand L/min when there is little resistance, and at least some appear to have enough pressure capacity to push air through a high-grade filter. Vacuum cleaner blowers are another promising candidate. Vacuum cleaners are very common in US households, and their fans are designed to pull air through high-resistance systems. Cordless vacuums may be easier to adapt because they already have batteries, though their runtime may be limited; corded vacuums are less portable, but their blowers are powerful and widely available[13].

The main design constraint is filter resistance. For a given amount of airflow, a small filter forces air to move through the filter quickly, creating a large pressure drop. A larger filter spreads the same airflow over more area, lowering the air velocity and making the filter much easier for the fan to push through. This is why the same fan that fails with a tiny filter may work easily with a larger one.

In my rough calculations, this filter surface area appears to be quite influential on performance. If the filter surface area is only the cross-section of a small hose, even car HVAC or vacuum fans may not deliver enough air for a loose-fitting PAPR. But if the filter is more like a small panel—say a few hundred cm2—then car HVAC and vacuum fans appear capable of delivering well above the ~170 L/min minimum used for loose-fitting PAPRs, and often above the 200-400 L/min range we think is a reasonable DIY target. This suggests that the basic fan-and-filter physics is not obviously limiting, though the exact design would need to be tested.

There are still important practical uncertainties. Very high airflow might improve safety by keeping the hood more strongly pressurized, but it could also create comfort or performance issues. The filter also has to be sealed extremely well, because even a small amount of air bypassing the filter could dominate exposure. And a DIY PAPR needs a workable battery, tube, blower housing, and headpiece. But at a high level, the materials picture is encouraging: many homes or cars contain powerful candidate fans, and filter resistance can likely be managed by using enough filter area.

Conclusion

A severe E2H pathogen may be one of the most concerning sources of biological risk, with the potential to directly cause human extinction. The most important way to reduce biological risk is to prevent mirror bacteria or other possible E2H pathogens from ever being developed. If these efforts fail, however, it may still be possible for humanity to survive. With bioshelters, PPE, disinfectants, and strict hygiene practices, people could limit their exposure to the environmental medium that might contain E2H pathogens (food, water, surfaces, air). If most people avoid infection for long enough, society could keep operating, potentially allowing for the development of medical countermeasures or a way to eliminate the pathogen.

The options for bioshelters and PPE that I laid out here are not exhaustive. While I focused on people turning households into bioshelters with readily available materials because this may be particularly robust to failures in coordination or supply chains, governments or companies may be able to provide public shelters that offer better and more reliable protection. Indeed, several countries, such as Israel, Switzerland, and Finland, already have shelters in public in private buildings designed to provide protection against chemical and biological attacks (though they’re not necessarily designed for long term occupation). Societies could also prepare by stockpiling key inputs to bioshelters or PPE, making it much easier to provide robust protection if an E2H pathogen is released.

We’re excited that some are already exploring ways to defend against environmental biological threats, and hope to see much more work in the future.

  1. People inhale about 0.5-1 m3 of air per hour, and there are around 104–106 bacteria per m3 of outdoor air; in footnote 3, I describe how indoor air has similar bacterial concentrations (often ~30-40% lower) as outdoor air. ↩︎

  2. This approximation follows a similar logic as that discussed in footnote 2; a given bacterial species, including potential E2H pathogens, is unlikely to represent a large share of the total bacteria one inhales, so even though a person’s daily exposure to total bacteria could be quite high, their exposure to an E2H pathogen would be low if they had a consistent 3-log protection level. ↩︎

  3. Outdoor-originating particle concentrations are lower indoors than outdoors because indoors has a mix of unfiltered air from outdoors, such as air let in through open windows or doors or air that infiltrated through building cracks, and filtered air from a building HVAC system. Three US studies—one in Virginia, one in Colorado, and one looking at a national sample—all found average bacteria concentration ratios for indoors vs outdoors of ~0.6, with lower ratios for healthcare facilities in Virginia. ↩︎

  4. While air that infiltrates into a building through cracks is treated here as unfiltered, since it doesn’t pass through a filter, treating this airflow as having the same particle concentration as the outdoors likely slightly overstates the contribution of contaminants. Infiltration can attenuate outdoor aerosol concentrations through incidental “passive filtration” mechanisms, such as particles depositing in cracks and wall cavities, getting captured by porous envelope materials, and gravitational/Brownian losses during transport. Various studies have shown that these mechanisms can effectively filter a large fraction of small particles, perhaps 40-85% of the small particles we likely most care about for an E2H pathogen. This means that treating infiltration as totally unfiltered is a bit conservative. ↩︎

  5. While the most challenging part of combating an E2H pathogen release is that the infectious agent may spread through an environmental medium like outdoor air or contaminated food or surfaces, the pathogen may also be able to transmit person to person. This means that large shared shelters may be difficult to keep safe, since they may have to be designed to both keep infectious particles out and minimize transmission risk. If they didn’t control transmission between occupants, they could lead to correlated failures and raise the bar for the protection threshold required. As a toy example, consider a shelter that provides 3-log protection against outdoor aerosols, such that each occupant has a 1% likelihood of contracting an infection per year, and infections are independent of one another. This may be a tolerable individual risk, but if there were 10 occupants, then the chance that at least one of them gets an infection would be ~10%, and if the shelter was not good at mitigating person to person spread, everyone would be at an elevated risk of infection. ↩︎

  6. This is simplified in a few ways. Wind pressure isn't uniform across a building, and the table values approximate the maximum (stagnation) pressure on the windward face, but leeward and side walls actually experience suction, so portions of the building exterior can see net inflow even at wind speeds below the threshold the table implies. Interior pressure also varies. For example, it varies with height (warm air rises and accumulates near the ceiling, the so-called "stack effect"), with proximity to the supply fan, and with the distribution of leaks. A "+75 Pa" pressurization target is best read as an average; individual cracks may sit at smaller pressure differences, or experience brief reversals during gusts. The threshold above is a rough guide to the wind regime in which a given pressurization budget is broadly adequate, not a sharp cutoff. ↩︎

  7. The supply airflow needed to hold a building at pressure ΔP scales as Q ≈ Cd · ELA · √(2ΔP/ρ), where ELA is the effective leakage area, Cd is the discharge coefficient (usually around 0.6), ΔP is the pressure differential, and ρ is the density of air. For a 425 m3 US house at typical tightness (ELA ≈ 0.1 m2, roughly ACH50 ≈ 5), that's ~1,400 m3/h to maintain +25 Pa and ~2,400 m3/h for +75 Pa; a leakier home around ACH50 ≈ 10 needs roughly double. A weak residential furnace blower (free airflow ~1,550 m3/h, stall pressure ~370 Pa) pushed through a clean HEPA + MERV-8 stack (~90 Pa pressure drop) delivers only ~1,400 m3/h, falling to ~1,160 m3/h as the filters dust-load to ~3× their clean pressure drop. So a normal-tightness house can just barely hit +25 Pa with a weak furnace fan, and +75 Pa is generally out of reach without first reducing ELA — halving ELA roughly halves the airflow needed at any pressure target. ↩︎

  8. I go through more detailed calculations in these research notes, arguing that common household materials, such as the fiberglass used in insulation for most US houses, may be able to function as high grade filters, similar to high efficiency particulate air (HEPA) filters. To evaluate this, I used a standard model (based on Darcy’s law for fluid flow through a porous medium) that treats a mechanical filter as a random mat of fibers, where particles are captured by diffusion, interception, and inertial impaction. Given assumptions about fiber diameter, particle diameter, packing density, and airflow velocity, this lets us estimate both how thick the filter needs to be to reach a target efficiency and how much pressure drop the filter creates. Using insulation-grade fiberglass as a reference material, the calculations suggest that a layer on the order of 1-2 cm thick could plausibly reach HEPA-like efficiency, meaning >99.97% removal of ~0.3 µm particles, while somewhat thicker layers could reach even higher efficiency for PAPR-style applications. To be useful, filters don’t just need to be very efficient at capturing particles but should also have a low enough resistance to flow that fans can still pull a meaningful amount of air through them, and I walk through some pressure drop calculations in the research notes as well. Filters become much easier for household fans to push air through when their surface area is large, because the same total airflow is spread over a lower face velocity. For a house-scale pressurization filter, this is why a large filter area—say, comparable to a large HVAC filter—is important. In the research notes’ illustrative calculation, a HEPA-like filter plus prefilter with ~0.3 m2 of area, moving ~850 m3/h of air, creates a pressure drop that appears compatible with many residential HVAC supply fans, especially before heavy dust loading. These numbers should not be treated as validated design instructions. Real filters need to be tested for efficiency, pressure drop, durability, fiber shedding, sealing, and performance as they load with dust. But the calculation makes the basic tradeoff clear: high efficiency is possible in principle, and low pressure drop is mainly achieved by using enough filter area. More recent experimental work also suggests that fiberglass insulation may not be the only candidate material; some common textiles may also be able to form high-efficiency DIY filters, though this is still preliminary and not yet something we would rely on without further testing. ↩︎

  9. Window fans have a typical airflow of about 850 m3/h, and floor fans might have an airflow of anything from a typical desk fan (~500 m3/h) to a box fan (usually ~3900 m3/h). However, this is just the flow when there’s no resistance imposed by a filter. Looking at some tests of CR boxes using box fans, researchers have recorded pressure drops (through MERV-13 filters) of about 5-10 Pa, which reduced clean air delivery rates by about 60-80% relative to their free airflows. ↩︎

  10. One rough way to size this lapse is to treat a door opening as a short pulse of “equivalent infiltration.” Unfortunately, I haven’t found a lot of experimental work to bench mark estimates on, there is a study of a negative-pressure hospital isolation room that can serve as a loose analogy: air escaping from a -20 Pa isolation room into the surrounding space is analogous to air entering a +20 Pa shelter from the surrounding space when the shelter door is opened. In that study, door opening produced about 1.5 m3 of counter-pressure-gradient exchange, and the reported relationship between imposed exhaust flow and escaping air can be extrapolated to the positive-pressure case. Applying that relationship to an indoor shelter pressurized with ~60 m3/h of filtered supply air gives an estimate of roughly ~1.5-1.6 m3 of air entering the shelter per door opening. This should be treated as an OOM estimate, not a validated model: the geometry, pressure field, thermal gradients, door motion, human movement, and surrounding airflow could all differ substantially from the hospital-room experiment. But it gives a useful scale. For a ~15 m3 indoor shelter, one opening might introduce air equal to ~10% of the shelter volume; four openings per day would be ~6 m3/day, or an average “equivalent infiltration” of ~0.02 ACH. That could matter a lot if this air were fully outdoor-contaminated and not quickly removed, but the effect is mitigated if the air outside the shelter is already partly cleaned by the larger house, and especially if the shelter has high recirculating clean-air delivery and occupants keep PPE on for a short purge period after the door opens. The research notes similarly flag door openings as a weak link and use the same negative-pressure-room study as a rough reference point, while emphasizing that the result may not generalize well.) ↩︎

  11. This can be accomplished by arranging a, where pressure falls as air gets less clean: clean shelter > airlock > outdoors. If the main shelter is pressurized with filtered air, some of that clean air will leak into adjacent spaces through gaps around doors, walls, ducts, etc. If the airlock also leaks to the outdoors, its pressure will settle somewhere between the shelter and the outside, so the default direction of airflow is from the shelter into the airlock and from the airlock outdoors. This can happen “passively,” without a separate airlock fan, if leakage from the shelter into the airlock is large enough relative to leakage from the airlock outdoors. The research notes suggest this may be common in houses because interior partitions are often much leakier than exterior walls; for example, this study found 10x higher leakage per unit area between residential interior zones than through the exterior envelope. A separately powered airlock can also work, but the key design constraint is that it should not be pressurized above the cleaner interior space or else, when the inner door opens, the airlock could push less-clean air into the shelter. The exact pressures will depend on the geometry and leakage paths, and wind or door opening can temporarily disrupt the cascade, but the core idea is simple: each boundary should leak in the safer direction. ↩︎

  12. A simple purge model treats the airlock as a well-mixed volume and assumes that clean air delivery removes contaminated air exponentially. If the airlock has an effective air change rate A in eACH and we want an L-log purge, the time required is approximately t = L/(A×log10(e)). So 10 eACH gives a 3-log purge in ~40 minutes, while 40 eACH gives a 3-log purge in ~10 minutes. In a small airlock, these eACH levels may be achievable with ordinary fan-filter setups: a PAC delivering 500 m3/h of clean air provides 50 eACH in a 10 m3 entryway, but only ~7 eACH in a 70 m3 garage. This calculation is only an approximation. It assumes the airlock is well-mixed and that no additional contaminated air is entering during the purge. In reality, some outdoor air may still leak in, especially if the airlock is windy, poorly sealed, or not positively pressured. More exactly, the concentration would decay toward a nonzero steady state set by the ratio of incoming contaminated airflow to clean-air delivery. But if the PAC’s clean airflow is much larger than ongoing infiltration, the simple purge calculation should be a decent approximation; the main error may be that the airlock is not actually fully contaminated after a door opening, since only some fraction of its volume may have been replaced by outdoor air. ↩︎

  13. Car cabin HVAC fans look promising because there are ~280 million cars in the US, almost all have HVAC systems, and these fans generally deliver several thousand L/min with little resistance; in one representative spec sheet I looked at, the fan’s stall pressure was >1 kPa, which is roughly in the range needed to push air through a clean high-grade DIY filter. Vacuum cleaners are also promising: a YouGov survey finds that 88% of US households own a vacuum cleaner, and vacuum blowers are designed specifically to move air through high-resistance systems. Cordless vacuums may be easier to adapt because they already include batteries, but corded vacuums may be more robust for continuous operation; cordless vacuums appear to account for ~40% of the US vacuum market by value, but because they are more expensive, perhaps closer to ~25% of units. In the research notes, I estimated airflow through clean 5-log fiberglass filters by intersecting fan curves with filter pressure-drop curves. The resulting illustrative table gives car-HVAC-fan airflow of ~12, 59, 1,056, and 3,742 L/min through filter areas of 2, 10, 200, and 1,000 cm2, respectively; the analogous vacuum-fan estimates are ~15, 71, 766, and 1,098 L/min. These numbers should not be taken as validated design outputs—the fan curves are based on representative examples, real filters load with dust, tubing adds additional pressure losses, and high airflow may create comfort or performance issues—but they support the key design intuition: car HVAC and vacuum fans are plausibly strong enough for DIY PAPRs if paired with sufficiently large filter area, while tiny filters or narrow tubes can make even strong fans underperform. ↩︎

It May Be Possible to Improvise A High Grade Bioshelter