The Case for Case Fans
With a ton of time in Covid isolation, I was researching air filtration. I was surprised to learn that we'd been using our air filters incorrectly and they weren't providing much benefit.
We are Covid cautious, and have two of the Coway Airmega AP-1512 on our 1000 sq ft first floor in an attempt to reach the CDC guideline of 5 air changes per hour (ACH). But since it's so loud to run on full speed, 67 decibels (dB) (louder than a normal conversation), we keep it at the lowest speed --- especially when guests are over! Ironically that's when it's most important to be running it on high. We were getting less than 2 ACH total from running them on medium, or less than one ACH from running them on low!
This led me down a rabbit hole of research on how much filtration we need, and how we can achieve it in a cost-effective and quiet way.
What matters for filtration?
Filtration is measured in the Clean Air Delivery Rate (CADR), which is the airflow (Cubic Feet per Minute or CFM) x the effectiveness of the filter.
However, filtration is not the only way to increase ACH. We can also bring in fresh air by opening windows and creating a cross-breeze, which we can validate is working with a CO2 sensor. We can add up the ACH created by each mitigation to calculate the total ACH. And each ACH doubling halves the risk.
How much filtration do we need?
We're not sure what our house's baseline ACH is, or the exact number created when we keep windows open to maintain CO2 below 1000.
We can start with a goal of achieving hospital levels of filtration, as those are the highest recommendations for the most risky situations. We can calculate this with two different measures: CFM per-person (of clean air, so same as CADR), and air changes per hour (ACH), but both lead to around the same result. A hospital waiting room should have 90 CADR/person. For 4 people at dinner, that's 360 CADR. Likewise, a hospital should have 14 ACH, which in our 180 ft² dining room, is 336 CADR.
This is still not enough to eliminate the risk if we're having dinner with an infected person. According to the Safe Air Spaces estimator, if five people are having a three hour dinner and one of those people has Covid, 5 ACH from ventilation and 14 ACH from filtration would lower the risk to 40% chance per-person, meaning 2 people would likely become infected.
If we tried for 1500 CADR, we'd still have a 15% chance per-person!
Another approach may be to put a filter in the middle of, or above, the table to try to immediately scrub exhaled air. This is worth validating and exploring further.
The answer seems to be to have as much filtration as we can get and targeting a minimum of hospital levels of ventilation -- around 350 CADR.
What's the quietest and most cost-effective filter?
HEPA vs MERV 13
Evaluating the effectiveness of filters is somewhat counterintuitive. At first glance, you would assume that a HEPA filter that captures 99.97% would appear to be far more effective than a MERV 13 filter that only captures 85%.
But since the air is being mixed in the room, what we actually care about is the Clean Air Delivery Rate (CADR), which is the airflow x the filtration effectiveness of the filter. A MERV 13 filter allows so much more air through, and requires so much less pressure, that we can get equivalent CADR at much lower noise and power consumption. Check out this excellent comparison, but basically the 3M Filtrete MERV 13 is ideal.
(An exception would be if you're directly breathing the clean exhaust, then you do want the highest filtration possible.)
Why not use the Coway HEPA filter?
We'd need two Coways, one at full blast (250 CADR, 67 decibels (dB)), and one at medium (110 CADR, 49dB) to reach 350 CADR -- too noisy for a dining room. And we can do better by building our own. Here are more stats on the Coway.
Corsi-Rosenthal with a box fan and 4 MERV 13
A Corsi-Rosenthall box is a cheap and effective way to get a high CADR, but they rely on box fans, which are too loud.
An Air King box fan at full power, with 4 MERV 13 filters, would give 425 CADR @ 68 dB - the same sound level as the Coway option. It's hard to estimate how much they would deliver at lower speeds, but perhaps two fans at lower speeds would be enough, but even on low they are 61dB.
PC case fans and MERV 13 filters
Arctic P14 case fans are $33 for 5, and max out at 44dB even in configurations with 10 fans and over 500 CADR. 45 dB is about the background noise level of a library. This is the clear winner on noise.
The "tower of power" configuration (2x 20x30 filters, 10 PC fans) gives the best price/performance ratio, while the Corsi-Rosenthal configuration with 9 fans comes in second.
Full spreadsheet with stats and citations
The tower of power has the advantage of taking up minimal floor space, and also can be more easily positioned away from walls for better airflow.
Conclusion
We'll be building several Towers of Power to have around the house, and move as much filtration as we can into the room when guests are over. There's no way to truly eliminate risk, but this plan would move the likelyhood of someone getting Covid from 45% with one Coway on blast today down to 21% with two Towers of Power.
If you end up going this route, I'd love to hear about it.
Special thanks to Rob Wissman for all his incredible work and research.
Construction and Parts List
Building a PC Fan Corsi-Rosenthal Box
Sleek and Quiet: Building a Better Box
Quiet DIY Air Purifier Build Tutorial
More Resources