Longevity test Day 200

HEPA Longevity Test: Day 200

Longevity test Day 200

One of the biggest questions when I started doing the DIY was how long the HEPAs last. Rather than just give people a nice-sounding answer, I wanted to see what the data says, so Gus has been dutifully turning on his Original DIY every day in his 12.3m2 Beijing bedroom and tracking how effective it is each day in real Beijing air

Around Day 100, effectiveness dropped 4%. Around another Day 170, effectiveness dropped another 5-10%. But for the benefit of science, Gus has risked his own lungs by continuing the test for another 60 days.


Gus turns on the Original DIY while he sleeps each night and tracks effectiveness using a Dylos particle counter. I calculate effectiveness as the percentage decrease in the number of particles .5 microns (and above) over the course of the night.

Here’s what a sample day looks like:


To smooth out variability over time, I averaged the effectiveness over every 10 days. For the full details on the methods, check out my earlier post.


The death crawl continues. At Day 200, effectiveness dipped to just below 50%.


At the risk of Gus’s health (especially given the coming onslaught of winter air), I voted for Gus to stop the test. So the test is over!


In real Beijing air, the Smart Air HEPA lasted at 100 days at about 8 hours a day at full effectiveness (729 hours to be exact). It lasted through 140 days (1,028 hours) with a slight 4% drop, which is when I would change the filter.

Recommendation: I recommend changing the HEPA every 140 days at about 8 hours a day or every 1,000 hours. You can adjust the number based on how many hours you use it per day.

Limitations: We did the test in Beijing, so HEPAs probably last a few more weeks in places with cleaner air (but still not “clean air”) like Guangzhou and Hong Kong.

The results may also be different for the Cannon because it pushes more air. The next longevity test will be with a Cannon!

Open Data

As always, I’m publishing the raw data–all 200 days!


How Long do Smart Air HEPAs last? Longevity Test Day 130

How long do Smart Air HEPAs last? My quest to give a solid empirical answer has now completed 130 days! Gus is still dutifully tracking the effectiveness of the HEPA he’s been using in his bedroom since January (with a trip to the US in the middle), calculating the percent reduction in particulates every night.

I reported a few weeks ago that the HEPA was just as effective at Day 1 and Day 90. Now after 130 days and 951 hours of usage, here’s what the HEPA looks like:


Results: At 100 days, the overall effectiveness declined by about 4%. Even so, the HEPA is still removing 80% of .5 micron particles from the overall room air:



  1. On the Original DIY, the Smart Air HEPA lasted about 100 days without losing any effectiveness. That was for a total of 729 hours of usage or an average of about 8 hours a day (7.29 to be precise).
  2. Between days 100-130, the effectiveness dropped by about 4%. I think it’s up to you to decide whether a 4% drop is enough to make you change the HEPA.
  3. You can adjust these numbers to the number of hours you use the machine. For example, if you run the DIY 24 hours a day, you can get about 30 days with full effectiveness and at least 40 days with a roughly 4% reduction.
  4. As I’ve argued before, I prefer this type of test over the context-less numbers that some big purifier companies give because (1) it is based on air in China and (2) it gives the concrete number of hours you can use it rather than just the number of months. Of course, there will be some variation across different seasons, apartments, and cities in China.Limitations:
    1. This number doesn’t necessarily apply to the Cannon. On the one hand, the Cannon filter might have a shorter life because it pushes more air and therefore should collect particles faster. However, our Cannon HEPAs have a pre-filter, which extends the life of the HEPA. The next step is to do longevity tests with the Cannon!
    2. Gus changed apartments on Day 102, which is not ideal. However, the average effectiveness didn’t seem to be affected by his move (see the points at Day 100 and Day 110). We’ll repeat the tests in the future, and that will give us more reliable estimates.

    For now, the test continues! Stay tuned for the next report.

    I’m posting the raw data and detailed methods below for fellow nerds.


    Because there is so much data, I’m making the raw data available here on my personal site. The methods are the same as in my previous tests.

    Basics: Gus did the tests overnight while he slept. The room door was closed during the tests. I calculated effectiveness by averaging the number of .5 micron particles over the last three hours before Gus woke up. Then I divided that number by the baseline measurement before Gus turned on the purifier.

    Effect of Apartment-mate’s Purifier: As I wrote in my last update, we discovered on Day 57 that Gus’s roommate was running a purifier 24 hours a day in his room. This may artificially boost our estimate of the lifespan. It also artificially lowered the baseline particle counts and thereby lowered the estimates of effectiveness for the first 57 days.

    Therefore, the raw data points for days 0-60 would give people a false sense that the HEPA increases in effectiveness after 60 days. To correct for that, I averaged difference between Days 0-57 and Days 58-90. The original graph is in my earlier post, and all of the raw data is available here.

    Outdoor Fluctuation: On Day 94 and Day 113, the outdoor PM 2.5 concentration increased by almost 100 micrograms during the test, so I removed these days from the calculation. Keeping those datapoints lowers those two 10-day estimates by about 3%. This data is retained in the raw data file.

    Room Change: For days 1-101, Gus was in his old apartment, which had a 12.3 m2 bedroom. For days 102-130, he was in his new apartment. Obviously that’s not ideal, but his landlord was taking back his old apartment! We will repeat the tests in his new apartment from scratch after this test ends.

    Calculating room size for Gus’s new bedroom is a little tricky because the ceiling is slanted. The floor area is 13.7 m2–slightly larger than his old room. But the slope of his ceiling gives it a volume of 26.3 m3 compared to 33.5 m3 if it were perfectly rectangular and 30.1 m3 in his old room.

    Thus, his new room may be easier to purify, although I suspect the roof is not very well sealed, so it may leak more than his old room.


passport4adragon 提问: Hi, I’m in Shanghai and recently put together your DIY purifier and bought the same fan and filter you suggested – I think it’s working! A couple of questions – 1. I bought the exact same filter you did on Taobao – how often should it be replaced if I run the purifier every night? 2. The filter came with 3 extra thinner, foamy filter pieces (if you revisit the Taobao page, I think it’s item #1, I don’t know about #3) – should I stick that in between the actual filter and the fan, or not? Thanks!

Great questions!

  1. I don’t have good data on how often the filter should be replaced. I’ve been replacing mine after the filter turns dark, but that’s just based on my intuition, not data. What I need to do is use the particle counter to test effectiveness over time and figure out at which point it starts to lose effectiveness. Hopefully I’ll have more on that later!
  2. You’re right! The filter comes with some “pre-filters.” I don’t use them, but you could use them. On the plus side, they’d probably help your HEPA last longer. On the downside, I’d be worried the pre-filters would cut down on the air flow too much, but I haven’t tested that possibility with the particle counter.