Should air purifiers have the HEPA filter on the front or the back–is it better to push or pull air? The Smart Air lab in Beijing ran tests of the same fan with the HEPA filter on either side, and the results contradict most people’s intuition.

When I made my first DIY purifier, I put the filter on the front:

Within a few weeks, several people told me I was an idiot. Why’d you put the filter on the front? It should go on the back!

Many people have a strong intuition the filter should go in the back

Here’s a more-respectful-than-usual suggestion on my live test video:

I’m fascinated by how many people have this intuition. For example, this commenter on the website Quora said it’s like trying to push a rope.

But when I push further, most people struggle to come up with a good reason. For some people, I think it’s purely an intuition about having air come from a pure source. But here are the two most coherent reasons I’ve heard:

Claim 1: The HEPA Filter on the Front “Protects” The Fan.

This is probably the best reason I’ve heard. Yes, putting the filter behind the fan will prevent dust from hitting the motor and collecting on it. I think there’s a small, non-zero benefit of that, although I haven’t seen anyone actually test it.

But here’s a gut check: people run “unprotected” fans (we just call them “fans”) all the time, and it doesn’t seem to cause much harm.

Claim 2: The HEPA Filter on Front Improves Performance.

This is an odd idea to me. I’ve asked several physics PhDs, and they’ve said it shouldn’t make a difference. The amount of air coming in is equal to the amount of air coming out.

But who knows? Maybe there’s something about the air flow process that I don’t understand. My pet peeve is when people point to a theory and assume that proves the matter, without actually testing it. So I went out and tested it!

Air Purifier Filter Front vs. Back Test

It’s hard to test with that first fan I used (above) because putting the filter on the back would require me to tape around the back screen. That would alter the fan, so we can’t compare apples to apples.

Fortunately, I tested lots of different fans.

The results? In both tests, putting the HEPA on the intake side actually did slightly worse! Here are the more-controlled CADR test results.

I was surprised. I thought they’d perform the same.

Now this next part is hard for me to understand, but the Smart Air team is fortunate to have an aeronautical engineer. Here’s an important detail about the Cannon fan I didn’t mention:

The fan isn’t in the middle. Instead, it’s about 2/3rds of the way down the tube. According to our engineer, it helps to have more distance between the fan and the filter so that a column of pressure can build up.

Again, I still haven’t grasped the physics, so I invite anyone with more expertise to add their explanation here. However, two independent datasets show that putting the HEPA on the front leads to higher reductions in particulate.

If the fan were in the exact middle, we hypothesize that performance would be equal with the HEPA on the front or the back.

Input-Output Testing

I think this is a fascinating demonstration of how I bumbled into a good design. I didn’t know much about airflow dynamics, but I did lots of input-output tests before I chose this setup. In other words, I was agnostic about what was going on inside the fan, but I tested different fans and measured the output I cared about (reduction in particulate), and that led me to a good design—even though I couldn’t explain why at the time.

Speculation: Does turbulent airflow explain why “filter first” is worst?

Now we’re not 100% sure on this but based on research on guide vanes, putting a filter before the fan will make the air going into the fan turbulent.

Making the fan operate in turbulent airflow (as opposed to laminar flow) has two consequences:

A. The efficiency of the fan is reduced.

B. The fan will be noisier.

However, we don’t know whether this effect would be large enough to be meaningful.

Putting the filter after the fan means turbulent air (from the fan) is going into the filter. We’re unsure what effect that has on the efficiency of the filter.