The filter media inside a wearable air system is a patch you could cover with two fingers, and every litre of air you breathe on the ride has to pass through it. When researchers pushed air through fifteen mask and filter materials at two different speeds this year, the efficiency moved — non-woven media lost 1 to 17% of its filtration efficiency for no reason other than the air going faster (Quecke et al., Infection, Disease & Health, 2026). That is the part no product tells you. A helmet air filter's "95%" is not a property of the filter. It is a reading taken at one particular air speed, and nobody prints the speed.
Key takeaways
- A filter's efficiency rating is measured at one stated face velocity, and in a 2026 peer-reviewed test non-woven filter media lost 1 to 17% of its filtration efficiency when that velocity was increased (Quecke et al., Infection, Disease & Health, 2026).
- Raising the air speed from 10 to 25 cm/s cost improvised furnace-filter media 11.2 points of efficiency at 0.15 micrometres, while a certified N95 lost only 1.1 points — the grade tells you the best case, the media tells you how fast you lose it (Quecke et al., 2026).
- The same researchers concluded that the ASTM face-velocity test range is too wide to certify against, because filtration efficiency can vary considerably within it (Quecke et al., 2026).
- A passive helmet vent's face velocity is set by road speed rather than by design, so it has no fixed operating point and therefore no single honest efficiency figure.
What Does a Helmet Air Filter's "95%" Actually Measure?
It measures the fraction of particles the media captured at one stated face velocity — the speed of the air moving through the media, given in centimetres per second. Change that speed and the same material returns a different number.
Face velocity is the variable that links a lab bench to a moving motorcycle, and it is the one specification consumer filtration products never publish. Filter-media datasheets quote efficiency and pressure drop at a reference face velocity of 5.3 cm/s — a slow, standardised trickle chosen so that different media can be compared like for like. It is a benchmark, not a description of service conditions. Any real wearable filter, moving enough air for a person to breathe through a patch of media the size of a beer mat, runs at several times that reference speed.
So the honest reading of any filtration claim is conditional. "95% at 0.3 micrometres" means 95% at 0.3 micrometres, at the velocity the test rig was set to. Nobody is lying. The number is simply incomplete — and until this year, there was no published measurement of how incomplete.
Why Does the Same Filter Score Differently at Different Airflows?
Because filtration is a race between the particle and the media. Very small particles do not travel in straight lines — they wander by Brownian diffusion, and that wandering is what drives them into a fibre. Diffusion needs time. Speed the air up and each particle spends less time inside the media, so fewer of them get caught. Larger particles, captured by impaction and interception, are less affected. The loss therefore concentrates around the hardest particle sizes to catch — which are exactly the sizes that dominate haze and traffic smoke.
Quecke et al. (2026) put numbers on it. They took 15 mask and filter materials, built a rig based on ASTM F2299/F2299M, and tested every one of them at two face velocities — 10 and 25 cm/s — with four of the masks carried out to two further speeds, 17.5 and 32.5 cm/s. Scaled onto a real N95's 175 cm² of media, that main pair corresponds to roughly 105 and 263 litres per minute of through-flow, and the extended set reaches 341 litres per minute — the span from resting breathing to hard exertion.
Going from 10 to 25 cm/s at 0.15 micrometres:
| Material | At 10 cm/s | At 25 cm/s | Change |
|---|---|---|---|
| Certified N95 respirator | 99.0% | 97.9% | −1.1 pts |
| Level-3 surgical mask | 99.1% | 95.4% | −3.7 pts |
| Furnace filter media used as a mask insert | 86.9% | 75.7% | −11.2 pts |
Averaged across all the materials tested, efficiency fell 4.4 points at 0.15 micrometres and 1.8 points at 0.3 micrometres. The headline range across the non-woven media was a loss of 1 to 17%. Woven and knitted materials, and two of the commercial cotton masks, did not lose efficiency as the velocity rose — the penalty is specific to the non-woven media that every real filter is made from.
Read that table by the spread, not by the rows. The certified respirator barely moved. The improvised media collapsed. A filter's grade tells you its best case; its media tells you how fast it falls off that best case. And the researchers' own conclusion goes further than any product does: the ASTM face-velocity range, they argue, is too large and should be revised, because filtration efficiency can vary considerably inside it. A certificate is being issued against a band within which results genuinely differ.
The same three materials, two air speeds
Filtration efficiency at 0.15 micrometres. Quecke et al. (2026), rig based on ASTM F2299/F2299M.
At the slow test speed, all three materials look broadly alike. Switch to 25 cm/s to see which one holds.
| Material | At 10 cm/s | At 25 cm/s |
|---|---|---|
| Certified N95 | 99.0% | 97.9% |
| Level-3 surgical | 99.1% | 95.4% |
| Furnace filter media insert | 86.9% | 75.7% |
Why a Passive Helmet Vent Has No Rated Efficiency
Here is where this stops being an academic point and becomes a helmet problem.
Face velocity is volumetric flow divided by the open area of the media. Bolt a filter across a passive helmet vent and you have fixed the open area but handed control of the flow to the road: the air is pushed through by forward motion, so face velocity is road speed divided by open area. Sitting at a red light in still air, the flow through the media approaches zero and the rider is breathing whatever leaks in around the edges. At 60 to 80 km/h the same mesh is being driven far above any velocity in the published test band. Helmet-ventilation researchers run their thermal-manikin work at 39 to 59 km/h precisely because flow through a helmet is a function of speed, not a constant.
The consequence is sharper than "passive filtration is weak". A passive vent's filtration rating is not merely low — it is undefined. There is no single number that describes it, because there is no operating point to measure it at. It delivers a different efficiency at every second of the ride.
Illustrative schematic, not measured data. Because a passive vent's face velocity is road speed divided by a fixed open area, its filtration efficiency has no single operating point to be rated at — and measured efficiency moves with face velocity (Quecke et al., 2026).
Powered systems are not automatically exempt from this — they are just answerable. Motorcycle air-purification systems described in the literature circulate roughly 15 to 40 litres per minute, a 2.7× spread. Two products both printing "H11, 95%" can sit at opposite ends of that range and deliver measurably different air to the same rider.
There is a second axis, and it is a helmet problem specifically. The same study also exposed masks to simulated exhaled breath condensate for 1 to 24 hours and re-tested them immediately afterwards. For the cloth masks, filtration efficiency fell by up to 20% after that exposure, and the authors' recommendation is blunt: a filter should be removed after a maximum of 8 hours of wear to stop breath condensate degrading it. A helmet is the most humid enclosure a filter media will ever work in — the rider's own breath, at close range, for the whole commute, every day. That finding is about cloth masks rather than helmet units, so treat it as a mechanism to respect rather than a number to transfer. It is one more reason a single day-one figure is a poor description of what you are breathing in month three.
Will the Gap Get Better or Worse by 2031?
The projection below is synthesised — our arithmetic on published trends, not a measured forecast. The claim is that the gap between a helmet filter's printed rating and the air a rider actually breathes gets wider by 2031, because the two inputs move in opposite directions.
Input one: the air gets dirtier, so filters load faster. In the IQAir 2025 World Air Quality Report, only 14% of world cities met the WHO annual PM2.5 guideline, down from 17% in 2024 — a reversal driven by wildfire smoke reaching regions that used to be clean. That is a three-point drop in a single year. Hold that direction even at a decelerating rate and compliance is in single digits by 2031. None of this is abstract in Southeast Asia: Malaysia declared an air-quality emergency in Serian at API 521 on 4 September 2026, Metro Manila was rated "acutely unhealthy" from the same Kalimantan fires between 30 August and 2 September, and Indonesia's Health Ministry has recorded around 15.6 million acute respiratory infection cases nationwide since January 2026, with more than 5 million people on Borneo and Sumatra directly exposed to the smoke.
Input two: a loading filter raises its own face velocity. As dust blocks the media, open area falls while the system still tries to move air. Velocity through the remaining clean area therefore rises — and Quecke et al. have now priced what that costs: up to 17 points of efficiency, worst on the cheapest media. A filter that is honestly 95% on day one is delivering measurably less by the end of its service interval, and that interval shortens every haze season. Which is the unglamorous case for replacement discipline over grade-chasing: a fresh filter at the right interval beats a loaded higher-grade one, because a loaded filter is no longer operating where it was rated.
The counter-case deserves stating. ASEAN's transboundary haze roadmap targets a haze-free ASEAN by 2030, and if peatland restoration hits its targets, the first input reverses. Indonesia's 1.2 million-hectare restoration goal is currently behind schedule, and regional haze now looks closer to a near-regular feature of strong El Niño and positive IOD years than to an occasional shock. So the projection is arguable rather than certain — but the evidence points one way.
The prediction worth owning: by 2031, "what is it rated at?" will be the wrong consumer question, and "what does it deliver at its operating flow, at end of life?" will be the one the category is forced to answer.
What Should a Filtration System Be Designed To?
Not to a grade — to a duty point. The fix for a rating that moves with airflow is not better media. It is a known and constant airflow, so the media is held near one operating point instead of swinging from nothing at a red light to a gale at 80 km/h.
That is what an active induction system buys you. Easi Breezi clips onto the helmet a rider already owns and runs its own blower, drawing power from the bike, so the air speed through the filter is set by the design rather than by the traffic. It supplies filtered air continuously — stationary or moving — through H11 HEPA media rated at 95% or better on PM2.5, in an IP67-sealed unit (patent pending). Pre-orders are $199, and the first 500 include 10 filters.
For the grade half of the specification, our companion piece on H11 versus H13 media covers what the letters actually buy you; if your question is the buying-side one of what an air filter helmet setup looks like in practice, start there instead. Grade and flow are two halves of one specification, and the industry has only ever published one of them.
Frequently Asked Questions
Do motorcycle helmets have air filters?
Standard motorcycle helmets do not. Their vents are open ducts designed to move heat away from your head, not to clean the air, and they are sized for cooling airflow rather than filtration. Filtration only arrives as an add-on — either a passive filter laid over a vent, or an active unit with its own blower.
Do motorcycle helmets have ventilation?
Almost all full-face helmets do: intake vents at the chin and brow, exhaust vents at the rear. That ventilation is driven entirely by forward motion, which is why a helmet that feels cool at 70 km/h feels stifling in stopped traffic — and why those same vents cannot hold a filter at any steady air speed.
What face velocity is a helmet air filter rated at?
Almost no product tells you, which is the whole problem. Filter media datasheets typically publish at a reference face velocity of 5.3 cm/s, but a wearable system moves far more air than that through a small patch of media. If a product quotes an efficiency without an airflow, the number is incomplete rather than wrong.
Does a filter get worse as it fills with dust?
Yes, in two ways at once. It restricts more airflow, and — because open area is falling while the system still tries to move air — the velocity through the clean media that remains rises, which is exactly the condition that costs efficiency. Replacing on schedule matters more than buying a higher grade and leaving it in.
Is a higher HEPA grade the answer?
Only partly. A higher grade raises the best-case number, but the 2026 data shows that a material's behaviour under changing airflow varies independently of its grade — a certified respirator lost 1.1 points where improvised media lost 11.2. Ask what airflow the system is designed to hold, then ask about the grade.
Ready to Breathe Cleaner on Every Ride?
Ask any filtration product one question before you buy it: at what airflow was that number measured? If the answer is a vent, there isn't one — because a vent's air speed is set by the road rather than by the design, so its rating has nowhere to stand. A system that controls its own airflow can answer the question.
See how the Easi Breezi unit delivers filtered air at a designed flow, stationary or at speed.
Written by Ash — mechanical engineer and founder of Easi Breezi, building an active HEPA filtration system for motorcycle helmets (patent pending). Based between Hong Kong and Bali, riding daily in the traffic this blog writes about.