AHU Filter Classes After ISO 16890 — ePM1, ePM2.5 and ePM10 Explained

Specify "F7" on an AHU quote today and a consultant will send it back. EN 779 was withdrawn years ago, yet the habit of ordering filters by a letter-and-number code from a standard that no longer exists is still the single most common rejection reason on technical data sheets. ISO 16890 replaced it with a system that rates filters against actual particle sizes — and it changes what "good enough" means for every application on your desk.

What changed, in one sentence

EN 779 rated filters by a single efficiency number against a synthetic dust loading. ISO 16890 rates filters by fractional efficiency across three particle size bins — PM1, PM2.5, PM10 — measured against real atmospheric particulate, then reports the class as the worst-performing bin the filter still qualifies for. A filter is not "ePM1 60%" because it looks good on average; it is ePM1 60% because it clears the ePM1 threshold specifically, which is the hardest bin to pass.

The four classes and what they mean physically

ISO 16890 class	Efficiency threshold	What it targets	Old EN 779 rough equivalent
Coarse	Arrestance-based, no PM rating	Lint, fibres, insects	G3–G4
ePM10 ≥ 50%	≥50% of PM10 particles	Dust, pollen	M5–M6
ePM2.5 ≥ 50%	≥50% of PM2.5 particles	Combustion particles, mould spores	F7 (rough)
ePM1 ≥ 50% (to ≥80%)	≥50–80% of PM1 particles	Fine combustion soot, viral aerosol carriers	F7–F9

The "rough equivalent" column is deliberately approximate — ISO 16890 and EN 779 test different particle populations, so a filter that was F7 under the old standard can land anywhere from ePM2.5 65% to ePM1 55% depending on its actual particle-size efficiency curve. This is precisely why cross-referencing old spec sheets by letter code is unreliable: two filters both formerly labelled "F7" can have meaningfully different ePM1 performance.

Mapping classes to real applications

• Coarse / ePM10 — pre-filtration ahead of a finer stage, or standalone for warehouses and light industrial spaces with no fresh-air quality requirement beyond keeping the coil clean.
• ePM10 ≥ 60–70% — general offices, retail, schools. Meets EN 16798-3 category ODA 2 outdoor air handling in most European urban sites.
• ePM2.5 ≥ 50–70% — offices in higher-pollution urban areas, better commercial buildings, the default second stage ahead of heat recovery in most European new-build AHUs.
• ePM1 ≥ 50–80% — hospitals (non-critical areas), laboratories, and any application where the outdoor air quality is ODA 3 (high pollution) under EN 16798-3. Increasingly the default minimum requested by consultants post-2020 for occupied commercial space, independent of a hard regulatory requirement.
• ePM1 ≥ 80%, or HEPA downstream — operating theatres, cleanrooms, pharmaceutical manufacturing — outside what a standard ePM filter stage covers; treat as a specialist terminal filtration problem, not a casing filter bank selection.

The trade-off you cannot skip: filter class vs pressure drop vs SFP

A finer ePM1 filter has a denser media pack, and denser media means higher initial pressure drop for the same face area — which feeds straight into specific fan power. Upgrading a filter bank from ePM10 to ePM1 without recomputing SFP is a common way to blow an ErP compliance margin that looked comfortable on the coarser filter. The fix is not to avoid fine filtration — building owners increasingly demand it — but to size the filter bank's face area (and therefore face velocity) large enough that the finer media doesn't push absorbed fan power past the SFP_int threshold in EN 16798.

Loading behaviour matters too: an ePM1 filter run at 2.4 m/s face velocity will load — and its dP will climb — noticeably faster than the same class at 1.8 m/s, shortening the practical replacement interval even though both configurations meet the nominal ISO 16890 rating on day one.

What a compliant data sheet needs

Since 2018, a technical data sheet that only states a filter class without the tested efficiency percentage and the initial/final pressure drop at design airflow is incomplete by most consultants' checklists. Expect to state: ISO 16890 class, initial and recommended final dP, and — for ePM1 stages in critical applications — the discharge (electrostatic) status of the media, since some low-cost electrostatically enhanced media lose efficiency as they load in ways mechanical media does not.

Where the tooling matters

In AirSelect3D, filter selection pulls real manufacturer ePM1/ePM2.5/ePM10 performance curves and initial/final dP at the actual face velocity your casing produces — not a nominal catalogue number — and feeds that pressure drop straight into the SFP and ErP compliance check live as you swap filter class. Cross a threshold going from ePM10 to ePM1 and the SFP verdict updates before you leave the filter selection screen, not after the dossier is printed.

See filter class changes recalculate SFP and ErP compliance live →