HVAC and Commercial Kitchen Filtration Technologies

We often think of filters as simple sieves that catch dirt the way a net catches fish. In reality, filtration in HVAC systems and commercial kitchens works through fluid dynamics. Particles get trapped based on their mass, their speed through the air, and the geometry of the filter fibres. Understanding these mechanisms helps us design systems that protect equipment, improve indoor air quality, and reduce fire risk in kitchens.

1. Mechanisms of Particle Capture

The model below shows one fibre with air flowing past it. Change the particle size and watch which mechanism takes over.

1.1 Straining (Sieving)

Think of straining as pure physical blockage. When a particle is larger than the gaps between the filter fibres, it simply cannot pass through.

We use this method mainly for gross particulate larger than 10 µm.

You will find straining at work in basic pre-filters rated MERV 4 to 8. Their job is not to clean the air for people but to protect the more expensive coils and high-efficiency filters further downstream from getting clogged too quickly.

1.2 Inertial Impaction

Air streams bend and curve around filter fibres. Heavy particles, however, have enough mass and speed that they cannot change direction quickly. They keep flying in a straight line and crash into the fibre.

This works best on particles between 1.0 µm and 10 µm.

Effectiveness rises as air velocity increases. We describe this behavior with the Stokes number, written as $Stk$.

$Stk = \frac{\rho_p d_p^2 C_c U}{18 \mu d_f}$

Here $\rho_p$ is the particle density and $d_p$ the particle diameter. $C_c$ is the Cunningham slip correction, $U$ the air velocity, $\mu$ the viscosity of air, and $d_f$ the fibre diameter. A small $Stk$ means the particle follows the air. A value near 1 or above means it cannot.

Take a coarse medium with 20 µm fibres at 1.5 m/s. A 1 µm particle gives $Stk = 0.27$. A 5 µm particle gives 5.9, and a 10 µm particle gives 23. This is why the useful range is 1 µm to 10 µm.

You will see inertial impaction in baffle grease filters in kitchen exhaust hoods and in MERV 8 to 11 filters inside air handling units.

1.3 Interception

Here the particle follows the air streamline almost perfectly. Yet its own physical size is large enough that it brushes against the fibre surface and sticks. Van der Waals forces hold it in place.

Interception targets particles from 0.1 µm to 1.0 µm.

Unlike impaction, this mechanism does not depend much on air speed.

We rely on interception in secondary filters such as bag filters and rigid V-bank filters.

1.4 Diffusion (Brownian Motion)

The smallest particles, those below 0.1 µm, have almost no mass. Gas molecules constantly bombard them, causing a random, zig-zag motion we call Brownian motion. This random walk pushes the particle off the streamline and raises the chance it will hit a fibre.

The strength of the wandering follows the Stokes-Einstein diffusion coefficient:

$D = \frac{k_B T C_c}{3 \pi \mu d_p}$

A 0.1 µm particle has $D$ near $6.9 \times 10^{-10}$ m²/s. A 1 µm particle wanders about 25 times more slowly. Smaller particles wander more.

Diffusion works better when air moves slowly, because the particle spends more time near the fibres.

You find this principle engineered into HEPA and ULPA filters.

1.5 Electrostatic Attraction

Some filter media carry a permanent static charge, called electret media. The charge attracts particles that have the opposite charge or that become temporary dipoles.

There is an important limitation. As dust builds up on the filter, the charge gets masked. Efficiency then drops sharply over time, back towards what the fibres can do on their own. We must always remember this when we specify charged media.

1.6 The Most Penetrating Particle Size (MPPS)

Impaction and interception become less effective as particles get smaller. Diffusion becomes less effective as particles get larger. These two curves cross between 0.1 µm and 0.3 µm. That size range is the hardest for any filter to capture. We call it the Most Penetrating Particle Size, or MPPS. HEPA filters are tested against this size because it represents the toughest challenge.

For fine and HEPA media the MPPS usually falls between about 0.15 µm and 0.25 µm. Drop the face velocity and it moves to a larger size, because diffusion then does more of the work.

The 0.3 µm test size is not special in itself. It simply sits close to the worst case. The European standard EN 1822 goes further and tests each filter at its own measured MPPS. It classifies filters from E10 up through H13 and H14 to the ULPA grades U15 to U17. An H14 filter must reach 99.995 % at that measured worst case.

2. Global Rating Standards

2.1 ASHRAE 52.2 (MERV)

In North America we use the ASHRAE 52.2 standard. It assigns a Minimum Efficiency Reporting Value from 1 to 16. The test measures how well a filter captures synthetic dust in three size bands:

• E1: 0.3 to 1.0 µm
• E2: 1.0 to 3.0 µm
• E3: 3.0 to 10.0 µm

Appendix J of the standard offers an optional step. The filter is first conditioned with potassium chloride aerosol to remove any electrostatic charge. The resulting MERV-A rating shows the true mechanical efficiency after the charge has faded.

2.2 ISO 16890 (Global Standard)

ISO 16890 has replaced the older European EN 779 standard. It rates filters according to the mass of real atmospheric particles they capture and links the results directly to World Health Organization health categories.

The main classes are:

• ISO Coarse
• ePM10 (particles that stop in the upper respiratory tract)
• ePM2.5 (particles that reach deep into the lungs)
• ePM1 (particles that can enter the bloodstream)

To earn a rating, the filter must remove at least 50 % of the particle mass in that size range. Every test requires isopropanol vapor conditioning so electrostatic effects are eliminated. This gives us a realistic picture of how the filter will perform after months of service.

One detail matters a great deal. ISO 16890 assigns the ePM1 and ePM2.5 classes on the discharged result. The ePM10 class uses the mean of the initial and discharged values. So the charge problem is already built into the answer. A plain MERV number does not do this unless we ask for MERV-A.

The gap between the charged and discharged figures is not a laboratory curiosity. It plays out over the first months of every service interval, while the pressure drop only ever climbs.

Two things are worth noticing. The efficiency falls early, but the pressure drop rises late. So the filter is at its weakest long before a gauge suggests changing it. The dust cake slowly wins some efficiency back, but it charges us fan power for the favour.

3. Primary HVAC Filter Types

3.1 Pre-filters (Pleated or Panel)

These filters usually carry a MERV 7 to 8 or ISO Coarse rating. Initial pressure drop sits around 30 to 50 Pa.

We install them as the first line of defense. They are not meant to improve indoor air quality for occupants. Their sole purpose is to keep dust away from the more expensive filters and coils that follow.

3.2 Secondary Filters (Bag and V-Bank)

Bag filters reach MERV 11 to 15, or ePM10 to ePM1. The deep pockets create a huge surface area. Face velocity drops, pressure drop stays moderate (about 80 to 120 Pa at the start), and the filter lasts longer. We avoid bag filters in variable-air-volume systems because the bags can flutter and release dust when airflow falls.

Rigid V-bank filters use mini-pleat media arranged in a V shape. They deliver MERV 13 to 16 or ePM1 performance. The rigid construction handles variable airflow, high humidity, and velocities up to 3.8 m/s without problems.

3.3 Final Filters (HEPA)

A true HEPA filter must capture at least 99.97 % of particles at 0.3 µm, which is the MPPS.

We achieve this by packing 20 to 30 m² of media into a frame that is only about 0.36 m². Internal velocity falls to 0.02 to 0.05 m/s. The long residence time lets Brownian motion do its work inside a dense mat of borosilicate glass fibres.

At 0.020 m/s a particle takes about 28 ms to cross half a millimetre of media. That is the long residence time the mechanism needs.

The price we pay is a high pressure drop, typically 250 to 300 Pa when the filter is new. Dedicated fan power is usually required.

3.4 Putting the Stages Together

Each stage cleans the air a little more. Each stage also adds resistance that we pay for every hour the plant runs.

4. Molecular and Gas-Phase Filtration

Mechanical filters cannot stop gases or vapors. For those we need a different process called adsorption. Adsorption means molecules stick to a solid surface. It is not the same as absorption, where one substance dissolves into another.

4.1 Media Types

Physical adsorption uses activated carbon. The carbon has an enormous network of microscopic pores. Van der Waals forces hold volatile organic compounds and heavier odor molecules inside those pores.

Chemical adsorption, or chemisorption, uses impregnated media such as potassium permanganate. These media chemically oxidize lighter gases such as hydrogen sulfide, sulfur dioxide, and formaldehyde and turn them into solid residues that stay on the media.

4.2 Engineering Constraints

Residence time is the critical design parameter. We calculate it as

$t_r = \frac{V_m}{Q}$

where $V_m$ is the volume of media and $Q$ is the volumetric airflow. For ordinary HVAC applications we need 0.06 to 0.10 seconds.

Because $V_m = A d$ and $Q = v_{bed} A$, we can also write it as

$t_r = \frac{d}{v_{bed}}$

So residence time is bed depth against bed velocity. The face area drops out. This is the more useful form when we are sizing a bed.

We must also protect the carbon with a good particulate pre-filter, at least MERV 13 or ePM1. If dust reaches the carbon, it quickly blocks the micropores and the gas filter fails.

This failure is silent. The bed keeps its pressure drop and simply stops working. Gas-phase media is also consumed, so there is no rising resistance to warn us. We replace it on a schedule, or on a measured breakthrough.

5. Commercial Kitchen Exhaust and Grease Filtration

Kitchen exhaust systems face extreme conditions: hot air loaded with aerosolized grease and strong odors. The design priority is fire safety as much as air cleaning.

5.1 Baffle Filters (Primary)

Baffle filters rely on inertial impaction. Air makes sharp 180-degree turns between stainless-steel vanes. Grease droplets larger than about 10 µm cannot follow the turns; they hit the vanes and drain away.

The same geometry also acts as a flame arrester. It breaks the straight line of sight into the ductwork and helps stop flames from spreading.

Notice that the velocity rule runs the opposite way to an air handling unit. Raising the face velocity improves capture, because impaction is a Stokes number mechanism. We size a grease filter for enough velocity, not against it.

At 2 m/s across the filter, capture of a 10 µm droplet is around 70 %. A 20 µm droplet is caught almost completely. A 1 µm droplet passes through untouched. That surviving sub-micron grease is the reason for the stages that follow.

5.2 Electrostatic Precipitators (Secondary)

An electrostatic precipitator charges particles and then collects them on grounded plates. An ionizer section first gives the grease and smoke particles a positive charge. Collection plates then pull those particles out of the airstream.

These units capture sub-micron grease very well, but they need frequent washing. If grease builds up on the plates, the electric field collapses and efficiency collapses with it.

5.3 UV-C Technology (Chemical Degradation)

The lamps give out two wavelengths. Light at 254 nm breaks the long-chain bonds in grease molecules (photolysis). Light at 185 nm splits oxygen and creates ozone, which further oxidizes the grease vapor into dry microscopic ash, carbon dioxide, and water vapor (ozonolysis).

The ash that remains is dry, crumbly, non-combustible, and non-abrasive. It either leaves the building harmlessly or settles as a fine powder. This dramatically reduces the fire load inside ducts and keeps the exhaust fan impeller in balance.

5.4 Activated Carbon and Ozone Mitigation

When odor control must be extremely tight, we add activated carbon at the discharge.

In UV-C systems with short duct runs (residence time less than 0.5 seconds), residual ozone can still be present. Ozone is a strong respiratory irritant. Carbon converts leftover ozone back into ordinary oxygen before the air leaves the building.

6. Regulatory and Design Framework (Ireland Context)

In Ireland we follow several interlocking standards and laws.

BESA DW/172 sets the rules for kitchen ventilation systems. It defines capture velocities, exhaust rates, and how electrostatic precipitators and UV-C units should be integrated.

BESA TR/19 Grease governs the internal cleanliness of extract ducts and how often they must be cleaned. Failure to comply can void fire insurance.

ISO 16890 is the accepted standard for specifying particulate filters in European HVAC systems. It also supports the indoor-air-quality requirements of Building Regulations Part F.

The Air Pollution Act 1987 is enforced by local Environmental Health Officers. It requires effective odor control (usually a combination of electrostatic precipitation plus UV-C or carbon) so that cooking smells do not become a statutory nuisance at the nearest residential boundary.

When we understand both the physics of particle capture and the practical standards that govern our work, we can design filtration systems that protect people, equipment, and the surrounding community.