The how and why of removing odours from commercial kitchen exhaust

by Robert Lord

Many designers (& council officials) appreciate the effects of commercial kitchen exhaust and the odorous discharge that occurs. What is sometimes misunderstood is that odour complaints can still occur with code-compliant systems and prosecutions can still occur.

What this paper sets out is how SEED derived our data-based understanding of kitchen odour, and how we have applied that understanding to a variety of projects over the years. The paper concludes with a methodology as to why we select specific treatment technologies.

Quantifying Odour

Whenever we smell something, our nose and brain work together to make sense of hundreds of very tiny invisible particles, known as molecules or chemicals, that are floating in the air. If we sniff, more of these molecules can reach the roof of our nostrils and it is easier to smell a smell.

The sense of smell is so important because it is related to the sense of taste. In our ancient history, taste and smell were important for survival: to find good food, to avoid being eaten, and to pick proper mates. Smell warns us about rotten food, a burning house, or an unhappy baby, and it can also capture our attention if looking for potential mates.

Even people who claim to have a poor sense of smell are still capable of smelling odours at levels as low as 10ppb. The particularly pungent smells of some sulphides can be detected at levels below 4ppb.

Our noses vary in sensitivity from person to person, based on genes, familiarity, hormones, and time of day. The sense of smell also varies with wellness, and exposure to respiratory illnesses. Unfortunately, COVID victims, who have suffered a loss of smell, have found that the sense of smell can take a long time to return.

When we register an odour, it is usually described in terms of taste or flavour, and potency. What some noses believe is a pleasant smell may be unpleasant to other noses, and this determination is related to memory.

Figure 1 Smell is related to taste. Our noses are particularly sensitive and detect flavours within flavours.

For example, the familiar smells of the Milton Brewery or the late Arnott’s Biscuit factory could be fondly described by many Brisbane residents, despite several kilometres of separation from the source. Yet, even pleasant smells can cause nuisance over time, particularly when they are intense or unwavering. An example of an initially pleasant smell becoming unpleasant over time are some of our flowering trees.

Australian Standards

In 2001, an Australian Standard for measuring odour concentrations was released. It was AS4323.3: Stationary source emissions – Determination of odour concentration by dynamic olfactometry and it was based on the European CEN standard, which was also very similar to the German standard VDI 3882.

The methodology of these standards was to determine how much dilution had to be applied to an air sample so as to render it detectable by only half of the population. This is conducted in a laboratory using “sniffers”. The sniffer has a simple resume – they need only possess an “average” sense of smell and odour recognition. However, they are regularly trained and tested to ensure they are an average nose in the best sense of the word.

The extent of dilution is the odour intensity measured in “odour units” (or OU). If an air sample has an odour intensity of 1000 OU, that means it needs to diluted by 1000, to render it barely noticeable by half of the population.

When a formal odour sample is reported, the OU rating is stated, and usually a simple description of the taste and character is provided. Note that these descriptions are usually written up well before the odour reaches a low concentration. This is because the human nose is so powerful, it can detect an odour at low strength but not recognise it.

Permitted Odour Intensity

Given that odour strength can be quantified, and it’s hedonistic quality described, authorities now had a tool by which they could judge whether an odour complaint was real. Several authorities formed regulations around odour nuisance based on odour readings measured to AS4323.3. For example, the Queensland EPA considers an odour nuisance will exist at 5OU and it is worth prosecuting at 10OU whilst NSW OEH has a sliding scale of 2 OU to 7 OU and the decision to prosecute rests with the complainant. The City of Melbourne has the tightest standards in Australia, seeking to prohibit any odour detected at 1OU or above.

The standard which we work with most is the NSW OEH table presented below. It is a sophisticated, nuanced table that adjusts odour level thresholds based on the density of the population.

Table 1: Impact assessment criteria for complex mixtures of Odorous Air Pollutants (nose-response time average, 99th percentile. Source: Department of Environment and Conservation (NSW), 2005.

Population of Affected Community Impact Assessment Criteria (OU)
Urban (≥2000 people) 2
500 people 3
125 people 4
30 people 5
10 people 6
Single rural residence 7

One of the victims of this style of regulation is that some industries rely on odour for advertising. An obvious example is a croissant bakery, which would regularly create environments with 25-40 OU.

In European and Asian nations, where land is scarce, there is more opportunity for odour nuisance from the colocation of farming or industry near residential developments. Various technologies have been developed where a chemical sampling device can predict an OU reading insitu, leading to a more informed response for odour remediation. The benefits of this type of sensor is that it can assign a proportion of odour nuisance to a source.

Kitchen Odours Intensity

Kitchen cooking odours arise from the release of hydrogen sulphides & aromatic carbons[1], during the cooking process. Aromatic carbons is a term I use to describe all of the various carbonyls, ethers, alcohols (such as butanol), and carbamides that are the source of many cooking odours. Carbon itself is odourless, but carbon attached to hydrogen, oxygen, sodium and/or sulphides is not odourless and responsible for most of our sensations of food-related odours.

Cooking drives the release of these chemicals and a useful surrogate for the stink of cooking is the release rate of fat content. This is affected in equal parts by the fat content prior to cooking and the temperature at which the cooking is undertaken. The higher the temperature, the faster the release of fat content.

There are other cooking odours that are not regularly attached to grease cooking, namely smoking (solid fuel cooking), or the additions of spices and sauces. These particles do not necessarily attach to grease particles, which is a point I will return to later within this paper.

In one of our earliest projects (2002), we were to design a kitchen exhaust system which by necessity discharged 12m away from an apartment balcony. The developer for whom we worked was most concerned about nuisance, despite the compliant discharge. The style of cooking was a South East Asian restaurant. We arranged for an air sample and derived an existing odour level (890 OU/m3). Using CFD, we could also determine how to dilute that odour so that it avoided nuisance levels. What was striking was that a compliant system could still cause considerable odour nuisance.

We were later approached for similar odour nuisance studies, which prompted me to investigate what was the empirical relationship between air treatment systems and odour reduction.

We purchased testing data and arranged our own testing data to compile a set of 38 tests. From those tests, we derived the tables below:

The tables above are some of the post processing of the raw data we had collected. In the collection of the data, we determined that the 5-micron particle size was an effective surrogate for odour intensity. Note also that the baffle filter was not removing much of the grease by weight.

That is, if a filtration system effectively removed 50% of the 5 micron particles, the odour levels would reduce by 50%. I stress that this is an empirical finding. And should eb treated as such.

Armed with that empirical relationship, we began to design street-level discharge kitchen exhaust systems by paying particular attention to the filtration technologies that performed well in that particle size.

We have found that the “5 micron” strategy consistently predicts the onset of the odour nuisance. It appears to be neither conservative nor optimistic.

Kitchen Filtration Tactics

From the data we had collected, we could see that the following filtration and design tactics were successful:

  • Filtration with cyclonic filters (e.g. Veritech filters) to remove 5 micron grease
  • Filtration with wool inserts (e.g. Shepherd Filters) to remove 5 micron grease
  • Filtration with Electrostatic Precipitators (e.g. AOS ESP) to remove 5 micron grease
  • Dosing ozone using an ozone generator to break down grease
  • Dosing ozone using UV lighting to break down grease
  • Discharging at high velocity to quickly dilute any ozone or odorous air.

We still use each of these methods but our past projects have informed us on the risks of each of these methods.

For example, we have found ESP maintenance costs to be largely unaffordable for many F&B outlets. If we improve the upstream filtration of grease, it enables the maintenance costs of ESPs to be greatly reduced. Hence, we still promote ESP use, particularly where spices or solid fuel cooking is involved as the spices and solid fuel smoke respond well to ESP filtration.

Care must be taken when specifying ESP velocities. Often HVAC specifications have no actual velocity nominated, and some suppliers allow layouts on a velocity of 3.6m/s. As per the research provided by Dalmon & Lowe, this velocity is too fast. (refer image below).

Figure 2 Dalmon & Lowe research into particle velocity vs capture.

We regularly specify filters that are cyclonic so as to lift grease removal in the 5-micron range. Alternatively, for existing hoods, we will specify wool inserts that can also reduce 5-micron grease significantly.

Figure 3 This is a useful table for selecting the odour reduction rate based on the 5 micron size removal efficiency.

We have found ozone generators to be particularly effective at reducing odours but use of UV lighting to be less effective. In the data for kitchen exhausts treated with UV lighting ozonation, we have seen the description of odours change to acidic, “burnt butter”, etc. What this implies is that the ozonation has been sufficient to form butanol but not completely “cold burn” the odour. The odour intensity has reduced but the taste has become more acrid, until the odour has been cold-burned.

Ozonation for kitchen exhaust odours needs to be approached in a particular manner if it is to be reliable. Firstly, the discharge air must be filtered by a filter capable of removing 80% of the 5 micron grease. This is so that dosing is not consumed oxidising copious amounts of grease.

Once the air has been preliminarily cleaned, it should be dosed at 8ppm approximately and have at least 2 seconds contact time. This contact time allows for the odour to be “cold burned” or ozonated to oxidisation.

Whilst doing to 8ppm sounds simple, one of the complexities is that our NH&S exposure standards for ozone mean that the discharge should be diluted below 0.5ppm before a predictable nose or point of interest (POI) has the chance to smell the discharge. When there is very little load for the ozone, it is possible for unconsumed ozone to be discharged at a level higher than desirable for exposure.

This advice is in line with following excerpt is from the article “ASHRAE Position Document on Filtration and Air Cleaning”:

Experience suggests that control of a moving airstream does not provide favourable killing rates because of the short dwell time. Under ideal conditions, inactivation and/or killing rates of 90% or higher can be achieved but depend heavily on the dosage rate and air cleanliness.

Discharging at velocity and with high aspiration will permit both the safe accommodation of ozonation and further dilution for odour control.
One filtration technology that is not pursued is the use of carbon filters. Carbon filters are regularly prescribed as a “catch-all” but there are some reasons for caution and why they fail. Firstly, the filter uses adsorption and hence it is quickly consumed when the air stream is moist or heavy with grease load. The main reason they fail though, is that the adsorption “let’s go” of the dirt particle at temperatures above 500C, which is a common kitchen exhaust temperature.

Adsorption is a reversible process due to the relatively weak (van de Waal) forces involved i.e. odours once adsorbed can later desorb back into the airstream. Hence, there are regular precedents where carbon filters have been specified as an emergency measure but they failed.

Pleasant Surprises

We accidently experienced one pleasant surprise recently which is worth sharing. The kitchen exhaust system was a low-level discharge, discharging near a seating area. The system utilised “good” filtration and ozonation but no ESPs.

The cooking was high intensity Asian cooking and we would often use a ESP for this type so as to address spices. In this instance, there was no room nor budget so we compensated with additional filtration at the 5 micron level. The result appears to have removed the grease odour, but the spice odour can still be faintly determined. The proprietor is pleased as he claims the “clean” smell of the spices is attractive to those who have grown up with this odour in their local markets.

Another pleasant surprise is that ozonation does break down grease particles into dust. See the photo below from a project with a 5 second contact time, where the ozone has broken the grease down into a dusty surface that is easier to clean and does not support combustion.

Figure 4 Downstream photo of hood dosed with ozone. Note the dusty appearance, This actually indicates ozonolysis.

Another pleasant surprise is the manner in which aspirated diffusers can be used to diffuse kitchen exhaust air at low levels. Aspiration is a process where an air stream is separated so as to induce dilution air into the middle of the air stream.

Some of the projects where this technique is employed:

Rashays (Indian) Restaurant, Dee Why – discharge above entry.

Fish restaurant, Melbourne Street, South Brisbane. Discharge is above restaurant entry.

Happy Boy (Chinese/Malay) restaurant, Fortitude Valley. Discharge is directly below apartment balcony.

Mrs Luus,(Vietnamese), Tank Street Café. Discharge above seating and onto street


A caution for designers is warranted. These systems do have useful advantages for owners to enable them to do business where previously it was not permissible. However, there is a need for the owner to conduct maintenance and a great temptation for the owner to not execute maintenance.

For this reason, a discharge near the public is ironically a good feature. It motivates the café or restaurant owner to maintain and clean their system.

It is imperative that restaurant owners and landlords are educated about the costs and implications of maintenance. Where centralised systems require landlord works, the savings for the landlord come about from reducing the extent of duct cleaning and from reduced fire exposure.

The Association of British Insurers has been briefed on the relationship of grease filtration to fire load and associated losses. Understandably, there is concern about the maintenance regime of any filtration system.

One tactic that we have found useful is to provide a schedule for kitchen exhaust maintenance that resembles the essential fire maintenance schedules presented in AS1851. That has been useful in ensuring that maintenance is executed appropriately.


One regular conversation is the conversation about capital and maintenance costs of kitchen exhaust air treatment. Air treatment costs are largely proportional to the exhaust air flow rate and all measures that reduce exhaust air quantities are in effect assisting with odour control.

One measure that we have found effective has been to remove the cooking processes that have a recirculation or low flow option. Many ovens and dishwashers have options for recirculation where the appliance itself comes with the filtration hood. These devices generally manage the kitchen internal environment quite well and have the benefits of removing fire risk and reducing energy costs. Simply avoiding three duct cleaning exercises will pay for the additional recirculation hood, whilst avoiding the conditioning of 1500L/s make up air is also a powerful commercial argument.

There are now recirculation options for a wide array of the cooking devices (Halton’s Mobi-Chef is one example).


The tactics presented herein are focussed on making F&B facilities more cost effective in situations where a roof discharge is not feasible. Many of the tactics are synergistic with other desirable goals. For example, over half of the fires faced by urban Fire Brigades are started in commercial kitchens. Hence the reduction of grease fire load (by a figure of >90% by weight) is fire safety measure that benefits the community. It is estimated that over 23 tonnes of kitchen grease is deposited on USA rooves every day.

Another desirable outcome is the reduction of greenhouse gas emissions associated with kitchen exhaust and treating kitchen make up air. By using recirculation systems to reduce kitchen exhaust air quantities, air conditioning capacity and greenhouse gas emissions can be saved to the extent of 30W/Lps of exhaust as an average across Australia.

There will not be a SEED TV episode that accompanies this paper. We wish you well in your design endeavours.