Case Study: Osteria Du, a Toronto Restaurant

“We need to manage the pandemic with data and evidence, not fear.”


Imagine you have the unthinkably bad luck to wander into a restaurant where all the other customers are infected with COVID-19. How likely are you to become infected from the air in that space if you dine there for two hours? We wondered that too, and with the co-operation of Roger Yang, a Toronto restauranteur that owns both Avelo and the soon-to-be launched Osteria Du we were able to find the answer. The results may surprise you.

Photo by Dan Burton on Unsplash

Like many, Roger is concerned about preserving customer and employee safety. That means careful attention to hygiene and disinfection, but also with managing the risks of airborne infection — an obvious consideration for a respiratory illness that we’re already wearing masks for. Reducing airborne risk means ensuring a steady supply of fresh air that acts as an extra layer of protection within a space. That way, even if someone infected with COVID-19 enters the space, adequate fresh air and filtering means that any exhaled infectious particles are evacuated from the space quickly — ensuring they don’t collect and build up to higher, more infectious, concentrations. To understand and mitigate those risks, Roger engaged to help measure the indoor air at Toronto restaurant Osteria Du (the subject of this post), and then took additional steps such as additional air filtering to increase safety even further (the subject of a future post). AirQ found that, as with many restaurants, the cooking ventilation hood was a huge asset for Osteria Du, helping to boost the rapid fresh air exchange for the space overall (to a respectable 2.15 ACH). That’s in contrast to well-known superspreading events like the much more crowded restaurant “X” in Guangzhou, China (where airflow was a paltry .56 ACH). The resulting air flow at Osteria Du provides an estimated “safe” occupancy level of approximately 12 people.

“the amount of fresh air coming in means that an uninfected person in a worst-case scenario — where they spend 2 hours surrounded by others with COVID — is only 2% likely to get infected by airborne particles”

So what does “safe” mean in this case? While there’s no such thing as a 100% safe place (someone can become infected by as little as a single particle of the virus), we can estimate relative risk. So “safe” at restaurant Osteria Du means that even in a truly unlucky scenario where 11 people in the space have COVID-19 and one doesn’t, the amount of fresh incoming air means their risk of infection is still low. More specifically, the unlucky person in our worst case scenario — who spends 2 hours surrounded by people with COVID— has a 2% chance of getting infected by airborne particles. In fact, on average, they’d need to spend 58 hours or more without a mask in the space before they have a 50% chance of catching COVID via airborne particles. Add masks (even home-made cloth ones) and that rises to 90 hours. That’s with the existing ventilation capabilities, and BEFORE any additional additional safety measures like filters are put in place. For those who are technically inclined, in the next section below we’ve included a much more detailed description of the measurement process at Osteria Du and the steps that were involved.

Measuring Safe Air in Detail

One of the questions that many venues have is about aerosol infection risk. How bad is the risk? Is it something they ought to be worried about? The answer frequently depends on the indoor space and its HVAC system. The risk of airborne infection has everything to do with airflow — the more fresh air, the better. The speed at which you swap out old air for fresh is measured in terms of Air Changes per Hour (ACH). Stale air and poor ventilation is something many superspreading events have in common—for example Wuhan Wet Market (.1 ACH), Skagit Valley Chorale (.7 ACH), and a restaurant in Guangzhou (.56 to .77 ACH).

One of the ways to manage these airborne risks is with a sensor that continuously monitors the air at your location, providing an indication of risk and how it varies throughout the day. That’s a great way to manage your air: just keep an eye on how stale the air is getting and ensure it doesn’t get too high (our sensor also blinks red and chirps). But what if we want to analyze our air in more detail, and estimate safe room capacities?

That’s where an air audit comes in, allowing AirQ to measure the actual airflow (ACH) directly and the number of people it will safely support. Using a harmless tracer gas lets us measure how rapidly your HVAC system brings in fresh air, while clearing out lingering exhaled air. The result is your venue’s Air Changes per Hour (ACH) which allows us to estimate a venue’s safe room capacity (and risk) in greater detail. Let’s take a closer look at the process in more detail, using Toronto Restauranteur Roger Yang’s new restaurant Osteria Du as an example (you may also want to take a look at some of the risk management measures Roger has taken at his other restaurant, Avelo).

To measure ACH at Osteria Du, the first thing we did was to flood the space with a tracer gas. This allowed us safely simulate high exhaled air concentrations, without needing to bring a crowd of people into the space. Roger’s restaurant was large enough that we emitted gas at several interior locations then used fans to help mix it adequately (we also measured how plexiglas barriers slowed the spread and mixing of gas in adjacent areas). Here’s what that looked like:

Once we obtained a sufficiently high concentration of gas within the space, we then turned on the sensors and began collecting data, monitoring how quickly the HVAC system is bringing in fresh air and removing the the tracer gas (or exhaled air in normal conditions). We ended up with a chart you can see in Fig 1 (below), where tracer gas levels decline over time as the fresh air clears it out (each coloured line is a sensor located at a different spot in the restaurant).

Figure 1: HVAC reduces the concentrations of our tracer gas

By logging the sensor data over time, we were able to calculate the Air Changes per Hour (ACH) at multiple locations in the space, giving us a much better understanding of airflow at different locations (as well as the volume of air at each of them). The result is a detailed map of the indoor space, providing an indication of the airflow at each spot. Here’s what that looked like at Roger’s restaurant (each green dot is a different ACH measurement):

Once we have the ACH results, that’s where things gets interesting… It’s a powerful way to model the space and help estimate safe room capacities (and understand how additional measures like portable filters boost them — more about that later). We began with a simple high-level estimate of safe room capacities. Based on recommendations from the Federation of European Heating, Ventilation, and Air Conditioning associations (REHVA) for example, we can calculate that a conservative capacity limit of approximately 12 people will keep exhaled air concentrations below REHVA recommended levels (assuming no masks or other filtering measures are in place).

With more additional modelling we can also take that recommended figure and make approximate estimates on how “likely” an infection might be to pass from one person to another, given how much time they spend in the space.

So for example, lets start with what might be a worst-case scenario. Assume that you’re incredibly unlucky and that you’re one of 12 people in a space where everyone else at the venue has COVID-19 and nobody there is wearing a mask (for purposes of our model we’ll use a quanta of 5 as the average rate of infectious particle generation). How likely is it that you’ll contract COVID-19 from airborne spread if you spend a couple hours at this location? We plugged that into our model and…

Or put another way, on average you’d need an exposure of around 58 hours to that set of conditions before you become likely (as in 50% likely) to get infected — 90 hours if everyone is wearing masks.

You might be wonder about those numbers, especially in light of superspreading examples such as the Guangzhou restaurant. The answer is that crowding under conditions of poor ventilation is exponentially more risky. Even more so if the crowded conditions lead to loud talking, or even shouting — which again increases risk substantially. By taking the above numbers (again the ultra-unlucky risk of 11 infected people present), and assuming: poor ventilation, a large crowd, and much louder talking, the 50% risk level declines to 3.5 hours. In a room with 90 people you could end up with 40 cases in that 3.5 hours of exposure (or in the case of only 1 infected person, 5+ new cases).

Now, the reality of estimating one’s infection risk when dining in a restaurant can get far more complex of course. Much of it depends on exactly whom you’re dining with, and who else is closest to you. Key factors include: whether those nearby you are infected or not, how loud they’re talking, and exactly how close they’re sitting to you (i.e. droplet infections from people within arms-length are an additional risk that’s not modelled here). Also, if someone near you happens to be infected, it also matters what stage they’re at in their illness, and the amount of viral particles they’re shedding (which can vary a lot from person to person). All these things matter. Ideally, you’re already dining with a member of your household (or agreed upon social bubble in areas that are currently practicing that), and that means you’re already eliminating one of the biggest sources of risk, especially if everyone in your bubble has been practicing safe physical distancing. If you’re dining in an area with significant plexiglass barriers, thats another factor, meaning that most of the local exhaled air is going to be mainly from your dining partner(s), and much less from those further away.

The simple truth is that you can ignore most of the complexity and focus on three really simple things that matter: stay in your bubble, spend your time in safe air, and wear a mask.

If it all of this sounds like a complex calculus, well, it kind of is. But don’t let that scare you. The simple truth is that you can ignore most of that complexity and focus on three really simple things that matter a lot:

(1) Keep to your family and bubble as much as possible, and reduce close contact with others

(2) Spend your time in areas with safe air, whether it’s your home, outdoors, or public/private venues with known safe air, and

(3) Wear a mask, it’s an extra layer of protection that’ll help keep you and the people you love safe. Even home-made masks decrease risk significantly.

At AirQ, we’re here to help you with the second factor: shedding light on indoor air safety. Just knowing whether indoor air is safe, or not, is often a big relief. It helps us understand actual airborne infection risks and act accordingly. We need to manage the pandemic with data and evidence, not fear. Measurement of indoor air is the key to responding more intelligently to risk with simple and precise mitigation efforts.

In a future post, we’ll talk about risk reduction for businesses…. looking at additional measures that Roger and others have been taking using tools such as portable air filters, plexiglas barriers, and the impact they have on reducing airborne infection risks even further.



Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store