Air Changes Per Hour Heat Loss in a Retrofit Home

Air Changes Per Hour Heat Loss in a Retrofit Home

Introduction

This post aims to investigate the effect of ventilation losses in total house heat loss calculations, looking specifically at air changes per hour (ACH) and using my own home to drive the data.  A subject Nathan Gambling and I spoke about on the BetaTalk podcast.

I will use historical heat pump usage tracked via Open Energy Monitor, which logs heat output and electricity use from the heat pump in real time. I will also use the Spruce modelling tool to show different total house heat loss figures at various ACH values to see if we can infer the actual ACH from real data.

The article then looks at ways of measuring actual ACH in my house and compares the results of those physical tests with the modelled data.

You may not have come across the phrase air changes per hour before, but it plays a surprisingly big role in heat pump design. Get it wrong and heat loss can be overstated, leading to unnecessary cost. Get it right and you end up with a system that’s better sized and cheaper to install.  Think less radiators to swap out, less disruption, potentially a smaller heat pump and maybe higher efficiency and lower running costs.

Who this article is for

This article is written for homeowners, retrofitters and heating engineers who want a better understanding of how ventilation and air changes per hour affect heat loss calculations in real homes.

It assumes a basic familiarity with heat pumps and heat loss concepts, but you do not need to be an expert. Where tools, standards or acronyms are used, they are explained as they appear. The examples and data are drawn from a lived in 1930s retrofit home, so the focus is on real world behaviour rather than textbook assumptions.

Table of Contents

Use the Table of Contents below to jump straight to the section that matters most or scroll through at your own pace.

Diving into the data

I have heat pump usage data going back to October 2022 from my 5 kW Vaillant aroTHERM. However, because I made the last of my significant fabric changes in the summer of 2024, when the dining room floor was upgraded, I will focus only on the coldest days since July 2024 for this analysis.

Here is a photo of that final project being completed.

If you’re interested in the rest of the fabric changes I’ve made over the years, you can check those out here: Fabric Retrofit in a 1930s Semi-Detached House.

Fabric Retrofit in a 1930s Semi-Detached House

The 35 coldest days since summer 2024

Using Open Energy Monitor I was able to isolate the 35 coldest days since the summer of 2024 (to end of January 2026).  I thought this would make for a good working data set to analyse.

For the days chosen we only wanted to know the heating usage (kWh), so we again used Open Energy Monitor to eliminate the hot water usage.

We also track the average outdoor temperature for the whole day as well as the average internal temperature.

From all this data we can calculate the HTC (heat transfer coefficient) for each day, which describes how much heat the house loses for each degree of temperature difference.

In most cases the average outdoor temperature is warmer than −3 °C and the indoor average less than 21 °C.

Using the indoor temperature, outdoor temperature and heating kWh, we can calculate a HTC value and then scale that to the MCS design conditions (set by the Microgeneration Certification Scheme, which governs UK heat pump standards) of −3 °C outside and 21 °C inside.

Sample HTC calculation

This calculation is shown for one day (14th February 2025), but the same calculation was applied across all the 35 coldest days data.

• Heat out (64.2 kWh) ÷ 24 × 1000 = 2675 W
• Average inside (19.9 °C) minus average outside (2.0 °C) = 17.9 °C
• 2675 W ÷ 17.9 °C = 149 W per °C

Result: 24 (difference between −3 °C and 21 °C) × 149 W per °C = 3587 W

So for this day the HTC is 149 W per °C and the total calculated heat demand (heat loss) is 3587 W (or ~3.6 kW).

This means that for every 1 °C difference between inside and outside, the house loses around 149 watts of heat.

Once we’d done this calculation for all 35 days, we could arrive at an average and sensible ballpark heat demand figure.

You can view the raw data behind 14th February 2025 via my own Open Energy Monitor Link or via the image below.

Daily heat loss data example (Top 5 days)

Other factors affecting heat demand

On first viewing, the dataset was pretty messy with no consistent trends.  So I started to look whether solar gain comes into play.  I used the kWh generated each day from our 5.1kW solar PV array to see if there was any correlation.

I also looked at wind data for each day. What the maximum and average wind speed and wind gusts were.

I was looking for any nice and simple cause and affect.

But it turns out heat demand is a minefield as there are so many different things that can sway it on any single day.

• Outdoor temperature

• Target indoor temperature

• Time spent producing hot water (so not heating the house)

• Temperature the day before

• Solar PV and solar gains

• Wind driven infiltration

• People and pets in the house

• Cooking and household appliances

• Door opening and window ventilation

• Party walls (particularly in semi detached or terraced properties)

When you consider all these variables it’s not hard to see why the data I collected was a bit messy.

It is also worth noting that all of the airtightness and ventilation measurements discussed in this article were purposely taken during the winter. Winter is when heat loss actually matters and when windows are typically closed.  For heat loss purposes I don’t think it makes sense to do some these types of tests in the summer, especially ones that analyse over longer periods, as you’re more likely to leave windows open.

Whole-house heat loss from measured data

Based on the cleanest data days, the whole house heat loss is around 3.6 kW at a notional design condition of −3 °C outside and 21 °C inside, with day to day variation of roughly ±0.1 kW.

At −3 °C outside and a 20 °C indoor setpoint, the equivalent heat loss is around 3.4 kW, which better reflects how we actually run the house.

But for the purpose of this article, so we can compare to the values the MCS accredited application Spruce gives us, we are going to use the 3.6 kW (−3 °C outside and 21 °C inside) whole house heat loss figure.  i.e our MCS “design conditions”.

Introducing Spruce

Spruce is an online heat pump design and modelling tool used by installers to calculate room by room heat loss, size emitters, and build a whole house view of how a property performs. It also allows installers to combine enquiry handling, heat loss surveys, system design, estimates and paperwork into a single workflow.

I used the design section of Spruce to model my entire house. I input the construction details, fabric elements and their U values, along with the existing radiators in each room. That provided a full house heat loss calculation and an emitter survey.

One of the things I found most useful is how easy it is to test changes. You can adjust a single element, such as upgrading a wall or changing window performance, and immediately see how that affects the heat loss of an individual room and the property as a whole.

What makes up heat loss?

Total heat loss is made up of two parts.

• Total heat loss = fabric heat loss + ventilation heat loss.

Fabric heat loss is the heat that escapes through the physical structure of the building. That includes walls, windows, doors, floors and the roof.

Ventilation heat loss is the heat carried out of the house by moving air. This includes intentional ventilation, such as opening windows, trickle vents or using extractor fans, and unintentional air leakage through gaps and cracks.

MCS minimum room air change rates

For each heated room, the default minimum air change rate used in calculating ventilation heat loss is determined by room type and property age band.

• Age Bands A to I (pre-1996)
• Age Band J (1996 – 2002)
• Age Band K onwards (2003-Present)

My house falls into Age Band A to I (pre 1996), as it is a 1930s property.

These figures are derived from the CIBSE Domestic Heating Design Guide (2021).  You can find these figures listed in the MCS Heat Load Calculator.

Older properties are assigned much higher default ACH values, often between 1.5 and 3.0, compared to modern properties which commonly use 0.5 due to better airtightness.

Minimum room air change rates

Fabric heat loss from Spruce

When you put all the house elements into Spruce, so U values etc for the fabric of the building and the ACH for each of the rooms, it gives the whole house heat loss in watts along with room by room losses too.

It also gives an individual breakdown of fabric losses and ventilation losses.

Modelled to −3 °C outside and 21 °C indoors throughout, the fabric heat loss of the house is 2756 watts or 2.75 kW.

Any remaining watts must therefore be ventilation losses.

Whole house heat loss – fabric heat loss = ventilation losses.

Ventilation losses at different ACH values

Using the MCS default ACH values for each room type in a pre 1996 property:

The total ventilation losses of my house calculated using those values is 2612 W or 2.6 kW.

So if we add these 2.6 kW of ventilation losses on top of the 2.75 kW fabric losses Spruce gave us, gives us a total of 5.35 kW heat loss.

But as we’ve already seen, the reality is that the actual recorded total heat demand of the house (from Open Energy Monitor heat meter data and HTC calculations) is only around 3.6 kW total.

Comparing ventilation losses in Spruce

Spruce is absolutely brilliant at seeing recalculated values instantly. You only have to change the numbers and the heat losses are updated immediately.

Fabric losses are static at 2.75 kW, so it’s just the ventilation losses that change as we play with different ACH values.

We can see the result of those ACH changes and ventilation losses in the table below (values shown in watts).

These numbers suggest that my ACH is much closer to 0.6 than the 1.6 implied by the MCS and CIBSE defaults.

With a total heat loss of ~3.7kW seen with 0.6 ACH, which aligns with the real world data from Open Energy Monitor.

But the headline here is the difference between the results at 1.6 ACH and 0.6 ACH (5.3kW total heat loss versus 3.7kW), that’s close to 2kW.

How ventilation heat loss is actually calculated

Ventilation heat loss is driven by two simple things:

• How much air is being replaced (ACH / air changes per hour)

• How much air there is to replace (the volume of the house)

That second point is easy to overlook, but it matters a lot. Ventilation losses scale directly with house volume, not floor area. Bigger houses move more air, even at the same ACH.

In my case, the internal volume of the house is around 231 m³, coming from roughly 90 m² of floor area. Using Spruce, we can see how much heat is lost purely through air movement at different ACH values.

• At 0.6 ACH, the ventilation heat loss is around 966 W
• At 1.6 ACH, that rises to 2575 W

Nothing else has changed. Same house, same fabric, same temperatures. The only difference is the assumed number of air changes per hour. Increasing ACH by a factor of around 2.5 has increased the ventilation heat loss by a similar amount.

Now imagine applying the same assumptions to a house twice the size.

If the internal volume doubled from 231 m³ to around 460 m³, the ventilation losses would also roughly double. At 0.6 ACH that would be close to 2 kW, and at 1.6 ACH it would be over 5 kW. That is before you even consider fabric losses.

The key point is that ACH is not a minor tweak. It is multiplied by the full volume of the house and the temperature difference. That’s why getting it wrong can add or remove multiple kilowatts from a heat loss calculation, even in a relatively modest-sized home like mine.

So you can see why just accepting the MCS / CIBSE defaults without question can get you into big trouble.

Measuring real world ACH

Reverse engineering ACH from heat demand data has been amazingly useful, but I wanted to find out how accurate these numbers were compared to real physical tests.

There are three main tests that I know of

• Build Test Solutions Pulse Test
• Purrmetrix CO2 decay testing
• Blower Door Test

I have been lucky enough to have used two of these methods on our house and I will describe all 3 below.

Build Test Solutions pulse air permeability test

I had the pleasure of welcoming Ross Greenwood into our house to conduct a Pulse Test using his Build Test Solutions kit.

Ross’s company (Vayu), based in Yorkshire provide a whole host of renewable services; heat pumps, solar, batteries, EV chargers as well as air tightness testing.

A pulse test checks how much air leaks through the envelope of a home or building by using a controlled burst of compressed air at a low pressure level, typically around 4 pascals.

Instead of sealing up a door frame and running a big fan like you do in a blower door test, a pulse tester sits inside the building, charges an internal air receiver and then releases a short “pulse” of air. Sensors track how the internal pressure rises and falls over a few seconds and record the airflow needed to restore equilibrium.

From this data the test calculates leakage rates in cubic metres per hour and related values such as air permeability and effective leakage are and results are available as soon as the test finishes

You can see how it works on this video or look at this page on the Build Test Solutions website.

The process started with Ross surveying the house so calculate the volume and envelope area as these are used to calculate the air leakage rate and air permeability.

Ross then charged up the two pressure cylinders and placed them in the centre of the house before final connections to the control unit. The test itself only takes around 20 seconds to complete and you can see the data on the screen straight away.

Here is a copy of the report showing the result.

I have to admit, there are so many numbers here, most of which I didn’t understand.  So I asked Google for some help.

How to read this air permeability report

The headline result on the report is air permeability at 4 Pa, shown here as 1.52 m³/h/m². This tells us how much air leaks through each square metre of the building envelope when the house is held at a small pressure difference. Because this is a low-pressure pulse test, it is already much closer to real conditions than a traditional 50 Pa blower door test.

The report also gives us the two numbers we need to turn this into something more practical: the envelope area (263.3 m²) and the internal volume (231.1 m³). From these, the tester has already calculated the whole-house airflow rate at 4 Pa, shown as 400 m³/h, which equates to 1.73 air changes per hour at 4 Pa. In other words, under test conditions, the entire volume of air in the house would be replaced about 1.7 times per hour.

From test result to usable ACH (0.44 ACH)

Heat loss surveys, MCS sizing and CIBSE guidance do not use test-pressure ACH directly. They need an estimate of natural ventilation and infiltration, which reflects everyday conditions driven by wind and temperature difference, not fans.

To get there, the measured ACH at 4 Pa is converted using the SAP infiltration method. SAP assumes that only a fraction of the airflow seen during a pressure test occurs on average in real life. Applying that standard conversion:

• 1.73 ACH @ 4 Pa

• becomes 0.44 ACH (natural infiltration)

This 0.44 ACH is the figure that should be used in heat loss calculations and room ventilation assumptions.

Real-world ventilation measurement with Purrmetrix

Another way to estimate air changes per hour (ACH) is by observing how the building behaves in everyday use rather than during a short, controlled test.

The Purrmetrix system uses a network of environmental sensors placed throughout the home to record carbon dioxide (CO2), temperature and humidity over a period of weeks. CO2 levels rise when people are present and fall when air is replaced by fresh air. By tracking the rate at which CO2 concentrations decay after peaks, the system can estimate how quickly air is exchanging in the space. The data is collected wirelessly into a gateway and sent to the Purrmetrix cloud portal for analysis, where ventilation rates and patterns are visualised over time.

Hermione Crease from Purrmetrix sent me this pack of sensors and monitoring kit (oh how I love the little cat logos!).

What makes this approach different from a pulse or blower door test is that it is in situ and continuous. The sensors sit in the occupied home while doors are opened, windows are used and people go about their normal routines. That gives you a sense of how ventilation actually operates in real life, including behaviour-driven patterns that a pressure test can’t capture. The cloud analytics convert the observed CO2 decay into ventilation estimates that can be expressed as ACH, providing another way to quantify and compare ventilation performance alongside the more traditional test methods.

Purrmetrix results

The Purrmetrix system analysed 10 days of data, covering the period 10th January 2025 to 20th January 2025.

The average ventilation rates recorded were:

• Living room: 0.68 ACH

• Loft bedroom: 0.56 ACH

• Front bedroom: 0.55 ACH

• Back bedroom: 0.78 ACH

• Dining room: 0.74 ACH

From their site: “Ventilation rate is calculated using an exponential CO₂ decay model and is expressed as the number of air changes per hour. Each marker on the graph represents an estimate of the decay rate during a period where no additional CO₂ is being introduced into the room. The system then displays the average, best and worst decay rates across the analysed period.”

Example CO₂ decay graph showing calculated ventilation rates for our dining room.

Relating room-level results to whole-house ACH

What’s reassuring here is how closely these room-level figures line up with the earlier reverse-engineered whole-house result of around 0.6 ACH. Most rooms sit either side of that number, with two of the bedrooms clustering tightly around 0.55–0.56 ACH and more active spaces like the dining room and living room coming in a little higher.  The back bedroom at 0.78 is the boys bedroom.  So if anyone with a teenage son will know, that this ‘gaming room’ needs the window cracking open more than others. LOL.

When averaged out across the whole house, these values land very comfortably in the same ballpark suggested by the heat loss data and Spruce modelling.

Why ventilation rates vary by room

It’s completely normal for ACH to vary from room to room. Occupancy patterns, door usage, room volume, window opening habits and even how often the cat wants in and out all play a part. Bedrooms tend to be quieter spaces with long periods of stable conditions, which makes CO₂ decay easier to observe and often results in lower apparent ventilation rates.

By contrast, living spaces and dining areas see more movement, more door openings and more connection to the rest of the house. That naturally pushes their measured ventilation rates higher. This variation is exactly why a whole-house average can be potentially more useful for heat loss calculations than relying on any single room measurement.

It begs the question, is one sensor enough for this test?  or are you better with a few to see those differences?

My thoughts are that if you are confident that the build (and air tightness) of the various parts of your home are similar, then you may get away with one sensor.  But if you have drastically different elements of the house, maybe a newer built extension attached to an older part of the house, maybe more sensors would be required?

Comparison with the pulse test result

The Purrmetrix results also sit neatly alongside the pulse test outcome. The pulse test conversion gave a natural infiltration rate of 0.44 ACH, which represents a sealed, closed-house condition under controlled testing. The Purrmetrix data, by contrast, captures the reality of daily life, with doors opening, people moving around and windows occasionally being used.

Seen together, the two methods are in the same ballpark. The pulse test shows what the building envelope is capable of when everything is shut, while the Purrmetrix results show what actually happens in use. A whole-house working value around 0.6 ACH makes sense when you consider both perspectives, and explains why the measured heat demand of the house lands where it does.

Blower door test

A blower door test is the more established method of measuring airtightness and has been used for many years in building regulations and compliance testing.

It works by fitting a large calibrated fan into an external doorway using a temporary airtight frame. The fan pressurises or depressurises the building while instruments measure the airflow needed to hold a fixed pressure difference, normally 50 pascals.

The test is more intrusive than a pulse test but has the advantage that it can be used to find draughts and leaks. With the fan running, air movement through gaps is exaggerated, making problem areas easy to detect.

Having already undertaken the Pulse Test and used the Purrmetrix sensors I saw no need to bother with a blower test.  If I did, I’d be pretty confident the results would come out around the same as the other tests.

 Photo credit: Glyn Hudson

Different ways to measure ACH, and why they all matter

All three routes are simply different ways of answering the same question: how fast air is changing in the building.

The method varies, and you are obviously measuring different things, but ultimately you are just trying to find out an accurate air changes per hour rate.

The downside of airtightness in retrofit homes

Airtightness on its own is not automatically a good thing, especially in older homes that do not have mechanical ventilation like MVHR. When you tighten up a building without giving air a planned way in and out, you can end up trading heat loss for stuffy rooms, lingering smells and rising CO₂ levels. From a heat loss point of view low ACH looks great. From a living-in-the-house point of view, it can feel less great.

Most retrofit homes rely on what I’d call manual ventilation. Windows get cracked open, trickle vents do their thing and every so often the house needs a proper “burp” by opening a door and flushing the air through. That works, but it is inconsistent and entirely down to occupant behaviour. If no one opens a window, ventilation drops. If it’s cold or windy, people tend not to. The result is that airtightness numbers alone can give a false sense of comfort if they are not considered alongside how fresh air is actually being replaced.

In our case, we have an unexpected bit of help. Our cat Romeo insists on going in and out multiple times a day turns out to be a surprisingly effective ventilation strategy.

Every door opening dumps warm air and pulls fresh air back in. It is not controlled, it is not efficient, but it is very real. That is why it matters to understand what ACH actually represents in day-to-day life.

Airtightness needs ventilation to go with it. Otherwise, you just end up with a very well sealed box that still needs the doors and windows opening regularly to breathe.

There isn’t a single “dangerous” ACH number, but homes operating much below about 0.3 – 0.4 ACH under normal conditions often start to rely heavily on occupants opening windows or using mechanical ventilation to maintain air quality. Older retrofit homes without MVHR can end up with condensation, stale air and mould risk if airtightness is improved but ventilation isn’t.

Remember, newer build houses that are targeting 0.3 – 0.4 ACH will likely have some automatic ventilation in place.

No heat loss numbers are ever perfect

At this point it’s worth reiterating that you can never be 100% with home heat loss.  Whether it’s using reverse engineered data or a specialist coming round to do a survey.

As we saw from working with the figures from the heat meter there are so many variables at play.

And it’s the same with an in person heat loss survey.  You may think you know the materials that have been used to build the house, but you don’t know for certain how well they have been installed behind a plasterboard or how well or complete a cavity is filled.

But throughout the process you are trying your best to eliminate or reduce the unknowns to get you closer to a realistic heat loss figure.

Do you size the heat pump to perfect ACH tests or real world?

The pulse test shows what the house can achieve in a sealed, best-case scenario, giving a lower bound around 0.44 ACH. The Purrmetrix data shows how the house actually behaves when it’s lived in, with doors opening, people moving around and normal ventilation habits, clustering around 0.6 ACH. That real-world figure also aligns closely with the reverse-engineered heat loss from the heat meter data.

Taken together, these results suggest that sizing purely from a “perfect” airtightness test risks underestimating ventilation losses, while blindly following MCS defaults risks oversizing. A whole-house ACH around 0.6 reflects how the house genuinely operates day to day. That makes it a far more sensible figure to use when sizing a heat pump in a lived-in retrofit home.

So which set of results do you use to size a heat pump?

Well, the heating engineers I’ve spoken to said they always add a little “real world” onto the results of a blower test or pulse test.  They do this to mimic the additional air changes that come with real life.

And even then, this heat loss total is not always what is then used to choose the heat pump. Investigations have shown that oversizing a heat pump by around 20-30% over the heat loss is not a bad thing.  Obviously, we are talking 20-30% over an “accurate” heat loss.

So my 5kW heat pump matches my 3.6kW heat loss pretty well in that regard.

And this is not a problem as most of the ‘better’ heat pumps out there (like my Vaillant Arotherm) all handle cycling really well due to their internal algorithms, so a little head room is fine, allowing the heat pump to work down to even colder temperatures than the MCS design. 

I wrote an article about Heat Pump Cycling and Minimum Modulation that could be of interest.

Heat Pump Cycling and Minimum Modulation

And you may also be sizing the heat pump with the hot water demand in mind, as much as the space heating, especially in a property with a small heat loss.  So again, a little oversizing is not a bad thing if you need quicker hot water recharge times.

The key takeaway then from the whole article is to not blindly accept default ACH values, but to challenge them with evidence wherever possible.

The more accurate you can get the heat loss, the better your heat pump design will be.  An accurate heat loss compared to an inflated one could well mean less radiator changes, less disruption replacing pipework and a smaller heat pump.  It could also mean higher efficiency and lower running costs as you could operate the system at lower temperatures.

I hope you found this article interesting and useful.

Why accurate heat loss matters

An accurate heat loss of the individual rooms and the whole home is key to getting the best heat pump design for your property.

Oversized heat loss can lead to unnecessary costs.  Potentially swapping radiators that don’t need swapping or ripping up floorboard to have primary pipework upgraded that might have been okay.

Using Gas data to get ball park heat loss

If you currently have a gas boiler and a smart meter, you can estimate your home’s whole-house heat loss using real energy data from cold weather days.

The heat loss survey tool takes one or more cold 24-hour gas readings, adjusts for boiler efficiency and hot water use, and converts them into a design heat loss figure (kW). 

It’s not a replacement for a proper full house and room-by-room survey by a skilled engineer, but it’s a solid, reality-based starting point.

And as the article / calculator describes, run lots of cold days data through it.  Just as I found with real heat demand data, it can get messy for a variety of external factors.

Free Heat Loss Survey Tool – What Size Heat Pump?

Conclusion – Always challenge MCS air changes (ACH)

So we have definitely proven here that my 1930 semi detached house does NOT have around 1.6 to 1.8 air changes per hour as per the MCS / CIBSE guidelines. Real world for us is around 0.6 ACH.

And the many engineers I’ve spoken to are finding the same thing as they test more and more homes.

But I am certainly NOT suggesting that everyone in a 100 year old house instantly drops their ACH to 0.6.  We have undertaken some modifications and retrofit work, so perhaps we’ve positively affected our ACH.

Your old house may well be a leaky and draughty home with ACH closer to 1.6.

Get multiple quotes / surveys

If possible, the more eyes you have on your property the better.  The more data points you have the better.  Even the gas data calculation and survey tool can help.

A friend of mine had heat loss surveys done by two reputable big name suppliers.  One heat loss came in at 7.5kW and the other at 12.5kW.  Such a massive difference.

I then helped him use the gas data calculation on his house and we came to a figure around the 7.4kW mark.  Which helped him eliminate the 12.5kW quote and supplier.

You can only assume that something went awry with the calculations that arrived at the larger 12.5kW estimate?  Either from the fabric side (wrong U values in construction elements) or the air changes values were too high. 

Were default ACH changes used?

Whenever you are having a heat loss survey done, always ask to see the calculations.  Ask what U values and ACH figures have been used.  Ask if the surveyor is using the default numbers or is using their own experience and knowledge to make a judgement.

If you (or the installer) are unsure about the ACH values chosen, think about getting professional help.  Consider a pulse test, a blower test or using the CO2 decay solution.

Test, don’t guess.

The financial outlay of the testing could well be saved in not having to upgrade radiators, have pipework replaced or buying too large a heat pump.

Heat loss surveys are estimates based on lots of assumptions.  The more assumptions you can turn into the fact, the better the outcome of the calculations will be.

Further Reading

There is a helpful section on the Open Energy Monitor heat pump guide.

https://docs.openenergymonitor.org/heatpumps/air_change_rate_calculations.html

And subsequent article by Trystan Lea.

https://www.linkedin.com/posts/trystan-lea-983019206_air-change-rate-calculation-methods-activity-7315290180293177344-eFmp/