Getting Hands on with What Affects Air Quality

As the inquisitive kind of person I am, a few years ago I started to build my own portable air quality sensors, to look for two of the most important pollutant groups, particulate matter that is known as PM2.5 and PM10.

I was keen to understand why I felt there to be a difference in what I could sense in the air around me, and what was being shown on the various online air quality tools I was consulting.

Related: 8 million people will die from this – how good is your air quality?
(includes links to online Air quality tools)

My conclusion was that while the models and tools that were being used to generate these wide-area estimations of air quality were impressive and incredibly intricate, they couldn’t possibly take into consideration hyper-local events and activities that would have a significant impact on the local air quality.

My findings: Those trendy wood burning stoves aren’t so great for air quality, especially in quaint villages. Read my findings below.

For example – how could a data model know that my neighbour has just installed a wood burner and now heats his home burning wood instead of gas central heating? Could the model know that two doors down on the other side of the street they are having their driveway dug up, or even that the local fitness centre’s car park acts as a drop off/pick up location for the school, and since they closed the car park, there is a lot more slow moving traffic?

Of course not. How could a model that is looking at air quality across the country, county or city take those micro factors into consideration? But hang, aren’t those the most important factors for you and me? The actual air quality in the places I spend my time, not the predicted for the wider area I happen to be a part of.

What I wanted to achieve

I wanted a portable device that would measure particular matter pollution automatically, log it with time, date and location in a way that could be mapped and then compared against the reference data available on both local government and other third party tools.

There are many off-the-shelf devices I could purchase if I wanted to shell out thousands of pounds, but quite honestly, I didn’t really want to do that.

Then there were a number of other, cheaper devices that I could have picked up for a few hundred pounds – but these often tended to be intended for static locations. Even still, many of these would have required other devices to log location, let alone doing all of this automatically, for hours at a time without human intervention.

So the answer was obvious, I needed to build something myself. (Ok, maybe again if I am completely honest I set out with the intention of building one anyway and used the price tags as a reason why not to buy 😉 )

Indicative not reference quality

It became quite clear, quite quickly that there is a reason why these handheld devices, along with the static monitoring stations that local councils deploy (see here) are so expensive.

Measuring air quality is tricky and in order to be able to compare apples to apples, not only are the things you are measuring very small so you need high precision equipment, but you also need to eliminate many of the environmental factors that cause variance in accuracy and readings. In particular, controlling temperature and humidity is incredibly important.

With this in mind, I decided what I wanted to do is create something that would give me indicative information that would allow me to understand whether my environment was kind-of-ok, kind-of-bad, getting worse, or getting better. Making this decision was good because it helped me take the cost down from potentially thousands to less than £100.

What did I build?

I started the project early 2018, and by May 2018 I was ready to post a snippet of what I was doing on Twitter.

This weekends project is a #raspberrypi connecting to #android for #gps as another #sensor that’ll I’ll keep secret for now 🙂 pic.twitter.com/ZkXrXigvOj

— 5c0tt (@5c0tt) May 26, 2018

It wasn’t that it took that long to build. It took that long to find and test components, whilst maintaining my day job and family commitments.

I built a simple system using a Raspberry Pi (£9), and old android phone that had GPS (kind of free) and a particulate matter laser-scatter sensor (an “SDS-011” for about £15). I then put all of that in a box, with a big fat tube poking out of it, added a solar powered battery pack (£10) and it was good to go. Later on I added a couple of buttons and an LCD read out to the box so I could see what was going on without connecting to the in-built web server to check out the readings.

The whole thing was connected to the Internet via my mobile phone, and would send the data at a frequent intervals to another application I built on my server that then ploted the data using Google Maps.

Fast forward quite a few months of writing code, swearing a lot (writing bus interfaces for the USB controller was not something I was ever very patient with), much trial and error and I had a portable, Internet connected device that would take air samples every 2 minutes, log them against date, time and position, and would last for up to 10 hours.

My first trials – social awkwardness

This thing was portable but a little bulky.

It had four or five wires poking out of it, a couple of big red and yellow buttons and a long tube. I loaded it into a rucksack and thought I’d get some readings whilst picking the kids up from school and taking them to the park.

I didn’t get too far before I realised my mistake.

Here I was walking around with a rucksack that had wires and tubes hanging out of it. Every now and then I would stop, take it off, open it up, take out the black box, and press a red or yellow button. I would then stare at the screen for a bit, look around (I was trying to see if there were any visual clues for the air quality measurements) then put it back in the rucksack and carry on walking.

It wasn’t until I took off the rucksack and left it on a bench in the park, that I turned round to see this very dubious package of wires and tubes that I realised how alarming my behaviour might have been.

Covering more ground

Whilst I wanted to have a system that could be walked, or even cycled around, I realised I needed to cover more ground (and not be carrying around a dodgy looking bag), so I set about installing this in my car.

So that’s what I did, and I drove hundreds of miles taking air quality samples all the way. Below is just one example of the map and data I generated.

What did I learn ?

Primarily I learnt that my rudimentary device worked, but had limitations. The biggest limitation was to do with environmental variations and calibration. This is something I could fix with time and investment, but for now, this was the biggest obstacle to furthering my plans.

I also learnt more about what affects air quality, and the local variations compared to the more broad reaching data models.

So what affects air quality?

I discovered that local events really did have an impact on the immediate air quality I was experiencing. Some of the factors I noticed I already covered in the intro above – neighbours burning wood to heat their homes, congestion near schools and shops, construction, and mud on the roads.

Additionally, I noticed that air quality tended to get worse at pinch-points. For example, whilst driving through some lovely quaint villages I would find high streets with multiple road junctions, pedestrian crossings and traffic calming (speed bumps, chicanes etc). These pinch points seemed to funnel and concentrate pollution along the high street, and the road features meant cars were slowing down, turning, braking, stopping, and starting – all of which create particulate matter.

My take away is that air quality models are valuable, but local and hyper local data points need to be captured to make these valuable and accurate.

It’s 04:34, Weather data says Light level is 0.00 lux, the temperature is 11.0C. The remainder of the hour will be Possible Drizzle and 11C, changing to Drizzle, and 11C. Overnight temperature will drop to 3.7C. #AQI PM2.5 is currently ‘Good’ with a value of 25.0 [6.3ug].

— DIY Smart Home (@SmartHomeDiy) February 15, 2020

Ongoing project

My little air quality measurement device is still up and running, I have it hooked up to my house, which tweets about this and the status of other IoT contraptions I’ve built. It’s a riveting read, so please do follow it @SmartHomeDiy.

More to come

Next I will cover the latest device I am using and my findings from that. I’ll be running my new gadget 24 x 7, taking it out walking, at home, on trains, planes and automobiles to learn more about what affects air quality. Find out more and keep up to date by subscribing using the form below.

Don't Miss Out.

Join 1,000s of others, sign up for alerts. No spam! Just interesting tech, doing good.

Leave this field empty if you're human: