How can environmental sensors help us understand our homes’ current thermal performance, influence retrofit decisions and quantify the benefits of individual measures? Here Matt Fawcett explores how the use of low-cost sensors in combination with energy usage and weather data can inform our retrofit journey.
Each month Carbon Coop facilitates EcoHome Lab – an informal meet-up of environmental tech and data enthusiasts – below is a write-up of our April session on installing low-cost environmental sensors alongside some thoughts on it’s implications. This guide is separated into three sections, the first on the value of home monitoring and the last where next are suitable for a general audience, while the main setup section requires some technical knowledge (though we’ll link throughout to further guides).
Understanding your home’s thermal performance
Through the roll-out of smart meters we have a better understanding than ever before of how much energy we are using to heat our homes, but without environmental sensing it is hard to quantify the impact of that energy usage. Paired with constant readings of the temperature in each room of our home, we can easily see not just how much we increased the temperature but also how long that increase was maintained.
While we can use modelling to estimate this, we know that all modelling is error prone, especially where we have incomplete data. For example when assessing existing glazing, if the specifics (in terms of coatings, spacers and particular gas filling etc) are unknown, the worse-case scenario must be modelled – this might therefore lead us to over-estimate the potential energy savings from glazing replacement. This difference between these modelled and actual savings is called the ‘performance gap’ and it can be significant, particularly in projects using innovative building materials and heating systems or requiring complex detailing. We hope that by base-lining the pre-works energy performance, we can improve modelling and refine our recommendations.
Similarly monitoring of post-works can be equally important in closing the gap, as we know that looking at bills alone is insufficient in assessing whether the promised performance improvements have been achieved. This is due in part to ‘comfort take-back’, where residents heat their homes to a higher temperature after retrofit than before, as having the heating on is no longer so costly or guilt inducing. Real time monitoring can help us diagnose which elements of any gap are down to design and construction, which to handover and operation error and which to residents simply opting to live more comfortably.
Creating healthier homes
Indoor Air Quality (IAQ) is increasingly recognised as a serious health concern. We spend up to 90% of our time indoors and increases in the airtightness of buildings risks increasing the build up of indoor pollutants. While it is possible to buy monitoring for the full range of environmental pollutants, such sensors are often extremely costly, so in this case we will use humidity as a proxy. This is not only because humidity monitoring helps identify inadequate ventilation but because humidity exacerbates a whole range of air pollutant sources.
Impact of Relative Humidity on Indoor Air Quality
For more on air quality and the importance of addressing ventilation as part of the retrofit process look out for our semi-regular ‘Ventilation for Householders/Home Health Check’ workshops. A recording of the most recent Home Health Check webinar is available here. So enough about the why, let’s look at the how…
An example setup
I won’t add step-by-step instructions and configuration files here, as such things rapidly date, rather I’ll refer you to the up-stream guides and resources I used for each step.
Mounting guidelines for sensors
The guidelines below are for indoor environmental and air-quality sensors, they are taken from the Siemens Sensor Installation Guidebook which also covers most other sensor types.
- Mount sensors in rooms at a height of approx. 1.5 m and a distance of at least 50 cm from the nearest wall
- Do not expose to direct sunlight
- Do not mount on external walls
- Do not place in alcoves or on shelves
- Avoid locations near to air flows and heat sources
- To get accurate per-room readings keep interconnecting doors closed
- Do not cover with curtains
The sensor: Xiaomi Aqara
The sensor we’ve been using is the Aqara Temperature and Humidity sensor (Model: WSDCGQ11LM, data-sheet), which incidentally also monitors atmospheric pressure. According to the specifications they are accurate to within ±0.3℃ and 3% relative humidity, not good enough for the SpaceX but probably more accurate than the average domestic heating thermostat. I ran the sensors in a row, next to my Remeha iSense thermostat, for 10 days and all remained within that accuracy range.
The Aqara hardware uses the Zigbee communication protocol which is also used by other home automation products such as Philips Hue, Samsung SmartThings and the IKEA range of smart devices. There are a number of protocols competing to be the ‘smart home’ protocol and I won’t go into the debates here, other than to say that at present Zigbee seems to strike a good balance between range, reliability and affordability. Zigbee is a ‘mesh’ protocol meaning that in theory each device on the network can act as a repeater for other devices, extending range and reliability. In practice, this is only for mains-powered devices such as wall sockets or light bulbs rather than our battery powered sensors and devices from different manufacturers aren’t always compatible. I’ve had no range issues but other users report that the IKEA range of bulbs and sockets work well as repeaters for these sensors.
Why they’re a good start
- Discrete – not normally my first criteria when selecting tech, but these things don’t look out of place even in the most minimalist living space.
- Good battery life – at least a year on the replaceable CR2030 battery and as battery charge level is also reported, it’s easy to plan a time for replacement.
- Cheap – especially when purchased directly from China – I got them here on Aliexpress at under £10 each (delivered).
- Popular – meaning that they are easy to source and that many different home monitoring and management systems have documentation or forum posts on integrating them.
The hub for data collection
Xiaomi produce their own Aqara home hub and it’s also possible to link the sensors with a Samsung SmartThings hub (guide here) and I’m sure many other propitiatory hardware. If you’re already running one of these existing systems, you should use that. If you’re starting out, we tend to recommend local-network solutions where a loss of internet connection doesn’t mean a loss of data and you don’t have to rely on someone else’s server to get to your own information.
Home Assistant as a local-network hub: The software I’m running is Home Assistant (a popular open-source home automation system), as I already had it running at home, it gives me local control and it’s easy to make additions to. As with much at EcoHome Lab the centre of this monitoring system is the Raspberry Pi single board computer, if you don’t already have a pi running at home (or sat in a draw), you can get a full kit of the latest Pi with everything you need for under £60. If you’d like to follow the same route follow these installation instructions.
You need a ‘gateway’ to allow the pi to speak to the Zigbee network, either in the form of a USB stick or extension board that plugs directly onto the headers of the Pi. There are a number of options available, the cheapest being the CC2531 zigbee sniffer that you can pickup for two or three quid (see Zigbee2mqtt for info on going this route). I chose the more expensive Conbee II solution from Phoscon.de at around £35, which I’ve found to have very good range and support.
Necessary Home Assistant Add-ons
deCONZ – A visual GUI for managing zigbee devices
The deCONZ add-on gives you a simple interface for managing your sensors and other zigbee devices, with permission it can also automatically add all the sensors as entities in in Home Assistant. While what you see in the add-on is actually the Phoscon app, the full deCONZ software is also installed but is only accessible via VNC. Most users probably won’t need this but it is an impressive bit of software, especially for troubleshooting, mapping how Zigbee devices are meshed, showing (and auto-updating) firmware version etc.
InfluxDB for long term data storage
Great, so we’ve got all the data visible on our dashboard, we can view the data as graph for the past week, but we can’t go back beyond that, as by default Home Assistant will purge the data every 10 days. Home Assistant’s default database is about enabling home control rather than long-term data logging, for that task we install InfluxDB (another core add-on), a performant time-series database.
Grafana for data visualisation
This is an optional step which integrates very well with InfluxDB to allow you to create beautiful Dashboard. It’s an endlessly powerful piece of software for data visualisation, but is quite easy to get started with, for example it takes a couple of clicks you can graph all sensors with values in °C. There are a wealth of tutorials to take it further.
Adding context – outdoor environment data
In order to give context to the environmental data inside my home it is vital to get accurate data for the conditions outside, primarily temperature and humidity but also other areas which might have an impact on internal conditions such as rainfall, wind speed and direction. There are a plethora of high accuracy data sources for current conditions available for free, in this case I chose the Met Office, as I’m traditional like that.
In order to pull that automatically into Home Assistant you need to setup a free account with the Met Office Datapoint, to get an API key and then copy it alongside the Met Office Sensor integration settings into your configuration.yaml file.
On reloading my Home Assistant, the sensor feeds above popped up on my dashboard. Each feed updates every 30 minutes which is fine for our purposes, if you were going deep and trying to use wind-speed measurements to estimate airtightness you might need to pay for higher resolution data. Home Assistant uses your postcode to find the most appropriate weather station, in my case ‘Manchester Youth Hostel‘ less than a mile from me and at the same elevation (elevation checker).
Personally: like most Carbon Coop members I’m interested in trying to minimising my carbon emissions while maximising my comfort, but homes (especially older ones) are complex systems and in the past I’ve found it hard to isolate the impact of specific changes. I have my home monitoring system connected to an OpenTherm Gateway, so I can log what my heating and hot water is doing and how efficiently they are operating. By collecting half hourly data for energy usage, boiler activity and external and internal temperatures I hope to be able to optimise my home and answer the following questions.
- Will the air-tightness works I’m currently doing necessitate mechanical ventilation?
- Boiler optimisation: what further energy savings are possible by enabling weather compensation, reducing cycling or flow temperature?
- Is the remote control user-friendly enough to move to on-demand heating for the evening rather than scheduling heating regardless of occupancy, what will the impact be?
- Are we nearly there yet? I’ve been on this ‘carbon decent’ journey for almost two decades, how’s it actually going?
Professionally: I’m interested in how far this type of constant year on year monitoring can help us to provide definitive answers to controversial retrofit debates, refine our approaches and ensure the best outcome for our members. Some initial questions.
- What role is there for the atmospheric pressure data, might it help us to identify airtightness issues or to balance input/extract ventilation?
- What role is there for monitoring as part of quality assurance? What about energy performance contracts?
- Will having monitoring in place (to spot unintended consequences) enable us to roll out more innovative solutions or speed up the process of iterative improvement?
- Once we have the infrastructure for sensing and control in place, what else might become possible?