How do you feel about this article? Help us to provide better content for you.
Thank you! Your feedback has been received.
There was a problem submitting your feedback, please try again later.
What do you think of this article?
We're developing a human-centred energy disaggregation system to empower UK households to reduce their energy consumption and save money. Through a single CLIP device, we will utilise deep learning to detect household appliances in real time and show users their device-specific electricity usage. Alongside this, we will provide personalised insights generated from the CLIP data to enable users to make informed and cost-effective choices regarding their electricity consumption. This will allow for optimised household electricity use by detecting unknown power drains and diagnosing faulty appliances.
We're facing a cost-of-living crisis. Disposable income has been falling at the highest rate the UK has ever seen and everyday costs like food and energy keep rising. Last October, National Energy Action estimated 4.5M UK households were unable to afford to heat their home. To save money, households are forced to reduce their energy use. However, they don't have the information needed to do this effectively without impacting their quality of life. Smart meters attempt to provide this information, but instead only make people anxious about their energy bills.
User Engagement and Initial Exploration
We started with consumer interviews, surveys and AB testing to answer key questions and test our assumptions. How familiar are people with smart home products? What problems do people have with smart meters? What do people really want to know about their energy usage and appliances?
This showed us the user pain points and needed product features:
1. Disaggregation: Appliance level data so people can see what is contributing to their bills.
2. Easy Install: No electrician needed, with a safe and comfortable clip.
3. Private Data: Not shared with businesses setting the cost of energy.
Prototyping, Form and UI Development
Iteration 1 - Focusing on approachable smooth curves and colour indication to match often brown live wires.
Iteration 2 - Form improvement based on feedback, including mechanism development and more realistic current transformer casing.
Hardware - Using a Heltec Wifi kit, Arduino and Matlab Thingspeak to sample current data from a house for several months. The firmware was iteratively upgraded: from 1Hz sampling to 1000Hz, from 15s interval single upload to half a second bulk upload, from only current sampling to power approximation using voltage. The graph shows the trends that can be seen each day in a household, with clear periods overnight with no use and typically larger spikes in the mornings and evenings.
UI - Experimenting with what is best to show the user: electricity saving, CO2e saving or money saving. Starting to work out how to use behavioural change research to start designing interventions including notifications, insights and appliance ranking.
The hardware prototype allowed us to test data sampling, but could not simulate giving users insights without further machine learning development. To trial with users, we instead implemented sets of smart plugs into several willing test houses and manually analysed appliance data to provide recommendations. This is where we found that one of our test houses was streaming netflix through their playstation, and by switching to a smart TV stick they will now save over £120 per year. For one of our testers, this would be an additional 7% disposable income each year!
Detailed Summary and Big Picture
CLIP is a self-installable device with an app subscription which tells you how much energy your individual household appliances are using. We provide users with live data on their energy usage and actionable insights on how to reduce it. Users simply need to clip the lightweight monitor onto the household power line by their fuse-box and log into our app. On the dashboard, they will be able to see a breakdown of their electricity consumption by appliance (e.g. fridge, washing machine, microwave etc.) and their total electricity usage, as well as money and CO2e saving comparisons. As well as the accessible no-electrician install, the provided insights will be based on behavioural change research, so people can act on the information provided. With this, they are empowered to reduce their consumption, their bills, and their impact on the environment.
Past individual benefits and granular demand side information are critical for the future energy system, especially as a higher ratio of electricity sources are renewable. The data can be used for sustainable grid management, electricity forecasting and effective tariff development. There are further opportunities in the home appliance sector, for appliance development, product marketing and market research. From an environmental perspective, scientific research indicates a potential 20% savings with disaggregated energy data which would lead to a saving of 192 kgCO2e per household per year.
Whilst prototyping continues, to bring the product to market we need to spread our focus between other business development activities. We'll update you when enough progress has been made on the hardware and software, stay tuned!
In the meantime, you can stay up to date at www.clipenergy.co.uk