Inspiration

The cheapest, cleanest electricity of the day usually arrives in the middle of the night, when wind is strong and nobody is using it. By 5pm the grid leans on gas, and the same power costs more than twice as much. On 13 November it went from 29.1p at 3:30am to 61.9p at 5pm. Households are paying more and polluting more simply because their appliances run at the wrong time. We wanted to fix that without asking anyone to set alarms or change their habits. Plug things in as normal, and GreenSwitch quietly waits for the right moment.

What it does

GreenSwitch is a smart switch that decides when your appliances should run. You pick a mode on your phone: Green for the cleanest hours, Cost saver for the cheapest, or Urgent for power right now. Then you say when it needs to be done, and GreenSwitch switches the power on at the best times. If you plug in a phone or laptop charger at a bad time, it pauses it and charges it later. A clamp sensor measures what each appliance really uses, so you can see exactly what you spent and what you could have saved. In our sample data, an EV charge cost £6.29 instead of £12.68, with about 79% less CO₂.

How we built it

We used both halves of the Arduino UNO Q. The Linux side runs Python: it downloads live carbon data from National Grid and prices from Octopus Energy, plans the best half-hours, and serves the web page. The microcontroller reads the clamp sensor and switches a relay connected to an extension strip. The two halves talk to each other through the Arduino Bridge.

Challenges we ran into

Our biggest puzzle was that no sensor can tell whether electricity came from a wind farm or a gas plant, because the electrons are identical. We solved it by pairing what we could measure, exactly how much power an appliance draws, with what National Grid knows: how clean the grid is every half-hour.

Next came a relay that wouldn’t let go. Even when we told it to switch off, it kept leaking enough current to stay on. We worked out that this relay is active-low, so the only reliable way to cut power was to switch the pin into a high-impedance state. After that it finally clicked off cleanly.

Getting honest numbers from the clamp was harder than we expected. The raw readings drifted with electrical noise, and a single spike could make a kettle look like it had finished, or an idle socket look like it was drawing power. We calibrated the sensor against real loads, set a noise floor to zero out phantom readings, and added a filter that ignores one-off spikes.

Accomplishments that we're proud of

GreenSwitch isn’t a mock-up. It reads live grid data, measures a real kettle at 2.5 kW, and physically switches a real extension strip on and off. In our tests it cut the cost of an EV charge almost in half and its carbon by about 79%, just by waiting a few hours. We’re also proud that anyone can use it. The app tells you in plain English whether now is a good time (“Clean and cheap right now”), and the smart part happens automatically.

What we learned

We learned how much electricity prices and carbon change through the day, and how closely the expensive hours match the dirty ones. We also learned how to split work between a computer and a microcontroller on one board, and that setup problems can take as long as the code itself.

What's next for GreenSwitch

We want to build a certified smart-plug version for real homes. We also want it to recognise appliances from how they use power and to set daily budgets. A camera could spot when an EV parks and start its charge automatically.

Built With

Share this project:

Updates

Submission history