HeatSpace
Reads a floor plan and works out where to put heaters so a home heats evenly.
- +25%
- HVAC efficiency, simulated
- <2s
- analysis time
Hackathon team
Next.js, Flask, OpenCV, NumPy, SciPy, Redis
What it is
HeatSpace takes a floor plan and recommends where to put heaters so the space heats evenly instead of leaving cold spots. We built it as a team at a hackathon, and it won 1st place out of 35+ teams.
Why we built it
Most homes put a heater wherever there's an outlet. We wanted to see how much a smarter placement could actually save.
Architecture
- client
- service
- database or storage
- request
- A floor-plan image comes in.
- OpenCV thresholds it and finds the contours of the walls.
- The rooms become a graph: rooms are nodes, shared walls are edges.
- A diffusion solver simulates heat spreading between rooms, and the result picks the heater placement.
- The heat map is drawn in the browser.
The interesting part
The solver is a discretized Laplace equation in NumPy and SciPy, run over the room graph. Redis caches its intermediate results, so an unchanged floor plan gets its heat map back instantly. In simulation, the placements improved HVAC efficiency by 25%.
Results
- hvac efficiency
- +25% (simulated)
- analysis time
- <2s
What I'd change
I'd check the simulation against a real room. Everything we measured was simulated, which is fine for a hackathon and not much further.