Building heat-loss calculation from your thermostat data
EcoEdge measures how fast every room in your home loses heat — automatically, from the thermostat data you already have in Home Assistant. No blower doors, no thermal cameras, no manual surveys: the measurement comes from physics your home performs every night.
The k value: your room's thermal fingerprint
When the heating switches off, a room cools toward the outdoor temperature. How fast it cools is governed by one number — the heat-loss coefficient k — in the classic Newton cooling relationship:
dT/dt ≈ −k · (T_indoor − T_outdoor)
A room with k = 0.3 loses 0.3% of its indoor–outdoor temperature difference
per hour... in plain terms: the higher the k, the leakier the envelope.
EcoEdge fits this coefficient continuously for every room, using the idle
periods between heating cycles as free experiments.
What the numbers mean
| Band | k value | What it tells you |
|---|---|---|
| Well insulated | k < 0.3 | The room holds heat — setpoints can relax further, savings headroom is large |
| Average | 0.3 – 0.6 | Typical building envelope — moderate savings headroom |
| High loss | k > 0.6 | Heat escapes fast — worth investigating insulation, drafts, or glazing |
Because k is a physical property of the room, it doubles as an objective insulation score. Comparing k across rooms shows exactly where your building bleeds energy — the attic bedroom versus the ground-floor living room — and comparing k before/after renovation shows whether the insulation work actually paid off.
How the fit works (and why you can trust it)
Two physics models compete for every room:
- RC (Newton cooling) — fits k from pure cooling periods when heating is idle or off.
- KQ — extends RC with a heating-power term, so it also learns from mixed heating/idle periods.
Each fit carries a confidence score based on how much clean data supports it, and the better-supported model wins. Weather matters, so the fit accounts for outdoor temperature (from your own outdoor sensor when available, otherwise from hourly forecast data), wind exposure, and solar gain through windows — a sunny south-facing room warms itself, and ignoring that would bias k.
Low-confidence values are shown as low-confidence, not hidden and not overstated. Every recommendation on the dashboard traces back to these measured coefficients — physics you can check, not a black box.
What heat loss unlocks
The k value is the foundation of everything else EcoEdge does: predicting how your rooms will behave over the next 24 hours of weather, recommending the lowest comfortable setpoint, and estimating what each degree of setback actually saves in your specific building rather than an industry average.
Create a free account and connect Home Assistant — after a day or two of data, every room in your home has a measured heat-loss value.