Introduction¶
Climate Orchestrator turns a pile of independent TRVs and air conditioners into one coordinated whole-home climate system: each device works toward the same comfort band, from the sensor in its own room. This page is the feature tour; if you just want it running, jump to Installation.
What it does¶
- One control surface. A single
climate.climate_orchestratorentity with two setpoints (a heating edge and a cooling edge) controls all your TRVs and ACs. - Configurable presets, plus boost. Pick which presets (home/away/sleep/ boost) the thermostat offers — only those create tuning entities. Boost temporarily pushes the band in the right direction (heat or cool) and auto-reverts after a set duration.
- Manual changes are respected. Adjust a TRV or the AC directly (on the device or its own entity) and the orchestrator stands back from that device for a configurable window instead of silently fighting you.
- Area-matched sensors. Each device reads the temperature/humidity sensor on its Home Assistant area (Settings → Areas → Related sensors); a home-wide average is computed across all of them.
- Asymmetric trigger. A device engages when its area OR the home average crosses an edge, and disengages only when both are back at target.
- Comfort index. Control runs off a humidity-adjusted "feels-like" (apparent) temperature, with a dew-point guard that can run the AC's dry mode.
- Per-area offset. A per-area number can run one room warmer or cooler than the rest, biasing only that area without changing the whole-home band.
- Coordinated heat/cool. A neutral band between the two setpoints keeps heating and cooling from fighting; frost protection, window-open (with a grace delay), and outdoor-temp gating are layered guards.
- AC drive. ACs are commanded a setpoint biased below the real target so the AC's own sensor doesn't satisfy before the room does — and that bias self-tunes with integral feedback so it lands the room on target over time. Fan/swing controls are surfaced and forwarded when the AC supports them.
- Adaptive cooling comfort (opt-in). When it's genuinely hot outside, the cooling setpoint is allowed to drift gently upward (the heat edge never moves), saving energy on extreme days without chasing a setpoint you wouldn't pick anyway. Needs an outdoor sensor.
- MPC calibration (opt-in). TRVs can be driven by a learning MPC valve controller or a local-temperature offset; learned thermal parameters persist across restarts. With a weather entity, forecast preconditioning feeds the hourly forecast into the optimiser so a radiator pre-heats a room ahead of a cold spell.
- Observability. Per-device diagnostics (action, runtime, cycles, valve %,
learned MPC model), a whole-home status sensor (
initializing/ok/degraded), Repairs notices that flag silent misconfigurations (including a watchdog for devices that accept commands but ignore them), bus events for every notable transition, and self-clearing bell notifications for the important ones. - Resilient. One device or sensor going offline never takes the whole-home entity down, and a frozen ("stale") sensor is detected and dropped rather than trusted.
Requirements¶
- Home Assistant
2025.2or newer. scipy(declared in the manifest; Home Assistant installs it automatically on first setup).- Each managed device's HA area should have a temperature (and ideally humidity) sensor assigned under Related sensors; otherwise that device falls back to the home-wide average.
Project status¶
The integration is feature-complete against the design and heavily unit- and
integration-tested. The hardware-specific TRV valve/offset writes are not yet
validated on real devices, so the default target
calibration mode is the safe baseline.
Warning
As with any early-stage thermostat replacement, keep an independent fallback for heating/cooling until it has proven itself in your setup.
Next: Installation