archetype

Smart-HVAC startup

Decentralized Climate Control

A room thermostat with no room temperature sensor: motorized smart vents estimate each room's temperature from duct readings.

IoTBuilding AutomationConsumer Electronics
100+
sensors across 8 zones
−75%
server latency, through memoization
+20%
device battery life, through HILHardware-in-the-loop (HIL) testing Testing real controller hardware and firmware against a simulated plant in real time, so risky, rare, or costly conditions can be exercised safely. Glossary → modeling
0003100
Living 70.0°F est – · meas – –
Kitchen 68.0°F est – · meas – –
Bedroom 66.0°F est – · meas – –
Office 71.0°F est – · meas – –
Filter life –
Tap a room · drag to set Living · estimated from duct temperature — no room sensor

  • section cut
  • supply air · vent calling
  • room satisfied
  • test sensor (validation only)
Each motorized vent becomes the room's thermostat without a room temperature sensor: the room's temperature is estimated from duct temperature, and server-side machine learning answers requests from every room at once. A test home with 100+ sensors over eight zones validated the estimate, with load and flow checked by flow hood, anemometer, thermal camera, and current clamp. Filter life is counted by airflow odometry, not runtime. The thermal model and numbers shown are a simulation.

Challenge

Turn a motorized smart vent into a room thermostat, without a room temperature sensor.

Approach

We instrumented a test home with 100+ sensors across eight zones, measuring temperature, pressure, humidity, airflow, ambient light, presence, and power. Flow hoods, anemometers, thermal cameras, and current clamps verified ACCA, ANSI, and ASHRAE load and flow calculations. From that data we built a thermodynamic model, an algorithm that estimates room temperature from duct temperature, and a distributed control system that stabilizes zones with the vents. Server-side machine learning processes control requests from many rooms at once, behind a virtual-thermostat mobile UI. API integrations reach major thermostat, ceiling-fan, smart-hub, and energy-monitoring platforms.

Highlights

  • A patented algorithm calculates remaining air-filter life from flow odometry instead of runtime estimates.
  • Server latency cut 75% with memoization; device battery life up 20% through hardware-in-the-loop kinematics modeling.
  • An interactive booth at a major consumer-electronics show demonstrated compatibility with a major smart-thermostat platform.
Ask an engineer