An adaptive façade that balances comfort, daylight, glare and energy
A deep-learning surrogate is trained in your browser on building-performance simulation data, then deployed inside a model-predictive controller that re-optimises the shading configuration in closed loop — responding in real time to the occupant.
clock 10:00
energy 0.00 kWh
Environment
Simulated sensor readings for the operating location
Sensor · T_out
24.8 °C
Sensor · GHI
565 W/m²
Solar elevation
59.7°
Solar azimuth
237°
Surrogate model
Neural network trained in-browser on simulation data
Initialising
Operating room · east façade · Wellington NZ
Simulated sensor feedback keeps the loop closed
Shade 50% · Slat 30° · Vent 30%
Current control decision
Surrogate-predicted vs simulated response
Shade deployment50%
Slat angle30°
Vent opening30%
Objective cost0.000
Surrogate predicts—
Simulation confirmsE 512 · DGP 0.15
Indoor T21.6 °C
Last actionAwaiting control model
Occupant preferences
Express a desired indoor state — the controller models the next-best façade action
At the energy-efficient operating baseline.
Closed-loop performance
Recorded telemetry across control steps
HVAC energy — W (cooling / heating)
Work-plane illuminance — lux
Daylight glare — DGP
Thermal comfort — PMV
Actuation log
Recent control steps — observe, predict, optimise, actuate
Waiting for the first control step…