Modeling Building Thermal Response to HVAC Zoning Virginia Smith Tamim Sookoor Kamin Whitehouse April 16, 2012 CONET Workshop (CPS Week)
Dec 24, 2015
Modeling Building Thermal Response to
HVAC ZoningVirginia SmithTamim Sookoor
Kamin Whitehouse
April 16, 2012CONET Workshop (CPS Week)
Homes are ~30% vacant
* National Academy of Science, 2006
Homes are ~30% vacant
Smart Thermostat: 28% savings--Sensys 2010
Homes are ~50% usedwhen occupied
Ongoing work:Occupancy-driven
Zoning
Ongoing work:Occupancy-driven
Zoning
Homes are ~50% usedwhen occupied
Outline
•Zoning Overview
•Coordination Approach
•Results
Outline
•Zoning Overview
•Coordination Approach
•Results
“Snap-in” Zoning Retrofit
“Snap-in” Zoning Retrofit
•Low cost
•DIY: no configuration
•Focus on forced air
•Other systems are similar
•Central Heat
•One sensor
•One heater
Snap-in ZoningZoned Heat
•K sensors
•K heaters
•K sensors
•One heater
•K+1 Control Signals
Q: When the system turns on:
Which damper configuration will achievethe desired temperature distribution?
Outline
•Zoning Overview
•Coordination Approach
•Results
Weather:• Has a large effect on temperature• Is not fully observable• Rarely repeats
Q: Can we learn the effect of dampers on temperature sensors without knowing the
weather?
T D
dTk/dt = aT + ßD
When OFF:Train a
dTk/dt = aT + ßD
When ON:Use a; Train ß
Outline
•Zoning Overview
•Coordination Approach
•Results
Experimental Approach
•Deployed zoning in a 7-room house
•7 sets of dampers
•12 thermostats
•Controlled based on occupancy
•21 days of data
Time
T
Conclusions•“Snap-in” Zoning
•Cheap, easy, & energy saving
•Coordination btwn objects is needed
•Learning is complicated by weather
•ON/OFF separates weather/system
Credits & Questions
Ginger Smith Tamim Sookoor Kamin Whitehouse