Smart Buildings for Smart Grid Bing Dong, Ph.D. Assistant Professor Department of Mechanical Engineering University of Texas at San Antonio, USA
Where are Our Energy Consumed? 70% U.S. Energy Bill for Buildings: $410 billion per year
Elements Interact with Buildings Climate weather CO 2 emission water energy Equipment heating cooling ventilation lighting Social Occupancy IEQ Productivity Energy Systems generator smart grids renewable Design site architecture facade Transportation subway bus EV Image Credit: Untied Technologies Center
BMS networks Zonal temperature controls Economizer Cooling staging Heating staging Indoor fan controls Performance Monitoring Energy Diagnostics System Prognostics Cost Structure Materials Installation Commissioning Operation Built Environment Science and Technology (BEST) Lab HVAC Cost Power Sensor Virtual Sensors Energy Diagnostics System Prognostics oposed Reduction Anomaly scorescore Diagnostics Flag Flag Fault Fault Alarm Alarm Energy Impact Equipment Fault Detection & Diag. Low cost, Diagnostic and Prognostics System Dynamic Life Cycle Assessment 60% 80% 30% 40% System Health Index System Slow Degradation Degradation Detected Time Utility/Smart City Artificial Intelligen ce Multi- Physics Modeling Sustainable Building Design HVAC Control Behavior Science Advanced Sensing Infrastructure Modeling Occupancy Behavior Energy Measurement & Verification
Smart Building Load Forecasting Illustration of energy savings for retrofit Sponsor: Collaborator: Dr. Zheng O Neill
Smart Building Load Forecasting NMBE =4.3% Prediction of hot water energy consumption for an office building based on data-driven models Hourly Gaussian Mixture Model (GMM) regression with 95% confidence interval Hourly model prediction vs measurement on time horizon NMBE: Normalized Mean Bias Error R 2 =0.9 *Dong et al. Comparisons of Inverse Modeling Approaches for Predicting Building Energy Performance, Building and Environment. 2015
monthly Total KWh Smart Building Load Forecasting Physics-based Energy Model based on EnergyPlus 140000 120000 100000 80000 60000 Model Bill 40000 20000 0 1 2 3 4 5 6 7 8 9 10 11 12 Sponsor:
Smart Building Load Forecasting Sponsor: Collaborators: Dr. Les Shepard, Dr. Rolando Vega, Dr. Hongjie Xie Students: Zhaoxuan LI, Gaelen Mcfadden, S M Rahmam, Tuan Le
Smart Building Load Forecasting Hybrid load forecasting model for residential building integrating physics and machine learning techniques improves forecasting accuracy by 10%
Smart Building Load Forecasting MAPE=3.1% + 15 Min ahead Load Forecasting MAPE=2.23% 15 Min ahead Solar PV Forecasting
Smart Building Load Forecasting UTSA Building Footprint from LiDAR data Total power estimation (in MW) produced from all rooftops LiDAR: Light Detection and Ranging
Smart Building Load Forecasting LiDAR-based solar map of a neighborhood UTSA Building Footprint from LiDAR data Total power estimation (in MW) produced from all rooftops LiDAR: Light Detection and Ranging
Occupancy Behavior in Smart Buildings Sponsor: Source: Natural Resources Canada Student: Zhaoxuan Li
IEA EBC Annex 66 Definition and Simulation of Occupant Behavior in Buildings. 22 countries and regions
Occupancy Behavior in Smart Buildings-- UTSA Test-Bed Those four houses are one floor, three bedrooms and 1,100 ft 2 each. Each house is built in one kind of material (Wood, ACC, ship container and SIP) 15 different power measurement channels
April Energy (kwh/m 2 ) Occupancy Behavior in Smart Buildings-- Motivation 12.00 10.00 8.00 6.00 4.00 2.00 0.00 SIP Conventional AAC Container DOE Benchmark Model Number of Occupants 2 4 4 1 3
Behavior 1: Interactions with Thermostat 85 80 Stick AAC Container SIP Temperature(F) 75 70 65 60 MON TUE WED THU FRI SAT SUN Typical Weekly Thermostat Settings for Four Houses
Behavior 2: Usage of Major Appliances
Behavior 3: Occupancy Presence 1 Month Occupancy Presence in Living Room
Living Room 1 0 Modeling Occupancy Behavior in Smart Buildings Markov Models for Day Ahead Occupancy Status Prediction For x = s 1,, s n the state vector, prediction is: x k+1 = P ij x k. Where P ij = P(x k+1 = s j x k = s i ) Observation Prediction 1 Mon Tue Wed Thu Fri Sat Sun Time
Building Grid Integration Behavior Driven Transactive Energy for Residential Buildings Vol ron Agents Outside building nodes Transac ons Devices Weather U lity Company Dynamic Energy Price Neighboring residen al Building Produced and saved Energy Price and Availability Sta s cal Occupancy Behavior Model Building RC Model HVAC Model SMPC Residen al building Vol ron message bus Driver 1 Driver 2 Driver 3 Driver 3 Sponsor: Electric Vehicle Plug Ligh ng HVAC Collaborators: Dr. Les Shepard, Dr. Rolando Vega Photovoltaic Panels Student: Amin Mlrakhorli Ba eries Building Produced and saved Energy to sell
Existing Controller Utilize AC energy use based on electricity price Grid signal Utilize building mass as Thermal energy storage Optimal pre heating before occupancy Model Predictive Controller(MPC) Occupancy 10% Energy saving Method Energy (Wh) Bill ($) Bang Bang 8781 3.62 MPC minimizing Energy 8502 3.55 MPC Minimizing Bill with Occupancy Prediction 7969 (10% reduction) 3.34 (10% reduction) MPC+Occupancy+Grid Signal
3.8 2.5 1586 2300 7737 8200 MWh er CO 2 Energy 10.5 MTons of CO 2 Energy Lighting Computation CFD al Design Decision Acoustics Support Control Dynamic Life-cycle Building Information Model Economy Design Construction Commissioning Acknowledgement Occupancy Detection Computation al Operation Decision Support Renewabl e Energy Operation MPC Performance Benchmark Data Visualization Eco City 6.9 5335 5654 HVAC Equip. Construction Management Solar House 1 Retail Box 3 Intelligent UTSA AET Workplace,CMU Building 2 Mix-use Marriott Hotel Commercial 4 High-rise NREL Residential Microgrid 5 TBP Design and Procurement Structural System Envelope System BUILDING DELIVERY PROCESS Interior System Mech & Elec Systems External/Landscape Framework ensures comprehensive and integrative design consideration for high performance and sustainable solutions Knowledge-based Design Total Building Performance Mandates Spatial Thermal Indoor Air Quality Visual Acoustical Building Integrity Physiological Psychological Sociological Economical Diagnostic methodology for measuring building performance, evaluating impacts and establishing benchmarks Knowledge Production Total Building Performance TBP Commissioning and POE Our Sponsors 0.004k 0.8k 7k 140k 152k 250k 100M ft 2