Track: HVAC&R Systems and Equipment
Sponsor: 6.7 MTG.OBB Occupant Behavior in Buildings
Chair: Da Yan, Tsinghua University, Beijing, China
Occupant behavior is one of the major drivers of energy consumption in buildings, yet there is currently little integration of occupancy-estimation and feedback control systems. These savings can be achieved through occupant-based, operation or retrofit strategies. Accurate predictions of occupant behavior are needed to inform MPC algorithms to improve their efficacy. Conversely, the model used within an MPC controller can be used to test the energy (and peak power) implications of different occupant behavioral scenarios, and use this insight to inform the occupant about how to better interact with the building systems.

1. Occupant-Integrated Model Predictive Control of Building HVAC Systems: Benefits, Drawbacks and Challenges
David Blum, Ph.D., Associate Member, LBNL, Berkeley, CA
Within the last decades, needs for building control systems that reduce cost, energy, peak demand and that facilitate building-grid integration, district-energy system optimization and occupant interaction have come about. Model Predictive Control (MPC) is a control technique that utilizes system models and forecasts to predict performance and optimize control inputs in real-time. This presentation discusses in detail the ways occupant interaction with MPC-controlled building systems can occur, particularly as related to the control of HVAC systems, including benefits, drawbacks and challenges. This presentation discusses progress on current work that is exploring and implementing these interactions in demonstration buildings.

2. Behavior Driven Model Predictive Controls for Future Smart Buildings
Bing Dong, Associate Member, University of Texas at San Antonio, San Antonio, TX
This presentation reviews current occupancy behavior (OB) based MPC control projects for smart building at the University of Texas at San Antonio. Challenges and opportunities of OB-MPC for smart buildings are presented and discussed, particularly on models to use, occupancy data and optimization algorithms. Simulation results show that OB-MPC can achieve up to 24% energy cost reduction in residential buildings and 17% in commercial buildings.

3. Fault and Occupant Tolerant Model Predictive Control of Building HVAC System
Pengfei Li, Ph.D., United Technologies Research Center, East Hartford, CT
This presentation covers the collaborative research work from cross-functional team effort behind the journal paper that was recently honored with ASHRAE Research Journal Best Paper of the Year Award. The development and application of a fault and occupant tolerant control technology, its online implementation, and results from several tests conducted for a large-sized HVAC system are discussed. The performance and limitations of the fault detection and diagnosis, model predictive control as well as the fault and occupant tolerant control algorithms are illustrated and discussed using measurement data recorded from multiple field tests.

4. The Combination and Application of Model Predictive Control and Occupant Behavior
Da Yan, Tsinghua University, Beijing, China
MPC is a new approach to controlling building systems to optimize equipment operation. Meanwhile, occupant behavior is a key contributor to the uncertainty of energy consumption. The combination of MPC and OB would foster great building energy saving potential. The integrated network to measure occupancy and behavior related with environmental parameters is introduced. With the measured data, the occupant behavior predicting models will be built and integrated with the building system model to improve the control logic. A demonstration of the predictive model application is presented to see the effect of occupant behavior based model predictive control.

Presented: June 25, 2017, 11:00 AM-12:30 PM
Run Time: 90 min.

This is a zip file that consists of PowerPoint slides synchronized with the audio-recording of the speaker (recorded presentation), PDF files of the slides, and audio only (mp3) for each presentation.