Empirical Methodologies for Improving HVAC Efficiency

September 21, 2012 - 12:00pm
Seminar Host/Point of Contact: 

This talk describes the use of empirical methodologies that we have developed for the purpose of improving heating, ventilation, and air-conditioning (HVAC) efficiency through better control algorithms and configuration.   We show that semiparametric regression can both identify simplified models of thermal HVAC dynamics while also estimating time-varying heating loads using only real-time temperature measurements from thermostats.  These models can be used with our learning-based model predictive control (LBMPC) method in order to improve the energy-efficiency of HVAC.  Experiments on testbeds with different types of HVAC show the ability of our approach to achieve large energy reductions.   As part of our efforts in comparing and quantifying the differences in energy consumption and occupant comfort of various HVAC controllers and building configurations, we present a statistical comparison methodology that utilizes nonparametric statistical methodology to rigorously perform hypothesis testing for comparisons.   We conclude the talk by discussing the design and analysis of incentive schemes for encouraging better configuration and commissioning of large buildings.  Game theory is used to show that "baselining" incentives lead to perverse outcomes, and so we propose a different incentive scheme that has the interpretation of a performance-based bonus.  A game-theoretic analysis and numerical simulation show that this may lead to large energy savings in buildings while also improving occupant comfort

Add event to Google Calendar

Announcement List

For notification about future seminars, contact Erin Bertiglia.

Site Access

Schedule subject to change without notice. If you are coming from off-site, please call first to verify. UC staff and guests are welcome. LBNL shuttle buses stop every few minutes at marked sidewalk locations along Bancroft and Hearst Avenues and Rockridge BART.