PFI-RP: Architectural design and intelligent control tools for decarbonizing space cooling and heating in buildings

University Of Oregon Eugene, Eugene OR

Investigators

Abstract

The broader impact/commercial potential of this Partnerships for Innovation - Research Partnerships (PFI-RP) project will be realized through the creation of new tools to dramatically increase the energy independence of buildings, improve their thermal survivability during power outages, as well as reduce their energy demands and greenhouse gas emissions. Space heating and cooling comprise the single greatest energy end-use in US buildings, contributing to climate change. Vigorous efforts are now underway to design, build, retrofit, and operate a new generation of green buildings. The green building sector is one of the fastest-growing markets in the US. Recent work has shown that well-designed, intelligently controlled ‘dynamic passive’ systems, which use shading, natural ventilation, and night insulation to recruit climatic resources such as solar heat, cool air, and cold night skies, can reduce space heating and cooling loads extensively across US. These dynamic passive systems contribute to the urgently needed gas emission reduction effort and provide resilience to heatwaves and cold snaps. Although the value of these potentially transformative systems is recognized, effective tools for optimizing their design and control are lacking. This project will, therefore, develop the needed tools, by creating new marketable design software, design services, and system control hardware for dynamic passive systems. The proposed project will address the complexity of dynamic passive systems design and control generated by the diversity of climatic resources, e.g., solar heat, cool air, and cold night skies, in different US regions. All US climates possess extensive resources for passive heating and cooling in different seasons, presenting valuable opportunities nationally. At the same time, these systems require climate-specific materials, configurations, and control logics. The first project goal is to develop a practical design tool for specifying system components. This goal will be accomplished by integrating structured, prioritized numerical optimization procedures with existing building energy modeling programs. The second goal is to develop a robust, practical control tool for operating the systems’ dynamic elements. This goal will be accomplished by using deep reinforcement learning methods to train and control climate-, weather-, and occupant-responsive agents. The resulting policies will be used both in design, to inform configuration refinements, and in operation, to alert occupants in manual systems or to signal actuators in automatic systems. The project will generate proof-of-concept and prototypes. The prototypes will be evaluated as the tools’ abilities to reduce heating and cooling loads in simulations and testbeds. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.