A Data-Driven and Physics-Based Framework for Integrated Energy-Air Quality (iE-AQ) Modeling

Abstract

To date, limited work has been implemented to integrate energy and air-quality models in a unified system of built environment design. This limitation can be more critical for performance evalu- ation of naturally ventilated buildings in which understanding the trade-offs and synergies between energy-saving goals and indoor air quality is of primary importance. The main objective of this research is to develop a framework for integrated Energy-Air Quality (iE-AQ) modeling to support data-informed decisions for reducing human health risks and energy consumption for the built environment. This framework hybridizes data-driven and physics-based platforms, and brings the power of artificial intelligence (AI) techniques into the conventional simulation workflows to enable a more reliable and efficient approach. The proposed framework identified spatiotem- poral factors that explain outdoor air quality variation across urban areas and localized outdoor air pollution, herein PM2.5, through the Land Use Regression (LUR) based using Gradient Boosting Machine (GBM) approach (LUGB). The proposed framework was tested on the prototype large-size office buildings provided by the U.S. Depart- ment of Energy (DOE) across the City of Chicago. The obtained R2=0.71 from LUGB suggests the power of this approach compared with the traditional LUR model with multiple linear regression (MLR) (R2=0.43) for localizing hourly outdoor PM2.5 concentrations. The variations of energy-saving potentials were obtained 6.4% to 15.6%, showing the significance of the proposed approach for evaluation of naturally ventilated buildings by generalizing outdoor conditions, not compromising human health. This research has the potential to aid designers, engineers, planners, and policymakers with a better awareness of the existing profiles of urban air quality variations across urban districts to achieve sustainable built environment goals.