While one goal was to create a framework that other researchers could replicate and build upon, another was to help the City of Pittsburgh meet its ambitious energy goals. "We wanted to determine if an urban model for commercial buildings could be accurate based on acceptable errors, and it was." They're bigger and have more diverse uses," explained Mohammadiziazi. "A lot of good work has already been done in this field, but there are fewer studies focusing on commercial buildings, because data about them is more difficult to capture than residential buildings.
The focus on commercial buildings, as well, was an important addition to the field of research. Though it's currently mostly guided by the researchers looking at the images, the researchers hope that this modeling framework can eventually take advantage of machine learning to more quickly analyze and categorize building images. When they validated their findings using other publicly available data, they had just a 7 percent error rate.
Geological Survey to determine building height, the researchers were able to simulate and map the annual energy use intensity of 209 structures in Pittsburgh. With street-level images to determine the building material, window-to-wall ratio, and number of floors, and LiDAR data from the U.S. The researchers used publicly available Geographic Information System (GIS) data and street-level images to develop their UBEM, then created 20 archetypes of buildings that comprised eight commercial use types.The buildings were sorted into the groups based on categories including use type and construction period. Our hope is that by using image processing, we can build a framework that reduces some assumptions." "Researchers need to rely on assumptions based on when buildings were built or what the mechanical and electrical systems look like. It's cumbersome or even impossible to find and process detailed information about hundreds or thousands of buildings in an urban environment," said Rezvan Mohammadiziazi, lead author and graduate student in the Swanson School's Department of Civil and Environmental Engineering. "We found that in the existing literature, the scale of commercial buildings was always one of the challenges. Their findings were recently published in the journal Energy & Buildings.
While other models may be hindered by a scarcity of data in public records, the researchers' Urban Building Energy Model (UBEM) uses street-level images to categorize and estimate commercial buildings' energy use. Researchers at the University of Pittsburgh Swanson School of Engineering and the Mascaro Center for Sustainable Innovation used the City of Pittsburgh to create a model built upon the design, materials and purpose of commercial buildings to estimate their energy usage and emissions.