Using machine vision to enhance fire safety

Esther Saula (center) works with Negar Elhami-Khorasani (left) and Juan Gustavo Salado Castillo (right) on the digitized fuel load survey methodology project.

By Peter Murphy

Published February 24, 2020

“We were able to develop a methodology to collect data that helps structural engineers define the design fuel load,” says Esther Saula, a junior civil engineering student.

Enhancing data collection in structural fire engineering

Print

The research team, led by Department of Civil, Structural and Environmental Engineering assistant professor, Negar Elhami-Khorasani and assistant professor Thomas Gernay from Johns Hopkins University established a Digitized fuel load surveying methodology using machine vision. The project was funded by the National Fire Protection Association (NFPA), and the final report is available on the Association’s website.

“The fuel load is an essential parameter in structural fire engineering,” Elhami-Khorasani says, “it quantifies the amount of energy available in a room to fuel the fire.”

As part of performance-based methodologies, the NFPA 557: Standard for Determination of Fire Loads or Use in Structural Fire Protection Design provides fuel load data for structural fire protection design. Engineers either use occupancy-based fuel load values reported in the code, or conduct surveys to collect fuel load data. According to Elhami-Khorasani, current tools for collecting new data could be enhanced.

“Fuel load surveys using current methodologies are a substantial undertaking,” Elhami-Khorasani says, “there is a limited availability of fuel load data to support occupancy-based statistics.”

The team in this project researched a new methodology to harness recent developments in mobile electronic devices, cloud storage and machine vision to efficiently complete fuel load surveys in buildings. Existing surveys rely on various methods to calculate fuel load density, including weighing the physical items, or looking for information by searching through catalogs.

This new method uses an interactive electronic surveying form, accessible on any mobile device with an internet connection, including an iPad or iPhone.

“This method gives you a way of basically taking an image of a furniture item, and finding the same item online using a search engine like Google Images to quantify the item weight,” says Saula, “one of the parameters for calculating fuel load density is the weight of the item.” The material composition of the furniture also needs to be quantified. Using reported data online or judgment of the surveyor, researchers can estimate the material composition of an item an convert weight to energy in order to define the fuel load.

Elhami-Khorasani and Gernay both conduct research on advancement of performance-based methods for structural fire engineering. According to the researchers, their goal for this new surveying methodology is to allow engineers to perform more building fuel load density surveys to improve the currently outdated database of fuel load density, a critical input for structural design of a building under fire.