- Waterloo researchers used up to 175 sensors to study how fires develop in modern residential environments.
- The research focuses on the changing behaviour of fires as buildings become more energy efficient and increasingly sealed.
- The findings could have implications for evacuation planning, fire engineering and future building safety requirements across the Middle East.
As Middle Eastern cities continue to develop increasingly energy-efficient and technologically advanced buildings, understanding how fire behaviour changes within modern building environments is becoming an increasingly important challenge for fire professionals.
Researchers at the University of Waterloo in Canada are using artificial intelligence (AI), data science and advanced mathematics to investigate that challenge, developing a framework that could eventually help predict fire behaviour and the toxic gases produced during a fire.
The research is particularly focused on modern furniture materials and increasingly airtight, energy-efficient homes. These conditions can alter how a fire develops, including the point at which a fire becomes oxygen-starved and begins producing different and potentially more toxic combustion products.
That has implications beyond the experimental burn house in which the research was conducted. As building design across the Middle East places greater emphasis on energy efficiency and controlling cooling demand, fire engineers and authorities face the broader challenge of ensuring that changes in building performance do not create unintended consequences for fire and smoke behaviour.
The University of Waterloo team conducted 15 experimental fires in a campus burn house, using up to 175 sensors to measure temperature, airflow, humidity, burn rate and numerous gases.
The resulting datasets were enormous. A single set of sensors in one location, sampling four times per second, generated millions of data points, representing only a fraction of the information collected across the experiments.
The researchers developed a system combining mathematical analysis, data science and machine-learning techniques to identify relationships within this high-dimensional dataset and understand how fires evolve.
One of the areas the system can identify is when a fire begins to become under-ventilated, or runs short of oxygen. This represents a critical change in combustion chemistry and is linked to changes in the volume and composition of toxic gases.
“We want to create smart systems that model and anticipate what a fire will do and then route people to get out of the building safely,” said Dr Beth Weckman, professor of mechanical and mechatronics engineering at Waterloo.
For Middle Eastern fire professionals, the potential application is significant. Saudi Arabia’s Building Code, for example, brings together requirements covering fire protection, ventilation, energy conservation and residential buildings, reflecting the need to consider building performance and life safety as interconnected issues.
The Waterloo researchers say their framework could ultimately support evacuation planning, inform building codes and help emergency responders better understand how fires develop and how toxic smoke is produced.
Dr Joshua Pulsipher, a chemical engineering professor at Waterloo, said the objective is to deepen understanding of fire behaviour and the gases released so that smart systems can support safer and more effective evacuations.
The team plans to extend the framework to more complex fire scenarios in future research.
The study, A framework for high-dimensional fire sensor data analysis, was recently published in the Fire Safety Journal.
