Innovative collaboration transforms future of fire investigations

Original Article: https://www.engineering.unsw.edu.au/news/computer-simulations-unlock-mystery-of-quakers-hill-fire

The partnership between UNSW Engineering and Fire and Rescue NSW (FRNSW) began just before Christmas 2011 – two months after the tragic fire in an aged-care facility in Quakers Hill, in Sydney’s west, that killed 14 residents and injured dozens more. Investigators discovered the arsonist, night-shift registered nurse Roger Dean, lit a fire in one room to divert the fire brigade. While this fire was extinguished quickly, Dean lit a second fire in another room that prevented the fire brigade from gaining access to the rest of the nursing home. This devastating fire burnt a large section of the facility and created a smoke layer that endangered residents and made fire-fighting efforts difficult.

Tragic fire in aged-care facility in Quakers Hill, Sydney

Tragic fire in aged-care facility in Quakers Hill, Sydney

FRNSW investigators needed to present evidence to the coronial enquiry that established how the fire had spread. That’s when Superintendent Bob Alexander, the FRNSW’s Investigation and Research Unit Operations Officer, contacted Associate Professor Guan Heng Yeoh from the UNSW School of Mechanical and Manufacturing EngineeringGuanand his team of UNSW researchers set about creating large-scale modelling and computer simulations.

Reverse engineering

FRNSW needed to establish a fire-ignition source that could completely destroy the contents of the second room. “Bob asked me ‘can you run the simulations before carrying out the experiments?’” Guan says. Over Christmas and throughout January 2012, the UNSW researchers ran computer simulations of the second room that would be used later to validate field experiments and fire modelling.

Laura Elbourne-Binns, graduate, School of Mechanical and Manufacturing Engineering

Laura Elbourne-Binns, graduate, School of Mechanical and Manufacturing Engineering

Laura Elbourne-Binns, who graduated from the School of Mechanical and Manufacturing Engineering last year, was the industry liaison for the student group. She worked with a team using Fire Dynamics Simulator (FDS) software to model ignition sources against markers based on fire debris observations and witness evidence.

“Changing the input parameters, such as fire size and intensity of the ignition source, changed the way the fire travelled,” Laura says. Using FDS to simulate the fire, students were able to “reverse engineer” the ignition to match the fire that caused the contents of the room to combust. This gave them a library of ranges from which they could work backwards to reconstruct the Quakers Hill fire.

Reconstructing the fire

Initial investigations focused on a window, which broke due to the fire’s intensity and size. Using physics to decipher forensic clues, researchers found the broken window drew in a large amount of air to create a fireball. This very quickly burnt the contents of the room.

Researchers tried various fire-source locations, and then the computer simulations were matched with experiments based on modelling. “Once again, we started from something bad and worked backwards,” says Guan.

Researchers painstakingly reconstructed the room to find the ignition source. The known contents – mattresses, couches, chairs and wardrobes – were used in the reconstruction, as well as foam, pinewood, gypsum plaster walls and steel. Various combinations were tried in an effort to match the intensity of the fire.

Model Image 1: Initial investigations focused on a window in an effort to find the ignition source.

Model Image 1: Initial investigations focused on a window in an effort to find the ignition source.

“They didn’t know how to make a model for a burning bed, so we had difficulty digging out properties of the materials to burn,” says Guan.

Adding to the challenge was the researchers didn’t know what started the fire, although the fireball theory told them the ignition source didn’t need to be big. “We started with just a match,” says Guan. “And we would tell Bob: ‘It didn’t burn!’”

In March, after an unsuccessful attempt to recreate the fire’s intensity using an ignition point close to the entry door, it was clear that one burning bed wasn’t enough. “The simulations were re-done with two beds lit using correct ignition sources,” says Guan. “That confirmed what was found in the experiments.”

The investigators finally had answers for how the Quakers Hill fire started.  The evidence was presented to the coroner and was used by the prosecution to convict Dean, and led to his admission of guilt. In 2013, he was sentenced to life imprisonment without the possibility of parole.

Meanwhile, Guan’s colleague and friend Bob Alexander received the Australian Fire Service Medal ­– the most prestigious award for a fire fighter.

Future fire modelling and training

Model Image 2: Simulations showed the rapid development of the fire, which sets the foundation for proactive fire modelling in fire-training situations.

Model Image 2: Simulations showed the rapid development of the fire, which sets the foundation for proactive fire modelling in fire-training situations.

Computer simulations showed the fire developed within 10 minutes, after which the fire was uncontrollable. This short time frame shows how little time firefighters had to extinguish the flames and deal with evacuations.

“The Quakers Hill fire was a very, very interesting case,” says Guan. “This project sets the foundation for moving fire-fighting training into the digital era.”

The researchers created a simulated environment where items in a room could burn using various input parameters. Computer simulations can now be used for proactive fire modelling in training situations or for quick diagnostics of ignition sources.

Guan is optimistic about the future of fire-source detection. “Bob has mentioned it would be his dream if we could convert one fire-fighting scenario into a virtual fire-fighting program,” says Guan. “We could let fire fighters watch this particular fire and then ask how they would respond.”

Digital fire training could save costs and lives. Current fire-fighting training facilities are small and in metropolitan areas near residential developments.

Guan says that using digital insights as a predictor would provide more reliable and realistic fire modelling. “If we can understand fires better, and create different fire scenarios, we may not need to rely on live experiments to train fire fighters.”