Visual analysis of firebrand generation from a large mass timber compartment fire
File(s)
Published version
Author(s)
Mitchell, Harry
Geng, Yingfan
Amin, Rikesh
Kotsovinos, Panagiotis
Rein, Guillermo
Type
Journal Article
Abstract
Firebrands, the solid burning fragments transported by fire and wind, are a key mechanism of wildfire spread.
Firebrands are also generated by building fires, but are not typically considered in building fire protection,
and can be a more prevalent hazard in mass timber buildings. This paper presents the first study of firebrands
generated from a large mass timber compartment fire (CodeRed #01) and ejected from openings. Firebrand
location and velocity was studied using particle Streak Velocimetry. Commercial cameras were used as a simple
and field-deployable alternative to more complex approaches, whilst also providing a method of approximating
the gas flow-field. Over 15600 firebrands were ejected from one opening over 17 min, reaching up to 300
firebrands in one instance. Maximum firebrand velocities of 14 𝑚 𝑠−1 and mean velocities between 2 − 6 𝑚 𝑠−1
were observed. Firebrands had greater velocities along the external venting flame centreline, while the peak
of firebrand number densities were at the outer region of the external venting flame. Computer vision was
trained and validated to rapidly and automatically detect firebrands whilst minimising false negatives and false
positives. This paper finds that firebrands can be generated in significant quantities from mass timber building
fires, posing a significant hazard to nearby structures
Firebrands are also generated by building fires, but are not typically considered in building fire protection,
and can be a more prevalent hazard in mass timber buildings. This paper presents the first study of firebrands
generated from a large mass timber compartment fire (CodeRed #01) and ejected from openings. Firebrand
location and velocity was studied using particle Streak Velocimetry. Commercial cameras were used as a simple
and field-deployable alternative to more complex approaches, whilst also providing a method of approximating
the gas flow-field. Over 15600 firebrands were ejected from one opening over 17 min, reaching up to 300
firebrands in one instance. Maximum firebrand velocities of 14 𝑚 𝑠−1 and mean velocities between 2 − 6 𝑚 𝑠−1
were observed. Firebrands had greater velocities along the external venting flame centreline, while the peak
of firebrand number densities were at the outer region of the external venting flame. Computer vision was
trained and validated to rapidly and automatically detect firebrands whilst minimising false negatives and false
positives. This paper finds that firebrands can be generated in significant quantities from mass timber building
fires, posing a significant hazard to nearby structures
Date Issued
2025-10-01
Date Acceptance
2025-06-18
Citation
Fire safety journal, 2025, 156
ISSN
0379-7112
Publisher
Elsevier
Journal / Book Title
Fire safety journal
Volume
156
Copyright Statement
© 2025 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
License URL
Subjects
Compartment fire
Computer vision
Engineering
Engineering, Civil
Firebrands
Mass timber
Materials Science
Materials Science, Multidisciplinary
Particle streak velocimetry
PARTICLE TRACKING
Science & Technology
Technology
Publication Status
Published
Article Number
104458
Date Publish Online
2025-07-03