PHD STUDENT / RESEARCH ASSISTANT

Abdelmoutaleb
Noumeur

Fire Safety Researcher specializing
in evacuation and human behavior
during emergencies

ABOUT MY CAREER
I am a PhD researcher in fire safety engineering, specializing in human behavior, evacuation dynamics, and data-driven safety analysis. My work focuses on understanding how people perceive risk and respond during emergencies, with the aim of improving evacuation strategies and building safer environments

EXPERIENCE

My research combines experimental testing, behavioral modeling, statistical analysis, and machine learning to study fire alarm interpretation and evacuation decision-making. During my doctoral work at Universiti Putra Malaysia, I developed the FAIR framework to better understand early evacuation behavior in buildings

AUTONOMY

I work with strong autonomy, managing complex research tasks, data analysis, and technical problem-solving with efficiency and precision. My background also includes Python, SPSS, ArcGIS, and web technologies, which helps me turn research data into practical results

INVOLVMENT

I value collaboration and contribute actively within multidisciplinary teams. Through publications, conference presentations, and ongoing academic work, I aim to bridge the gap between human behavior, engineering, and real-world safety solutions

EXPERIENCE & STUDIES

2012 - 2015

BACHELOR’S DEGREE

UNIVERSITY OF BATNA 2

2015 - 2017

MASTER’S DEGREE

UNIVERSITY OF BATNA 2

2020 - 2025

PhD

UNIVERSITI PUTRA MALAYSIA

2020 - 2025

RESEARCH ASSISTANT

UNIVERSITI PUTRA MALAYSIA

Bachelor’s degree in Health and Safety (Prevention and Industrial Safety), at University of Batna 2, Algeria

Master’s degree in Occupational Health, Safety, and Environment,
University of Batna 2, Algeria

Thesis title: DEVELOPMENT OF THE FIRE-ALARM INTERPRETATION RESPONSE (FAIR) GUIDELINE FOR EARLY EVACUATION IN MALAYSIAN BUILDINGS.
Supervisor: Dr. Mohd Zahirasri bin Mohd Tohir
Status: Completed (Viva Voce passed: December 15, 2025)
Research Methods: A mixed-method approach combining in-situ acoustic measurements, machine learning modeling for alarm audibility, behavioral sequence analysis during fire drills, and statistical modeling of survey data on risk perception and evacuation decisions within the FAIR (Fire Alarm, Interpretation, Response) framework.

Member of the Safety Engineering Interest Group (SEIG), Department of Chemical and Environmental Engineering. Contributing to research on fire safety engineering, focusing on fire alarm audibility and human behavior during evacuation under the supervision of Dr. Mohd Zahirasri bin Mohd Tohir.

EXPERTISE & SKILLS
I combine fire safety engineering, human behavior analysis, and machine learning to develop data-driven solutions for evacuation and risk assessment in complex environments.
80 %
LEVEL Advanced
EXPERIENCE 5+YEARS

Specialized in fire risk assessment, evacuation strategies, and performance-based safety design. Experienced in analyzing fire alarm systems, audibility, and real-world evacuation dynamics to improve safety outcomes.

70 %
LEVEL Advanced
EXPERIENCE 5+ YEARS

Focused on understanding human decision-making during emergencies, including risk perception, response behavior, and pre-evacuation actions. My work bridges behavioral science with engineering applications.

90 %
LEVEL ADVANCED
EXPERIENCE 5+ YEARS

Skilled in Python, statistical modeling, and machine learning techniques to analyze evacuation data, predict safety outcomes, and support data-driven decision-making in fire safety engineering.

ACADEMIC LIFE

MY PUBLICATIONS

Predicting California Wildfire Damage to Structures Using Machine Learning: A Comparative Study of Random Forest and XGBoost

LJUBLJANA - SLOVENIA

This study develops and compares machine learning models (Random Forest and XGBoost) to predict wildfire damage to structures in California using pre-fire building attributes such as construction type, roof materials, and property value. Both models achieved high predictive performance (~90% accuracy), with XGBoost showing slightly superior results, particularly for minority classes. The findings highlight key factors influencing structural resilience, including building age, assessed value, and fire-resistant design features such as enclosed eaves and ember-resistant vents. The research demonstrates the effectiveness of data-driven approaches in wildfire risk assessment and provides valuable insights for emergency planning, resource allocation, and the development of evidence-based fire mitigation strategies.
Conferences Selected

Exploration of Trends and Characteristics in Malaysian Building Fires and Wildfires (2000-2019)

BARCELONA - SPAIN

This study investigates trends and characteristics of building fires and wildfires in Malaysia between 2000 and 2019 using nine key fire indicators, including incidents, fatalities, injuries, and economic losses. The findings reveal a significant increase in fire occurrences and associated risks over time, with arson identified as the leading cause, followed by electrical and smoking-related fires. The analysis also highlights regional and temporal variations in fire impacts, as well as the growing economic burden of fire incidents. This research provides valuable insights into fire patterns in Malaysia and supports the development of more effective prevention strategies, data-driven policies, and future comparative studies at both national and international levels.
Conferences Selected

Fire incidence and relative risk analysis in Malaysia (2000–2019): patterns, causes, and implications for safety and prevention

New Delhi - India

This study analyzes fire incidents in Malaysia from 2000 to 2019, highlighting a significant rise in fire outbreaks and their impact on life safety and property. Using statistical analysis and relative risk assessment, the research identifies arson, electrical faults, and smoking as the leading causes of fires. It also reveals regional disparities in fire-related fatalities and injuries, emphasizing the need for targeted, state-specific prevention strategies. The findings contribute to improving fire safety policies and support global efforts aligned with the UN Sustainable Development Goals to reduce injury and mortality from fire incidents.
Conferences

A study of staff pre-evacuation behaviors in a Malaysian hotel

Fire and Materials

This study investigates the pre-evacuation behavior of hotel staff during fire emergencies through an unannounced fire drill conducted in a Malaysian hotel. By analyzing staff responses, behavioral sequences, and influencing factors such as fire training, awareness, drill participation, and prior experience, the research provides valuable insights into decision-making processes before evacuation begins. The findings reveal that most staff members do not evacuate immediately after receiving an alarm, instead engaging in actions such as investigating or delaying response, which increases pre-evacuation time. The study highlights that higher awareness and participation in fire drills significantly improve response times, while other factors such as prior fire experience and training show limited influence. These results contribute to improving evacuation modeling, enhancing fire safety strategies, and supporting the development of more realistic, behavior-based evacuation simulations.
Journal Paper Selected

Predicting spatial sound attenuation in buildings with machine learning: Implications for fire alarm placement

Process Safety and Environmental Protection

This comprehensive research publication explores the pivotal role of early childhood education (ECE) in fostering the holistic development of young learners. Grounded in developmental psychology..
Journal Paper Selected

From Risk Perception to Action: Exploring Factors Influencing Evacuation Behavior in a Real Fire Scenario

Fire Technology

This study explores the pre-evacuation behavior of hotel staff during fire emergencies through an unannounced fire drill conducted in a Malaysian hotel. The research analyzes how staff respond to fire alarms and identifies key factors influencing their decision-making, including fire training, awareness, drill participation, and prior experience. Findings show that most staff do not evacuate immediately, often choosing to investigate or delay their response, which increases pre-evacuation time. The study highlights that higher awareness and prior participation in fire drills significantly improve evacuation response, while factors such as previous fire experience and alarm familiarity have limited impact. These insights contribute to enhancing evacuation modeling, improving fire safety strategies, and supporting the development of more realistic, behavior-based emergency planning.
Journal Paper

DEVELOPMENT OF THE FIRE-ALARM INTERPRETATION RESPONSE (FAIR) GUIDELINE FOR EARLY EVACUATION IN MALAYSIAN BUILDINGS

Malaysia

Fires in buildings remain an unpredictable hazard, where the minutes between the initial alarm and the first evacuation movement frequently dictate survival outcomes. Current fire safety practices are limited by two persistent gaps: corridor-only alarm layouts often fail to achieve awakening-level audibility in residential sleeping rooms, and existing engineering models severely under-represent the cognitive and behavioral drivers of pre-evacuation delay, particularly within the Malaysian context. To bridge the isolation between physical acoustic engineering and cognitive behavioral science, this thesis developed the Fire-Alarm, Interpretation, and Response (FAIR) guideline. The FAIR guideline serves as the primary product and main conceptual output of this research. It functions as a unified, sequential continuum that integrates three core research objectives: (1) establishing the physical acoustic boundary conditions (Signal), (2) evaluating trained cognitive processing (Interpretation), and (3) analyzing untrained psychological reactions (Response). The mixed-methods research design investigated this continuum sequentially. First, in-situ acoustic measurements and machine-learning predictive modeling (Random Forest and Gradient Boosting) were utilized to map spatial sound attenuation and optimize alarm placement. Second, sequence-based mapping of staff actions during an unannounced hotel drill was conducted to quantify how prior training affects pre-evacuation interpretation timing. Finally, a post-incident resident survey from a real high-rise fire was analyzed using latent profile and mediation models to examine how risk perception and fear shape initial evacuation responses. The empirical findings validate the integrated FAIR guideline. At the Signal layer, machine learning predictions and physical measurements confirmed that corridor alarms routinely fall 15–21 dB below the required 75 dBA waking threshold when doors are closed, establishing that without in-unit sounders, subsequent behavioral processes cannot initiate. At the Interpretation layer, drill data demonstrated that once an audible signal is present, procedural memory developed through regular drills cuts the odds of a 1–3 minute delay by 73%. Finally, at the Response layer, real-incident data revealed that only 27.4% of residents evacuated immediately, with mediation analysis confirming that perceived severity elevates fear, which partially drives the final decision to move. Ultimately, this thesis delivers the algorithm-flexible FAIR guideline as a cohesive diagnostic tool for fire safety engineering. By explicitly demonstrating that physical signal delivery dictates cognitive interpretation, which in turn structures the final behavioral response, the guideline provides practitioners with actionable audit metrics to systematically eliminate evacuation delays. Keywords: Building fire; Alarm audibility; Pre-evacuation; Human behavior; Machine learning. SDG: SDG 3: Good Health & Well-Being SDG 9: Industry, Innovation & Infrastructure SDG 11: Sustainable Cities & Communities
Thesis Selected
RESEARCH PAPERS
Explore my research papers, where I present studies on fire safety engineering, human behavior, and data-driven evacuation analysis.
MARCH 2026

From Risk Perception to Action: Exploring Factors Influencing Evacuation Behavior in a Real Fire Scenario

Understanding human behavior in fire is crucial for improving fire safety strategies and the effectiveness of the evacuation process. This study examines the factors influencing evacuation behavior during a real fire scenario in a Malaysian residential building, focusing on emotional, environmental, and social factors. Data were collected through a post-fire questionnaire distributed to occupants, assessing their perceptions, preparedness, and responses to the fire. The analysis employed descriptive and inferential techniques to identify key behavioral patterns and demographic differences. The findings reveal that emotional responses, environmental conditions, and social influence significantly shaped evacuation behavior. Preparedness levels varied across demographic groups, highlighting vulnerabilities in specific populations. Emotional responses, such as fear, significantly influenced decision-making. These findings emphasize the need for tailored interventions, clear communication strategies, and enhanced fire safety protocols inside the building to support effective evacuations.

March 2025

Predicting spatial sound attenuation in buildings with machine learning: Implications for fire alarm placement

Audible and intelligible fire alarms play a critical role in ensuring occupant safety in emergencies. Although various experimental and theoretical methods exist to measure and predict alarm sound levels, the variability and limitations of these methods, especially in complex layouts, remain underexplored. Building on both established fire engineering research and advanced alarm management concepts from process safety, this study evaluates the effectiveness of fire alarm placement within residential units by comparing in situ measurements, calculation-based estimates, and machine learning predictions. The findings show that open doors result in higher sound levels than closed doors, and corridor-based alarms typically fail to meet the recommended 75 dBA threshold needed to awaken sleeping occupants. Moreover, established calculation methods show an average error rate of about 9 %, especially in geometrically complex or acoustically variable settings. By contrast, the machine learning model achieves a notably lower error rate at around 2 % underscoring its potential to integrate uncertainty factors such as distance, partitions, and acoustic attenuation more effectively than traditional formulas. From a risk management perspective, these results highlight the value of data-driven, risk-based alarm design, aligning with hybrid alarm modeling approaches seen in process industries. The study concludes that installing fire alarms within each dwelling unit, coupled with interconnected sounders in sleeping areas, significantly enhances occupant alertness and system reliability in residential buildings.

October 2024

A study of staff pre-evacuation behaviors in a Malaysian hotel

Simulating fire and evacuation scenarios is crucial for engineers to assess building safety during fire incidents. Accurate simulations require data on occupants' behaviors, particularly during the pre-evacuation phase as these decisions significantly impact evacuation duration. Gathering comprehensive data from diverse regions while considering cultural and regional variations is necessary to understand how occupants' behavior is influenced. Thus, this study focuses on examining the behavior of Malaysian hotel staff during unannounced fire drill to gain insights into factors affecting their behavior during pre-evacuation stage, such as fire experience, fire alarm, drill participation, fire training, and awareness. The study categorizes the actions performed by the hotel staff into sequences and analyses them based on influencing factors. The findings indicate that instead of immediately evacuating in response to emergency notification, the hotel staff engage in various actions. Most staff members initially investigate or ignore the emergency, resulting in longer pre-evacuation times. Moreover, the results suggest that previous drill participation and high awareness levels contribute to shorter pre-evacuation times. Conversely, previous fire experience, fire training, and fire alarm familiarity have no effect on pre-evacuation time.

LABORATORY TEAM

Mohd Zahirasri Mohd Tohir

Head of Research Group

Abdelmoutaleb Noumeur

RESEARCH ASSISTANT

BLOG & NEWS

MY DIGITAL DIARY

What Algeria’s Fires Tell Us About WUI Evacuation Safety

Watching images of recent wildfires in Algeria is difficult. Beyond the visible destruction of forests, homes, infrastructure, and landscapes, these events raise a more difficult question for fire safety professionals:…

What Algeria’s Fires Tell Us About WUI Evacuation Safety

August 29, 2026

Watching images of recent wildfires in Algeria is difficult.

Beyond the visible destruction of forests, homes, infrastructure, and landscapes, these events raise a more difficult question for fire safety professionals:

What happens when a wildfire reaches a community faster than people can understand the threat, make a decision, and evacuate?

This is one of the central challenges of Wildland–Urban Interface (WUI) fire safety.

A wildfire that remains in a remote forest presents one set of challenges. A wildfire that moves toward populated areas creates a much more complex emergency involving fire spread, smoke, buildings, roads, emergency services, communications, traffic, and—critically—human decision-making.

The question is no longer simply How do we stop the fire?

It becomes:

Can people recognize the threat, receive a credible warning, decide what to do, and reach safety before conditions become untenable?

Algeria Is Not New to the Wildfire Problem

Wildfire is not an emerging issue for Algeria.

Research examining the country's historical fire activity has shown that wildfires are strongly concentrated in northern Algeria, particularly in humid and subhumid areas where vegetation provides greater fuel availability. Fire activity is also associated with human presence, including croplands and built-up areas located alongside shrublands and forests. [1]

This geographical relationship is important because it brings wildfire and human settlements into the same risk environment.

Research on Algeria's wildfire history has also identified significant challenges in wildfire management and argued that suppression alone is unlikely to be sufficient for the country's future wildfire risk. Prevention, fuel management, and broader wildfire-management strategies are also required. [2]

This is where the concept of the Wildland–Urban Interface becomes particularly relevant.

What Is a Wildland–Urban Interface Fire?

The WUI describes areas where human development meets or is intermingled with wildland vegetation.

These environments can include:

  • houses located close to forests;
  • villages surrounded by vegetation;
  • suburban development expanding into fire-prone landscapes;
  • roads and infrastructure crossing forested areas; and
  • communities where buildings and natural fuels are closely connected.

When wildfire enters such an environment, the emergency becomes multidimensional.

Fire behaviour can change rapidly.

Smoke can reduce visibility.

Road capacity can become a limiting factor.

Power and communications can be disrupted.

Residents may have incomplete information.

And evacuation routes that appear safe at one moment may become hazardous later.

This makes WUI evacuation fundamentally different from treating evacuation as simply a matter of moving people from Point A to Point B.

Evacuation Is a Human Decision Before It Is a Traffic Problem

One of the most important findings from wildfire evacuation research is that people do not necessarily evacuate immediately after receiving a warning.

They first have to interpret the situation.

Research on WUI fire behaviour identifies a range of factors influencing protective-action decisions, including risk perception, previous experience, social and environmental cues, preparation, family responsibilities, location, and perceptions of the credibility of the threat. [3]

In other words, receiving information is not equivalent to acting on it.

A person may receive an evacuation message and still ask:

Is the fire actually close to me?

Is the warning serious?

Where is the fire moving?

Should I wait for more information?

Should I protect my home first?

Are my family members ready?

Which road should I take?

These questions can consume valuable time.

The "Wait and See" Problem

One particularly important behaviour identified in wildfire research is the tendency to wait and see.

Rather than immediately evacuating, some residents observe how the situation develops before deciding whether to leave.

This behaviour is not necessarily irrational.

People may be attempting to reduce uncertainty. They may be trying to confirm whether the fire actually threatens them, waiting for information from authorities, checking environmental conditions, contacting relatives, or assessing whether evacuation is necessary.

However, in a rapidly developing wildfire, waiting can reduce the amount of time available for evacuation.

Research comparing WUI residents in France and Australia found that "wait and see" responses were influenced by the type of cues people received, their preparedness, previous wildfire experience, and their perceived risk during the event. [4]

This has an important implication for emergency planning:

The objective of a warning system should not simply be to deliver information. It should help people understand the threat well enough to take appropriate protective action.

Warnings Matter but So Does Trust

Research from the 2016 Chimney Tops 2 fire in Tennessee found that both fire cues and warnings from trusted sources influenced people's risk perceptions before evacuation decisions. Risk perception, in turn, was strongly associated with evacuation decisions. [5]

This illustrates a fundamental principle of emergency communication:

People need more than information. They need information they can interpret and trust.

A warning saying that a wildfire exists may not be enough.

Residents may need to know:

  • where the fire is;
  • how rapidly it is moving;
  • which areas are threatened;
  • what action is expected;
  • where they should go;
  • which evacuation routes remain available; and
  • how urgently they need to act.

The communication problem becomes especially difficult when the fire is developing faster than official information can be updated.

Evacuation Delay Can Become a Safety-Critical Variable

Recent research examining the 2021 Marshall Fire in Colorado provides another important perspective.

The study investigated risk perception, evacuation decisions, and evacuation delay among people affected by the fire. It found that several factors influenced whether people evacuated and how long they delayed before leaving. Fire cues, previous risk perception, warnings, household characteristics, preparation activities, and the timing of alerts all played roles. [6]

This is significant for fire safety engineering.

Traditional evacuation analysis often focuses heavily on the physical movement of people :

How quickly can people travel along an evacuation route?

But that calculation can miss an earlier stage of the problem:

When do people actually begin moving?

If a household spends 20 minutes interpreting the threat, collecting belongings, contacting family members, or deciding whether the warning applies to them, those 20 minutes become part of the evacuation problem.

The evacuation process therefore begins before people enter the road network.

From Fire Spread to Human Response

This is one reason WUI evacuation modelling is becoming increasingly important.

A useful WUI evacuation model cannot consider only wildfire spread.

It needs to consider the interaction between:

Fire → Information → Human decision → Movement → Traffic → Safety

Researchers have developed integrated modelling approaches specifically for this problem.

For example, the WUI-NITY platform couples wildfire spread, pedestrian behaviour, and traffic movement to simulate evacuation scenarios in WUI communities. The objective is to provide a more comprehensive representation of how a wildfire emergency evolves and how people respond to it. [7]

This type of approach is important because the fire and the population do not operate independently.

As the fire changes, people's perceptions change.

As warnings change, decisions change.

As people begin evacuating, traffic conditions change.

And as traffic conditions deteriorate, the feasibility of evacuation changes again.

The emergency is therefore a dynamic system.

What Does This Mean for Algeria?

Algeria presents an important case for further research.

The country already has a substantial wildfire history, particularly across northern forested regions. Existing research has identified both environmental and human factors contributing to the spatial distribution of wildfire activity. [1]

At the same time, research on wildfire management in Algeria has highlighted the need to move beyond a predominantly suppression-oriented approach and strengthen prevention and proactive management. [2]

But prevention should not stop at vegetation management or firefighting capacity.

Community evacuation preparedness deserves equal attention.

We need to better understand questions such as:

  • How do Algerian residents perceive wildfire risk?
  • How quickly do residents recognize that a wildfire threatens their community?
  • Which warning channels do people trust?
  • How do residents respond to official evacuation warnings?
  • How common is "wait and see" behaviour?
  • How much time do households need to prepare before departure?
  • How do families make evacuation decisions?
  • What happens when evacuation routes become congested?
  • How do elderly people, children, and people with mobility limitations affect evacuation timing?
  • How does previous wildfire experience influence future decisions?
  • How should warnings change as fire conditions deteriorate?

These are not merely communication questions.

They are life-safety questions.

The Importance of Pre-Evacuation Time

From a fire safety perspective, one variable deserves particular attention:

pre-evacuation time.

This is the period between the point at which people become aware of the threat and the moment they actually begin evacuation movement.

It can include:

  1. receiving or noticing a warning;
  2. interpreting the information;
  3. confirming the threat;
  4. deciding whether evacuation is necessary;
  5. informing other household members;
  6. collecting essential belongings;
  7. assisting vulnerable people;
  8. preparing vehicles;
  9. selecting an evacuation route; and
  10. finally leaving.

Every one of these actions consumes time.

And when wildfire conditions are changing rapidly, time is a finite safety resource.

Research on wildfire evacuation has repeatedly demonstrated that human behaviour, rather than simply physical movement capacity, can substantially influence evacuation outcomes. [3,5,6]

This means that improving evacuation safety requires understanding why people leave when they do—not simply how fast they travel once they leave.

We Should Not Wait Until the Next Fire

The loss of life associated with wildfire disasters should not be viewed only through the lens of firefighting response.

It should also prompt questions about preparedness.

Could residents recognize the threat earlier?

Could warnings reach them sooner?

Could the warning be more specific?

Could evacuation decisions be supported with clearer information?

Could vulnerable households be identified before an emergency?

Could evacuation routes be tested under realistic wildfire scenarios?

Could communities practice evacuation before they actually need to evacuate?

These questions shift the focus from reacting to wildfire toward building community resilience before wildfire occurs.

WUI Fire Safety Needs a Broader Approach

Wildfire risk cannot be reduced to a single intervention.

Fire suppression is important.

Early detection is important.

Fuel management is important.

Land-use planning is important.

Building resilience is important.

But so are human behaviour and evacuation preparedness.

The research increasingly shows that people are active participants in wildfire emergencies. Their perceptions, experiences, information sources, decisions, and actions all influence how an evacuation unfolds. [3–6]

For fire safety engineers, emergency planners, authorities, and researchers, this creates an important opportunity.

We need to connect wildfire science with human behaviour.

We need to connect warning systems with evacuation modelling.

And we need to connect community preparedness with the actual time available for people to reach safety.

Algeria's Wildfire Risk Is Also a Human Behaviour Problem

The images coming from Algeria are a reminder of the destructive power of wildfire.

But they should also encourage a deeper discussion about preparedness.

The central question should not only be:

How do we fight the next wildfire?

It should also be:

How do we make sure people have enough time, information, and capacity to survive it?

That means treating evacuation as an integral component of wildfire safety—not as something that begins after the fire reaches a community.

For communities exposed to WUI fire, the difference between warning and evacuation can be measured in minutes.

And sometimes, those minutes determine the outcome.

Wildfires are a tragedy first.

But they are also a warning.

As wildfire risk continues to challenge communities across the Mediterranean and beyond, WUI fire safety deserves substantially more attention—not only in how we manage fire, but in how we prepare people to respond when the fire arrives.

References

[1] Curt, T., Aini, A., & Dupire, S. (2020). Fire Activity in Mediterranean Forests (The Algerian Case). Fire3(4), 58. https://doi.org/10.3390/fire3040058

[2] Meddour-Sahar O (2015). Wildfires in Algeria: problems and challenges. iForest 8: 818-826. - doi: 10.3832/ifor1279-007

[3] Folk, L.H., Kuligowski, E.D., Gwynne, S.M.V. et al. A Provisional Conceptual Model of Human Behavior in Response to Wildland-Urban Interface Fires. Fire Technol 55, 1619–1647 (2019). https://doi.org/10.1007/s10694-019-00821-z

[4] Sandra Vaiciulyte, Lynn M. Hulse, Edwin R. Galea, Anand Veeraswamy, Exploring ‘wait and see’ responses in French and Australian WUI wildfire emergencies, Safety Science, Volume 155, 2022, 105866, ISSN 0925-7535, https://doi.org/10.1016/j.ssci.2022.105866.

[5] Kuligowski ED, Walpole EH, Lovreglio R, McCaffrey S. (2020) Modelling evacuation decision-making in the 2016 Chimney Tops 2 fire in Gatlinburg, TN. International Journal of Wildland Fire 29, 1120–1132. https://doi.org/10.1071/WF20038

[6] Ana Forrister, Erica D. Kuligowski, Yuran Sun, Xiang Yan, Ruggiero Lovreglio, Thomas J. Cova, Xilei Zhao, Analyzing Risk Perception, Evacuation Decision and Delay Time: A Case Study of the 2021 Marshall Fire in Colorado, Travel Behaviour and Society, Volume 35, 2024, 100729, ISSN 2214-367X, https://doi.org/10.1016/j.tbs.2023.100729

[7] Jonathan Wahlqvist, Enrico Ronchi, Steven M.V. Gwynne, Max Kinateder, Guillermo Rein, Harry Mitchell, Noureddine Bénichou, Chunyun Ma, Amanda Kimball, Erica Kuligowski, The simulation of wildland-urban interface fire evacuation: The WUI-NITY platform,
Safety Science, Volume 136, 2021, 105145, ISSN 0925-7535, https://doi.org/10.1016/j.ssci.2020.105145.

When Fire Alarms Fail to Be Heard: A Lesson from Wang Fuk Court

Reading accounts from the recent fire at Wang Fuk Court in Hong Kong is deeply distressing. One resident reportedly described…

When Fire Alarms Fail to Be Heard: A Lesson from Wang Fuk Court

March 29, 2026

Reading accounts from the recent fire at Wang Fuk Court in Hong Kong is deeply distressing. One resident reportedly described the moment this way: “As soon as I went downstairs, I saw his building was already on fire.” That kind of testimony is a stark reminder that, in life-threatening emergencies, seconds matter and warning systems must perform reliably.

From a fire safety engineering perspective, this event highlights a critical issue that is sometimes underestimated: audibility. A fire alarm is only effective if the warning signal actually reaches the occupants in time and at sufficient clarity to prompt action. When that does not happen, the protection system loses one of its most essential functions.

This is precisely why our recent work, Predicting Spatial Sound Attenuation in Buildings with Machine Learning: Implications for Fire Alarm Placement, focuses on the acoustic side of life safety. Traditional methods can be limited in complex residential layouts, where sound does not travel uniformly and local “deaf spots” may leave some occupants insufficiently alerted. Our research explores how machine learning can improve the prediction of sound attenuation in built environments, with the goal of supporting better alarm placement and more reliable warning coverage.

The broader lesson is simple: life safety cannot depend on chance, proximity, or the hope that occupants will notice indirect cues in time. Fire alarm design must account for how people actually experience warning signals in real buildings. Robust acoustic modeling is therefore not a technical luxury; it is a fundamental part of ensuring that an alarm, once triggered, is actually heard.

Load more