.png){kind=icon}
.png){kind=icon}
Return
Emerging trends in human-centered fire evacuation modelling: a global review of advanced and hybrid methods for building safety design
.png){kind=icon}
PRE
AI
H
M
S
S
DOI：10.1080/17452007.2026.2649357.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
As buildings grow increasingly complex and design decisions rely more heavily on performance-based assessment, conventional fire evacuation models have limited capacity to represent human behaviour, decision-making, and dynamic interaction under emergency conditions. These limitations reduce reliability for early design-stage evaluation, particularly in complex layouts with diverse occupants. This review synthesises 93 peer-reviewed journal studies published between 2018 and 2024, drawn from global research across Europe, North America, Asia, and other regions. It examines emerging simulation technologies, including Artificial Intelligence (AI), agent-based and hybrid modelling, Building Information Modelling (BIM) integration, and immersive Virtual Reality (VR) environments, used to improve behavioural realism and the practical value of evacuation analysis. Previous reviews have often emphasised conventional formulations or isolated methods and behavioural factors; in contrast, this study provides an integrated, design-oriented synthesis of advanced and hybrid frameworks including all introduced categories. The studies are categorised by modelling methodology, integration strategy, application context (design-stage evaluation versus real-time management), and behavioural fidelity. Results indicate a shift from prescriptive, rule-based approaches toward adaptive, data-driven architectures capable of representing cognition, perception, social interaction, and heterogeneous occupant responses. Hybrid models show promise for informing architectural decisions, improving design coordination, and supporting scenario-based safety management, but they require clearer validation protocols and shared benchmarks to enable deployment. However, gaps remain in empirical behavioural datasets, representation of vulnerable and neurodivergent populations, and validation of VR-based findings against computational models. Overall, the review highlights methodological challenges and research opportunities for strengthening performance-based design and developing safer, more resilient, human-centred buildings.
Keywords:
Artificial intelligence
building information modelling (BIM)
evacuation modelling
fire safety engineering
human-centred design
performance-based design
virtual reality (VR)
Journal
IF:
2.5
Papers:
183
Citations:
1.3K
Organization
Cited Papers