Fire

30 pages, 4258 KB

Open AccessArticle

Experimental–CFD Optimisation of Hydraulic Jet Reach for the GS Mark III Peatland Firefighting System

by Wan Mohd NurulHisam Wan Nawang, Azfarizal Mukhtar, Mohd Zamri Yusoff, Ahmad Faiz Tharima, Adam C. Watts, Zarina Itam and Muhammad Nuruddin Zulkifle

Fire 2026, 9(9), 411; https://doi.org/10.3390/fire9090411 (registering DOI) - 19 Sep 2026

Abstract

The use of Computational Fluid Dynamics (CFD) with surrogate-based optimisation is increasingly becoming common in the design of fluid delivery systems although there are very few cases where it has been applied to fixed water-based fire suppression systems. This paper seeks to propose [...] Read more.

The use of Computational Fluid Dynamics (CFD) with surrogate-based optimisation is increasingly becoming common in the design of fluid delivery systems although there are very few cases where it has been applied to fixed water-based fire suppression systems. This paper seeks to propose an integrated four-stage engineering methodology incorporating field experiment, internal flow CFD using the Shear Stress Transport (SST) k-ω turbulence model in ANSYS Fluent, coherent-stream trajectory analysis and response surface methodology based on third-order polynomial regression. The proposed method is used to design the GS Mark III which is a fixed sprinkler nozzle system used to extinguish subsurface smouldering peatland fires. The validation of the coupled CFD and trajectory analysis model for 3, 5, and 7 bar using the field measurement shows the error of the model ranging from 2.55 to 3.58%. The coherent-stream trajectory is modelled by direct integration of the equations of motion, with aerodynamic deceleration represented by a single lumped coefficient calibrated against the field data, since neither a bluff-body drag coefficient nor a skin-friction closure reproduces the measured reach. The design of a parametric model involving 20 nozzle geometries in terms of diameter (5–15 mm) and discharge angles (0–67.5°) yields a surrogate with R² = 0.988, a root mean square error (RMSE) of 1.19 m and a mean absolute error (MAE) of 0.92 m. The surrogate locates the optimum at a diameter of 15 mm and a discharge angle of 38.26°, although the fitted response is flat between approximately 31° and 46°, so any angle within that band performs equivalently within the resolution of the model. The CFD simulation of the optimum (exit velocity = 32.121 m/s) gives a coherent-stream jet distance of 46.91 m, within 1.01% of the surrogate prediction. This represents a 25.9% improvement over the baseline configuration within the same modelling framework (37.25 m against 46.91 m), obtained at 2.28 times the baseline discharge, which the water supply must be able to sustain. Full article

(This article belongs to the Section Mathematical Modelling and Numerical Simulation of Combustion and Fire)

► Show Figures

Figure 1

28 pages, 2778 KB

Open AccessReview

Toward a Behavior-Aware, Adaptive, and Socially Responsive Fire Evacuation Training Framework: A Systematic Review of Serious Game Applications

by Flavia-Ioana Patrascu and Amirhosein Jafari

Fire 2026, 9(9), 410; https://doi.org/10.3390/fire9090410 (registering DOI) - 19 Sep 2026

Abstract

Fire evacuation training is essential for reducing casualties during building emergencies, yet conventional approaches such as drills and instructional materials are often limited in realism, scalability, and their ability to capture human behavior. Serious games (SGs), particularly those using immersive virtual environments, have [...] Read more.

Fire evacuation training is essential for reducing casualties during building emergencies, yet conventional approaches such as drills and instructional materials are often limited in realism, scalability, and their ability to capture human behavior. Serious games (SGs), particularly those using immersive virtual environments, have emerged as promising alternatives for interactive fire safety training. This study systematically reviews 31 peer-reviewed SG-based fire evacuation training studies identified through Scopus and Google Scholar, with searches completed on April 18, 2026. Eligible studies addressed game-based approaches to fire or general building evacuation and reported primary research involving system development, implementation, application, or evaluation. The studies were analyzed across five dimensions: simulation and environment design, interaction and game mechanics, behavioral representation, performance evaluation, and social and collaborative aspects. The review identifies three recurring limitations: simplified representations of occupant decision-making, fragmented evaluation approaches, and limited integration of social dynamics and adaptive training. Given the heterogeneity of study designs and outcomes, the overall effectiveness of SG-based fire evacuation training cannot yet be determined consistently. Based on these findings, the study proposes a roadmap for next-generation SG-based fire evacuation training systems and introduces the Behavior-Aware, Adaptive, and Socially Responsive (BAS) framework. The proposed roadmap and framework provide a conceptual foundation for designing more realistic, personalized, and behavior-aware fire evacuation training systems by integrating behavioral modeling, social interactions, multidimensional assessment, and AI-enabled adaptive learning. The review is limited by English-language inclusion, database coverage, and single-reviewer screening and coding. Full article

(This article belongs to the Special Issue Building Fire Dynamics and Fire Evacuation, 2nd Edition)

► Show Figures

Figure 1

28 pages, 3054 KB

Open AccessArticle

A Lightweight Forest Fire Detection Model with Multi-Granularity Vision-Language Enhancement

by Yifan Ma, Weifeng Shan, Yanwei Sui and Mengyu Wang

Fire 2026, 9(9), 409; https://doi.org/10.3390/fire9090409 (registering DOI) - 19 Sep 2026

Abstract

In recent years, Unmanned Aerial Vehicle (UAV)-based object detection technology has demonstrated immense potential for forest fire monitoring in complex environments. However, constrained by the drastic multi-scale variations in fire targets, severe background interference, and the limited computational resources of edge devices, existing [...] Read more.

In recent years, Unmanned Aerial Vehicle (UAV)-based object detection technology has demonstrated immense potential for forest fire monitoring in complex environments. However, constrained by the drastic multi-scale variations in fire targets, severe background interference, and the limited computational resources of edge devices, existing object detection models struggle to strike a balance between detection accuracy and inference efficiency. To systematically address the aforementioned issues, this paper proposes MVLFireNet, a lightweight and real-time forest fire detection model driven by multi-granularity vision-language enhancement. First, a Multi-scale Spatial-Aware attention (MSA) module is proposed to capture global long-range dependencies while explicitly preserving the high-frequency two-dimensional spatial features of weak fire spots and smoke edges. Second, a Cross-Modulation Fusion (CMF) module is designed to replace the traditional passive feature concatenation with bidirectional nonlinear conditional modulation, thereby achieving active denoising and compensation for both deep high-level semantics and shallow spatial details. Finally, a Multi-granularity Vision-Language Enhancement (MVLE) branch is innovatively introduced to inject robust semantic discriminative capabilities into visual features via a hierarchical text alignment enhancement mechanism covering both global scenes and local targets. Furthermore, FSDataset-VL, the first large-scale multi-granularity text-image dataset tailored for forest fire detection, is constructed. Extensive experiments on FSDataset-VL demonstrate that MVLFireNet, with merely 2.41 million parameters, achieves mAP@0.5 and mAP@0.5:0.95 of 86.3% and 54.8%, respectively, both representing the highest performance among all comparative models. Under the mAP evaluation framework, it attains an optimal balance between detection accuracy and computational efficiency, thereby providing an efficient solution for UAV-based forest fire monitoring in complex environments. Full article

(This article belongs to the Special Issue Intelligent Forest Fire Prediction and Detection: 2nd Edition)

► Show Figures

Figure 1

19 pages, 2711 KB

Open AccessArticle

Data and Knowledge Dual-Driven Inversion of Heat Release Rate in Tunnel Fires

by Juncun Chen, Yufei Zhu and Chao Guo

Fire 2026, 9(9), 408; https://doi.org/10.3390/fire9090408 (registering DOI) - 19 Sep 2026

Abstract

The heat release rate (HRR) indicates the scale of a tunnel fire, and inverting it in real time from ceiling sensors supports fire detection and ventilation control. Purely data-driven (deep learning) models are accurate within the training range but cannot extrapolate to larger [...] Read more.

The heat release rate (HRR) indicates the scale of a tunnel fire, and inverting it in real time from ceiling sensors supports fire detection and ventilation control. Purely data-driven (deep learning) models are accurate within the training range but cannot extrapolate to larger fires, whereas a purely physics-based formula is less accurate and fails during the fast-growth transient. This paper proposes a data and knowledge dual-driven HRR inversion method. The data component is an encoder-only Transformer on ceiling thermocouples, and the knowledge component is a slope-corrected plume-scaling inversion. The two are coupled by a training-time soft constraint and an inference-time two-layer gate: a magnitude gate raising the physics weight beyond the training power ceiling, and a steady-state gate down-weighting it during transients. On 24 simulated cases (six slopes × four powers, 0.5–4 MW), data are split by slope and power into mutually exclusive training, validation, and test subsets, the test covering unseen slopes and powers. The method outperforms the physics formula at every power tier; on power extrapolation it far surpasses the pure deep learning model (R² = 0.84), and on slope extrapolation it matches that model (R² = 0.94). The results demonstrate, within the present single-geometry FDS tunnel configuration and the investigated working conditions (0–5% slopes, 0.5–4 MW, t² growth, natural ventilation), that physics-guided gated fusion can improve HRR estimation under a 4 MW single-power extrapolation test while retaining the accuracy of the data-driven model under slope extrapolation; the conclusions are not claimed to be directly transferable to other tunnel configurations. Full article

(This article belongs to the Special Issue Experimental and Numerical Investigations into Fire Dynamics in Enclosed and Open Spaces)

► Show Figures

Figure 1

28 pages, 12090 KB

Open AccessArticle

Self-Heating Behavior of Raw Linseed-Oil-Contaminated Cotton Gloves Under Non-Isothermal Ramp-Heating and Covering Conditions

by A-Young Choi, Ki-Hun Nam, Ji-Won Yoon, Sin-Dong Kang, Seong-Min Lim and Seo-Young Kim

Fire 2026, 9(9), 407; https://doi.org/10.3390/fire9090407 (registering DOI) - 18 Sep 2026

Abstract

Cotton textile waste contaminated with drying oils such as raw linseed oil poses a self-heating fire hazard, but previous work has mainly relied on milligram-scale thermal analysis (DSC/TGA) or isothermal oven-basket exposure, both of which provide limited insight into the reproducibility and covering-dependent [...] Read more.

Cotton textile waste contaminated with drying oils such as raw linseed oil poses a self-heating fire hazard, but previous work has mainly relied on milligram-scale thermal analysis (DSC/TGA) or isothermal oven-basket exposure, both of which provide limited insight into the reproducibility and covering-dependent behavior of macroscale textile waste. The present study investigated the self-heating behavior of raw-linseed-oil-contaminated cotton gloves under a scenario-based, non-isothermal ramp-heating protocol (5 °C/min) in which the sample and oven atmosphere were heated simultaneously, reproducing the gradual thermal exposure conditions typical of workplace waste storage. Three oil loadings (5, 10, 15 g), four oven set temperatures (100–140 °C), and three covering conditions (open, closed, towel-covered) were tested in triplicate (108 tests total). Significant self-heating was defined as a maximum excess temperature ΔT_max ≥ 60 °C. All 27 tests at 100 °C were negative, whereas self-heating became increasingly frequent from 120 °C onward. The towel-covered condition produced the most severe and reproducible response, satisfying the criterion in all replicates at 120–140 °C, with mean ΔT_max reaching 116.0 ± 10.1 °C (10 g, 130 °C), while open and closed conditions showed mixed, condition-sensitive responses near the 120–130 °C transition region. Oil loading did not increase self-heating monotonically; this non-monotonic response points to a trade-off between fuel availability and oxygen transport within the cotton matrix. These reproducible, triplicate-based results demonstrate that covering or piling oil-contaminated textile waste with other porous materials markedly increases self-heating severity and consistency, even before ignition is reached. This finding supports the practical recommendation that such waste be stored uncovered, ventilated, and separated in non-combustible containers. Full article

(This article belongs to the Special Issue The Thermal Decomposition and Combustion Behavior of Combustible Materials)

► Show Figures

Figure 1

14 pages, 5516 KB

Open AccessArticle

Unlocking Self-Extinguishment Behavior in Cotton Fabrics Through Eco-Friendly Chitosan/Halloysite Cationic/Anionic Assembly

by Hamid Hassan, Zeeshan Ur Rehman, Naveen Yadav, Subin Jung and Bon Heun Koo

Fire 2026, 9(9), 406; https://doi.org/10.3390/fire9090406 (registering DOI) - 18 Sep 2026

Abstract

A hybrid cationic system comprising bio-based chitosan and synthetic branched polyethyleneimine (BPEI) was employed to facilitate the flame-retardant deposition of halloysite nanotubes (HNTs) onto cotton fabrics through a layer-by-layer (LbL) assembly process. The FTIR results revealed new peaks of O-H stretching and Si-O-Si [...] Read more.

A hybrid cationic system comprising bio-based chitosan and synthetic branched polyethyleneimine (BPEI) was employed to facilitate the flame-retardant deposition of halloysite nanotubes (HNTs) onto cotton fabrics through a layer-by-layer (LbL) assembly process. The FTIR results revealed new peaks of O-H stretching and Si-O-Si groups on halloysite nanoclay at the respective positions of 3699.84 cm⁻¹, 3625.60 cm⁻¹, and 527.57 cm⁻¹, which suggests successful layering of the anionic species. From surface microstructural analysis, bridging features between adjacent fibers of coatings were found; however, an increase in homogeneity and randomness was also observed as the number of layers increased. Thermogravimetric analysis further exposed that the incorporation of nanoclay tubes engenders substantial char residue at 700 °C up to 32.98%, while the decomposition rate decreased from 0.235% to 0.149%. Additionally, the flame-retardant properties were rigorously assessed using a vertical flame test: self-extinguishment behavior emerged in the C-25 and C-30 coated samples, while the other samples effectively resisted flame propagation. The results confirm the effectiveness of hybrid flame-retardant coating and offer a sustainable strategy for eliminating reliance on hazardous chemical additives. Full article

► Show Figures

Figure 1

32 pages, 3243 KB

Open AccessArticle

Relative Ultrasonic Pulse Velocity-Based Prediction of Residual Compressive Strength in Thermally Damaged Loess-Substituted Concrete with Different Target Strengths

by Youngjin Nam, Taegyu Lee and Sikuk Kim

Fire 2026, 9(9), 405; https://doi.org/10.3390/fire9090405 (registering DOI) - 17 Sep 2026

Abstract

This study investigated the elevated-temperature deterioration of loess-containing concrete with different target strengths and evaluated ultrasonic-pulse-velocity (UPV)-based models for predicting residual compressive strength. Six mixtures combining target strengths of 30 and 45 MPa with loess replacement levels of 0, 15, and 30% were [...] Read more.

This study investigated the elevated-temperature deterioration of loess-containing concrete with different target strengths and evaluated ultrasonic-pulse-velocity (UPV)-based models for predicting residual compressive strength. Six mixtures combining target strengths of 30 and 45 MPa with loess replacement levels of 0, 15, and 30% were exposed to 23, 100, 200, 300, 500, and 700 °C. The dataset comprised 108 individual measurements representing 36 mixture-temperature conditions. Bulk density, UPV, and compressive strength were measured after natural cooling. Two normalization schemes were distinguished: a normal-concrete-based relative performance index, which retains both the initial penalty caused by loess replacement and subsequent thermal deterioration, and a mixture-specific residual ratio referenced to the initial value of each mixture. Experimental variability was quantified using standard deviations, coefficients of variation, and 95% confidence intervals. The effects of target strength, loess replacement, and temperature were examined using three-way ANOVA and Kruskal–Wallis tests. In addition, leakage-free condition-wise group cross-validation was performed so that the three replicates from each mixture-temperature condition were never divided between training and validation sets. UPV and compressive strength decreased markedly between 300 and 500 °C. Absolute compressive strength was significantly affected by all three factors, whereas exposure temperature was the dominant main effect for the mixture-specific residual strength ratio. Under condition-wise cross-validation, the normal-concrete-based relative model retained R² = 0.906, MAPE = 10.81%, and MPE = 0.60%, while the mixture-specific residual-ratio model achieved R² = 0.957 and MAPE = 7.94%. The proposed models are therefore suitable as preliminary screening-level tools within the investigated material and temperature ranges, but not as stand-alone bases for final structural safety decisions. Full article

(This article belongs to the Section Fire Risk Assessment and Safety Management in Buildings and Urban Spaces)

17 pages, 5567 KB

Open AccessArticle

Real-Time Tiny Fire-Spot Detection in Farmland Scenes Based on an Improved YOLOv11

by Lin Zhang, Xinnian Yang, Mingyang Wang and Yunhong Ding

Fire 2026, 9(9), 404; https://doi.org/10.3390/fire9090404 (registering DOI) - 17 Sep 2026

Abstract

Farmland straw burning can produce incipient fire spots that occupy only a few pixels in UAV images and are easily confused with straw reflections, soil highlights, smoke, and illumination changes. This study proposes FireFly-YOLOv11, a real-time detector for tiny fire-spot detection under edge-device [...] Read more.

Farmland straw burning can produce incipient fire spots that occupy only a few pixels in UAV images and are easily confused with straw reflections, soil highlights, smoke, and illumination changes. This study proposes FireFly-YOLOv11, a real-time detector for tiny fire-spot detection under edge-device constraints. A stride-4 P2 detection head is added to the YOLOv11 prediction hierarchy to preserve fine spatial details for small targets. A Warm-Contrast Cue Attention (WCCA) module enhances fire-related saliency by jointly modeling local contrast variation and learnable warmth-inspired appearance cues. An Adaptive Asymmetric Atrous Feature Fusion (A3F) module adaptively aggregates multi-scale context while retaining low-level details through asymmetric gated residual fusion. Experiments on the StrawBurning UAV dataset show that FireFly-YOLOv11 achieves 88.6% Precision, 92.3% Recall, 91.7% mAP@0.5, and 41.0% mAP@0.5:0.95 at 50 FPS on an NVIDIA Jetson Orin Nano Super Developer Kit. Compared with baseline YOLOv11, it improves the four accuracy metrics by 2.1, 1.8, 2.7, and 2.0 percentage points. Ablation results confirm that P2, WCCA, and A3F provide complementary gains for UAV-based farmland fire monitoring. Full article

(This article belongs to the Special Issue Machine Learning (ML) and Deep Learning (DL) Applications in Wildfire Science: Principles, Progress and Prospects (2nd Edition))

► Show Figures

Figure 1

22 pages, 28587 KB

Open AccessArticle

Pyrolysis and Pyrolysis Reoxidation Combustion Characteristics of Lignite

by Xin He, Yujia Huo and Weiwang Chen

Fire 2026, 9(9), 403; https://doi.org/10.3390/fire9090403 - 17 Sep 2026

Abstract

Coal, as the main energy source for building heating in northern cities of China, features large reserves, low cost, and high stability in heat supply. Low-temperature pyrolysis reactions are prone to occur within the coal pile; the coal after low-temperature pyrolysis is more [...] Read more.

Coal, as the main energy source for building heating in northern cities of China, features large reserves, low cost, and high stability in heat supply. Low-temperature pyrolysis reactions are prone to occur within the coal pile; the coal after low-temperature pyrolysis is more prone to spontaneous combustion. From the point of chemical change, there is little discussion on the oxidation and combustion characteristics of coal affected by heat conduction. This paper takes the lignite of Xilin Gol League in Inner Mongolia as the research object and prepares mixed large-particle coal samples, raw coal samples, and pyrolysis samples. The pyrolysis and reoxidation characteristics of the samples are analyzed using experimental equipment such as a TG, tubular furnace, GC, and detection methods such as elemental analysis, ¹³C-NMR, XPS, and in situ FTIR. The results show that the thermal weight loss of the raw coal sample heated to 900 °C in an N₂ atmosphere is 33.74%. The activation energy and pre-exponential factors of pyrolysis are calculated by a one-dimensional diffusion model. The non-isothermal pyrolysis can be divided into heat conduction, drying, active decomposition, and pyrolysis equilibrium stages. The concentration of characteristic gases has a quasi-exponential relationship with temperature, and the trend conforms to the characteristics of the secondary pyrolysis stage. The content of oxygen-containing functional groups (OCFGs) decreased during pyrolysis, and the active sites generated new OCFGs after contacting oxygen, which increased the risk of coal spontaneous combustion. The combustion performance and activation energy at 200 °C are lower than those of other samples, which are more prone to combustion reactions and have a greater tendency for spontaneous combustion. This paper explores the coupling relationship between macroscopic aspects such as stage division and gas release patterns during the thermal decomposition of lignite, as well as microscopic structures such as functional group evolution and aromatic condensation. It reveals the quantitative impact of the thermal decomposition reaction on the characteristics of reoxidation combustion, providing theoretical guidance for preventing and reducing the occurrence of coal spontaneous combustion accidents. Full article

(This article belongs to the Special Issue Innovative Methods and Insights into Coal Mine Fire Prevention)

► Show Figures

Figure 1

18 pages, 5097 KB

Open AccessArticle

Passenger Ship Evacuation Time Prediction Based on Sobol Sequence Sampling and the Bayesian-Optimized Random Forest Method

by Shihai Wang, Zhonghui Li, Qimiao Xie, Zhangyu Chang, Zuoqi Xu and Jing Zhang

Fire 2026, 9(9), 402; https://doi.org/10.3390/fire9090402 - 15 Sep 2026

Abstract

With the rapid growth of global waterborne tourism, the passenger capacity of ships continues to increase, which means passenger evacuation safety is becoming a critical concern. Existing passenger ship evacuation prediction methods mainly rely on computationally intensive evacuation simulations, which suffer from high [...] Read more.

With the rapid growth of global waterborne tourism, the passenger capacity of ships continues to increase, which means passenger evacuation safety is becoming a critical concern. Existing passenger ship evacuation prediction methods mainly rely on computationally intensive evacuation simulations, which suffer from high computational cost. This limitation hinders rapid evacuation assessment and emergency decision-making for large passenger ships. The objective of this study is to develop a rapid and interpretable surrogate prediction model for passenger ship evacuation time. To achieve this objective, Sobol sequence sampling is employed to efficiently construct representative evacuation scenarios, while Bayesian optimization is used to improve the prediction performance of the Random Forest model. The results show that the Sobol sequence sampling method efficiently designs multi-scenario evacuation cases for the passenger ship. The optimized model achieves high prediction accuracy with an R² of 0.901, effectively capturing the nonlinear relationship between evacuation factors and total evacuation time. Feature importance analysis demonstrates that stairways near the embarkation stations and passengers positioned farther away from them significantly affect total evacuation time, highlighting the spatial disparity in evacuation efficiency. This study provides a reliable data-driven approach and technical support for emergency decision-making and safety management in passenger ships. Full article

(This article belongs to the Special Issue Behavioral Research on Fire Evacuation and Decision-Making Processes)

► Show Figures

Figure 1

27 pages, 10330 KB

Open AccessArticle

Assessment of Hazardous Substance Exposure During Electric Vehicle Fires

by Charlotte Wierling, Susanne Lott, Alexander D. Gelner, Simone Krüger, Tim Rappsilber, Tina Raspe and Hans-Georg Schweiger

Fire 2026, 9(9), 401; https://doi.org/10.3390/fire9090401 - 15 Sep 2026

Abstract

Internal combustion engine vehicles and electric vehicles share many similarities in the event of a vehicle fire but also differ in certain aspects. While both vehicle types release pollutants typical of vehicle fires, fires involving electric vehicles also emit battery-specific emissions. To investigate [...] Read more.

Internal combustion engine vehicles and electric vehicles share many similarities in the event of a vehicle fire but also differ in certain aspects. While both vehicle types release pollutants typical of vehicle fires, fires involving electric vehicles also emit battery-specific emissions. To investigate hazardous-substance exposure of emergency responders, two fire tests were carried out on identical electric vehicles. In each test, one battery cell was mechanically short-circuited, resulting in thermal runaway and propagation within the battery. Gas and time-resolved aerosol measurements, wipe samples, extinguishing agent samples, and urine samples were analyzed. A primary finding was the pronounced variability between the two fire events: despite comparable test conditions, the two vehicle fires exhibited markedly different development patterns, highlighting the inherent variability of electric vehicle fire behavior. Wipe samples showed increased metal concentrations after the tests, with maximum post-extinguishing surface concentrations of 7.4 mg/m² nickel, 17.45 mg/m² cobalt, and 14.25 mg/m² lithium. Extinguishing agent samples showed increased concentrations of inorganic contaminants and polycyclic aromatic hydrocarbons. Urine samples showed no indication of a systematic increase in creatinine-adjusted metal concentrations among protected participants. Toxic gases were released immediately after cell short-circuiting, with hydrogen fluoride, hydrogen cyanide, and acetylene reaching approximately 500 ppm before visible flames occurred. Aerosol measurements identified transiently elevated particle number concentrations in the exterior near-field environment and, in one test, inside the passenger compartment, where a shift towards smaller particle diameters was observed. The findings demonstrate multiple responder-relevant exposure pathways and substantial variability in electric vehicle fire events. Full article

► Show Figures

Figure 1

16 pages, 3098 KB

Open AccessBrief Report

Post-Fire Abundance Trajectories of Bird Species and Functional Groups Within an Atlantic–Mediterranean Ecotone

by Fernando García-Fernández, Jesús Domínguez, Alberto Gil-Carrera, Luis Tapia and Adrián Regos

Fire 2026, 9(9), 400; https://doi.org/10.3390/fire9090400 - 15 Sep 2026

Abstract

The Atlantic–Mediterranean ecotone of northwestern Iberia is one of the most fire-prone regions in Europe, yet post-fire avian dynamics there remain poorly documented. To characterise short- to mid-term post-fire avian trajectories in this ecotone, we combined a 16-year fire record with six consecutive [...] Read more.

The Atlantic–Mediterranean ecotone of northwestern Iberia is one of the most fire-prone regions in Europe, yet post-fire avian dynamics there remain poorly documented. To characterise short- to mid-term post-fire avian trajectories in this ecotone, we combined a 16-year fire record with six consecutive breeding seasons of annual point-count surveys. Between 2020 and 2025, we recorded 13,770 individuals of 67 bird species across 210 fixed locations in a mountain protected area of NW Spain dominated by frequent small- to medium-sized wildfires. We contextualised trajectories against both contemporary unburned reference conditions and historical baselines representing contrasting fire regimes. Community-level species richness and total abundance converged rapidly toward reference levels within 3–5 years, whereas functional-group responses showed marked divergence: shrubland and open-habitat species consistently overshot unburned reference abundances, while forest-associated guilds showed persistent deficits—most pronounced among canopy foragers. Conservation value peaked during early post-fire stages and remained above reference levels for at least 9 years. This study provides the first integrated characterisation of post-fire avian trajectories at the species, functional-group, and community levels in this Atlantic–Mediterranean transitional system, offering an empirical baseline for future mechanistic research and evidence-based fire management in fire-prone landscapes. Full article

► Show Figures

Figure 1

18 pages, 10169 KB

Open AccessTechnical Note

Toward a Remote Sensing System Architecture for Operational Wildfire Prevention Using Lightning Suppression

by Phillip M. Stepanian, Kiley L. Yeakel, Adonis F. R. Leal, Jhonys Moura, Timothy A. Bonin and Earle R. Williams

Fire 2026, 9(9), 399; https://doi.org/10.3390/fire9090399 - 15 Sep 2026

Abstract

Past field experiments have demonstrated multiple cloud seeding techniques for modifying the electrical characteristics of developing thunderstorms with the purpose of reducing or eliminating lightning strikes. One motivation for these weather modification research programs was the prospect of preventing wildfires in inaccessible locations [...] Read more.

Past field experiments have demonstrated multiple cloud seeding techniques for modifying the electrical characteristics of developing thunderstorms with the purpose of reducing or eliminating lightning strikes. One motivation for these weather modification research programs was the prospect of preventing wildfires in inaccessible locations or particularly hazardous conditions by temporarily suppressing a prolific ignition source: lightning. It has been nearly 50 years since the last large-scale field effort in lightning suppression, and the ensuing five decades of technological innovation hold the promise of supporting this novel hazard mitigation approach. This study outlines observational and forecasting requirements, as well as an associated remote sensing architecture, that would support an operational wildfire prevention program based on lightning suppression by chaff seeding clouds. Two capabilities enabled by remote sensing observations are highlighted: (1) identifying potential regions of extreme wildfire behavior based on wildland fuels, topography, and weather, and (2) predicting regions where dry lightning strikes are probable. In combination, the proposed nowcasting system would predict the most likely areas of lightning-initiated wildfire danger. Dependent on land management strategies, surrounding infrastructure, firefighting capacity, and risk to human wellbeing, this information could be used to deploy lightning suppression technology to reduce or temporarily prevent wildfire ignitions in these high-risk situations. Full article

(This article belongs to the Special Issue Integrative Approaches to Wildland Fire Research: From Fundamental Fuel Behavior to Advanced Technological Solutions)

► Show Figures

Figure 1

20 pages, 7624 KB

Open AccessArticle

Integrating Spatial Dependence into Machine Learning to Quantify the Impacts of 2D/3D Built Environment Features on Fire Risk

by Zelong Xia, Zhouxi Zhao, Guofang Zhai and Yifan Zhang

Fire 2026, 9(9), 398; https://doi.org/10.3390/fire9090398 - 14 Sep 2026

Abstract

Clarifying the relationships between built environment characteristics and urban fire risk is important for developing effective fire prevention and planning strategies. However, spatial dependence and nonlinear relationships between the built environment and fire risk remain insufficiently understood. Accordingly, this study presents a geographically [...] Read more.

Clarifying the relationships between built environment characteristics and urban fire risk is important for developing effective fire prevention and planning strategies. However, spatial dependence and nonlinear relationships between the built environment and fire risk remain insufficiently understood. Accordingly, this study presents a geographically enhanced machine learning (GE-ML) framework that incorporates spatial adjacency into machine learning models through spatially weighted feature construction. Specifically, contiguity-based spatial weight matrices were used to derive spatially weighted features from 2D and 3D built environment variables. The Optimal Parameter-based Geographical Detector (OPGD) was applied to assess scale sensitivity and compare the explanatory power and interactions of the original and spatially weighted features. Six candidate models, including Ordinary Least Squares (OLS), K-Nearest Neighbors (KNN), Multilayer Perceptron (MLP), Random Forest (RF), Light Gradient Boosting Machine (LightGBM), and eXtreme Gradient Boosting (XGBoost), were then evaluated under different feature configurations, followed by SHapley Additive exPlanations (SHAP) analysis of the selected model. Results show that: (1) spatial weighting generally increased the explanatory power of major built environment factors and their interactions, with Queen contiguity yielding higher q-values than Rook contiguity; (2) spatially weighted features improved predictive performance across different models, and GE-XGBoost achieved the highest R² (0.7067) and lower residual spatial autocorrelation than GWR and GWRF; and (3) 2D and 3D built environment features accounted for 59.55% and 40.45% of the total SHAP importance, respectively, with Geo-TPD, Geo-BVD, Geo-PS, and Geo-LUI identified as the most important features. SHAP analysis further revealed nonlinear relationships and interactions between these features and predicted fire risk. These findings highlight the value of incorporating spatial adjacency information into fire risk modeling and support spatially differentiated fire risk management. Full article

(This article belongs to the Section Fire Risk Assessment and Safety Management in Buildings and Urban Spaces)

► Show Figures

Figure 1

29 pages, 1626 KB

Open AccessArticle

Research on Game-Theoretic Behavior of Collective Emergency Evacuation in Wildfire Under the Drive of Individual Risk Perception

by Yueqiao Yang, Mingyuan Li, Yuanhong Bi, Zhixiang Yuan, Liang Zhao, Zewen Song and Gege Gai

Fire 2026, 9(9), 397; https://doi.org/10.3390/fire9090397 - 14 Sep 2026

Abstract

The increasing frequency of wildfires has made large-scale collective emergency evacuation increasingly critical. However, existing studies provide limited understanding of how information structures shape the interaction between individual risk perception and collective evacuation behavior. This study develops a collective evolutionary game-based evacuation framework [...] Read more.

The increasing frequency of wildfires has made large-scale collective emergency evacuation increasingly critical. However, existing studies provide limited understanding of how information structures shape the interaction between individual risk perception and collective evacuation behavior. This study develops a collective evolutionary game-based evacuation framework under ambiguous and clear information conditions. Under ambiguous information, individual heterogeneity in risk sensitivity, mobility, and resource endowment is incorporated into social interaction payoffs. Under clear information, observable evacuation consequences, including travel time, risk exposure, and congestion effects derived from route-choice interactions, are incorporated into evacuation utility. Numerical simulations examine the evolutionary characteristics of collective evacuation behavior under different information conditions and population scales. The results show that social interactions play an important role in shaping evacuation decisions under ambiguous information, while congestion effects and route-choice interactions influence evacuation utility under large-scale demand. Sensitivity analyses further demonstrate that congestion representation affects evacuation utility across different population scales. These findings highlight the importance of considering information structure, individual heterogeneity, and collective interactions in evacuation modeling. Emergency management should therefore improve risk communication, evacuation capacity, and congestion mitigation strategies. This study provides theoretical insights into collective evacuation decision-making under heterogeneous information conditions. Full article

► Show Figures

Figure 1

32 pages, 2932 KB

Open AccessArticle

YOLO-FSD: A Deployment-Validation-Oriented Lightweight Fire Smoke Detection Network for Resource-Constrained ZYNQ7020 FPGA Edge Platforms

by Chaoyun Mai, Panrong Chen, Haipeng He, Hao Xie, Chongyi Huang, Tianlei Wang, Zhiyuan Su and Hongye Li

Fire 2026, 9(9), 396; https://doi.org/10.3390/fire9090396 - 13 Sep 2026

Abstract

Early-stage flame and smoke in forest fire scenes are often small, weakly textured, and easily confused with complex backgrounds, while many accurate YOLO-based detectors are difficult to deploy on resource-constrained FPGA devices. This study proposes YOLO-FSD, a lightweight detector derived from YOLOv4-tiny for [...] Read more.

Early-stage flame and smoke in forest fire scenes are often small, weakly textured, and easily confused with complex backgrounds, while many accurate YOLO-based detectors are difficult to deploy on resource-constrained FPGA devices. This study proposes YOLO-FSD, a lightweight detector derived from YOLOv4-tiny for FPGA-oriented deployment. It integrates inverted residual and depthwise separable structures, a lightweight semantic enhancement (LSE) block at the deep feat2 feature, a lightweight P₄ detection head, and a shallow detail compensation branch. On the combined test set of the D-Fire and New Fire and Smoke datasets, YOLO-FSD achieves a mean average precision at an intersection-over-union threshold of 0.5 (mAP50) of 69.36%, with 3.951 M parameters and 1.520 G multiply-accumulate operations (MACs). Compared with YOLOv4-tiny, mAP50 increases by 2.76 percentage points, while parameters and MACs decrease by 32.76% and 55.53%, respectively. For deployment validation, batch normalization (BN) fusion and 16-bit integer (INT16) parameter conversion are followed by fixed-point forward inference on a Xilinx Zynq-7020 FPGA, with a software-side parameter-quantization sensitivity analysis used as an intermediate check. FPGA raw output evaluation achieves 69.14% mAP50, only 0.22 percentage points below the PyTorch 32-bit floating-point (FP32) model, demonstrating the feasibility of FPGA-side forward inference. Full article

► Show Figures

Figure 1

37 pages, 11562 KB

Open AccessArticle

Fire Suppression Simulation and Risk Assessment for a Lithium-Ion Battery Energy Storage Station

by Junwei Shi, Ziyan Zhang and Ziming Xu

Fire 2026, 9(9), 395; https://doi.org/10.3390/fire9090395 - 12 Sep 2026

Abstract

Lithium-ion battery energy storage stations are being rapidly deployed for peak regulation, renewable energy integration, and emergency power supply in power systems. Their fire risk is governed by interacting factors, including cell thermal runaway, equipment failure, operating environment, personnel behavior, management systems, and [...] Read more.

Lithium-ion battery energy storage stations are being rapidly deployed for peak regulation, renewable energy integration, and emergency power supply in power systems. Their fire risk is governed by interacting factors, including cell thermal runaway, equipment failure, operating environment, personnel behavior, management systems, and information systems, and is characterized by coupling, dynamic evolution, and confined-space fire spread. Existing static risk assessment methods cannot fully represent feedback among multiple risk factors or connect risk assessment results with the physical-field evolution of fires in energy storage compartments. This study develops an integrated grey relational analysis, system dynamics, and FDS framework. Personnel, equipment, environmental, management, and information risk factors are first established, and their weights are calculated using grey relational analysis. A system dynamics model is then used to analyze the temporal evolution of overall risk and subsystem risk responses. Finally, FDS is applied to simulate fire spread in a 30 ft containerized lithium-ion battery energy storage compartment under no-suppression and water-mist suppression conditions. The results show that the central fire-source region and battery module layer are key areas of gas-phase high-temperature accumulation and potential fire spread. In the no-suppression scenario, the high-temperature region remains localized near the fire source at 3.0 s, expands along the module layer from 30.0 to 50.0 s, and approaches a relatively stable distribution after 70.0 s. Under the investigated simulation conditions, water mist reduces near-source heating, weakens smoke-layer development, and slows spatial fire spread through evaporative cooling, reduced thermal radiation feedback, and disturbance of the hot smoke layer. These findings provide a methodological reference for fire risk assessment and fire suppression design in containerized battery energy storage stations. Full article

► Show Figures

Figure 1

20 pages, 19725 KB

Open AccessArticle

Fire Safety Assessment of Insulation Boards Made from Recycled Cotton Fibers: Part I—Thermal Properties and Fire Behavior

by Tadeáš Zachara, Vlastimil Borůvka, Tomáš Kytka, Benjamín Petržela, Kryštof Kubista and Přemysl Šedivka

Fire 2026, 9(9), 394; https://doi.org/10.3390/fire9090394 - 11 Sep 2026

Abstract

Recycled cotton fibers offer a promising route for sustainable thermal insulation, but their flammability limits wider application. This study evaluated rigid recycled cotton insulation boards bonded with a polyvinyl acetate (PVAc)–starch system and modified with kraft lignin or aluminium trihydroxide (ATH) at dosages [...] Read more.

Recycled cotton fibers offer a promising route for sustainable thermal insulation, but their flammability limits wider application. This study evaluated rigid recycled cotton insulation boards bonded with a polyvinyl acetate (PVAc)–starch system and modified with kraft lignin or aluminium trihydroxide (ATH) at dosages of 10 and 15 wt.%. Board density, thermal conductivity, volumetric heat capacity, thermal diffusivity, cone-calorimeter behavior, and ignitability under single-flame exposure were assessed. Thermal conductivity remained within a narrow range of 0.071–0.075 W·m⁻¹·K⁻¹, indicating that the basic insulation function was preserved despite formulation changes. Greater differences were observed in volumetric heat capacity and thermal diffusivity, particularly in lignin-modified boards at higher PVAc contents. Fire behavior was strongly formulation-dependent and showed no uniform dose-dependent response. The lowest peak heat release rate (pHRR) was obtained for the formulation containing 10 wt.% ATH and 20 wt.% PVAc, while the lowest total heat release (THR) was obtained for the formulation containing 10 wt.% ATH and 18 wt.% PVAc, reaching 170.31 kW·m⁻² and 62.30 MJ·m⁻², respectively. Lignin produced a less consistent effect on heat release. Under single-flame exposure, edge application was generally more critical than surface application, while ATH, particularly at 15 wt.%, provided the most consistent reduction in ignition and early flame spread. Overall, ATH was the more promising fire-modifying additive, although further formulation optimization and formal reaction-to-fire classification are required. Full article

► Show Figures

Graphical abstract

27 pages, 4093 KB

Open AccessArticle

What a Euroclass Does Not Show: Specimen-Level and Between-Campaign Variability in Single Burning Item Testing of Polystyrene-Based ETICS

by Adrian Simion and Florin Ioan Bode

Fire 2026, 9(9), 393; https://doi.org/10.3390/fire9090393 - 11 Sep 2026

Abstract

The reaction-to-fire class printed on the Declaration of Performance of an External Thermal Insulation Composite System (ETICS) is often the only fire performance evidence available during early design. This paper reports, specimen by specimen, twenty-three Single Burning Item tests on nominally equivalent expanded-polystyrene [...] Read more.

The reaction-to-fire class printed on the Declaration of Performance of an External Thermal Insulation Composite System (ETICS) is often the only fire performance evidence available during early design. This paper reports, specimen by specimen, twenty-three Single Burning Item tests on nominally equivalent expanded-polystyrene ETICS, performed in one notified laboratory in two campaigns about two and a half years apart. A mineral silicate render on an inert substrate gave a fire growth rate index of 77.3 W/s; a complete ETICS with 100 mm of expanded polystyrene beneath an organically bound render gave 70.6 W/s, so that within the twenty-minute exposure, and while the rendering system retained its integrity, the assembly had a fire growth rate index comparable to, and slightly below, that of the render alone; because the two configurations differ in more than the presence of the core, this bounds the core contribution rather than isolating it. Coefficients of variation between nominally identical specimens ranged from 2% to 87%, and collapse and detachment carry no weight in the class. Individual specimens of one product spanned the B/C boundary, and one assembly was classified as s1 based on a mean of 0.4 m², below the threshold, although two of its three specimens lay above it. In the later campaign, nominally equivalent assemblies gave mean indices that were 3.7 to 7.0 times higher, with lateral flame spread in seven out of eight tests. The campaigns fall under different editions of EN 13823: the measured burner output places any residual bias of the heat-release chain opposite to the observed change, while the revised smoke correction makes cross-campaign smoke comparison indicative only. The recovered specimen-level dataset, including the time-resolved records, accompanies this paper. Full article

(This article belongs to the Section Fire Risk Assessment and Safety Management in Buildings and Urban Spaces)

► Show Figures

Figure 1

Journal Description

Latest Articles

Journal Menu

Journal Browser

Highly Accessed Articles

Latest Books

E-Mail Alert

News

Topics

Conferences

Special Issues

Topical Collections

Further Information

Guidelines

MDPI Initiatives

Follow MDPI