Fire Simulation Research Articles

A waypoint based approach to visibility in performance based fire safety design

In performance-based fire safety design, ensuring safe egress, e.g. by visibility of safety signs, is a crucial safety goal. Compliance with the building requirements is often demonstrated by simulations of smoke spread. Numerical models like the Fire Dynamics Simulator generally compute visibility as a local quantity using the light extinction coefficient, without the consideration of the actual light path to a safety sign. Here, visibility maps are introduced, providing an approach for post-processing fire simulation data. They indicate safe areas along egress routes, with respect to visibility. At each location, the available visibility is calculated using Jin’s empirical relation, as an integrated value of the extinction coefficient along the line of sight to the closest exit sign. The required visibility results from the distance between those points. Additional parameters like view angle or visual obstructions are considered. The presented method allows for temporal visibility assessment, e.g. in an ASET-RSET analysis.

An end-to-end framework for fire following earthquake simulation at regional scale: A case study on the 2024 Japan Noto Peninsula earthquake

Fire following earthquakes remains a significant and threatening hazard to communities in urban regions. Addressing this critical issue requires an effective simulation method that is suitable for widespread adoption. This paper presents a systematic end-to-end framework for simulating fire following earthquakes at a regional scale by integrating models and methods available in existing literature. Relevant theories and models are summarized and incorporated into the calculation workflow. The framework encompasses three key features: (1) GIS-based regional building data management, (2) physics-based fire simulation, and (3) high-fidelity visualization. The implementation of the framework is presented in a sequential and detailed manner, with a focus on its practicality and replicability. Each component of the workflow is designed to be adaptable, allowing for easy customization and improvement to accommodate unique site characteristics or the integration of new models through the introduction of sub-modules. To demonstrate the practicality of the proposed framework, a case study based on the recent 2024 Noto Peninsula earthquake in Japan is presented. The simulation results correspond well with the observed actual damage at the fire following earthquake disaster site, with an accuracy of 87.8 %, which validates the framework's accuracy and reliability in simulating fire following earthquakes.

Response of Fuel Characteristics, Potential Fire Behavior, and Understory Vegetation Diversity to Thinning in Platycladus orientalis Forest in Beijing, China

Objective: Active fuel management operations, such as thinning, can minimize extreme wildfire conditions while preserving ecosystem services, including maintaining understory vegetation diversity. However, the appropriate thinning intensity for balancing the above two objectives has not been sufficiently studied. Methods: This study was conducted to assess the impact of various thinning intensities (light thinning, LT, 15%; moderate thinning, MT, 35%; heavy thinning, HT, 50%; and control treatment, CK) on fuel characteristics, potential fire behavior, and understory vegetation biodiversity in Platycladus orientalis forest in Beijing using a combination of field measurements and fire behavior simulations (BehavePlus 6.0.0). Results: A significant reduction in surface and canopy fuel loads with increasing thinning intensity, notably reducing CBD to below 0.1 kg/m3 under moderate thinning, effectively prevented the occurrence of active crown fires, even under extreme weather conditions. Additionally, moderate thinning enhanced understory species diversity, yielding the highest species diversity index compared to other treatments. Conclusions: These findings suggest that moderate thinning (35%) offers an optimal balance, substantially reducing the occurrence of active crown fires while promoting biodiversity. Therefore, it is recommended to carry out moderate thinning in the study area. Forest managers can leverage this information to devise technical strategies that simultaneously meet fire prevention objectives and enhance understory vegetation species diversity in areas suitable for thinning-only treatments.

Prediction of temperature and structural properties of fibre-reinforced polymer laminates under simulated fire exposure using artificial neural networks

Load-bearing fibre reinforced polymer laminates soften and decompose when exposed to high temperature fire which may cause significant deformation and weakening, ultimately leading to failure. A combined experimental and modelling study is presented to predict the fire structural survivability of laminates using artificial neural networks based on machine learning. Multiple experimental fire-under-tension load tests are performed under identical conditions to determine the average values and scatter to the surface temperatures, deformation rates and rupture times for an E-glass/vinyl ester laminate. A data-driven modelling strategy based on artificial neural networks is presented that can predict the temperatures and fire structural properties for the laminate when subject to combined fire exposure and tension loading. It is shown that the model gives excellent agreement to the measured surface temperatures, deformations, and time-to-failure of the laminate when exposed to one-sided heating at a constant heat flux. It is envisioned that the ANN based model could be used to assess the fire structural survivability of load-bearing composite structures exposed to fire.

Prediction of flashover time in a compartment fire by CNN-LSTM based deep neural network considering wearable data collection equipment

Real-time prediction of flashover in a compartment fire is of great significance for rescue. This paper investigated a CNN-LSTM neural network to predict the flashover based on the temperature and radiative heat flux in 24 compartment fire cases simulated with FDS. Multiple conditions affecting compartment fires have been taken into account, like room size, room height, fire source location, opening size, number of vents, and fire load. Radiative heat flux meters and thermocouples were positioned at specific heights, to simulate sensors worn by firefighters entering a fire scene. The proposed model consisted of a convolutional layer and three LSTM layers. After training, the deep learning model effectively identified the flashover in compartment fires, with an average relative error of 4.1 % for temperature prediction and 11.2 % for radiative heat flux prediction. In addition, the model was validated on other simulated fire cases and full-scale experimental data, and the results showed good performance. In this paper, the sensitivity of the model to lead time was analyzed, and the application range of the model was determined. The model performed well within the lead time of 25s. In view of the strong performance of the proposed deep learning model in predicting flashover based on wearable sensors (temperature and radiative heat flux), it is believed that this model holds potential for application in a broader range of fire scenarios. This demonstrates the feasibility of using deep learning artificial intelligence models to predict compartment fire development, providing scientific guidance for smart firefighting technology.

A new strategy for developing a Nb microalloyed fire-resistant steel: Effects of boron and cooling rate

This study investigates the correlation among cooling rates, boron addition, and their subsequent effects on the microstructure and mechanical properties of niobium microalloyed steels, specifically with an intent to develop new fire-resistant steels. Slower cooling rates enhanced the ferrite transformation and (Nb,Ti)C precipitation, promoting a shift toward an equilibrium-phase microstructure. A boron addition of 30 ppm was sufficient to increase the fire resistance of niobium microalloyed steels when slowly cooled after thermocontrolled rolling. At rapid cooling rate boron induced the formation of upper bainite. Boron is mainly segregated at the interface of (Nb,Ti)C precipitates, possibly hindering the coarsening kinetics of nanosized carbides and contributing to the maintenance of a high yield strength during fire exposure simulations. The boron addition promoted a ratio between yield strength at 600 °C and room temperature greater than 66%., boron also shifted the strain-temperature curve to higher temperatures in transient tensile tests. These findings contribute to the potential use of boron additions in advancing cost-effective, high-performance, fire-resistant steel development, highlighting the central role of microstructural control and optimized thermomechanical treatment.

Static analysis-based rapid fire-following earthquake risk assessment method using simple building and GIS information

After the occurrences of large-scale earthquakes, secondary damage (e.g., fire following earthquake) can result in tremendous losses of life, properties, and buildings. To reduce these disaster risks, fire following earthquake assessment methods composed of ignition and fire-burned rate estimation models have been utilized. However, previous methods required for large amounts of building and GIS information, and complex modeling and analysis processes, leading to significant time consumption. This paper proposed a static analysis-based rapid fire following earthquake assessment method using simple information and implemented it in Pohang City, South Korea. Based on previous studies, the best-fit model for the ignition rate estimation was selected, and a cluster-based fire-burned rate estimation model was developed using simple building information (e.g., construction year, building occupancy, story, and total floor area) from the public building database (e.g., building registration data). For the fire-burned rate estimation model, fire-resistant structure types were defined using simple building information, and this was utilized to generate clusters of buildings at a regional level by comparing fire-spread distances for each fire-resistant structure type with adjacent distances among the buildings. This proposed method was applied to Pohang City, South Korea, and validated as follows: (1) the selected ignition rate model predicted similar ignition numbers to the actual reported number (actual number of ignitions = 4 vs. predicted number of ignitions = 3), and (2) the fire-burned rate model estimated fire-burned areas with a marginal difference compared to the fire spread simulation (fire-burned area using the proposed model = 13,703.6 m2 vs. results of fire spread simulation = 16,800.0 m2, with an error of approximately 18%).

Egress Safety for STUDIO Residential Buildings

In recent years, the number of studio residential buildings has increased significantly in Korea, as well as in many other countries, due to changes in living patterns. In Korea especially, there have been many fire accidents in studio residential buildings, which have caused a huge number of casualties and property damages, because the buildings were not adequately equipped for firefighting. In this study, the egress safety of a typical studio residential building in Korea is analyzed. Fire simulations were performed with variables of the fire location and the capacity of the smoke exhaust system to estimate the available safe egress time (ASET); egress simulations were also performed with the variable of egress delay time, and the required safe egress time (RSET) was determined. Then, the egress safety was evaluated, and the criteria for egress safety evaluation were proposed based on the simulation results. A studio residential building with a floor plan different from the prototype was used to validate the proposed egress safety criteria. Finally, a simple evaluation model is presented to estimate the required safe egress time (RSET) without simulation and to examine the impact of bottlenecks.

Carbon fibres: Effect of various thermo-oxidative environments on structural and performance damage, both alone and in composites

This study examines the effects of heat and fire on the physical, mechanical and electrical properties of carbon fibre and those in carbon fibre reinforced composites (CFRCs). Carbon fibres were exposed to controlled heating (thermogravimetric analysis (TGA) and a tube furnace) in inert and air (oxygenated) environments and simulated fire (cone calorimetry at 35–75 kW m−2 and jet fire (propane burner) of 116 kW m−2) atmospheres. In inert atmospheres there was a minimal effect on the properties of carbon fibres, but in an oxygenated environment, significant oxidation began at temperatures ≥550 °C, resulting in a reduction in fibre diameter, which reduced further with increasing temperature and exposure duration. Tensile strength and electrical conductivity of carbon fibre decreased with reduction in fibre diameter. CFRCs exposed to 75 kW m−2 in a cone calorimeter and direct flame in a propane burner (116 kW m−2) showed varying degrees of oxidation in CFRC plies, with surface ply fibres experiencing more oxidation and consequent reductions in fibre diameter and tensile properties compared to fibres in underlying plies, where oxidation was limited due to restricted oxygen availability. Fibres exposed to the propane burner exhibited notable damage, including pitting and internal oxidation. Despite this, the overall electrical properties of residual carbon fibres did not significantly decrease, indicating that they still pose an electrical hazard if exposed during a high heat or fire event.

Can green firebreaks help balance biodiversity, carbon storage and wildfire risk?

Green firebreaks (strategically placed plantings of low-flammability vegetation) are designed to reduce the rate of fire spread and thereby increase the suppressibility of fires. Successful examples have led to some fire-prone regions investing heavily in the establishment of green firebreaks as a method of reducing fire risk while improving biodiversity and carbon storage. However, beyond small-scale case studies there has been little research quantitatively exploring the interactions among biodiversity, carbon, and wildfire risk in relation to green firebreaks. Here, we combine a Bayesian Network (BN) analysis, and fire simulations in PHOENIX RapidFire (hereafter Phoenix), to identify planting designs that reduce wildfire risks while also providing positive biodiversity and carbon outcomes. Using a BN analysis, we prioritised optimal planting designs as the combination of elements (e.g., stem density, distance from houses, shrub design, age etc.) that delivered the greatest increase in biodiversity and carbon while reducing fire risk to people and property for eight sites across south-eastern Australia. We ranked combinations of planting designs, prioritising house, and life loss first, to identify optimal designs. Optimal planting designs varied among sites, although the design elements that best reduced risk to houses and lives were consistent. These elements included ‘scattered’ shrubs and planting densities of trees consistent with an open forest structure. Estimated fuel loads for the optimal planting design at each site were used to create a simulated revegetation area in Phoenix. We simulated fire behaviour in Phoenix across a grid of ∼1000 ignitions for each site, and for up to 54 historic weather conditions for a 'current fuel’ scenario (no green firebreaks present) compared with a ‘green firebreak fuel’ scenario. We found that the establishment of a green firebreak did not result in significant changes to fire behaviour at most sites. In some cases, it reduced risk to people and property, and where fire behaviour did change in terms of intensity, frequency, ember attack and overall risk, the differences relative to the current fuel scenario were less than two percent. Overall, simulated green firebreaks in most cases were found to provide biodiversity and carbon benefits without increasing fire risk. These findings illustrate their potential as an effective nature-based solution for managing multiple priorities; however, further testing of real plantings is required to evaluate this potential as an at-scale solution.

Simulated fire video collection for advancing understanding of human behavior in building fires.

The goal of the present research was to develop a video collection of simulated fires to investigate how people perceive growing building fires. In fire safety science, a critical factor to occupant responses to building fires is the pre-movement period, determined by how long it takes an individual to initiate taking protective action with an incipient fire. Key to studying the psychological processes that contribute to the duration of the pre-movement period is presenting human subjects with building fires. One approach used in previous research is to present videos of building fires to individuals via scenarios. The numerical simulations used to model fire dynamics can be used to render videos for these scenarios. However, such simulations have predominantly been used in fire protection engineering to design buildings and are relatively inaccessible to social scientists. The present study documents a collection of videos, based on numerical simulations, which can be used by researchers to study human behavior in fire. These videos display developing fires in different types of rooms, growing at different rates, different smoke thickness, among other characteristics. As part of a validation study, participants were presented with subsets of the video clips and were asked to rate the perceived risk posed by the simulated fire. We observed that ratings varied by the intensity and growth rate of the fires, smoke opacity, type of room, and where the viewpoint was located from the fire. These effects aligned with those observed in previous fire science research, providing evidence that the videos could elicit perceived risk using fire simulations. The present research indicates that future studies can utilize the video library of fire simulations to study human perceptions of developing building fires as situational factors are systematically manipulated.

Simulated fire injury: effects of trunk girdling and partial defoliation on reproductive development of apple trees (Malus domestica)

Fire damage can significantly impact fruit productivity in orchards. However, the effects of nonlethal fire injuries on the reproductive development of apple trees remain poorly understood. To investigate these effects, we implemented three treatments: trunk girdling to simulate fire injury to xylem, defoliation of a third of the canopy (simulated crown fire injury), and a combined treatment (simulated surface fire injury), alongside a control. The experiment was conducted during the 2021–22 growing season using a randomised block design with four biological replicate plots. Girdling was less effective than crown and surface fire treatments in influencing fruit composition during the current growing season, and flowering and fruiting in the following season. The crown and surface fire treatments induced localised detrimental effects on fruit sugar and titratable acidity while stimulating peel blush. Additionally, these treatments led to reduced starch reserves by harvest, which likely disrupted subsequent flowering and crop load near the previously defoliated sections of the canopy. When surface fires damage leaves near the base of the canopy in addition to the trunk, fruit production in the lower part of the canopy is more likely to be compromised in the following season. Crown fires, which cause leaf loss near the apex of the canopy, appear to be particularly detrimental to tree productivity, as the top defoliation treatment impaired carbohydrate reserves in shoot terminals and roots. In conclusion, fire-induced loss of leaf area during fruit growth alters fruit composition in the current growing season and may lead to lower yields in the subsequent season.

Numerical simulation of high-pressure jet fire in on-board hydrogen storage cylinders under fire conditions

Hydrogen fuel cell vehicles (HFCVs) may cause fires in the event of an accident. When the high-pressure hydrogen storage cylinders are exposed to fire for a long time, there will be a risk of catastrophic consequences such as leakage, jet fire and explosion. Fire test is the main method to study the safety performance of hydrogen storage cylinders. During the firing process, the wall and internal temperature of the hydrogen storage cylinder rise gradually, and the thermally activated pressure relief device (TPRD) of the cylinder reaches a certain temperature and then releases high-pressure hydrogen to prevent the cylinder from bursting. As the hydrogen storage cylinder is in the fire environment, the release of high-pressure hydrogen will be ignited to form a jet fire and cause damage to personnel and equipment. In this paper, CFD software Fluent was used to numerically simulate the local fire test of the hydrogen storage cylinder with high-temperature air as the fire source. The variation law of cylinder wall temperature, gas temperature and pressure inside the cylinder, and the activation of TPRD were analyzed. At the same time, on the basis of the fire test simulation, a high-pressure hydrogen jet fire simulation numerical model caused by the TPRD action was established. The velocity field, temperature field and hydrogen concentration distribution of the jet fire were obtained, and the overpressure generated by the jet fire and the flame length size were analyzed. The results show that the high-pressure hydrogen released by TPRD was ignited under the fire condition to produce a jet flame and generated an overpressure in the jet direction, and the influence range of the overpressure was smaller than the influence range of the jet flame.

Evaluating crown scorch predictions from a computational fluid dynamics wildland fire simulator

BackgroundCrown scorch—the heating of live leaves, needles, and buds in the vegetative canopy to lethal temperatures without widespread combustion—is one of the most common fire effects shaping post-fire canopies. Despite the ability of computational fluid dynamic models to finely resolve fire activity and buoyant plume dynamics including heterogenous 3D distributions of forest canopy heating, these models have had only limited use in simulating fire effects and have not been used to evaluate crown scorch. Here, we demonstrate a method of evaluating crown scorch using a computational fluid dynamics model, FIRETEC, and validate this approach by simulating the experiments that were used to develop Van Wagner’s 1973 crown scorch model.ResultsThe average scorch height prediction from FIRETEC compares well with the empirical model derived by Van Wagner, which is the most widely used empirical model for crown scorch. We further find that the 3D buoyant plume dynamics from a steady and homogeneous idealized heat source on the ground results in a spatially heterogenous crown scorch pattern reflecting complex heating dynamics that are best represented by percent scorch rather than height of scorch.ConclusionsThe ability of the computational fluid dynamics model to capture variation in crown scorch due to 3D buoyant plume dynamics provides direct links between forest structure, fire behavior, and fire effects that can be used by forest managers and researchers to better understand how fires result in crown damage under various environmental and management scenarios.

Risk Assessment of Hydrogen Cyanide for Available Safe Egress Time in Fire Simulation

The majority of fatalities in building fires are attributed to asphyxiation caused by toxic gases. Hydrogen cyanide (HCN) is one of the toxic gases that can be released during a fire, posing a lethal risk to humans even at low concentrations. However, analysis of the risk posed by HCN in fire risk assessments using fire simulations is relatively rare. This study conducted fire simulations to examine the potential risks of HCN to occupants during a fire. The simulations considered various fire conditions in residential buildings by varying fuel types, fire growth rates, and HCN yields. The relative risk score (RRS) was derived based on the time to reach the threshold values of parameters considered critical for life safety. The results of the fire simulations indicated that the RRS for HCN was approximately 20–40 points higher than that of O2, CO, and CO2, reaching a maximum of 92 points. However, the risk posed by HCN was found to be limited in comparison to the risks associated with temperature and visibility. Nevertheless, considering that the primary cause of fatalities in fires is asphyxiation due to toxic gases, HCN must be regarded as a critical factor in fire risk assessments. Additionally, since HCN yield values can increase up to nine times depending on temperature and ventilation conditions, the risk posed by HCN could be significantly higher.

Fire behavior simulation of Xintian forest fire in 2022 using WRF-fire model

IntroductionThe behavior of forest fire is a complex phenomenon, and accurate simulation of forest fire is conducive to emergency response management after ignition. In order to further understand the characteristics of forest fire spread and the applicability of WRF-Fire in China, which is a coupled fire-atmospheric wildfire model, this study simulated a high-intensity forest fire event that occurred on October 17, 2022 in Xintian County, southern Hunan Province.MethodsBased on the fire-atmosphere coupled WRF-Fire model, we used high-resolution geographic information, meteorological observation and fuel classification data to analyze the forest fire behavior. At the same time, the simulation results are compared with the fire burned area observed by satellite remote sensing forest fire monitoring data.ResultsThe study found that, the simulated wind speed, direction and temperature trends are similar to the observation results, but the simulated wind speed is overestimated, the dominant wind direction is N, and the temperature is slightly underestimated. The simulated wind field is close to the actual wind field, and the simulation results can show the spatial and temporal variation characteristics of the local wind field under complex terrain while obtaining the high-resolution wind field. The simulated fire burned area is generally overestimated, spreading to the north and southwest compared with the observed fires, but it can also capture the overall shape and spread trend of the fire well.DiscussionThe results show that the model can accurately reproduce the real spread of fire, and it is more helpful to forest fire management.

Effect of air pressure and flow on steels materials in simulated wet fire drives

The corrosion behaviour of N80, 3Cr, and 5Cr steels were investigated under wet fire drive conditions using a high-temperature and high-pressure autoclave, weight-loss method, and scanning electron microscopy analysis. The findings indicated that elevated air pressure (4–6 MPa) and flow rates (0.5–1.5 m/s) significantly exacerbated corrosion, leading to both localised pitting and general corrosion on the steel samples. Predominant components of the corrosion product scales were identified as FeCO3 and Cr2O3. Notably, N80 steel exhibited higher pitting sensitivity compared to 3Cr and 5Cr steel, whereas 5Cr steel demonstrated superior corrosion resistance under the experimental conditions. Consequently, it is recommended to use cost-effective pipes with adequate corrosion resistance, such as various low-alloy pipes with differing chromium (Cr) contents, in oilfield applications. This study underscores the importance of selecting corrosion-resistant pipes as a primary means of corrosion prevention in oil and gas operations. Furthermore, it provides valuable insights into the corrosion performance of diverse steel materials in wet fire drive environments, offering pertinent implications for corrosion mitigation strategies within the industry.

Hakea decurrens invasion increases fire hazard at the landscape scale

Hakea decurrens subsp. physocarpa is an invasive fire-adapted shrub of Australian origin that is quickly expanding in Portugal with potential impacts on fire behavior and fire regime. In this study we examined the effects of H. decurrens on fire hazard by assessing fire behavior indicators at the landscape scale, using a modeling and simulation approach. Six fuel models for H. decurrens were developed through fuel characterization and experimental fires. The fuel models correspond to combinations of developmental stages of H. decurrens populations (Early, Intermediate and Mature) and management (Standing and Slashed fuels). These combinations were used with three levels of H. decurrens invasion, corresponding to 25%, 50% and 75% of cover of the landscape, applied to five real landscapes in northern Portugal (replicates) under three fuel moisture conditions (Low, Medium and High), used as surrogates of weather severity. Fire behavior simulations were conducted with FlamMap software. The relationships between fire behavior indicators (flame length, rate of spread and burn probability) at the landscape level and the four factors tested were analyzed using Generalized Linear Mixed Models. Standing fuels were found to be more hazardous than slashed fuels. Fire-hazard increased with H. decurrens stand maturity and slash, regardless of moisture conditions. The results of this study indicate that H. decurrens expansion might negatively affect the fire regime in the north of Portugal. Our findings add to other known negative impacts of the species on native ecosystems, calling for the need to reinforce its control.

Effect of different boundary conditions on the combustion properties and fire characteristics of typical woods

To provide guidance on the bioenergy potential of wood and its fire spread numerical simulation, insights into the combustion properties and fire characteristics of one hardwood (sassafras) and two softwoods (pine and fir) were investigated. The effect of different boundary conditions are focused. The condensed phase decomposition in thermogravimetric analysis indicated that the main degradation processes for all wood combustion could be divided into two stages, and the corresponding demarcation point of conversion rate is 0.4 and 0.35 for hardwood and softwood, respectively. Three specific regions could further be established based on the activation energies variations in Stage Ⅰ. The governing reaction mechanism of sassafras wood is determined to be the diffusion model connect with order-based model, while the reaction mechanism of softwoods (pine and fir) can be considered as the reaction order (1st to three-halves) type followed by diffusion model. The gas phase flaming in FPA combustion experiments suggest that fir and pine wood have a faster charring rate and produce less CO and CO2 during pyrolysis and combustion. Sassafras requires more time to be ignited owing to its high density, which causes higher thermal inertia therefore decelerating the temperature increase on the sample surface. In addition, the dense structure of sassafras leads to an early appearance of the second HRR and MLR peaks. As hardwood has a higher density than softwood, the thermal insulation in hardwood delays char cracking on sassafras wood surface and enhances the duration between ignition and flameout. Such findings should provide guidance for wood combustion modeling and flame spread simulation in future works.

Experimental study on nonlinear development process of oil-filled equipment fire under external fire source

Abstract The transformer is the key oil-filled equipment in the power system, and its fire behavior seriously affects the safe operation of the power grid. In this paper, in order to analyze the fire development process and combustion behavior of oil-filled equipment, a mesoscale model of transformer equipment was constructed, and fire simulation experiments of transformer equipment under the action of external ignition sources were conducted. The flame temperature, flame height, heat release rate, oil temperature, and pressure were measured. The experimental results show that the oil-filled equipment fire presents the characteristics of nonlinear development. The fire can be divided into three stages: ignition stage, stable growth stage, and combustion mutation stage. The transformer oil near the wall is pyrolyzed by the external heat source, and the combustible gas and transformer oil form a gas-liquid two-phase flow, which is the main reason for the nonlinear development of oil-filled equipment fire. The experimental results are of great significance for the safe operation and fire control of power system oil-filled equipment.