Flame Spread Research Articles

Horizontal flame spread behavior of densified wood: Effect of structural characteristics

Densified wood (DW) is a recently developed material with excellent mechanical properties, considered as an ideal construction material for high-rise timber buildings. However, its flame spread behavior has not been well understood limiting its applications. This work for the first time experimentally and analytically investigates the horizontal flame spread behavior of DW. The influences of material structural characteristics (e.g. density) on heat transfer characteristics are focused. Thermally thin DW with density ranges of 441–1025 kg/m3 and a constant thickness of 2 mm were tested. The flame spread rate, mass loss rate (MLR) and flame height decreased with the increased wood density. A power-law correlation between the dimensionless MLR and dimensionless density is proposed, which well predicts the experiments with an R2 of 0.922. The increased wood density decreases flame height and reduces flame radiation. The gas-phase heat transfer shifts from radiation-dominant to convection-dominant. The longitudinal and radial thermal conductivities of DW increased with wood density. Empirical equations are optimized to predict the thermal conductivities of DW with R2 of 0.870 (longitudinal) and 0.837 (radial). The preheating zone length and the temperature gradient per unit length don’t vary significantly with wood density. Their values are averaged to 25.62 mm and 9.312 K/mm respectively. The longitudinal heat conduction is increased with the wood density due to the increased conductivity. However, this conduction increase is lower than the radiation decrease as the wood density increases. Thus, the total heat flux is slightly decreased with the wood density. The gas phase heat transfer dominates the flame spread. However, the longitudinal heat conduction can account for more than half of the gaseous heat transfer, and thus cannot be negligible. Results of this work improve the understanding of DW flame spread, which helps to evaluate the fire behaviors of timber construction.

Large eddy simulations fire modeling of JIS A 1310 façade calibration test with respect to sidewall

AbstractThe full‐scale façade standard test is widely employed as a comprehensive method to assess the façade fire spread. Within this approach, the calibration test without combustible façade decouples the intricate interaction between gas‐phase combustion and material pyrolysis, which simplifies diagnostics and provides an ideal scenario for model validation. This paper presents large eddy simulations (LES) accompanied by comparisons of calibration tests in accordance with JIS A 1310. The calibration tests were conducted to obtain the flame morphologies, gas‐phase temperature, and heat flux of over‐ventilated façade fires, and the LES modeling is complemented by the modified eddy dissipation model for combustion, the one‐equation model for the sub‐grid scale turbulence, and the discrete ordinate method with the gray mean absorption‐emission approach for thermal radiation. The accuracy of LES data is discussed by comparing with measurements, and the mesh resolution is optimized as 2.5 cm for achieving mesh independency with good qualitative agreement. Furthermore, simulations are conducted to investigate the impact of sidewall distances on façade flame spread. The results highlight that the enhancement of sidewall in façade flame spread occurs under external heat release rate, and the 0.2 m sidewall distance for the designated JIS test is identified as a critical threshold increasing façade thermal load.

Fire modelling: The success, the challenges, and the dilemma from a modeller's perspective

The presentation will review the achievements of my team and the wider IAFSS community in advancing computational fluid dynamics (CFD) techniques for modelling a wide range of fire scenarios, including physics-based sub-models for combustion, soot, and radiative heat transfer as well as robust modelling approaches for flame spread over solids, liquid fuels, and fire suppression. These will be followed by revisiting some successful applications in which CFD simulations influenced design, development of fire protection strategies as well as fire safety in emerging energy technologies like hydrogen and batteries. Some challenges for which robust modelling techniques/approaches are still lacking will be flagged up. Examples in which codes and standards for fire protection should consider explosion safety and the role of consequence analysis using CFD to inform such consideration will be elucidated. Finally, the presentation will reflect on some dilemma faced by fire modelers.

Experimental study on the effect of sloping substrate on the spill fire in a well-confined ship engine room

Fuel leakage in the ship engine room tends to form the dynamically evolving spill fires instead of pool fires, which are full of danger and have been scarcely concerned. In this work, one-dimensional spill fire experiments with different fuel leakage rates and substrate slopes are tentatively conducted in a well-confined model-scale ship engine room to reveal their fire dynamics characteristics. It is found that the development of spill fires in the ship engine room is quite different from that in open space, and it, to a great extent, depends on the specific leakage rate and substrate slope. Typically, the spill fire in the ship engine room can be divided into five stages: continuous spreading, steady burning, burning shrinkage, quasi-steady burning, and decay stages, while the steady or quasi-steady stage does not always appear for all cases. In the quasi-steady burning stage, the thickness of the critical oil film is between 0.97 and 1.55 mm, smaller than that in the open space. The special fire phenomena are mainly attributed to the confined boundary condition, where the competing mechanisms of fuel control and ventilation control effects will collectively influence the development of the spill fire in the ship engine room. Characteristic parameters of spill fires in ship engine room, including flame spread rate, burning area, burning rate, and flame height, are also quantitatively analyzed.

Experimental study on the influence of restricted distance on the discrete flame spread of different widths

In this paper, the effects of varying material widths on the propagation of discrete fires within diverse confined space conditions are investigated experimentally. The width-induced mechanisms influencing the flame spread of discrete solid surfaces in distinct confined spaces are comprehensively discussed, encompassing flame characteristics, pyrolysis front dynamics, flame spread rate, and mass loss rate. The findings reveal that the dimensionless flame height and width conform to a negative power relationship at varying restricted distances. Furthermore, a positive correlation between the width of the PMMA in confined spaces and both the flame spread rate and the mass loss rate is established. Notably, when the width is fixed, the confined space plays a dominant role in the propagation of flames. As the restricted distance decreases, the flames become more slender. Additionally, flame height, flame spread rate, and mass loss rate increase with the widening of the material. However, as the restricted distance becomes smaller, the effect of width becomes less pronounced. Under conditions of same restricted distance, an increased width results in a higher radiation factor, thereby enhancing the thermal feedback of the wall to the material, which in turn accelerates the spread rate of discrete flames. Moreover, due to the dominant role of radiant heat feedback provided by the confined walls to the discrete PMMA panels, the proportion of heat transferred by convection in the flames is minimal. Consequently, the influence of the convective coefficient's width variation on flame spread is relatively minor within confined spaces. Ultimately, an energy conservation model under confined spaces was established, which facilitated the analysis of the comprehensive mechanism of flame spread influenced by varying confined spaces and material widths.

The limiting oxygen volume fraction for opposed flame spread extinction

The limiting oxygen volume fraction (LOC) for flame spread is a parameter used to determine flammability and fire hazard. The LOC depends on the solid material, geometry, and environmental conditions. It is important to understand the relationship between these conditions and the LOC, particularly for environments different than Earth atmospheres. Future spacecraft will have sub-atmospheric cabin pressures designed to reduce preparation time for extravehicular activities. This work attempts to provide further information on this aspect of solid fuel flammability by conducting experiments to determine the effect of ambient pressure and external radiant flux for flames spreading downward over cylindrical samples of black polymethyl methacrylate (PMMA). The experimental methodology is based on test methods ASTM 2863 (LOI) and ASTM E-1321 (LIFT), that are used to determine some aspects of material flammability, specifically extinction and flame spread under external radiant heating. Experiments are conducted in a pressure chamber at pressures from 40 to 100 kPa with radiant heating from 0 to 3.3 kW/m2. Results show that decreasing either the ambient pressure or the external radiant flux increases the LOC of the PMMA. The results are correlated in terms of the ratio of the partial pressure of oxygen, radiant flux, and ambient pressure with an equation that includes the LOC at atmospheric pressure. It is found that the heat transfer and chemical kinetic mechanisms that control the flame spread are primarily dependent on the oxygen molar fraction, explaining the experimental observations. The data from this work will be compared with experiments to be conducted in the International Space Station (ISS) under the SoFIE-MIST project to provide further understanding of the effect spacecraft environments have on the LOC of materials. The results will give further insight into the flammability of materials, particularly at sub-atmospheric ambient pressures found in spacecraft, aircraft, and high-altitude locations.

Effect of oxygen concentration, pressure, and opposed flow velocity on the flame spread along thin PMMA sheets

The flame spread of thin PMMA sheets has been studied both in opposed configuration under microgravity (9.3 s drop tower) and in downward configuration under normal gravity conditions. The position of the flame front has been determined with the help of an infrared camera. Propagation rates have been measured over a range of 30–200 mm/s forced opposed flow (only microgravity), 60–101.3 kPa ambient pressure, and oxygen concentrations of 21–35.4 vol. %, which corresponds to real environmental conditions on spacecraft. In addition to variations along the normoxic curve with constant oxygen partial pressure, the parameters oxygen and pressure were examined independently of each other in order to be able to determine their respective effects. The variation of the flow velocity was carried out at 60 kPa and 35.4 vol. % oxygen. The results for both microgravity and normal gravity are discussed separately on the basis of the respective atmospheric effect. Along the normoxic curve, a significant increase in the flame spread rate was found with increasing oxygen concentration and correspondingly decreasing pressure. This results in a significant increase in the risk of fire with regard to future exploration missions. The effect of the flow velocity cannot be neglected in the investigated velocity range and was linked to the influencing parameters in a correlation. Based on this parameterization, a linear function is given that reflects both the downward and the opposed results well. This function can be used to predict the flame spread in the given parameter space. Under normal gravity, the tested samples show a slightly increased propagation rate in most cases. This can essentially be attributed to the difference in velocity between the buoyant flow and the forced flow in µg.

Far-field signature of fire in low gravity: Influence of ambient oxygen content and pressure on size distribution of smoke particles

Fire detection in space habitats is hindered by the unique characteristics of smoke particles released by weakly buoyant flames. For the first time, non-invasive measurements of particle concentration, size, and light absorption properties are performed in the downstream flow of a flame spreading in microgravity. The studied flame spreads over Low Density PolyEthylene (LDPE) coated Nickel/Chrome (NiCr) wires in an opposed oxidizer flow. The far-field signature can then be contrasted with previous measurements performed over the same configurations and related to the morphology of soot particles sampled in and behind the flame.Following measurements regarding the impact of gravity, the influence of ambient pressure and oxygen content on the characteristics of the particles is reported in microgravity. Increased pressure boosts the soot production rate, leading to higher concentration levels of particles detected for any class of particle size. Higher pressure also causes larger aggregates that exhibit a higher level of polydispersity among the constituting primary particles. This tends to enhance light absorption, more specifically at larger particle size. Less intuitively, increased oxygen content accelerates the pyrolysis rate, therefore hastening soot particles’ maturity, and ultimately increasing the detected soot concentration. In addition, enhanced oxygen content reduces organic carbon within particles, boosting light absorption regardless of the particle size. The far-field signature and its sensitivity to the ambient conditions are in decent agreement with observations of soot particles collected at the trailing edge of the flame. As a result, combining higher oxygen content and lower pressure in future space habitats as projected by the main space agencies can cancel out the effect on the detection threshold of a given flame if this threshold only includes the concentration level.

Quantitative prediction of the flammability limits of filter paper in microgravity conditions

Under JAXA's FLARE project, ongoing since May 2022, long-duration microgravity combustion experiments have been conducted using filter paper. These experiments utilize the Solid Combustion Experiment Module (SCEM) in the ISS/Kibo module. The dimensions of the filter paper used were 130 mm in length and either 40 mm or 20 mm in width, with the experiments conducted in an environment of 298 K and 100 kPa. This paper highlights the results of 19 flame spread experiments performed in opposed-flow environments and compares the predetermined flammability limiting curve with the flammability map obtained in orbital experiments. One of the objectives of the FLARE project is to predict the flammability limiting curve in microgravity environments based on the results of tests that can be conducted on the ground. These experiments produced intriguing findings, demonstrating the effects of both radiation loss and blow-off during flame spread, as the opposed flow velocity was incrementally increased from 10 cm/s to 18 cm/s in 2 cm/s steps, within an environment containing 15.5 % oxygen concentration. When comparing with the combustible map obtained from orbital experiments, it was discovered that the original limit curve accurately replicated the blow-off branch but overestimated the flammable region in areas with opposed-flow velocity below 10 cm/s. This discrepancy was attributed to the inadequate consideration of the effect of opposed-flow velocity on the radiation quenching branch. Therefore, by considering the velocity distribution measurement results obtained in orbit and modeling the impact of the effective flow velocity opposing the flame front on both the radiation quenching branch and the blow-off branch, we successfully quantitatively replicated the results of the orbital experiments. These findings indicate that the flammability limits of thin flat plate samples can be accurately predicted by properly assessing the effect of the boundary layer on the sample surface.

Cyclic pattern along the downward flame spread over cylindrical samples in partial gravity

Downward flame spread over thin electrical wires is investigated at reduced gravity. The wire is made of a Nickel/Chrome (NiCr) core coated with Low Density PolyEthylene (LDPE). The flame spreads in an opposed flow and the conditions of the free stream, i.e. oxygen content, pressure, and forced flow velocity, are varied. Parabolic flights allow experiments to be performed at various gravity levels to reproduce conditions met on Earth, Mars, the Moon, or in microgravity. Past studies showed that dripping of the molten coating occurs at Earth gravity but disappears in microgravity. A new process is here systematically observed at intermediate gravity levels: while the flame front spreads at a steady rate, a molten droplet of the LDPE coating exhibits a cyclic motion ahead of the flame front. This is driven by the balance among the gravitational, viscous, and adhesion forces. Gravity primarily powers the force driving the droplet away from the flame front, while the adhesion force ensures the droplet’s attachment to the wire. The viscous force critically influences the droplet’s velocity, which shows an inverse relationship with viscosity. This specific cycle can be decomposed into two stages where the aforementioned balance can be evaluated to clarify the conditions of the cycle’s existence. Experimental results show that increasing the oxygen content tends to shorten the cycle by increasing the velocity of the flame front, while increasing the pressure also shortens the cycle by increasing the cooling rate of the droplet. This cyclic behavior can trigger flame extinction at Martian gravity levels, even in scenarios where flames propagate under normal and microgravity conditions. These findings can significantly impact fire safety strategies in environments with intermediate gravity levels.

The relative position of pyrolysis onset and flame front location for downward flame spread

The interaction of solid and gas phase phenomena in downward flame spread is studied under buoyant conditions. Phosphor thermometry (PT) was used to determine the spatiotemporally resolved surface temperatures both ahead and beneath the flame while CH* chemiluminescence allowed for the quantification of the flame location relative to the fuel surface. The combination of these measurements allowed for the superposition of the flame front location and the surface temperature distribution. The standoff distance (∼1.1 mm) and preheated distance along the surface (∼3.5 mm) were quantified. The application of PT allowed for the continued measurement of surface temperatures in the pyrolyzing region beneath the flame, where the surface temperature was found to remain relatively constant between 300 and 350 °C. The combination of measurements allowed for the calculation of the gas-phase heat flux distribution ahead of the flame front using a point-source approximation, which compared well with previous studies. PT and CH* measurements were used to expand previous work to quantify the thermal gradients through the solid both ahead of and behind the flame front. Surface temperature measurements were coupled with thermogravimetric analysis (TGA) and a simple pyrolysis model to determine that the onset of pyrolysis occurs approximately 1 mm ahead of the flame front. This approach was also used to illustrate that pyrolysis gases were generated in sufficient volume to result in a flammable mixture at the leading edge of the flame. This study highlights the benefit of optical diagnostic techniques in exploring the controlling mechanisms of flame spread, and provides insight into the interaction of solid and gas-phase phenomena in downward flame spread.

Numerical simulation study on combustion characteristics of square hollow oil pool fire in ship cabins

To study the combustion phenomenon of square hollow fire sources in ship cabins, PyroSim software was used to establish numerical models of different inner and outer edge lengths of the same fire source area. FDS software was used to numerically simulate fire temperature and smoke spread, observe and record the flame and smoke spread characteristics in cabins under different working conditions, and analyze the temperature distribution of flame plumes and the temperature field characteristics inside the cabin. The results indicate that: During the combustion process of a square hollow fire source in a ship’s cabin, a typical flame convergence phenomenon occurs, and the flame temperature of the fire source combustion first increases significantly and then gradually stabilizes. The smaller the inner edge length of a square hollow, the higher the combustion temperature of the fire source, the faster the heating rate, the more obvious the phenomenon of fire convergence, and the more intense the flame combustion. As the inner edge length increases, the influence of flame combustion at the center of the square decreases, and the peak temperature of the flame decreases with a smaller amplitude and tends to concentrate.

Flame Spread Experiments on a Horizontal Preheated Cable Layer

Flame Spread Experiments on a Horizontal Preheated Cable Layer

Characterizing the wave of discrete glowing smolder spots generated at the base of a concurrent microgravity flame spreading over a cotton-fiberglass fabric

Six Saffire experiments with SIBAL fabric (75 % cotton, 25 % fiberglass) were carried out in orbit aboard different Cygnus vehicles after unberthing from the ISS. In five tests the ambient atmosphere was ISS cabin air, which contained approximately 21 % O2 at ambient pressure. The sixth test was an exploration atmosphere (31 % O2 at 54 kPa). Relative humidity was measured but not controlled. In each experiment, after the concurrent flame passed over the fabric, glowing smolder spots were observed to spread over the hot reactive char. Characteristics of this smolder wave have been measured as a function of forced flow and relative humidity level. The spots showed a slight preference to spread concurrently in air but had no preference in lateral spread. An increase in the flow modestly increased the number of spots while an increase in ambient air relative humidity diminished the length of the smolder wave and linearly decreased the number of spots due to the endothermic heterogenous surface reaction C(s) + H2O > CO + H2. The smolder wave was found to have a lognormal distribution of glowing spots relative to distance from the onset of smolder. The smolder wave spread at the same speed as the flame after an ignition transient since the spots were generated just beyond the flame base when the hot reactive char was exposed to the fresh oxidizer flow. The smolder wave was made up of growth and extinction phases. The initial population growth occurs at an intrinsic growth rate, and after peaking, the population exhibits an exponential decay in the number of spots as extinctions become dominant. The populations follow a logistic population model with carrying capacity (K) and intrinsic growth rate (ro) parameters taken from the data and assuming a starting population of unity.

Strut-assisted injection of liquid fuel in a supersonic combustor

This paper investigates mixing and combustion of liquid fuel in a model scramjet combustor with strut-assisted injection adjacent to a cavity flame holder. Fuel/air mixing in a Mach 2.2 cross-flow and stagnation temperature of 1300 K was studied using filtered IR imaging. The insight gained by this method is discussed along with the method limitations. The impact of the strut assisted injection on the mixing of the liquid fuel was seen by the shorter vaporization distance, as the fuel vaporized upstream of the cavity, in contract to a liquid jet that expanded above most of the cavity when the strut was absent. Moreover, the fuel jet was seen to propagate laterally to the tip of the strut, which lead to lateral dispersion of the fuel up the strut height. The combustion experiments were conducted by piloting the cavity with ethylene and were imaged using CH* chemiluminescence. At stagnation temperature of 1300 K thermal choking and stable combustion occurred at equivalence ratio of 0.43, with the flame stabilizing at the strut wake, above the cavity. Intermittent thermal choking occurred at leaner condition. The flame stabilized on the cavity shear layer when the shock train moved downstream of the cavity and the flame stabilized at the strut wake when the shock train moved upstream. Thermal choking with the strut guided injection relative to a combustion only in the cavity without the strut is the result of the observed higher mixing efficiency, that enabled flame spreading into the supersonic flow and higher heat release. At a higher stagnation temperature of 1500 K the shock train moved downstream of the cavity, and the flame stabilized on the cavity shear layer. Finally, it is shown that the combustion efficiency could be further improved by injecting the fuel from the strut side walls.

Effect of silver inhibition on the ceramic foam as flame suppression

Aluminium dust explosions pose significant safety and economic challenges in various industrial processes. Due to this, the current research explores an innovative approach by inhibiting the silver nanoparticles (Ag NPs) to ceramic porous form substrate as a flame suppressant in order to mitigate the risks associated with these explosions. The antimicrobial and non-toxic qualities of silver are also attractive to be applied in medical and food technology. However, the interfacial adhesion between the metallic (nanosilver) and non-metallic (silica-based-ceramic) is still vaguely studied due to the mechanical and surface energy mismatch between the organic surface and the inorganic layers. From this study, the physicochemical and mechanical properties of the silver-coated ceramic foam were analyzed using X-ray diffraction, field emission scanning electron microscopy with energy dispersive X-ray, thermogravimetric analysis, and compression test. From the mechanical testing, it was found that the percentage increase of maximum load for silver-ceramic foam from the original ceramic foam was about 60%. The results indicate that silver-coated foam has a better compressive strength of 0.93 MPa as compared to 0.58 MPa by the original ceramic. The inhibition effect of Ag NPs powder on the explosion pressure evolution and flame spread mechanism of aluminium powder at different concentrations and particle sizes was tested using the Hartmann experimental system.

A Novel Flame-Retardant, Smoke-Suppressing, and Superhydrophobic Transparent Bamboo.

Silica glass, known for its brittleness, weight, and non-biodegradable nature, faces challenges in finding suitable alternatives. Transparent wood, made by infusing polymers into wood, shows promise but is hindered by limited availability of wood in China and fire risks associated with its use. This study explores the potential of utilizing bamboo, which has a shorter growth cycle, as a valuable resource for developing flame-retardant, smoke-suppressing, and superhydrophobic transparent bamboo. A 3-layered flame-retardant barrier, composed of a top silane layer, an intermediate layer of SiO2 formed through hydrolysis-condensation of Na2SiO3 on the surface, and an inner layer of Na2SiO3, has been confirmed to be effective in reducing heat release, slowing flame spread, and inhibiting the release of combustible volatiles, toxic smoke, and CO. Compared to natural bamboo and other congeneric transparent products, the transparent bamboo displays remarkable superiority, with the majority of parameters being notably lower by an entire order of magnitude. It achieves a long ignition time of 116 s, low total heat release (0.7 MJ/m2), low total smoke production (0.063 m2), and low peak CO concentration (0.008 kg/kg). Moreover, when used as a substrate for perovskite solar cells, the transparent bamboo displays the potential to act as a light management layer, leading to a marked efficiency enhancement of 15.29%. The excellent features of transparent bamboo make it an enticing choice for future advancements in flame-retardant glasses and optical devices.

Improving Fire Resistance in E-Vehicles: A Study on MPP-Enhanced S-Glass/Phenolic Hybrid Composites

The automotive industry has benefited greatly from the advancements in fireproof, high-strength fiber-reinforced polymer matrix composites during the last several decades. S-Glass/Phenolic hybrid composites made by hydraulic compression molding are the subject of this investigation into the effects of Melamine Polyphosphate (MPP) on their thermal and fire performance. In addition to studying the morphology of the samples, researchers analyzed the thermal and fire performances of hybrid laminates made with neat laminates and varying compositions of MPP (ranging from 3% to 15% by weight). Tests included UL 94, limiting oxygen index (LOI), and the impacts of accelerated heat ageing on the behavior of the hybrid specimens. The results demonstrated that specimens filled with higher MPP performed better in a fire compared to neat samples. According to the results in UL 94 horizontal burning test and vertical burning test, specimens filled with 12 and 15 wt. % MPP demonstrated a reduced flame spreading rate and fulfilled the V-0 criteria. Simultaneously, the energy absorption capacity of hybrid configurations is severely limited by heat ageing, due to its temperature and duration dependent. The research proved that the suggested S-Glass / Phenolic / Melamine Polyphosphate hybrid composites were suitable for creating environmentally friendly electric vehicle battery housings and fire-resistant automotive parts.

A Systematic Review on Cavity Fires in Buildings: Flame Spread Characteristics, Fire Risks, and Safety Measures

Fire spread scenarios associated with concealed cavity spaces have been relatively less discussed. The variation in studies with respect to geometry, influential parameters, and protection strategies has been an obstacle to deriving more generalized solutions in terms of cavity fire in buildings. A systematic literature review was conducted following the PRISMA method to identify the conclusive fire behaviour, safety risks, and protection strategies to enable future researchers to address cavity fire scenarios effectively, avoiding catastrophic disasters. This study identified that relative to open-fire scenarios, cavity fires could result in up to 10 times higher flame spread, up to 14 times higher heat exposure, and temperature conditions 13 times higher. Increased toxicity and smoke velocity are also found with cavity fires. Fire protection strategies and their efficiency were identified for a range of cavity geometries. Altogether, cavity spaces, especially narrow ones, cannot be neglected during fire safety, and proper risk identification is required to ensure the safety of the buildings and the occupants in a fire scenario.

Investigation on the burning behaviors of the combustible ceiling with the impingement of an incipient fire source

This paper presents experimental and theoretical research on ceiling jet, burn-through, and flame spread behaviors of the combustible ceiling with the impingement of an incipient fire source. Medium-density fiberboard was utilized as the combustible ceiling in experiments, and the ceiling height above the fire source was varied. The flame radial length (rf), ceiling surface temperature, flame shape and spread rate after burn-through, and radiative heat flux were measured. Results showed that rf beneath the combustible ceiling was significantly longer than that of the non-combustible ceiling. Theoretical analysis was conducted for rf based on energy distribution, and an expression of rf beneath combustible ceiling was proposed. The burn-through behavior of the combustible ceiling was explored by combining the ceiling temperature, small-sized experiment and Fourier law. The advancement of char front along the ceiling thickness was calculated, which was the key point during burn-through. Upper surface flame spread in a circular shape after burn-through, and an expression of spread rate was obtained. A three-dimensional radiation model was adopted to calculate the radiative heat flux received by the unburned ceiling surface during the dynamic flame spread process, the calculated results agreed well with the experimental data.