Learning from Real Fires (Forensic Highlights)

The present research aims to explore the effects of condition and memory on escape decision-making.Based on Sayegh et al.'s(2004) intuitive decision-making model and other related research findings,we assume that previous learning and the related memory would influence escape decision-making under crisis condition,and escape decision-making would be different under real fire and simulated fire conditions.We conducted a preliminary training and two main studies to test our hypothesis.We recruited mice rather than human-beings as participants,because it is dangerous to examine human reactions in real fire conditions.Previous research suggested that rodents show similar stressful reactions as humans when encountering crisis(Davis Whalen,2000;Lang et al.,2000).Moreover,rodents share almost the same escape behaviors with humans under crisis(Parr Gothard,2007).We used a 3(condition: real fire,simulated fire,common) × 2(memory: remembered and forgotten) mixed design,and tested a total of 96 mice in the study.Among them,72 received the preliminary training and 24 served as the control group which did not receive the training.Each of the two main studies included 24 mice,and the remaining 24 mice received the forgetting treatment.The 24 mice in each study were randomly and equally arranged into the three conditions(real fire,simulated fire and common),and the 24 mice of the control group are also arranged into such three conditions.The dependent variables for the two main studies are escape time and exit choices.Escape time is used because time or speed is the key in defining intuition and in distinguishing between intuitive and analytical decision-making(Bargh,1994;Betsch,2008;Sadler-Smith,2008).While exit choice is treated as another indicator because finding an exit is crucial for escape decisions(Altshuler et al.,2005;Helbing et al.,2000).Choosing the familiar exit shows the effect of learning and memory.In study one,escape time was examined when only the familiar exit is available(exit1 or 2).If escape time of the experimental groups after training was not significantly different from the memory baseline or even shorter,it would suggest the automatic retrieval from the memory(e.g.,Bargh Chartrand,1999;Dijksterhuis Nordgren,2006),which is an essential character of associative intuition(Glockner Witteman,2010).If escape time of the experimental groups after forgetting was still significantly shorter than that of the control group,it could suggest that forgetting happened only at the conscious level.Retrieval from unconsciousness is no doubt an automatic retrieval process(Dijksterhuis Nordgren,2006).In study two,exit choices were examined when the familiar and unfamiliar exits are both available(exit 1 and 2).It is assumed that if the mice choose familiar but smoky exit rather than unfamiliar but non-smoky exit,it would suggest the effect of learning and memory.After forgetting,if the mice still tended to choose the familiar exit,it would suggest the automatic retrieval from the unconscious,and therefore it is intuitive decision-making.The main findings are:(1) Escape time before forgetting under real fire condition is significantly shorter than the memory baseline,and also significantly shorter than that under simulated fire condition.(2) Escape time of mice under the three conditions is all significantly shorter than that of the control group.(3) When both the familiar and unfamiliar exits are opened and the familiar exit is in smoke,real fire group tends to choose the familiar exit,whereas the other two groups prefer to choose the unfamiliar exit.In conclusion,there is significant difference of the decision-making under real fire and simulated fire conditions.Mice under real fire conditions tend to adopt intuitive decision-making,whereas under simulated fire condition they do not prefer intuitive decision-making.

The present research aims to explore the effects of condition and memory on escape decision-making.Based on Sayegh et al.'s(2004) intuitive decision-making model and other related research findings,we assume that previous learning and the related memory would influence escape decision-making under crisis condition,and escape decision-making would be different under real fire and simulated fire conditions.We conducted a preliminary training and two main studies to test our hypothesis.We recruited mice rather than human-beings as participants,because it is dangerous to examine human reactions in real fire conditions.Previous research suggested that rodents show similar stressful reactions as humans when encountering crisis(Davis Whalen,2000;Lang et al.,2000).Moreover,rodents share almost the same escape behaviors with humans under crisis(Parr Gothard,2007).We used a 3(condition: real fire,simulated fire,common) × 2(memory: remembered and forgotten) mixed design,and tested a total of 96 mice in the study.Among them,72 received the preliminary training and 24 served as the control group which did not receive the training.Each of the two main studies included 24 mice,and the remaining 24 mice received the forgetting treatment.The 24 mice in each study were randomly and equally arranged into the three conditions(real fire,simulated fire and common),and the 24 mice of the control group are also arranged into such three conditions.The dependent variables for the two main studies are escape time and exit choices.Escape time is used because time or speed is the key in defining intuition and in distinguishing between intuitive and analytical decision-making(Bargh,1994;Betsch,2008;Sadler-Smith,2008).While exit choice is treated as another indicator because finding an exit is crucial for escape decisions(Altshuler et al.,2005;Helbing et al.,2000).Choosing the familiar exit shows the effect of learning and memory.In study one,escape time was examined when only the familiar exit is available(exit1 or 2).If escape time of the experimental groups after training was not significantly different from the memory baseline or even shorter,it would suggest the automatic retrieval from the memory(e.g.,Bargh Chartrand,1999;Dijksterhuis Nordgren,2006),which is an essential character of associative intuition(Glockner Witteman,2010).If escape time of the experimental groups after forgetting was still significantly shorter than that of the control group,it could suggest that forgetting happened only at the conscious level.Retrieval from unconsciousness is no doubt an automatic retrieval process(Dijksterhuis Nordgren,2006).In study two,exit choices were examined when the familiar and unfamiliar exits are both available(exit 1 and 2).It is assumed that if the mice choose familiar but smoky exit rather than unfamiliar but non-smoky exit,it would suggest the effect of learning and memory.After forgetting,if the mice still tended to choose the familiar exit,it would suggest the automatic retrieval from the unconscious,and therefore it is intuitive decision-making.The main findings are:(1) Escape time before forgetting under real fire condition is significantly shorter than the memory baseline,and also significantly shorter than that under simulated fire condition.(2) Escape time of mice under the three conditions is all significantly shorter than that of the control group.(3) When both the familiar and unfamiliar exits are opened and the familiar exit is in smoke,real fire group tends to choose the familiar exit,whereas the other two groups prefer to choose the unfamiliar exit.In conclusion,there is significant difference of the decision-making under real fire and simulated fire conditions.Mice under real fire conditions tend to adopt intuitive decision-making,whereas under simulated fire condition they do not prefer intuitive decision-making.

PurposeLight-Gauge Steel Frame (LSF) structures are popular in building construction due to their lightweight, easy erecting and constructability characteristics. However, due to steel lipped channel sections negative fire performance, cavity insulation materials are utilized in the LSF configuration to enhance its fire performance. The applicability of lightweight concrete filling as cavity insulation in LSF and its effect on the fire performance of LSF are investigated under realistic design fire exposure, and results are compared with standard fire exposure.Design/methodology/approachA Finite Element model (FEM) was developed to simulate the fire performance of Light Gauge Steel Frame (LSF) walls exposed to realistic design fires. The model was developed utilising Abaqus subroutine to incorporate temperature-dependent properties of the material based on the heating and cooling phases of the realistic design fire temperature. The developed model was validated with the available experimental results and incorporated into a parametric study to evaluate the fire performance of conventional LSF walls compared to LSF walls with lightweight concrete filling under standard and realistic fire exposures.FindingsNovel FEM was developed incorporating temperature and phase (heating and cooling) dependent material properties in simulating the fire performance of structures exposed to realistic design fires. The validated FEM was utilised in the parametric study, and results exhibited that the LSF walls with lightweight concrete have shown better fire performance under insulation and load-bearing criteria in Eurocode parametric fire exposure. Foamed Concrete (FC) of 1,000 kg/m3 density showed best fire performance among lightweight concrete filling, followed by FC of 650 kg/m3 and Autoclaved Aerated Concrete (AAC) 600 kg/m3.Research limitations/implicationsThe developed FEM is capable of investigating the insulation and load-bearing fire ratings of LSF walls. However, with the availability of the elevated temperature mechanical properties of the LSF wall, materials developed model could be further extended to simulate the complete fire behaviour.Practical implicationsLSF structures are popular in building construction due to their lightweight, easy erecting and constructability characteristics. However, due to steel-lipped channel sections negative fire performance, cavity insulation materials are utilised in the LSF configuration to enhance its fire performance. The lightweight concrete filling in LSF is a novel idea that could be practically implemented in the construction, which would enhance both fire performance and the mechanical performance of LSF walls.Originality/valueLimited studies have investigated the fire performance of structural elements exposed to realistic design fires. Numerical models developed in those studies have considered a similar approach as models developed to simulate standard fire exposure. However, due to the heating phase and the cooling phase of the realistic design fires, the numerical model should incorporate both temperature and phase (heating and cooling phase) dependent properties, which was incorporated in this study and validated with the experimental results. Further lightweight concrete filling in LSF is a novel technique in which fire performance was investigated in this study.

SummaryAccording to the concept of performance‐based design, the fire safety design of the timber structures should be focused on the actual design fire, which represents the realistic fire situation in a given building. However, the current prescriptive only provides a method for evaluating the fire resistance of timber components under standard fire. This paper proposed an energy‐based time equivalent approach (EBTEA) for evaluating the fire resistance of the timber components in realistic design fire scenarios. As an alternative, this method was based on the principle of energy equivalence so that the fire resistance design of timber components exposed to realistic fires was standardized, which can make fire resistance rating database available under standard fire exposure and promote performance‐based design. The validity of the method was verified by comparing the predicted values of EBTEA with the existing values obtained from the nonlinear finite element analysis. The proposed method was further simplified by analyzing the key factors of energy equivalence and using nonlinear fitting methods to facilitate the use of designers. The simplified EBTEA can predict equivalent fire resistance of timber components under realistic design fire scenarios.

Abstract. Wildfires are a complex phenomenon emerging from interactions between air, heat, and vegetation, and while they are an important component of many ecosystems’ dynamics, they pose great danger to those ecosystems, as well as human life and property. Wildfire simulation models are an important research tool that help further our understanding of fire behaviour and can allow experimentation without recourse to live fires. Current fire simulation models fit into two general categories: empirical models and physical models. We present a new modelling approach that uses agent-based modelling to combine the complexity possible with physical models with the ease of computation of empirical models. Our model represents the fire front as a set of moving agents that respond to, and interact with, vegetation, wind, and terrain. We calibrate the model using two simulated fires and one real fire and validate the model against another real fire and the interim behaviour of the real calibration fire. Our model successfully replicates these fires, with a figure of merit on par with simulations by the Prometheus simulation model. Our model is a stepping-stone in using agent-based modelling for fire behaviour simulation, as we demonstrate the ability of agent-based modelling to replicate fire behaviour through emergence alone.

With the adoption of performance-based fire design and the development of new engineered wood products, wood-based mid-rise and high-rise buildings are beginning to be constructed all around the globe. This trend has accelerated efforts to gain more understanding of the risks associated with combustible construction. To compare the differences between combustible and non-combustible construction, a series of full scale fire tests was conducted at Carleton University, the results of which are presented in this paper. These tests represented real bedroom fires with different room construction types: cross laminated timber (CLT) construction, light timber frame construction and light steel frame construction. Results showed that gypsum boards provided a better protection to CLT panels than to light frame walls. Compared with fires in the protected rooms, fire in the unprotected CLT room during the fully developed phase showed accelerated fire development and resulted in over 80% higher total heat release rate (THRR) but slightly lower room temperatures, and external burning contributed to about 64% of the THRR. It was also found that the light timber frame walls and light steel frame walls behaved very differently in the real fire. Furthermore, higher charring rates in both the unprotected CLT panels (1.0 mm/min) and wood studs (1.2 mm/min) were obtained in the real fire than those in the standard fire.

Full scale experiments have been conducted with telephone cables in an attempt to simulate the thermal and atmospheric conditions of a real cable fire. The cables used for the experiments had polyethylene sheathing and were of the same specification and layout as those in the real fire. Temperature, gas concentration, smoke concentration, and hot air velocity were measured and recorded every 15 sec. Thousand pairs of wires were monitored to record the times of their communication stogpages due to fire. Temperatures around the ignition region showed 800 C after 5 min, and the average flame spread over the surface of the cables was about 3 m/min along the tunnel. It was found that the stoppage against time showed a normal distribution with a standard deviation of 2 3 min. Based on the comparison of both stoppage behaviors of the experiments and the real fire, the start time and the propagation of the real fire were estimated.

Fire Detection System using Deep Learning

In this study, 1/1 scale Fire Experiment Laboratory was built to demonstrate the reality of fire with a real fire. In the prepared building, there are 2 fire test rooms measuring 300 × 350cm and 1 observation room measuring 190 × 620cm. In the fire tests, a living room and a bedroom were furnished by calculating the required fire loads. In the fire load calculation, a fire index value of 1.18 was obtained in the bedroom and a fire index value of 1.86 was obtained in the living room. Instant temperature values were taken for each second in the lower, middle and top zones of the fire rooms. The bedroom reached maximum temperatures around 1200 °C in the top zone, 800 °C in the middle zone and 400 °C in the lower zone. And in the living room, maximum temperatures were measured at around 900 °C in the top zone, 600 °C in the middle zone and 300 °C in the lower zone. Moreover, temperature changes outside the building were also recorded. According to the thermal camera images measured, the external wall temperature of the bedroom is around 25–31 °C at maximum temperature and the external wall temperature of the living room is around 41–46 °C at maximum temperature.

Approximately one billion people across the globe are living in informal settlements with a large potential fire risk. Due to the high dwelling density, a single informal settlement dwelling fire may result in a very serious fire disaster leaving thousands of people homeless. In this work, a simple physics-based theoretical model was employed to assess the critical fire separation distance between dwellings. The heat flux and ejected flame length were obtained from a full-scale dwelling tests with ISO 9705 dimension (3.6 m × 2.4 m × 2.4 m) to estimate the radiation decay coefficient of the radiation heat flux away from the open door. The ignition potential of combustible materials in adjacent dwellings are analyzed based on the critical heat flux from cone calorimeter tests. To verify the critical distance in real informal settlement fire, a parallel method using aerial photography within geographic information systems (GIS), was employed to determine the critical separation distances in four real informal settlement fires of 2014–2015 in Masiphumelele, Cape Town, South Africa. The fire-spread distances were obtained as well through the real fires. The probabilistic analysis was conducted by Weibull distribution and logistic regression, and the corresponding separation distances were given with different fire spread probabilities. From the experiments with the assumption of no interventions and open doors and windows, it was established that the heat flux would decay from around 36 kW/m2 within a distance of 1.0 m to a value smaller than 5 kW/m2 at a distance of 4.0 m. Both experiments and GIS results agree well and suggest the ignition probabilities at distances of 1.0 m, 2.0 m and 3.0 m are 97%, 52% and 5% respectively. While wind is not explicitly considered in the work, it is implicit within the GIS analyses of fire spread risk, therefore, it is reasonable to say that there is a relatively low fire spread risk at distances greater than 3 m. The distance of 1.0 m in GIS is verified to well and conservatively predict the fire spread risk in the informal settlements.

FRP strengthening is critically dependent upon the bonding adhesive. The adhesive used is typically an ambient cure epoxy with a glass transition temperature as low as 60°C. This paper describes the performance of bonded FRP strengthening within real compartment fires (the Dalmarnock Fire Tests), one of which was allowed to grow past flashover. The aim of these real fire tests was to complement the laboratory-based fire tests on FRP strengthened members that are currently being undertaken at various research centres. In this study, externally bonded plate and near-surface-mounted FRP strengthening were applied to the ceiling of a concrete structure. The FRP was protected using either an intumescent coating or gypsum boards, alongside FRP that was left unprotected. The tests demonstrated the vulnerability of FRP strengthening during a real compartment fire. The glass transition temperature was rapidly exceeded in the bonding adhesive for all samples. The near-surface mounted strengthening and the gypsum board protected strengthening was in a visibly better condition after the fire.

A critical analysis of the influence of architecture on the temperature field of RC structures subjected to fire using CFD and FEA models

In a cone calorimeter, the specimen receives uniformly distributed irradiance from the cone heater. Producing a heating environment simulating the heating intensity in real fires, this apparatus consequently is capable of providing information of materials relevant to their fire performance. Several previous upward flame spread models utilized the data as input with an assumption of uniformly distributed heat fluxes. Satisfactory flame spread rates were predicted. However, the heat flux in the heating region in upward flame spread is not uniform. This study introduces an alternative protocol of the cone calorimeter and a sample holder by which the following differences were made, including specimen turned 42° before ignition, lower ignition source before ignition, heater removed after ignition, and specimen moved back to vertical orientation after ignition. The heating environment is more consistent to real wall fire conditions. In addition, the prediction of flame spread rate using the alternative test protocol is closer to the measured flame spread rate than standard test methods.

Fire resistive materials (FRMs) are currently qualified and certified based on lab-scale fire tests such as those described in the ASTM El 19 Standard Test Methods for Fire Tests of Building Construction and Materials [1]. While these tests provide an "hourly" rating for the FRM, these ratings have no direct quantitative relationship to the performance of an FRM in an actual fire, e,g., a 2 h rating does not mean that the FRM will protect the steel (or other substrate) for 2 h in a real world fire. Computational heat transfer models offer the potential to bridge the gap between laboratory testing and field performance. However, these models, whether basic one-dimensional or more complex three-dimensional versions, depend critically on having accurate values for the thermophysical properties of the FRM (and substrate) as a function of temperature, to be used as inputs along with the system geometry and fire and heat transfer boundary conditions. Properties required include density, heat capacity, thermal conductivity, and enthalpies of reactions and phase changes. In this report, procedures for determining a consistent set of these input values are presented. Then, quantitative data for a variety of FRMs and several steel substrates that have been obtained from the literature or measured in the Building and Fire Research Laboratory are presented. The utilization of these properties to successfully simulate the thermal response of an FRM-steel layered system is demonstrated for the National Institute of Standards and Technology slug calorimeter experimental setup. Ultimately, similar performance simulations will be executed for El 19-type tests and even real fires.

This report is the second of a three-part series concerning the characterization and modeling of the thermal performance of fire resistive materials (FRMs). These materials are currently qualified and certified based on lab-scale fire tests such as those described in the American Society for Testing and Materials (ASTM) E119 Standard Test Methods for Fire Tests of Building Construction and Materials While these tests provide an "hourly" rating for the FRM, these ratings have no direct quantitative relationship to the performance of an FRM in an actual fire, e.g., a 2 h rating does not mean that the FRM will protect the steel (or other substrate) for 2 h in a real world fire. Computational heat transfer models offer the potential to bridge the gap between laboratory testing and field performance. However, these models, whether basic one-dimensional or more complex three-dimensional versions, depend critically on having accurate values for the thermophysical properties of the FRM (and substrate) as a function of temperature, to be used as inputs along with the system geometry and fire and heat transfer boundary conditions. In part I of this series, procedures for determining a consistent set of these thermophysical properties were presented. Now, in part II, a computational onedimensional multi-layer model for the heat transfer from the fire, through the FRM, to the substrate is developed and verified by comparison to the results of a series of slug calorimeter experiments, previously conducted in the Building and Fire Research Laboratory (BFRL). Ultimately, similar performance simulations will be executed for ASTM E119-type tests and even real fires.