Prediction of temperature and velocity profiles in a single compartment fire by an improved neural network analysis

A novel hybrid Artificial Neural Network (ANN) model, denoted GRNNFA, has been developed for fire studies. The major feature of the model is its ability to work in a noisy environment, which is usually the case with fire experiment data. The GRNNFA model is applicable in the determination of the location of the thermal interface in a single compartment fire. The performance of the GRNNFA has been proven to be comparable to that of the Computational Fluid Dynamics (CFD) model. In addition, the computational speed of the GRNNFA model is much faster than that of the CFD model. However, the original GRNNFA model is only capable of handling the training samples with scalar output. This shortcoming restricts the application area of the model. Hence, this paper presents a modification of the original GRNNFA model for multi-dimensional prediction problems. It also demonstrates the first application of ANN techniques to predicting the velocity and temperature profiles at the center of the doorway in a single compartment fire. These profiles are commonly used to benchmark the performances of CFD models. They are employed in this study to evaluate the performance of the modified GRNNFA model. The predicted profiles are compared with the experimental results and the results simulated by the CFD model. These results show that the prediction errors of the GRNNFA model are less than those of the CFD model in actual application.     

An advanced ANN model for predicting the rotational behaviour of semi-rigid composite joints in fire using the back-propagation paradigm

This paper describes an artificial neural networking (ANN) model developed to predict the behaviour of semi-rigid composite joints at elevated temperature. Three different semi-rigid composite joints were selected, two flexible end-plates and one flush end-plate. Seventeen different parameters were selected as input parameters representing the geometrical and mechanical properties of the joints as well as the joint’s temperature and the applied loading, and used to model the rotational capacity of the joints with increasing temperatures. Data from experimental fire tests were used for training and testing the ANN model. Results from nine experimental fire tests were evaluated with a total of 280 experimental cases. The results showed that the R² value for the training and testing sets were 0.998 and 0.97, respectively. This indicates that results from the ANN model compared well with the experimental results demonstrating the capability of the ANN simulation techniques in predicting the behaviour of semi-rigid composite joints in fire. The described model can be modified to study other important parameters that can have considerable effect on the behaviour of joints at elevated temperatures such as temperature gradient, axial restraints, etc.     

Numerical study on temperature distribution of high-strength concrete-filled steel tubes subjected to a fire

High-strength concrete-filled steel tube (CFST) columns offer a number of benefits and are widely used in high-rise building construction. This paper presents a brand new thermal modeling approach in comparison with tremendous test data for the heat transfer analysis. A heat transfer model has been developed to predict the thermal response of high-strength CFST columns under standard fire conditions with consideration of a number of parameters: steel and furnace emissivity, thermal interface conductance and concrete strength. The verified numerical models discussed the variation of emissivity of steel surfaces and thermal interface conductance in fire. It can be concluded that the influence of emissivity and thermal interface conductance is considerable in the numerical analysis. It is also demonstrated that thermal behavior of high-strength CFST columns subjected to fire during heating and cooling stages, and providing the guidance on predicting thermal response of high-strength concrete-filled steel tube columns.     

On Radiant Network Models of Thermocouple Error in Pre and Post Flashover Compartment Fires

Thermocouples are often used to obtain gas temperature measurements in compartment fires. Such measurements are subject to a thermal lag during fire growth, but the main problem is a steady-state error induced by radiant heat transfer at the thermocouple surface. This error is sensitive to thermal parameters of the flame, compartment structure, thermocouple surface and combustion products; and is also influenced by the size and position of both the flame and thermocouple. The literature contains models of varying sophistication to enable an assessment of steady-state error. A model is now proposed that makes use of the concept of radiosity. Developed from radiant network theory, the model can be applied to both pre-flashover and post-flashover conditions. Experiments have been performed using different sizes of thermocouple and the models compared. The simpler models pre-date the more sophisticated and predict much larger errors than the latest published and current versions.     

Turbulence statistics in a fire room model by large eddy simulation

Fire and smoke movement in a room is influenced by the turbulence characteristics (such as Reynolds stress, turbulent heat flux, etc.) of the flow and temperature fields. In order to accurately predict fire and smoke movement by computational fluid dynamics (CFD), it is necessary to verify these turbulence quantities. The purpose of this study is to predict the turbulence structure of the flow and temperature fields due to a fire in the compartment by large eddy simulation (LES) using detailed experimental data to verify the simulation results. The results show reasonably good agreement with experimental data for both the mean flow properties and the turbulence quantities with the exception of the region near ceiling. This study provides useful information for verifying LES technique when applied to compartment fires.     

Large eddy simulation of compartment fire with solid combustibles

For properly describing practical building fire processes with solid combustibles, the pyrolysis kinetics model of solid combustibles and the large eddy simulation (LES) approach are applied to the simulation of the thermal decomposition of the polyurethane foam (PUF) slab and the space fire spread in a compartment. The instantaneous variations of the heat release rate of the PUF slab, the smoke temperature, and the smoke interface height with time are obtained under different ventilation conditions. They are in agreement with the measured data. The ventilation conditions have distinct effects on the interactions between the pyrolysis of the PUF slab and the space fire spread. Influenced by the space fire spread, the heat flux on the top plane of the PUF slab exhibits a non-uniform distribution. The PUF slab is consumed in an asymmetric manner.     

An intelligent approach to assessing the effect of building occupancy on building cooling load prediction

Building cooling load prediction is one of the key factors in the success of energy-saving measures. Many computational models available in the industry have been developed from either forward or inverse modeling approaches. However, these models usually require extensive computer resources and lengthy computation. This paper discusses the use of the multi-layer perceptron (MLP) model, one of the artificial neural network (ANN) models widely adopted in engineering applications, to estimate the cooling load of a building. The training samples used include weather data obtained from the Hong Kong Observatory and building-related data acquired from an existing prestigious commercial building in Hong Kong that houses a mega complex and operates 24 h a day. The paper also discusses the practical difficulties encountered in acquiring building-related data. In contrast to other studies that use ANN models to predict building cooling load, this paper includes the building occupancy rate as one of the input parameters used to determine building cooling load. The results demonstrate that the building occupancy rate plays a critical role in building cooling load prediction and significantly improves predictive accuracy.     

Treatment of design fire uncertainty using Quadrature Method of Moments

The use of a single-design fire in a performance-based fire design code typically fails to account for the inherent uncertainty in knowledge of the future use of the space. Uncertainties in knowledge of intended use and the implications in terms of fuel loading and potential heat release rate can be bounded using probabilistic methods. Use of a cumulative distribution function (CDF) and the related probability density function (PDF) specify the best available estimate of the probability (likelihood) of a fire of given size to take place in a compartment. Monte Carlo simulation is a widely used computational method for treating uncertainty that might be described by a PDF. In this technique, one samples the uncertain variables from their underlying PDFs and runs a fire model for each sample. For complex fire models, this approach may be computationally intractable. In this work we present a computationally efficient technique called the Quadrature Method of Moments (QMOM) for propagating uncertainty bounds in distributions. In QMOM one solves for only the moments of a relevant uncertain parameter. The CDF of the uncertain parameter provides all the statistical information required for risk assessments. We consider a simplified propagation of uncertainty problem. Results using both the ASET and CFAST fire models indicate that computation of the moments of the PDF using QMOM and the reconstruction of the CDF by matching the moments with those of a four-parameter Generalized Lambda Distribution (GLD) give accurate results at a significantly smaller computational cost.     

Prediction of fire resistance of concrete filled tubular steel columns using neural networks

A functional relationship between the fire resistance of a concrete filled steel column and the parameters which cause the fire resistance is represented using an artificial neural network. Experimental data obtained from previous laboratory fire tests are used for training the neural network model. The model predicted values are compared with actual test results. The results indicate that the model can predict the fire resistance with adequate accuracy required for practical design purpose. The developed neutral network can be used to predict the fire resistance of similar columns under fire by observing various factors influencing the resistance such as: (a) structural factors, (b) material factors, and (c) loading conditions. The structural engineer is required to provide the magnitude of these influencing factors as inputs to the neural network and the network will predict the fire resistance, based on the combined effect of these factors. This system can be used by structural engineers to predict the resistance of fire in similar concrete filled steel columns without conducting costly fire tests, by using the known parameters such as column dimensions, column height, and loading conditions.     

PSO优化ELM在火灾探测中的应用

为提高火灾探测精度,避免标准ELM陷入局部最优,本文基于火灾特征值CO浓度、烟雾浓度、温度建构了一种基于粒子群(PSO)优化极限学习机(ELM)的火灾探测模型,通过PSO优化ELM输入层与隐含层权值以及偏置,利用最优值进行极限学习机网络训练,将训练好的网络对测试样本进行预测并验证方法有效性.研究显示,PSO-ELM的均...     

COMPBRN III — A fire hazard model for risk analysis

COMPBRN III is a deterministic fire hazard computer code designed to be used in a probabilistic analysis of fire growth in a compartment. Its primary application to date has been the assessment of fire risk in the nuclear power industry. COMPBRN III follows a quasi-static approach to simulate the process of fire growth during the pre-flashover period in an enclosure. Physical models which quantify the thermal hazard (including temperature and heat fluxes) during a compartment fire are developed. Simulations of experiments are performed to test the accuracy of the improved hot gas layer model used in this version of COMPBRN in predicting the behavior of large-scale fires.     

Numerical investigations on the thermo-mechanical behavior of steel-to-timber joints exposed to fire

A three-dimensional finite element model is developed to predict the thermo-mechanical behavior of steel-to-timber doweled joints in tension parallel to grain exposed to fire. To manage the plastic yielding of the materials, the mechanical model is based on the von Mises criterion for steel and the Hill criterion for timber. In fire, the material characteristics depend on the temperature. Two different meshes are used for the thermal and the thermo-mechanical models. The thermal model is continuous, to take account of the thermal continuity between the joint components. The thermo-mechanical model is discontinuous, to consider the contact evolution between the joint components. The thermal model is used to predict the evolution of the temperature field inside the joint which depend on the gas temperature. It is validated on the basis of measured temperatures during fire tests. The complex transformations in wood during fire are represented by apparent values of thermo-physical characteristics proposed in the bibliography and calibrated on the basis of the experimental measurements. The mechanical model is validated by comparison with the experimental results of joints in normal conditions. The thermo-mechanical model is validated by considering the experimental failure times of some joints. The numerical models showed a good capacity to simulate the behavior of the timber joints in cold and in fire situations. These developed and tested models can be used as a general tool to analyze the behavior of a large variety of joint configurations to constitute a data base that can be used in safe and economic practice of fire engineering of wood joints.     

A small-world network model for the simulation of fire spread onboard naval vessels

This work is devoted to the development of a small-world network model to predict real-time fire spread onboard naval vessels. This model takes into account short-range and long-range connections between neighboring and remote network compartments. Fire ignition and flashover, as well as fire transmissions through the walls and ventilation ducts are simulated using time-dependent normal probability density functions. Mean durations of fire transmission through the walls and ducts are determined by a three-zone model and a one-dimensional CFD code, respectively. Specific experiments are conducted in a steel room, representative of a naval vessel compartment, in order to validate the zone model. Then a proof of concept is developed by applying the network model to a full-scale vessel mockup composed of 113 compartments on 7 decks. A statistical study is conducted to produce fire risk maps, classifying the vessel compartments according to their propensity to burn.     

Dynamic wind response of tall buildings using artificial neural network

This paper presents an alternative approach for predicting the dynamic wind response of tall buildings using artificial neural network (ANN). The ANN model was developed, trained, and validated based on the data generated in the context of Indian Wind Code (IWC), IS 875 (Part 3):2015. According to the IWC, dynamic wind responses can be calculated for a specific configuration of buildings. The dynamic wind loads and their corresponding responses of structures other than the specified configurations in IWC have to be estimated by wind tunnel tests or computational techniques, which are expensive and time intensive. Alternatively, ANN is an efficient and economical computational analysis tool that can be implemented to estimate the dynamic wind response of a building. In this paper, ANN models were developed to predict base shear and base bending moment of a tall building in along‐ and across‐wind direction by giving the input as the configuration of the building, wind velocity, and terrain category. Multilayer perceptron ANN models with back‐propagation training algorithm was adopted. On comparison of results, it was found that the predicted values obtained from the ANN models and the calculated responses acquired using IWC standards are almost similar. Using the best fit model of ANN, an extensive parametric study was performed to predict the dynamic wind response of tall buildings for the configurations on which IWC is silent. Based on the results obtained from this study, design charts are developed for the prediction of dynamic wind response of tall buildings.     

The modeling of fire spread in buildings by Bayesian network

Fire spread modeling is very important to fire safety engineering and to insurance industries involved in fire risk–cost analysis of buildings. In this paper, the Bayesian network is introduced. The directed acyclic graph of a fire spread model is presented. When the fire ignition location is known, the fire spread model based on the Bayesian network from the compartment of fire origin to another compartment can be built, and the probability of fire spread can be calculated by making use of the joint probability distribution of the Bayesian network. A specific application for an office building is presented for a case without sprinkler and one with sprinkler installed.     

火灾计算机模型的应用与发展

从应用的角度将火灾模型分为室内火灾区域模型、室内火灾场模型、耐火极限模型、建筑疏散模型、探测器响应模型及其他模型6种。不同的模型有不同的假设条件、模化方式和应用范围;不同精度的模型对于火灾的描述有不同层次的要求。总结模型所需的输入参数、输出参数及各种不确定因素对模型输出的影响。从火灾计算流体力学模化技术的选择、模型解释、模型验证和模型维护等方面分析火灾计算机模型发展面临的问题。     

Extended Rankine approach for bi-axially loaded steel columns under natural fire

Up till now, there has been limited research work conducted on bi-axially loaded steel columns under fire conditions. Under normal ambient temperature, the load-bearing capacity of steel columns is governed by the interaction of strength and stability considerations, which gives rise to the Rankine method. The authors extended this method to predict the fire resistance of steel columns subjected to bi-axial loading under standard fire curve. Basically, the authors developed an interaction equation based on failure surface to account for the effects of axial load and bending moments in two directions. Predictions from the proposed approach were benchmarked against a well-established finite element program SAFIR for steel columns under standard fire conditions. The same approach is then extended to include natural fire curves. To model a compartment fire with different geometries, thermal characteristics of boundary walls, different fire loads and ventilation factors, a zone fire modelling program Ozone was used. Coupling Ozone to SAFIR, the failure times of steel columns in a compartment fire were predicted. These numerical predictions were compared with those from the proposed approach and reasonable agreement was obtained.     

Compartment fire near-field entrainment measurements

A widely accepted consensus on entrainment models for large fires in compartments does not yet exist. To obtain further information on such entrainment rates, 20 full-scale, near-field experiments were conducted. Near-field entrainment occurs when hot layer interface heights are beneath the burner mean flame height so that cold layer entrainment occurs only near the burner surface. A durable compartment, similar to the standard fire test compartment, was designed and used in conjunction with a 0·61 m × 1·22 m porous surface propane burner to produce compartment fires with heat release rates from 330 to 980 kW. Entrainment rates of 0·74–0·98 kg/s were calculated from temperature measurements made within the compartment and in the doorway. The entrainment rates determined here were correlated with values from the literature. This correlation led to two curve fits which modify Zukoski's far-field offset model and can be used to estimate near-field entrainment rates. An offset for the near-field model of Thomas was also developed. The fire plume model of Baum and McCaffrey was found to compare favorably with the entrainment rates determined here.     

真实火灾下单层门式刚架抗倒塌性能试验研究

设计完成了一个单层单跨门式刚架厂房的足尺火灾试验,得到了主要构件的温度及位移发展规律,分析了真实火灾下门式刚架厂房结构的受力响应。结果显示：真实火灾下门式刚架的温升曲线与标准升温曲线有较大差别,燃烧室中的上部构件达到较高温度而提前失效,下部构件温度较低;在火灾下,未做防火保护的钢结构很短时间内就会发生垮塌,在火场及构件到达峰值温度前结构已产生较大位移。试验研究发现,受火柱的柱顶出现了热膨胀伸长、轴向压缩、轴向破坏三个阶段,且受顶部热烟气聚集的影响,各柱的柱顶轴向位移均大于柱中位移。试验成果可为门式刚架结构抗火数值模拟研究及结构防火设计提供参考。     

基于LSTM网络的船舶机舱火灾多特征融合探测

针对船舶机舱火灾高效准确探测的需求,建立基于LSTM-ID3 判决的船舶火灾探测方法。首先确定采集船舶火灾特征的三类传感器,然后完成 LSTM 神经网络模型的构建、参数的优化,将 LSTM 神经网络输出的明火、阴燃火、无火的概率值与烟雾持续时间作为决策树的输入量,输出火灾探测结果。利用国家标准火典型数据进行训练,并开展相关试验,对船舶机舱火灾进行探测。试验结果表明,与其他算法进行对比,探测准确率达到97%以上,该方案能对机舱火灾做出有效探测,为船舶安全提供科学依据。