Fire detection, location and heat release rate through inverse problem solution. Part I: Theory

A proposed method of detecting, locating and sizing accidental fires, based on the solution of an inverse heat transfer problem, is described. The inverse heat transfer problem to be solved is that of the convective heating of a compartment ceiling by the hot plume of combustion gases rising from an accidental fire. The inverse problem solution algorithm employs transient temperature data gathered at the ceiling of the compartment to determine the location and heat release rate of the fire. An evaluation of the proposed fire detection system, demonstrating the limits on the accuracy of the inverse problem solution algorithm, is presented. The evaluation involves operating the inverse problem solution algorithm on transient temperature data from computer simulated compartment fires. The simulated fire data are generated assuming fires with quadratic growth rates, burning in a 20 m wide by 20 m deep by 3 m high enclosure with a smooth, adiabatic ceiling. The accuracy of the inverse problem solution algorithm in determining the location of a fire is shown to be insensitive to the errors in the fire model used in the forward problem solution, but sensitive to errors in the measured temperature data. The accuracy of the heat release rate of the fire is sensitive to both errors in the fire model and errors in the temperature data. The validity of the use of computer simulated data in the evaluation is verified with a second evaluation using fire data interpolated from published measurements taken in large-scale compartment fire burns.     

Further Validation of a Numerical Model for Prediction of Pyrolysis of Polymer Nanocomposites in the Cone Calorimeter

Nanocomposites have been increasingly used, as an alternative to traditional fire retardants, to improve the strength and fire retardancy of polymeric materials. A number of studies using the cone calorimeter showed that the nanoparticles used in small quantities (e.g., 3 wt%) reduce significantly the heat release rate (HRR). The formation of a surface layer on top of the unpyrolysed material is generally considered responsible for the reduced HRR. In a previous study, the global effects of the surface layer were examined by the present authors and a methodology was subsequently developed to predict pyrolysis of a polyamide nylon (PA6) nanocomposite in good agreement with the experimental data. This work presents further validation of the methodology for two more nanocomposites, namely polybutylene terephthalate and ethylene-vinyl acetate. Furthermore, the existing model is extended to explain the effects of change in the nanofiller loading on the HRR, and the modified model is applied to the experimental data obtained for a PA6 nanocomposite by Morgan et al. (Fire and polymers: materials and solutions for hazard prevention. American Chemical Society, Washington, DC, 2001, pp 9–23).     

Determination of the heat release rate inside operational road tunnels by comparison with CFD calculations

Usually, during a fire inside a tunnel, the average heat release rate (HRR) is estimated according to the type of vehicle. Frequently, the overall HRR is considered, however it is also necessary to know its time evolution to design real time systems, particularly ventilation, which respond to fire events or signals as fast as possible. Nowadays, there is not a well established and generally accepted procedure to know the power liberated at each instant of time inside an operational tunnel. That procedure could help in taking the correct actions to adapt the tunnel ventilation in order to diminish the effects of the fire and the smoke. This work shows a method to calculate the heat release rate using sensors that can be installed inside an operational road tunnel. Besides, the location of the fire could also be calculated accurately and quickly. To achieve the previous purposes, a stationary database that depends on HRR, its location, and the ventilation speed is calculated with CFD programs; the data are compared with temperatures measured by the sensors located inside the tunnel. The program used to generate the database is the simplified model UPMTUNNEL. The predictions of the model are compared with the results of calculations carried out using the general purpose code FLUENT, and with measurements done in a tunnel with a real fire, produced with a fuel tray.     

An Experimental Study of Smoke Movement in Multi-Storey Buildings

This paper presents the results of an experimental study of smoke movement in a 10-storey building. Eight full-scale experiments including four real fuel fires and four propane fires were conducted in the National Research Council Canada (NRCC)’s 10-storey experimental tower to generate smoke movement data that can be used for the validation of computer models. The heat release rate (HRR) of fire cannot be measured in this tower, so to estimate the HRR of fuel-package fires in this study, an approach using propane as a fuel was developed to reproduce the temperature distribution of various fuel-package tests.     

Investigation of the effect of tunnel ventilation on crib fires through small-scale experiments

This paper adopts a series of 1:20 scale tunnel experiments based on a series of large-scale tunnel experiments to study the influence of forced ventilation on fires. The small-scale tunnel has dimensions of 0.365 m (W)×0.26 m (H)×11.9 m (L). Cribs using a wood-based material provide the fuel source and forced ventilation velocities from 0.23 to 1.90 m/s are used. From the study of the measured heat release rate (HRR) and mass loss rate data it is found that the forced air velocity affects the fire spread rate and burning efficiency and further affects peak HRR values at different air velocities. A simple model to describe these influences is proposed. This model is used to reproduce the enhancement of peak HRR for cribs with different porosity factors noted by Ingason [1] and to assess the effects of using different length of cribs on peak HRR. The results from these analyses suggest that different porosity fuels result different involvement of burning surface area and result different changes in peak HRR. However, no significant difference to the enhancement on fire size is found when the burning surface area is similar. It is also found that the trend in the enhancement on fire size by using sufficiently long crib and available ventilation conditions matches the predictions of Carvel and Beard [2] for two-lane tunnel heavy goods vehicle fires.     

Variation of heat release rate with forced longitudinal ventilation for vehicle fires in tunnels

Many tunnels are equipped with longitudinal ventilation systems to control smoke in the event of a fire. However, the influence of such ventilation on fire development and fire spread has rarely been considered. This paper presents the results of a study using a Bayesian methodology to estimate the effect of forced longitudinal ventilation on heat release rate (HRR) for fires in tunnels. The behaviour of car and heavy goods vehicle (HGV) fires with a range of forced ventilation velocities is investigated. Results are presented and the implications are discussed. It has been found that forced ventilation has a great enhancing effect on the HRR of HGV fires, but has little effect on the HRR of car fires.     

Burning behavior of sedan passenger cars

Four full-scale fire experiments using 4-door sedan passenger cars were carried out. The cars were ignited either at the splashguard of the right rear wheel or at the left front seat in the passenger compartment with a gasoline spill. The temperature inside the burning car and the mass loss rate were measured. The burning of the 4-door sedan was composed of three compartmental fires: the engine compartment, the passenger compartment, and the rear part inclusive of the fuel. In the experiments where ignition was initiated at the splashguard, the flame spread in the following order: to the rear part of the car, to the passenger compartment, and to the engine compartment. Breakage of the window glass markedly affected the spread of fire into the passenger compartment. The quantity of gasoline in the fuel tank also affected the speed of spread of the fire, because the gasoline ignited at an early stage of the fire. In the experiment where ignition was initiated in the passenger compartment, the fire gained force after the windshield was broken entirely. The flame spread in the following order: to the passenger compartment, to the engine compartment, and to the rear part of the car. The temperature within the passenger compartment peaked at 1000 °C. The heat release rate (HRR) curves showed several peaks depending on the burning of the three compartments. The HRR increased markedly when the fire spread to several different parts of the car at the same time. The HHR peaked at 3 MW when the passenger compartment and fuel (gasoline) burned simultaneously. The measured HRR curves were characterized by superposition of a Boltzmann curve and a Gaussian curve in order to obtain a model, which allowed us to make a more precise prediction of the fire spread probability from a burning car to nearby structures. The HRRs of burning cars were described by the sum of HRR from each compartment.     

The Use of Fractional Factorial Design for Atrium Fires Prediction

In order to mitigate the excessive computational cost of atrium fire simulations, a novel methodology based on the use of the Fractional Factorial Design technique to obtain an experimental validated tool, in the form of a surface response model, capable to predict fire induced conditions is proposed. This methodology is supported by results from a Design of Experiments benchmark, which consists of a set of FDS simulations in the present work. Specifically, a \(2^{6-2}_{IV}\) approach has been considered and applied to a 20 m cubic atrium. Thus, six factors have been considered, namely the fire Heat Release Rate (HRR) and location, the exhaust flow rate, the exhaust location and activation time, and the inlet vents area. Furthermore, the smoke temperature at the roof and 15 m high and the smoke layer height have been considered the variables of interest. Subsequently, a multiple linear regression analysis has been performed to predict and compare the steady and non-steady temperature profiles and the smoke layer drop with six novel full-scale atrium fire tests, and also with specific adjusted FDS models. In addition, this methodology has been extended successfully to predict the non-steady behaviour of the fire tests. At the steady state, the HRR and the exhaust flow rate have been found to be the most relevant factors. The results obtained with the proposed methodology show a good fit both with the fire tests and with the adjusted FDS models, with discrepancies mostly below 14%. For non-steady conditions, a time analysis of the influence of the six factors has been carried out. Again, remarkable good agreement with the time-dependent experimental results is achieved, with average discrepancies below 12%, being the larger differences found in the prediction of local effects, such as the smoke ceiling jet, for high HRR or when the make-up air influence is significant. The results turn this methodology into a powerful and useful tool for fire safety designs.     

An engineering tool to calculate heat release rates of multiple objects in underground structures

Simple theoretical calculations of the overall heat release rate (HRR) of multiple objects have been carried out. The results were compared to fire experiments in a model tunnel using wood cribs placed at equal distances from each other. Three different methods are presented which are based on physical relations for fire spread between the wood cribs. The first method uses a critical heat flux as ignition criteria while the other two methods use an ignition temperature. The method using the critical heat flux as ignition criteria shows very good agreement with the corresponding experimental results used. The two methods using the ignition temperature as ignition criteria did not agree well with the corresponding experimental results. The prerequisites, that the methods should be kept relatively simple to be of practical use and that the burning objects should not necessarily have to be of uniform composition, were fulfilled.     

Using Sensor Signals to Analyze Fires

Building fire sensors are capable of supplying substantially more information to the fire service than just the simple detection of a possible fire. Nelson, in 1984, recognized the importance of tying all the building sensors to a smart fire panel [1]. In order to accomplish a smart fire panel configuration such as envisioned by Nelson, algorithms must be developed that convert the analog/digital signals received from sensors to the heat release rate (HRR) of the fire. Once the HRR of the fire is known, a multiroom zone fire model can be used to determine smoke layers and temperatures in the other rooms of the building. This information can then be sent to the fire service providing it with an approximate overview of the fire scenario in the building.This paper will describe a ceiling jet algorithm that is being developed to predict the heat release rate (HRR) of a fire using signals from smoke and gas sensors. The prediction of this algorithm will be compared with experiments. In addition, an example of the predictions from a sensor-driven fire model, SDFM, using signals from heat sensors, will be compared with measurements from a full-scale, two-story, flashover townhouse fire.     

Numerical studies on evaluation of smoke control system of underground metro rail transport system in India having jet injection system: A case study

The rail based urban transport system is being developed for national capital of India, New Delhi. The smoke control using ventilation in case of fire inside the tunnel, similar to Delhi Metro corridor has been investigated using computational fluid dynamics technique. A section of tunnel having dimensions 400 m long, 5.5 m wide and 6 m high is considered for simulation. The analysis has been carried out by assuming a variable fire source with a peak heat release rate (HRR) of 16 MW, located at the center of the tunnel. Ventilation ducts are located in the ceiling near the tunnel portals and are inclined at 10 degrees to the plane of the ceiling through which fans discharge air. The influence of the fire HRR curve slope on the smoke flow dynamics in a realistic tunnel model fitted with jet injection type longitudinal ventilation system has been investigated. In case of fire two cases are studied: (1) fans activated immediately after detection, (2) fans activated at delayed times to take into account the response time for the fans to achieve its maximum speed. The velocity of supply and exhaust fans necessary to remove smoke in 30 s from the upstream direction is determined. The velocities of fan required to produce desired critical velocity in the longitudinal direction for different HRR of fire is predicted.     

Assessing the Sprinkler Activation Predictive Capability of the BRANZFIRE Fire Model

This paper describes an investigation into the sprinkler response time predictive capability of the BRANZFIRE fire model. A set of 22 fire/sprinkler experiments are simulated where the sprinkler activation time and the heat release rate (HRR) for each individual experiment had been determined. The experiments provided data for use in validating the sprinkler activation prediction algorithms in the BRANZFIRE zone model. A set of base case values were chosen and input files constructed for the simulations. The experiments were then simulated by the fire model using both the NIST/JET ceiling jet and Alpert’s ceiling jet options (which are the two ceiling jet correlations available in the BRANZFIRE zone model). The fire model included a heat transfer calculation for the temperature of the heat sensitive sprinkler element. Different sprinkler operational parameters such as the conduction factor, response time index (RTI) and the sprinkler depth below ceiling were also varied to assess the sensitivity of their effect on the activation time. Results showed that using the NIST/JET ceiling jet algorithm gave a closer prediction of the sprinkler response time in a small room than Alpert’s correlation. This was expected, since the former includes the effect of a hot upper layer while the latter applies to unconfined ceilings. The experiments available for comparison had been conducted inside an enclosure with a developing hot upper layer. The findings also signified that changing the sprinkler operational parameters can change the predicted sprinkler activation time significantly.     

The application of a genetic algorithm to estimate material properties for fire modeling from bench-scale fire test data

A methodology based on an automated optimization technique that uses a genetic algorithm (GA) is developed to estimate the material properties needed for CFD-based fire growth modeling from bench-scale fire test data. The proposed methodology involves simulating a bench-scale fire test with a theoretical model, and using a GA to locate a set of model parameters (material properties) that provide optimal agreement between the model predictions and the experimental data. Specifically, a GA based on the processes of natural selection and mutation is developed and integrated with the NIST FDS v4.0 pyrolysis model for thick solid fuels. The combined GA/pyrolysis model is used with cone calorimeter data for surface temperature and mass loss rate histories to estimate the material properties of two charring materials (redwood and red oak) and one thermoplastic material (polypropylene). This is done by finding the parameter sets that provide near-optimal agreement between the model predictions and experimental data, given the constraints imposed by the underlying physical model and the accuracy with which the boundary and initial conditions can be specified. The methodology is demonstrated here with the FDS pyrolysis model and cone calorimeter data, but it is general and can be used with several existing fire tests and almost any pyrolysis model. Although the proposed methodology is intended for use in CFD-based prediction of large-scale fire development, such calculations are not performed here and are recommended for future work.     

Examination of WFDS in Modeling Spreading Fires in a Furniture Calorimeter

Validation of physics-based models of fire behavior requires comparing systematically and objectively simulated results and experimental observations in different scenarios, conditions and scales. Heat Release Rate (HRR) is a key parameter for understanding combustion processes in vegetation fires and a main output data of physics-based models. This paper addresses the validation of the Wildland-urban interface Fire Dynamics Simulator (WFDS) through the comparison of predicted and measured values of HRR from spreading fires in a furniture calorimeter. Experimental fuel beds were made up of Pinus pinaster needles and three different fuel loadings (i.e. 0.6, 0.9 and 1.2 kg/m²) were tested under no-slope and up-slope conditions (20°). An Arrhenius type model for solid-phase degradation including char oxidation was implemented in WFDS. To ensure the same experimental and numerical conditions, sensitivity analyses were carried out in order to determine the grid resolution to capture the flow dynamics within the hood of the experimental device and to assess the grid resolution’s influence on the outputs of the model. The comparison of experimental and predicted HRR values showed that WFDS calculates accurately the mean HRR values during the steady-state of fire propagation. It also reproduces correctly the duration of the flaming combustion phase, which is directly tied to the fire rate of spread.     

Forecasting fire growth using an inverse zone modelling approach

A new methodology to effectively forecast fire dynamics based on assimilation of sensor observations is presented and demonstrated. An inverse modelling approach with a two-zone model is used to forecast the growth of a compartment fire. Sensor observations are assimilated into the model in order to estimate invariant parameters and thus speed up simulations and recover information lost by modelling approximations. A series of cases of a compartment fire radially spreading at different growth rates (slow, medium and fast) are used to test the methodology. Spread rate, entrainment coefficient and smoke transport time are the invariant parameters estimated via a gradient-based optimization method with tangent linear differentiation. The parameters were estimated accurately within minutes after ignition and the heat release rate reproduced satisfactorily in all cases. Moreover, the temperature and the height of the hot layer are forecasted with a positive lead time between 50 and 80 s, depending on the fire growth rate. The results show that the simple mass and energy conservation equations and plume correlation of the zone model are suitable to forecast the main features of a growing fire. Positive lead times are reported here for the first time in fire dynamics. The results also suggest the existence of an optimal width for the assimilation window. The proposed methodology is subject to ongoing research and the results are an important step towards the forecast of fire dynamics to lead the emergency response.     

Modeling of fire suppression by fuel cooling

Fire suppression with water spray was investigated, focusing on cases where fuel cooling is the dominant suppression mechanism, with the aim to add a specific suppression model addressing this mechanism in Fire Dynamics Simulator (FDS), which already involves a suppression model addressing effects related to flame cooling. A series of experiments was selected, involving round pools of either 25 or 35 cm diameter and using both diesel and fuel oil, in a well-ventilated room. The fire suppression system is designed with four nozzles delivering a total flow rate of 25 l/min and injecting droplets with mean Sauter diameter 112 μm. Among the 74 tests conducted in various conditions, 12 cases with early spray activation were especially considered, as suppression was observed to require a longer time to cool the fuel surface below the ignition temperature. This was quantified with fuel surface temperature measurements and flame video recordings in particular. A model was introduced simulating the reduction of the pyrolysis rate during the water spray application, in relation to the decrease of the fuel local temperature. The numerical implementation uses the free-burn step of the fire to identify the relationship between pyrolysis rate and fuel surface temperature, assuming that the same relationship is kept during the fire suppression step. As expected, numerical simulations reproduced a sharp HRR decrease following the spray activation in all tests and the suppression was predicted in all cases where it was observed experimentally. One specific case involving a water flow rate reduced such that it is too weak to allow complete suppression was successfully simulated. Indeed, the simulation showed a reduced HRR but a fire not yet suppressed. However, most of the tests showed an under-estimated duration before fire suppression (discrepancy up to 26 s for a spray activation lasting 73 s), which demonstrates the need for model improvement. In particular the simulation of the surface temperature should require a dedicated attention. Finally, when spray activation occurred in hotter environments, probably requiring a combination of fuel cooling and flame cooling effects, fire suppression was predicted but with an over-estimated duration. These results show the need for further modeling efforts to combine in a satisfactory manner the flame cooling model of FDS and the present suggested model for fuel cooling.     

Estimating the Flashover Probability of Residential Fires Using Monte Carlo Simulations of the MQH Correlation

An important indicator of fire hazard in residential fires is the occurrence of flashover in the room of fire origin. Since the variability of residential fire scenarios is large, many different cases must be considered to evaluate the hazard of a given flammable product. Efficiently predicting the occurrence of flashover of a naturally ventilated compartment is possible using the correlation of McCaffrey, Quintiere, and Harkleroad (MQH). The large variability in United States (U.S.) living room fire scenarios is characterized from available data and propagated through the MQH correlation using Monte Carlo (MC) simulation. For the parameters, for which no relevant data was found, uniform probability distributions were assumed. The scenarios sampled in the MC simulations generally fell within the range of scenarios for which the MQH correlation has been validated. Flashover probabilities were estimated for fires up to 5 MW in heat release rate (HRR) and up to 8 min in duration. It was found that fires with HRRs less than 400 kW have a flashover probability of less than 0.01% regardless of their duration. Typical furniture fires were used as example cases, and it was predicted that a three seat upholstered sofa with a peak HRR of 2.15 MW has a 90% chance of flashing over a randomly chosen U.S. living room. The results of a global sensitivity analysis indicates that the fire location parameter and the vent opening width are the most important parameters affecting the prediction of the occurrence of flashover in U.S. living rooms. The methodology presented is generalizable, and the results can be readily improved by the collection of more data and the use of higher fidelity fire models.     

Assessment of the Use of Fire Dynamics Simulator in Performance-Based Design

This paper discusses a procedure for the use of fire modelling in the performance-based design environment to quantify design fires for commercial buildings. This procedure includes building surveys, medium-and full-scale experiments and computer modelling. In this study, a survey of commercial premises was conducted to determine fire loads and types of combustibles present in these buildings. Statistical data from the literature were analysed to determine the frequency of fires, ignition sources, and locations relevant to these premises. Based on the results of the survey and the statistical analyses a number of fuel packages were designed that represent fire loads and combustible materials in commercial buildings. The fuel packages were used to perform medium- and full-scale, post-flashover fire tests to collect data on heat release rates, compartment temperatures and production and concentration of toxic gases. Based on the experimental results, input data files for the computational model, Fire Dynamics Simulator (FDS), were developed to simulate the burning characteristics of the fuel packages observed in the experiments. Comparative analysis between FDS model predictions and experimental data of HRR, carbon monoxide (CO), and carbon dioxide (CO₂), indicated that FDS model was able to predict the HRR, temperature profile in the burn room, and the total production of CO and CO₂ for medium- and large-scale experiments as well as real size stores.     

Engineering Variables to Replace the Concept of ‘Noncombustibility’

The concept of noncombustibility evolved in the early days of building codes, before quantitative methods of measuring and assessing components of fire hazard were available. ‘Noncombustible’ lacks a technical definition of general scope, but in the US codes, which are the primary focus of this study, it is defined as a material which meets the criteria of the ASTM E136 test. The hazard variables underlying the noncombustibility concept are examined in this study. In view of today’s state of the art, it is shown that noncombustibility requirements, in most cases, constitute a misapplication of fire safety principles and that their use should be discontinued, in preference of using variables that express quantitative fire safety principles. Heat release rate (HRR) is the primary variable which correctly establishes the relevant hazard. In recent years, some regulations have been promulgated which use bench-scale HRR test results directly for this purpose. The ultimate hazard to be addressed, however, is the full-scale HRR behavior. When the hazard involves fires which may spread over surface linings, however, the full-scale HRR is not simply directly scaled to the bench-scale HRR. To quantify this hazard properly, additional properties of the material which govern the flame spread behavior need to be considered. A simple, easy-to-use method for this purpose are described, which is based solely on data obtainable from the Cone Calorimeter (ASTM E1354; ISO 5660) test. Validation of the concept against room-scale data is provided and is shown to be successful.