J. Adler, F.B. Carleton and F.J. Weinberg

Imperial College, London SW7 2BY


The conditions under which continuous wave laser irradiation of inert targets leads to the ignition of a surrounding flammable gas mixture are studied in relation to the hazard associated with the use of optical fibres in explosive atmospheres. Results for stoichiometric mixtures in air of eleven diverse fuels are presented in the form of minimum igniting irradiance, uniform over an area large by comparison with quenching dimensions as a function of the time to ignition. Thermal histories during radiant heating and cooling of the target in an inert atmosphere are used to establish its thermal characteristics. The variation of minimum igniting irradiance with ignition lag is universally such that the igniting energy density varies linearly with time - a relationship which is accounted for theoretically. The intercept gives the limit of minimum igniting energy surface density as time tends to zero, whilst the slope provides an accurate value of the minimum igniting irradiance at infinite time, from which the surface ignition temperature can be deduced. These temperatures are comparable to auto-ignition temperatures for rapidly reacting mixtures but greatly in excess of them when auto-ignition temperatures are low. This is accounted for theoretically in terms of the effect of natural convection and empirical correlations are proposed which allow surface ignition temperatures and hence minimum igniting irradiances to be estimated in terms of known combustion parameters. Conversely, these relationships may be applied to deduce unknown autoignition temperatures from our irradiance measurements and this is applied to calculating autoignition temperatures for mixtures of fuels, with some interesting results.


D. Morvanl, B. Porterie2, M. Larini2 and J.C. Loraud2

1IRPHE UMR CNRS 6594 60 rue J. Curie
2IUSTI UMR CNRS 6595 5 rue E. Fermi, Technopole de Ch‚teau Gombert, 13453 Marseille cedex 13 France


To study the basic mechanisms which govern the fire spread in heterogeneous fuel beds, a one dimensional numerical model based on a multiphase approach is proposed. The configuration studied reproduces a reverse combustion experiment which represents a simplifying situation including the main physical phenomena occurring in the development and the propagation of a surface fire. The evolutions of state variables in the gas and solid phases are calculated separately using a multiphase approach including interaction terms between the two phases. The set of partial differential equations which governs the physical behavior of the multiphase flow is solved using an accurate finite volume method along with an adaptive mesh algorithm to track correctly the thin region where fuel thermal degradation occurs. Both oxygen-limited and fuel-limited regimes are examined for inlet flow velocities ranging from 2 to 30 cm/s.


D.C. Walther, R.A. Anthenien+ and A.C. Fernandez-Pello

Department of Mechanical Engineering, University of California at Berkeley, Berkeley, CA 94720
+ Current Address: Air Force Research Laboratory / Propulsion Directorate, Wright- Patterson Air Force Base, OH 45433


Experiments have been conducted to study the ignition of both forward and opposed smolder of a high void fraction, flexible, polyurethane foam in a forced oxidizer flow. Tests are conducted in a small scale, vertically oriented, combustion chamber with supporting instrumentation. A Nichrome wire heater placed between two porous ceramic disks, one of which is in. complete contact with the foam surface, is used to supply the necessary power to ignite and sustain a smolder reaction. A constant power is applied to the heater for a given period of time and the resulting smolder is monitored to extinction. The gaseous oxidizer, metered via mass flow controllers, is forced through the foam and heater. Reaction zone temperature and smolder propagation velocity are obtained from the temperature histories of thermocouples embedded at predetermined positions in the foam with junctions placed along the fuel centerline. Tests are conducted with oxygen mass fractions ranging from 0.233 to 1.0 at a velocity of 0.1 mm/s during the ignition period and 0.7 or 1.0 mm/s during the propagation period. The results show a well defined smolder ignition regime primarily determined by two parameters: igniter heat flux, and the time the igniter is powered. The ignition regime is shifted to shorter times for a given igniter heat flux with increasing oxygen mass fraction. An additional requirement for ignition is that the fuel igniter interface must attain a minimum temperature. An energy balance model has been developed that well describes the observed ignition results. The model is based on the assumption that for ignition to occur the heater must provide a heat flux equivalent to a propagating smolder front at a given distance from the igniter.


Melissa K. Anderson, Randall T. Sleight and Jose L. Torero

Department of Fire Protection Engineering, University of Maryland College Park, MD20742-3031 USA


An experimental study has been conducted to determine signatures that will describe the onset of self-sustained downward smolder. The fuel used is polyurethane foam, mainly due to its common occurrence in households and thus its importance in fire scenarios linked with smoldering combustion. The polyurethane foam is subject to a constant heat flux imposed by a cone heater for different time periods. Temperature and mass loss rate measurements together with visual observations serve to determine serve to determine specific events that can be considered signatures of the onset of a self-sustained smolder reaction. There stages were identified in the ignition process: the warm-up stage and unsteady smolder stage, both controlled by external heat flux; and the self-sustained smoldering stage controlled by the heat generation by the smolder reaction. Each stage is characterized by a change in the signatures of temperature history and mass loss rate. It was observed that self-sustained smolder could only occur in an external heat flux window (6.1 kW/m2 < q"e < 6.8 kW/m2). A minimum duration of exposure, weakly dependent on the external heat flux, was also found. These observations agree well with previously reported work. A minimum temperature (575 K < Tc < 600 K) was found necessary for sustained smoldering combustion.


B. Mesli and I. Gokalp

Laboratoire de Combustion et Systťmes Rťactifs, Centre National de la Recherche Scientifique, 45071 Orleans Cedex 2, France


An experimental study on water mist extinction of turbulent premixed flames is described. The aim of the study is to compare the extinction limits of opposed jet turbulent methane/air flames with and without the addition of water mist, and to study the influence of several parameters including the structure of water mist in terms of droplet size and mass fraction of the condensed phase, mean strain rate and equivalence ratio. The effect of NaCl water solution on premixed flame extinction is also presented. An existing opposed jet turbulent premixed flame experimental set-up is modified to include a water mist production system. An air assisted atomizer is developed to produce and control the water mist. The structure of the water mist is characterized by a Phase Doppler Anemometer. Water mist interaction with three different configurations of the opposed jet premixed flames is explored and the results are discussed by introducing a parameter representing the water mist efficiency.

KEYWORDS: Water mist, Flame extinction, Turbulent premixed flames


Ahmet Arisoy* and Fehmi Akgun**

* Technical University of Istanbul, Faculty of Mech. Eng., Gumussuyu, Istanbul, Turkey
**Division of Energy Systems, Marmara Research Centre, P.O. Box 21, Gebze, Kocaeli, Turkey


The purpose of this study is to predict the safe storage height of coal stockpiles. To accomplish this, a one-dimensional non-steady-state model with the effect of stockpile height approximated by a heat sink in the one- dimensional energy equation has been developed. The model consists of conservation equa6ons for oxygen, water vapour, inherent moisture of the coal, and energy for both gaseous and solid phases. The heat sink term taken from literature is based on an assumed sinusoidal variation in temperature in the vertical direction.

Numerical solution of the model equations gives the time-dependent maximum temperature in the coal stockpile. The effect of stockpile height on the maximum temperature has been analysed parametrically for four different low-rank Turkish coals. When the safe storage time is specified, the critical height can be determined. By defining the storage time as three months it has been found that these critical values varied between 0.85 and 1.8 m depending on the liability of the coals to spontaneous heating.


M. Hanana1, M.H. Lefebvre2, P. Van Tiggelen

Laboratoire de Physico-Chimie de la Combustion, Universite Catholique de Louvain - Louvain-la-Neuve - Belgium
1Institut de Mecanique, Universite de Blida, Blida, Algeria
2Dept. of Chemistry, Royal Military Academy, Brussels, Belgium


Recently two types of 3D structure of gaseous detonation have been documented: rectangular and diagonal modes, easily distinguishable from soot records. This paper presents pressure measurements recorded along the central axis of the cellular structure. The pressure records are achieved by using piezoelectric gauges flush-mounted with respect to the surface of the soot covered plate located in the detonation tube. The low pressure reactive mixture used (H2, O2, Ar, Equivalence Ratio = 1) is ignited in a square cross-section tube. The detonation tube is operated in the shock tube mode. The time evalution of the local pressure exhibits several pressure peaks depending on the type of 3D structure and on the position in the detonation cell. The first peak characterizes the leading shock and the subsequent pulses correspond to the elaborate shock structure. The influence of the slapping waves is documented. The pressure profiles throughout the whole length of the detonation cell is reported for the individual types of 3D structure. The second pressure jumps can be rationalized in terms of the classical transverse waves structure and will be discussed.


M.Maremonti1, G.Russo2, E.Salzano3, J.H.S. Lee4

1Dipartimento di Processi Chimici dell'Ingnegneria, Universit‚ di Padova, Italy
2CNR-Istituto di Ricerche sulla Combustione, Napoli, Italy
3CNR-GNDRCIE, Napoli, Italy
4Department of Mechanical Engineering, McGill University, Montreal, Canada


It is well established that rapid flame acceleration can be achieved in an obstacle filled tube.
Depending on the fuel, mixture composition, tube diameter and obstacle configuration, the flame eventually accelerates to steady state velocities that range from tens to hundreds of meters per second and transition to "quasi- detonation" or C-J detonation can also occur for sensitive mixtures. The principle mechanism for this flame acceleration is due to the turbulence in the unburned gas and the obstacles provide a powerful means of "randomization" of the mean flow kinetic energy. Interaction of the flame with the obstacles also promotes strong mixing and hence rapid combustion in the turbulent flame zone.

Due to the importance of this flame acceleration mechanism due to obstacles in. an industrial environment and in congested offshore oil platform configurations, a number of computer codes have been developed for predicting the pressure profiles generated. These codes, however, have to be "validated" against experiments of up to full scale geometry in order to acquire confidence in their application.

In this study the "Autoreagas" code, where the turbulence is modelled by k-e type of model, is used to predict the turbulent flame speed achieved with different fuels in obstacle filled tubes of various diameters.

By comparing the code predictions with experimental data, some insight into the code accuracy can be determined. In particular the results obtained showed that the code is quite adequate in predicting the explosion behavior of stoichiometric methane- and propane-air mixtures. However, numerical convergence problems are encountered in the computation for off stoichiometric mixtures. The code also failed for highly reactive gas such as acetylene and hydrogen, particularly when adopting too refined a computational grid. The peak overpressures generated by large-scale explosions, however, can be reproduced with a good level of accuracy.


D. Morvan1, B. Porterie2 and J.C. Loraud2

1IRPHE UMR CNRS 6594 60 rue J. Curie
2IUSTI UMR CNRS 6595 5 rue E. Fermi Technopole de Ch‚teau Gombert 13453 Marseille cedex 13 FRANCE


The unsteady behavior of a buoyant methane diffusion flame is simulated numerically. The turbulent reactive mixture is evaluated using a RNG k-e-g turbulence modeling and a presumed shape (b-function) pdf approach. Radiation is considered via a differential Pl-approximation to calculate the irradiance field and the radiative flux. To obtain an accurate description of the unsteady behavior of buoyant diffusion flames, a Finite-Volume method including a high- order upwind convective scheme (Quick Scheme} with a flux limiter technique (Ultra Sharp method) and a second-order backward Euler scheme for time integration is used. The numerical results show that the buoyant flow above the flame is characterized by the development of large eddies on both sides of the column formed during the expansion of the hot gases. The symmetry broken of the flow pattern and the unsteady behavior of the flame which pulses vertically are observed.


P.Ciambelli1, A.Bucciero2, M.Maremonti3, E.Salzano4

1Dipartimento di Ingegneria Chimica e Alimentare, Universit‚ di Salerno, Italy
2EniChem S.P.A., S.Donato Milanese (Milano), Italy
3Dipartimento di Processi Chimici dell'Ingegneria, Universit‚ di Padova, Italy
4CNR-GNDRCIE, Napoli, Italy


In view of their particular geometry, road and rail tunnels have to be regarded as sites of major risk of devastating explosions. Therefore, appropriate design criteria and safety assessment procedures should be adopted for prevention purposes. However, only a few indications exist in the literature, due to the limited experimental results available at present on similar configurations.

Computational Fluid Dynamic (CFD) codes represent a suitable approach to this problem. Indeed, specific CFD codes have been developed, which are able to predict the pressure profiles generated during gas explosions in large-scale environments and in the presence of high obstacle congestion.

In this work, the Autoreagas code is used to predict the pressure, the blast wave propagation and the flame speed achieved in road and rail tunnels when varying the level of congestion and the amount of released fuel. Obtained results show that severe consequences can derive from the explosion of gas clouds in such environments, even if the fuel amount is relatively small, as that contained in a car tank.


Norman Alvares
Fire Science Applications

Carlos Fernandez-Pello
Department of Mechanical Engineering, University of California at Berkeley


Industrial ares originating in cable arrays are not common; but when they do occur, they can result in economic losses that far exceed the physical damage caused by the fire. This is particularly true for fires in telecommunication systems, where Central Offices necessarily contain enormous arrays of cables which are supported in multiple levels of elevated cable trays that are distributed throughout the facility. The insulation and protective covers most cables are combustible and the surface to volume ratio of the fuel array is typical of close packed cribs. In this paper an analysis is presented that was developed to describe fire growth that occurred in a Telecommunication Central Office fire, initiating at an intersection of cable trays in the toll terminal area. Though the building contained a complete fire detection system, the time between the first recorded fire alarm and notification of the local fire department was extensively delayed because of weather problems and misinterpreted signals. The building was not sprinkler protected. Consequently, the fire persisted for an extensive period of time. The heaviest fire damage involved an area of only 110 m2, but because of the local cable density in this area, the smoke and combustion gases caused potentially irreversible damage to telephone systems throughout the floor of origin.

The analysis developed here uses available test data on cable burning in a simplified model of fire development to predict the characteristics of the actual fire. These characteristics include: fuel burning and heat release rates, smoke and HCL generation, growth of the smoke layer, smoke and HCL concentration in the layer, smoke detector activation and sprinkler actuation. The results of the analysis are then used to estimate how alternate countermeasures could have effected the outcome of the initiating fire. This simple, empirically driven analytical model is sensitive to local conditions, fuel geometry and fuel composition.


Francesco Tamanini and Jeffrey L. Chaffee

Factory Mutual Research Corporation, Norwood, Massachusetts 02062


Testing was carried out in a 63.7-m3 chamber (4.57 by 4.57 by 3.05-m high) with stratified mixtures of propane in air, under both vented and unvented conditions. The tests are part of a program designed to quantify the explosion hazard from flammable liquid spills in industrial process and dispensing areas. Layers were formed by slowly injecting propane at the chamber floor through diffusers, at a rate of the order of the estimated rate of vaporization of a typical solvent such as acetone. The gas composition field was characterized through time-resolved concentration measurements obtained at twelve locations in the room. This was achieved using a single continuous analyzer receiving gas samples through a multiplexing valve. Several composition profiles were simulated, including layers in which the concentration of propane at the floor was above the upper explosive limit (UEL). The paper presents a discussion of the combustion behavior of the mixture following ignition. Estimates of the burning velocity are in substantial agreement with the results from previous works, which show that flame propagation velocities over stratified layers can be 3-5 times the fundamental burning velocity in a mixture of near- stoichiometric composition. A change in the internal geometry of the chamber was introduced to study the effect of blockages on the development of the explosion. These obstacles were found to increase the propagation velocity by about 50% over the values observed in the empty enclosure. The paper also discusses the role played in the process by the portion of the layer above the UEL, since this can be a significant contributor to the pressure development in the enclosure under certain accident scenarios.


L. di Mare1, T.A. Mihalik2, G. Continillo3, and J.H.S. Lee2

1Dipartimento di Ingegneria Chimica, Universit‚ di Napoli Federico II, Naples, Italy
2Department of Mechanical Engineering, McGill University, Montreal, Canada
3Facolt‚ di Ingegneria, Universit‚ del Sannio, Benevento, Italy and Istituto di Ricerche sulla Combustione CNR, Naples, Italy


The present work deals with an experimental and numerical study of flammability limits for hydrocarbon-air mixtures in porous media and their dependence on the physical and geometrical properties of the solid phase. Experimental data have been obtained in a standard flammability apparatus modified through the insertion of a packed bed of spheres. Numerical data have been obtained through integration of the equations of a one-dimensional model with single step kinetics. The model accounts for heat transfer to the solid phase in the porous medium. The kinetic parameters of the model are tuned in order to match flammability limits for freely propagating deflagrations; then, the model is applied to cases' with porous medium. Numerical results are compared with experimental results and at least qualitative agreement is found. Particularly, both sets of data show that flammability limits are more sensitive to the geometric properties of the porous medium than to its physical properties. An explanation is given through the analysis of the heat transfer process. The results show that, in modelling flammability limits within porous media, heat losses to the solid phase should be taken into account along with fluid-dynamic effects, to correctly understand extinction mechanisms and predict flammability data.

Future Meetings | Past Meetings | Proceedings on Sale | Related Links

ICHMT World Wide Web Administrator