Publications (75)109.81 Total impact
 [Show abstract] [Hide abstract]
ABSTRACT: The implosion technique has been used to extend measurements of turbulent burning velocities over greater ranges of fuels and pressures. Measurements have been made up to 3.5 MPa and at strain rate Markstein numbers as low as −23. The implosion technique, with spark ignition at two opposite wall positions within a fanstirred spherical bomb is capable of measuring turbulent burning velocities, at higher pressures than is possible with central ignition. Pressure records and schlieren high speed photography define the rate of burning and the smoothed area of the flame front. The first aim of the study was to extend the previous measurements with ethanol and propane–air, with further measurements over wider ranges of fuels and equivalence ratios with mixtures of hydrogen, methane, 10% hydrogen–90% methane, toluene, and ioctane, with air. The second aim was to study further the low turbulence regime in which turbulent burning coexists with laminar flame instabilities.Correlations are presented of turbulent burning velocity normalised by the effective rms turbulent velocity acting on the flame front, , with the Karlovitz stretch factor, K, for different strain rate Markstein numbers, a decrease in which increases . Experimental correlations are presented for the present measurements, combined with previous ones. Different burning regimes are also identified, extending from that of mixed turbulence/laminar instability at low values of K to that at high values of K, in which is gradually reduced due to increasing localised flame extinctions. 

Conference Paper: Vapour Cloud Explosions and Their Propensity to Create a Fire Storm
 [Show abstract] [Hide abstract]
ABSTRACT: Measurements of turbulent burning velocities in fanstirred explosion bombs show an initial linear increase with the fan speed and RMS turbulent velocity. The line then bends over to form a plateau of high values around the maximum attainable burning velocity. A further increase in fan speed leads to the eventual complete quenching of the flame due to increasing localised extinctions because of the flame stretch rate. The greater the Markstein number, the more readily does flame quenching occur. Flame propagation along a duct closed at one end, with and without baffles to increase the turbulence, is subjected to a onedimensional analysis. The flame, initiated at the closed end of the long duct, accelerates by the turbulent feedback mechanism, creating a shock wave ahead of it, until the maximum turbulent burning velocity for the mixture is attained. With the confining walls, the mixture is compressed between the flame and the shock plane up to the point where it might autoignite. This can be followed by a deflagration to detonation transition. The maximum shock intensity occurs with the maximum attainable turbulent burning velocity, and this defines the limit for autoignition of the mixture. For more reactive mixtures, autoignition can occur at turbulent burning velocities that are less than the maximum attainable one. Autoignition can be followed by quasidetonation or fully developed detonation. The stability of ensuing detonations is discussed, along with the conditions that may lead to their extinction.  [Show abstract] [Hide abstract]
ABSTRACT: This paper first briefly surveys the energy releases in some major accidents. It then examines the analyses of the explosion at the Buncefield fuel storage site in the UK, one of the most intense accidental explosions in recent times. This followed the release of approximately 300 tonnes of wintergrade gasoline, when a 15 m high storage tank was overfilled for about 40 min before ignition of the resulting flammable mixture. The ensuing explosion was of a severity that had not been identified previously in a major hazard assessment of this type of facility. It was therefore imperative to investigate the event thoroughly and develop an understanding of the underlying mechanisms to inform future prevention, mitigation and landuse planning issues. The investigation of the incident was overseen by the Buncefield Major Incident Investigation Board. A separate Explosion Mechanism Advisory Group examined the evidence and reported on the severity of the explosion. It concluded that additional work was necessary and recommended that a twostage project be initiated, phase 1 of which has been completed. The analyses of the damage and the derivation of explosion overpressures are described. Possible explosion mechanisms and the evidence for them at Buncefield are discussed, in the light of other major incidents. Mechanisms that are reviewed include highspeed turbulent combustion, quasidetonations, fully developed detonations, the generation of fireballs, flame instabilites, radiative heat transfer and aspects of twophase burning. Of particular importance is the acceleration of turbulent flames along the line of trees and hedgerows. A number of conclusions are drawn and suggestions made for further research.  [Show abstract] [Hide abstract]
ABSTRACT: A new technique is described for measuring turbulent burning velocities at higher pressures than is usual in fanstirred bomb explosions. Measurements are made during the final stage of inward propagation of two flames, initiated at diametrically opposite spark electrodes. Pressure records and schlieren high speed photography define the rate of burning and the smoothed area of the flame front. This implosion technique was validated at the lower pressures by the good agreement between the values of turbulent burning velocity it yielded and those obtained with the commonly employed central ignition. The new technique has the advantage of yielding values of turbulent burning velocity at pressures much closer to the safe working pressure of the explosion bomb. Subsequently, mixtures of ethanol–air and propane–air were investigated in the pressure range of 0.7–3.0MPa with a corresponding temperature range of 377–468K. For explosions with central ignition the maximum pressure was 1.2MPa.Findings over a wide range of conditions are generalised. Plots are presented of turbulent burning velocity normalised by the effective rms turbulent velocity, ut/u′k, against the Karlovitz stretch factor, K, for different strain rate Markstein numbers, a decrease in which, increases ut/u′k. At low values of K, ut/u′k is enhanced by flame instabilities, while at high values of K, it is gradually reduced due to increasing localised extinctions. 
Conference Paper: The Buncefield Explosion
 [Show abstract] [Hide abstract]
ABSTRACT: The turbulent burning velocity is defined by the mass rate of burning and this also requires that the associated flame surface area should be defined. Previous measurements of the radial distribution of the mean reaction progress variable in turbulent explosion flames provide a basis for definitions of such surface areas for turbulent burning velocities. These interrelationships. in general, are different from those for burner flames. Burning velocities are presented for a spherical flame surface, at which the mass of unburned gas inside it is equal to the mass of burned gas outside it. These can readily be transformed to burning velocities based on other surfaces.The measurements of the turbulent burning velocities presented are the mean from five different explosions, all under the same conditions. These cover a wide range of equivalence ratios, pressures and rms turbulent velocities for ethanol–air mixtures. Two techniques are employed, one based on measurements of high speed schlieren images, the other on pressure transducer measurements. There is good agreement between turbulent burning velocities measured by the two techniques. All the measurement are generalised in plots of burning velocity normalised by the effective unburned gas rms velocity as a function of the Karlovitz stretch factor for different strain rate Markstein numbers. For a given value of this stretch factor a decrease in Markstein number increases the normalised burning velocity. Comparisons are made with the findings of other workers. 
Conference Paper: Turbulent Burning Velocity and the Deflagration to Detonation Transition

Conference Paper: Correlation of the Height of Turbulent Choked Jet Flames
 [Show abstract] [Hide abstract]
ABSTRACT: A novel way is presented of deriving the proportionality constant, k (mm1), for flame surface density. This comprises simultaneous measurements, by Mie scattering, of mean reaction progress variable and the turbulent burning velocity, during explosions in a fanstirred bomb. Mean effective laminar burning velocities at the wrinkled flame surface are derived theoretically from the distribution of flame stretch rates, the Markstein numbers, and the possible generation of localised flame instabilities and flame quenching. The changing radial profiles of flame surface density have been obtained, as flames accelerate in different propaneair explosions. The effective value of the rms turbulent velocity acting on the flame is derived from a newly developed power spectral density function. The values of k are similar to those found by researchers employing stationary turbulent flames, but the present work suggests dependencies of k upon Markstein number. Insights are obtained on the interrelationship of k and turbulent burning phenomena. More research is required on the factors controlling the value of k, including its relationship to flame curvature.  [Show abstract] [Hide abstract]
ABSTRACT: The fast diminution of readily extractable sources of fossil fuels, particularly oil, and concerns about global warming are leading to the creation of many new potential fuels. Their practicability must be assessed in terms of a wide range of physicochemical properties, in relation to the operational aerodynamics in different engines. This article concentrates on those properties related to combustion and these are discussed in detail for some fuels with contrasting properties. Intrinsic fuel properties include volumetric energy, vapour pressure, heat of reaction, latent enthalpy of vaporization, and the relative volumes of energy that engines can breathe. Important combustion properties include the minimum ignition energy, laminar burning velocity, Markstein numbers for strain and curvature, flame extinction stretch rates for positive and negative stretch, stretch factor for flame instability, turbulent flame burning and quenching, autoignition delay time, excitation time for autoignition heat release, research and motor octane numbers and cetane number (CN). Such properties are manifest in a variety of aerodynamic contexts: for example, the nature of autoignition depends on spatial reactivity gradients and the acoustic speed. Particular problems can arise in characterizing engine knock when operational regimes are outside those in which the octane and CNs are determined. The approach adopted in this article does not assume any unique alignment between fuels, mode of combustion, and power unit, and different possibilities are discussed.  [Show abstract] [Hide abstract]
ABSTRACT: The principal burning characteristics of a laminar flame comprise the fuel vapour pressure, the laminar burning velocity, ignition delay times, Markstein numbers for strain rate and curvature, the stretch rates for the onset of flame instabilities and of flame extinction for different mixtures. With the exception of ignition delay times, measurements of these are reported and discussed for ethanolair mixtures. The measurements were in a spherical explosion bomb, with central ignition, in the regime of a developed stable, flame between that of an under or overdriven ignition and that of an unstable flame. Pressures ranged from 0.1 to 1.4 MPa, temperatures from 300 to 393 K, and equivalence ratios were between 0.7 and 1.5. It was important to ensure the relatively large volume of ethanol in rich mixtures at high pressures was fully evaporated. The maximum pressure for the measurements was the highest compatible with the maximum safe working pressure of the bomb. Many of the flames soon became unstable, due to DarrieusLandau and thermodiffusive instabilities. This effect increased with pressure and the flame wrinkling arising from the instabilities enhanced the flame speed. Both the critical Peclet number and the, more rational, associated critical Karlovitz stretch factor were evaluated at the onset of the instability. With increasing pressure, the onset of flame instability occurred earlier. The measured values of burning velocity are expressed in terms of their variations with temperature and pressure, and these are compared with those obtained by other researchers. Some comparisons are made with the corresponding properties for isooctaneair mixtures. (author) 
 [Show abstract] [Hide abstract]
ABSTRACT: A methodology is proposed for determining whether a deflagrationtodetonation transition (DDT) might occur for flame propagation along a duct with baffles, closed at the ignition end. A flammable mixture can attain a maximum turbulent burning velocity. If this is sufficiently high, a strong shock is formed ahead of the flame. It is assumed that this maximum burning velocity is soon attained and on the basis of previous studies, this value can be obtained for the given conditions. The increase in temperature and pressure of the reactants, due to the shock, further increases the maximum turbulent burning velocity. The gas velocity ahead of the flame is linked to onedimensional shock wave equations in a numerical analysis. The predicted duct flame speeds with the appropriate maximum turbulent burning velocities are in good agreement with those measured in the slow and fast flame regimes of a range of CH4–air and H2–air mixtures. DDTs are possible if autoignition of the reactants occurs in the time available, and if the projected flame speed approaches the Chapman–Jouguet velocity at the same temperature and pressure. Prediction of the first condition requires values of the autoignition delay time of the mixture at the shocked temperatures and pressures. Prediction of the second requires values of the laminar burning velocity and Markstein number. With the appropriate values of these parameters, it is shown numerically that there is no DDT with CH4–air. With H2–air, the onset of DDT occurs close to the values of equivalence ratio at which it has been observed experimentally. The effects of different duct sizes also are predicted, although details of the DDT cannot be predicted. Extension of the study to a wider range of fuels requires more data on their laminar burning velocities and Markstein numbers at higher temperatures and pressures and on autoignition delay times at lower temperatures and pressures.  [Show abstract] [Hide abstract]
ABSTRACT: Values of laminar burning velocity, ul, and the associated strain rate Markstein number, Masr, of H2–air mixtures have been obtained from measurements of flame speeds in a spherical explosion bomb with central ignition. Pressures ranged from 0.1 to 1.0 MPa, with values of equivalence ratio between 0.3 and 1.0. Many of the flames soon became unstable, with an accelerating flame speed, due to Darrieus–Landau and thermodiffusive instabilities. This effect increased with pressure. The flame wrinkling arising from the instabilities enhanced the flame speed. A method is described for allowing for this effect, based on measurements of the flame radii at which the instabilities increased the flame speed. This enabled ul and Masr to be obtained, devoid of the effects of instabilities. With increasing pressure, the time interval between the end of the ignition spark and the onset of flame instability, during which stable stretched flame propagation occurred, became increasingly small and very high camera speeds were necessary for accurate measurement. Eventually this time interval became so short that first Masr and then ul could not be measured. Such flame instabilities throw into question the utility of ul for high pressure, very unstable, flames. The measured values of ul are compared with those predicted by detailed chemical kinetic models of onedimensional flames. 
Article: The quenching of premixed turbulent flames of isooctane, methane and hydrogen at high pressures
[Show abstract] [Hide abstract]
ABSTRACT: Experimental and theoretical studies are reported of turbulent flame quenching, with premixed flames of methane–air, isooctane–air and hydrogen–air. Mixtures were exploded in a fanstirred explosion bomb in which the rms turbulent velocity was varied by changes in fan speed. Influences of Markstein number, pressure up to 1.5 MPa, and Karlovitz stretch factor, K, were studied. It was found that the ratio of the positive stretch rate for extinction of the laminar flame to the rms turbulent strain rate was an important parameter. Probabilities of flame propagation, pf, from an initial kernel were measured. The conditions for probabilities of 0.8 and 0.2 were correlated in terms of the Karlovitz stretch factor and Markstein number. However, at a given pf, the theoretical probability, Pbf, that the spectrum of stretch rates for a turbulent flame would support propagation, declined with increase in K. This theoretical probability assumed a positive flame stretch rate for quenching equal to that for the extinction of a steadystate laminar flame. Its declining value with K, at constant pf, was attributed, at least in part, to the inability of nonsteadystate flamelets of the turbulent flame to respond to transient excursions of flame stretch rates to values in excess of those that extinguish laminar flames. Another possibility was that the highly convoluted flames were able to follow a route through regions of sufficiently low stretch rate to avoid extinctions.  [Show abstract] [Hide abstract]
ABSTRACT: Experiments involving a spherical explosion bomb are reported, in which DarrieusLandau thermodiffusive, DL,TD, flame instabilities interacted with primary and secondary, selfexcited, thermoacoustic oscillations. Explosions with central ignition demonstrated that rich ioctane and lean hydrogenair mixtures generated strong pressure oscillations, a consequence of their negative Markstein numbers. Utilizing dual wall ignitions, the structures of high pressure flames were studied using appropriate optical techniques. The conditions that gave rise to the greatest increase in the rate of combustion were strong initial DL,TD, flame instabilities and a high rate of change of the heat release rate, sufficient to generate strong secondary pressure oscillations. These, in turn, generated RayleighTaylor instabilities that further wrinkled the flames. The bomb was equipped with four fans which showed that an rms turbulent velocity in excess of about 0.6 m/s was sufficient to reduce, and almost eradicate, the effect of these instabilities on the flame speed.  [Show abstract] [Hide abstract]
ABSTRACT: In an earlier mathematical model of laminar pulverized coal–air combustion, supported by added CH4, it was assumed that the volatiles from the coal consisted solely of CH4 and HCN. A revised model is introduced with speciated devolatilization rate constants for tar, CH4, CO, CO2, H2O, H2, and HCN. It is assumed that these rate constants can also be applied to the devolatilization of the tar. In addition, it is assumed that the soot is predominantly carbon and is oxidized by the attack of O, H, OH, and O2, in the same way as the coal char. Because the devolatilization rate is strongly dependent on particle temperature, the latter has to be determined accurately from the momentum and energy equations of the particle. The model is onedimensional, with axial radiative transfer. The introduction of soot formation and speciation of the volatiles results in much improved accuracy in the prediction of species and temperature profiles in subatmospheric combustion on a flat flame matrix burner. It is possible to derive an overall global devolatilization rate constant that agrees reasonably with the measurements. These computations suggest that the effective area of the assumed spherical coal char particles is four times greater than that of the assumed sphere. Modeling of atmospheric pressure flames suggests that in this case, the value of 4 should be reduced, probably because, as pressure increases, the diffusion flux of reactant is reduced. Subatmospheric pressure laminar burning velocities are predicted with satisfactory accuracy over the full range of overall equivalence ratios. Previous measurements of laminar burning velocity at atmospheric pressure are reviewed. However, the various means of supporting a stable coal flame and the associated uncertain geometries make it impossible to apply the present model to the different conditions. It is suggested that burning velocities measured on a flat flame burner, with a controlled amount of methane to support the combustion of a pulverized coal/air mixture, would provide a good test of the reactivities of different coals.  [Show abstract] [Hide abstract]
ABSTRACT: A flamelet approach is adopted in a study of the factors affecting the volumetric heat release source term in turbulent combustion. This term is expressed as the product of an instability enhanced burning rate factor, Pbi, and the mean volumetric heat release rate in an unstretched laminar flamelet of the mixture. Included in the expression for Pbi are a pdf of the flame stretch rate and a flame stretch factor. Fractal considerations link the turbulent burning velocity normalised by the effective rms turbulent velocity to Pbi. Evaluation of this last parameter focuses on problems of (i) the pdfs of the flame stretch rate, (ii) the effects of flame stretch rate on the burning rate, (iii) the effects of any flamelet instability on the burning rate, (iv) flamelet extinctions under positive and negative flame stretch rates, and (v) the effects of the unsteadiness of flame stretch rates. The Markstein number influences both the rate of burning and the possibility of flamelet instabilities developing which, through their ensuing wrinkling, increase the burning rate. The flame stretch factor is extended to embrace potential DarrieusLandau thermodiffusive flamelet instabilities. A major limitation is the insufficient understanding of the effects of negative stretch rates that might cause flame extinction. The influences of positive and negative Markstein numbers are considered separately. For the former, a computed theoretical relationship for turbulent burning velocity, normalised by the effective rms velocity, is developed which, although close to that measured experimentally, tends to be somewhat lower at the higher values of the Karlovitz stretch factor. This might be attributed to reduced flame extinction and reduced effective Markstein numbers when the increasingly nonsteady conditions reduce the ability of the flame to respond to changes in flame stretch rates. As the pressure increases, Markstein numbers decrease. For negative Markstein numbers the predicted values of Phi and turbulent burning velocity are significantly increased above the values for positive Markstein numbers. This is confirmed experimentally and these values are close to those predicted theoretically. The increased values are due to the greater stretch rate required for flame extinction, the increased burning rate at positive values of flame stretch rate, and, in some instances, the development of flame instabilities. At lower values of turbulence than those covered by these computations, burning velocities can be enhanced by flame instabilities, as they are with laminar flames, particularly at negative Markstein numbers. (c) 2005 The Combustion Institute. Published by Elsevier Inc. All rights reserved.
Publication Stats
3k  Citations  
109.81  Total Impact Points  
Top Journals
Institutions

19702013

University of Leeds
 School of Mechanical Engineering
Leeds, England, United Kingdom
