Bradley A. Williams

Comparative species concentrations in CH4/O2/Ar flames doped with N2O, NO, and NO2

Abstract Laser-induced fluorescence is used to monitor the relative concentrations of the species OH, CH, NH, CN, NCO, and NO in a 10-torr premixed stoichiometric methane/oxygen/argon flame doped with N2O, NO, and NO2 at 0.0086 mole fraction. The relative yields of combustion intermediates for equivalent mole fractions of the different dopants are measured, and comparisons are made with PREMIX calculations based on currently accepted reaction rates. This procedure allows the chemical pathways followed by the different dopants to be compared under nearly identical conditions. Comparison of radical concentrations between the NO- and N2O-doped flames shows that CN and NCO are underpredicted for the N2O-doped flame, while NO is overpredicted. Conversely, NH is underpredicted for the NO-doped flame. These findings suggest that reburn consumption of NO is roughly 50% higher than predicted by current models. Nevertheless, the fraction of NO that reacts with hydrocarbon radicals in the reburn reactions is found to be fairly minor (≤20%). Under the present conditions, NO2 is rapidly converted to NO, leading to quantitatively similar behavior for these two dopants.

Suppression of nonpremixed flames by fluorinated ethanes and propanes

Abstract Suppression of methane/air and propane/air nonpremixed counterflow flames, and n-heptane and methanol cup burner flames by fluorinated hydrocarbons was investigated. Four fluorinated ethanes, 10 fluorinated propanes, four bromine- or iodine-containing halons, and the inert agents CF 4 , SF 6 , and N 2 were tested in some or all of the flames. Laser Doppler velocimetry (LDV) determinations of peak velocity gradients in the oxidizer flow of the counterflow flames were found to be linearly correlated with the expression for global strain rate derived for plug flow boundary conditions. This correlation was used to estimate strain rate values at extinction. The bromine- or iodine-containing agents are more effective on a molar basis than the fluorinated propanes, followed by the fluorinated ethanes, and finally SF 6 , CF 4 , and N 2 . Agent effectiveness increases with the number of CF 3 groups present in the agent molecular structure. Numerical investigations of the flame speed reduction of methane/air mixtures doped with either CHF 2 CHF 2 or CF 3 CH 2 F predict that the latter is the better agent, in accord with experimental observations. Chemical contributions to suppression account for less than 35% of the total suppression offered by fluorinated hydrocarbons not containing bromine or iodine. At strain rates below 100 s −1 , suppression effectiveness rankings in methane and propane counterflow flames are similar to those obtained in n-heptane and methanol cup burner flames. Methanol flames are more difficult to extinguish than the alkane flames investigated, particularly with the chemical agent CF 3 Br.

The effect of nitric oxide on premixed flames of CH4, C2H6, C2H4, and C2H2

Abstract Nitric oxide was doped into premixed, stoichiometric, 10 torr flames of CH4/O2/N2, C2H6/O2/N2, C2H4/O2/N2, and C2H2/O2/N2. Spatial profiles of electronic ground state CH, OH, NO, CN, NCO, NH and metastable state 3C2 were obtained by laser-induced fluorescence. The peak temperatures of the flames were maintained at 1800 K by varying the N2 buffer gas mole fraction. Comparisons of the profiles among the different fuels illustrate differences in the hydrocarbon flame structure that lead to different NO reactivities in these flame systems. Concentrations of the nitrogen-containing intermediates are similar in the methane, ethane, and ethylene flames, but are considerably larger in the acetylene flame, indicating a much greater NO reactivity in this flame. The species profiles are modeled using three different kinetic mechanisms; all of these have significant shortcomings.

Dynamics and suppression effectiveness of monodisperse water droplets in non-premixed counterflow flames

In-situ measurements of velocity and size distributions of initially monodisperse water mists of initial diameters ranging from 14 μm to 44 μm seeded into the air stream of nonpremixed propane/air counterflow flames are reported. Droplets were generated piezoelectrically, and the size and velocity distributions and the number density were determined by phase-Doppler particle anemometry. Droplets having initial diameters of 18 μm underwent complete vaporization in a counterflow flame at a strain rate of approximately 170 s -1 , while droplets of 30 μm penetrated slightly beyond the visible flame zone. Measurements of the effect of water droplets on the extinction strain rates of propane/air counterflow flames were performed. Droplets of 14 μm and 30 μm were found to have similar suppression effectiveness, while droplets of 44 μm were substantially less effective at reducing the extinction strain rate. Both the 14 μm and 30 μm water droplets were found to be more effective, on a mass basis, than CF3Br. The present experimental results are in excellent agreement with the predictions of recent modeling studies exploring the behavior of various sized water droplets in counterflow

Intermediate species profiles in low-pressure methane/oxygen flames inhibited by 2-H heptafluoropropane: comparison of experimental data with kinetic modeling

Abstract Experimental profiles of the intermediate species H, OH, CH, CF, CF 2 , and CHF are obtained in a 10 torr premixed flat flame of methane/oxygen in a 1:2 molar ratio, inhibited by a 4% mole fraction of 2-H heptafluoropropane (HFP, CF 3 CHFCF 3 ). These data are compared to calculations using a recently published kinetic mechanism describing the consumption of this fire suppression agent. The profiles in the flame inhibited by HFP are compared to previously published data for flames containing CHF 3 and CH 2 F 2 under the same conditions of stoichiometry and flux of fluorine atoms. The species profiles relative to the flames containing the fluoromethanes are accurately predicted and atmospheric pressure flame speeds are fairly well predicted by the kinetic mechanism. Under equal fluorine loadings, profiles of temperature and of H and OH mole fraction are virtually identical between the flames containing HFP and CHF 3 . The flame inhibited by HFP, however, has approximately twice as much CH∗ emission as the flame containing CHF 3 . The kinetic model predicts that thermal decomposition, rather than H atom abstraction, is the primary destruction mechanism for HFP under the conditions studied.

Inhibition of premixed methane/air flames by water mist

The effect of submicron water drops on the buring velocity of methane/air mixtures was investigated. Results are compared to the suppression effect of water vapor, gaseous thermal agents N2 and CF4, and chemical agent CF3Br to determine if the theoretical thermal suppression effect expected for water mist can be experimentally achieved. These studies lay the groundwork (both experimental and modeling) for the effects of larger drops as well as for drops with solutes. Stoichiometric mixtures were stabilized at atmospheric pressure on one of two nozzle-type burners, producingcone-shaped flames. Burning velocities were determined using the total area method. An atomizer was used to produce water mist with a mean drop diameter of less than 1 μm, which was delivered as part of the reactant stream. Studies were carried out using both dry and humidified air. On a mass basis, the burning velocity of N2- and CF4-inhibited flames exhibited similar characteristics. Water vapor was observed to be more effective than N2 or CF4, but less effective than the same mass of water mist, consistent with thermodynamic analyses. Water mist inhibition results were well predicted by a recently developed multiphase flame model. Under these conditions, the burning velocity reduction effectiveness of water mist is approximately 3.5 times greater on a mass basis than for N2 and CF4 and is comparable to modeled data for CF3Br-inhibited flames.

Radical species profiles in low-pressure methane flames containing fuel nitrogen compounds

Abstract Relative concentration profiles of CN, NH, NCO, NH 2 , and NO are recorded for a 10 torr, premixed CH 4 /O 2 /Ar flame doped with the following compounds: 0.35% NH 3 , CH 3 NH 2 , and CH 3 CN, and 0.86% NO. Profiles were measured by laser-induced fluorescence and compared between the different dopants. The experimental data are compared to detailed kinetic calculations based on the Gas Research Institute Mechanism 2.11 for the NH 3 and NO dopants, with added submechanisms for CH 3 CH and CH 3 NH 2 . The kinetic model underpredicts the amount of CN and NCO formed from the ammonia dopant; if we postulate a reaction between CH 3 and NH, these discrepancies are largely resolved. Comparison of CN and NCO profiles between the NO and NH 3 dopants indicates that HCN is the primary product of the CH + NO reaction. Extrapolation of measured rates for CH 3 CN destruction reactions to combustion temperatures predicts a rate of CH 3 CN removal which is far too slow. Previously published mechanisms for methylamine combustion overpredict the amount of NH 2 produced, indicating that cleavage of the CN bond is less likely in the initial attack than the kinetic mechanisms predict. We propose modifications to the CH 3 CN and CH 3 NH 2 submechanisms which correct these deficiencies and lead to good agreement with the measured intermediate profiles for all dopants.

Sensitivity of calculated extinction strain rate to molecular transport formulation in nonpremixed counterflow flames

The counterflow flame is a commonly used geometry for experimental and chemical kinetic modeling studies of nonpremixed combustion. Because the flame structure is quasi-one-dimensional and more computationally tractable than inherently multidimensional flame geometries [1], detailed chemical kinetic modeling codes for counterflow flames have come into widespread use. An important characteristic of counterflow flames is the extinction strain rate, the maximum velocity gradient which a flame can support and still burn. At strain rates below the extinction value, there exist three solutions to the governing equations which determine the flame structure. Two of the solutions correspond to a stable flame and an essentially non-reacting cold flow, while the third branch is an unstable solution which cannot be physically realized [2]. Various computational approaches have been used to extend stable solutions onto the unstable branch, and thus determine the limits of bistability corresponding to flame extinction and autoignition [2, 3]. The extinction strain rate is important for modeling turbulent combustion using the laminar flamelet approach [4], and is a figure of merit for the effectiveness of fire suppressants in nonpremixed flames [5]. It is desirable for calculations to accurately predict extinction strain rates. Substantial variations in predicted extinction strain rates of nonpremixed methane/ air counterflow flames have been noted between different chemical kinetic mechanisms [6]. Here, the predicted extinction strain rate is found to also be sensitive to the treatment of molecular transport in the computational model. The effect of transport on premixed flame structure and extinction has been recently investigated by Paul and Warnatz [7], and by Ern and Giovangigli [8, 9]. To our knowledge, the effect of transport formalism on extinction of nonpremixed counterflow flames has not been previously reported in the literature. Furthermore, earlier studies reporting computational predictions of extinction strain rates have not always specified the manner in which the calculation handled species transport. In the present study, we employ two computer programs developed for counterflow flames incorporating the CHEMKIN program packages developed at Sandia National Laboratories. Initially, the chemical kinetics [10] and molecular transport [11] routines were applied

In situ determination of molecular oxygen concentrations in full-scale fire-suppression tests using tunable diode laser absorption spectroscopy

The fast and accurate determination of oxygen in air is important for fire research. Available O2 sensors (paramagnetic, electrochemical, ZrO2) are only of limited use because of significant errors caused by the specific sampling or measurement process, so that a purely optical, in situ detection is of great interest. Optical methods can account for the dilution of O2 by water (both vapor and drops) and are thus valuable tools for studying the effectiveness of water for replacing halogenated fire suppressants. To fulfill this need, a tunable diode laser based absorption spectrometer (TDLAS) has been developed for the in situ detection of molecular oxygen at 760 nm (A band, b1Σ+g←X3Σ−g). The device was successfully tested during fullscale fire-suppression tests carried out at the Naval Research Laboratory Chesapeake Bay Detachment facility in a 28-m3 rest compartment. A specially protected open-path Herriott multipass setup with an absorption length of 1.8 m was developed to restrict the probe volume to a base length of 30 cm. Various scenarios including water mist only, unsumppressed fires, and water-suppressed fires (methanol and n-heptane, both pool and cascading fires up to 400 kW) were investigated. O2 concentrations were measured at a 2.5 Hz repetition rate with a resolution of 0.01 to 1 vol % O2 depending on the transmission conditions. This demonstrated for the first time the capability for in situ oxygen measurements under fire-suppression conditions with large and rapid obscuration changes (transmission of as little as 0.8% of the emitted laser power). It also showed that the TDLAS results account for dilution of O2 by water vapor without any interference of other species.

Intermediate species profiles in low pressure premixed flames inhibited by fluoromethanes

Abstract We have investigated premixed 10 torr methane/oxygen flames containing CH 3 F, CH 2 F 2 , CHF 3 , and CF 4 . Profiles of temperature and CH∗ chemi-luminescence were acquired, and laser-induced fluorescence (LIF) was used to obtain profiles of the intermediate species H, OH, CH, CF, CHF, CF 2 , and CF 2 O. The fluoromethanes were added in amounts such that each flame had the same flux of fluorine atoms. In the flames containing CH 3 F, CH 2 F 2 , and CHF 3 , the methane flow was adjusted to give an equivalence ratio of 1.07 for all three inhibited flames. The experimental intermediate profiles were compared to predicted profiles calculated from a hydrofluorocarbon kinetic mechanism recently developed at NIST. No fluorinated intermediates were detectable in the CF 4 inhibited flame, indicating that this agent does not react significantly under the flame conditions studied. The temperature profiles, H atom profiles, and OH profiles for the other three fluoromethane inhibited flames are nearly identical, indicating that flames containing different fluorocarbon compounds, but identical proportions of fluorine atoms, have similar structures. The kinetic model correctly predicts the location of the reaction zone in the flames containing CH 2 F 2 and CH 3 F. In the CHF 3 flame, however, the location of the reaction zone is predicted to be too far above the burner surface, and concentrations of H and OH are too low. The discrepancy appears to be due to pressure dependence and third body efficiencies of the agent thermal decomposition. Furthermore, relative amounts of CF, CH, CF 2 , and CHF in the different flames are not very well predicted. In general, partially fluorinated methyl and methylene radicals appear to have a greater than predicted propensity to lose hydrogen atoms rather than fluorine. We propose modifications to the fluorine mechanism to correct the discrepancies observed in the low pressure experiments, while simultaneously achieving good agreement with atmospheric pressure flame speed data in CH 4 /air/CHF 3 flames, and, except in rich conditions (φ > 1.25) CH 4 /air/CH 2 F 2 flames.