Turbulent Nonpremixed Flames Of Methane Research Articles

Extinction of buoyant turbulent non-premixed flames under reduced oxygen concentrations

Local and global extinction of 10 kW buoyant turbulent non-premixed flames of methane from a 13.7 cm diameter burner under oxygen concentrations (OCs) from 20.9 % to 12 % are investigated using a combination of OH* chemiluminescence imaging and exhaust-gas sampling. Local flame extinction, indicated by holes or discontinuities in the OH* images, is only observed at OC lower than 17 %. For OC below 16 %, large-scale turbulent flow dynamics causes rapid creation of large extinction holes that do not reclose, and the overall combustion efficiency (η) decreases rapidly as local extinction becomes more probable. Direct observation of local extinction in buoyant turbulent flames provides important physical insights and validation data for models of extinction and re-ignition by distinguishing their respective contributions to η. We observe that local extinction events are associated with large-scale flow features, leading us to hypothesize that extinction behavior correlates with resolved-scale strain rates from large-eddy simulation (LES) in excess of the laminar extinction limit. This idea is investigated by performing LES for the target flame, extracting strain-rate statistics on the flame surface, and comparing these to both the extinction limits for canonical counterflow diffusion flames and to the experimentally observed extinction behavior. Results show that the cumulative distribution of turbulent flame strain rates up to the laminar extinction limit closely follows the measured OC dependence of η. This approach to η works well for different OCs and other fuels such as propane and ethylene, which suggests great potential for the development of computationally efficient extinction models for fire suppression.

Liftoff of a Co-Flowing Non-Premixed Turbulent Methane Flame: Effect of the Fuel Nozzle Orifice Geometry

This paper reports an experimental study on the effect of the fuel nozzle orifice geometry on the stability of turbulent non-premixed methane flame. Different internal geometries (orifice equivalent diameter, length to diameter ratio and contraction angle) of a circular and a rectangular nozzle with an aspect ratio of 2 were examined. The strength of the co-airflow was also varied to assess its impact on the jet flame stability. The experimental data revealed that the level of turbulence in the jet near-field is, in general, higher for the rectangular nozzle in comparison with the circular nozzle. This high level of turbulence was found to accelerate the liftoff transition of the attached flame. The results revealed also that there is a clear interplay between the flame liftoff height and the jet flow characteristics. That is, a rectangular jet, which spreads faster along the minor axis and generates higher near-field turbulence, results in a flame base sitting closer to the nozzle exit in comparison with that of its circular nozzle counterpart. Finally, the presence of a moderate co-airflow resulted in a higher flame liftoff velocity and height. It also led to the appearance of a hysteresis phenomenon in the low jet velocity range regardless of the exit orifice shape of the fuel nozzle.

On the Stability of a Turbulent Non-Premixed Methane Flame

This paper presents an experimental qualitative assessment of the stability of a turbulent, non-premixed methane flame. The burner consists of a central fuel nozzle surrounded by an annulus of co-airflow with varying swirl strength. Two distinct nozzle geometries having similar exit cross-sectional areas but different internal/orifice geometry—a rectangular and a contracted circular nozzle—were tested. They are referred to in this paper as RN and CCN, respectively. The main focus of the present study was on determining the flame liftoff and blowout velocities as well as the liftoff height and flame length, all of which can be used as indicators of the stability of non-premixed methane flame. The experimental data revealed that the blowout velocity of the RN nozzle's flame is remarkably higher than that of the CCN nozzle, and the liftoff velocity of the CCN is only slightly higher than that of the RN nozzle for identical swirl strength. In addition, the flame length of the RN nozzle is overall shorter than that of the CCN nozzle for identical test conditions, and the liftoff height of the CCN flame is higher than that of the RN flame. LDV velocity measurements were performed to determine the reacting flow overall dynamic along the centerline plane for typical jet and co-flow exit velocities. These results aimed at helping to explain the difference in flame stability between the two different nozzles' geometries in conjunction with the co-flow swirl strength.

Conditional moment closure calculations of a swirl-stabilized, turbulent nonpremixed methane flame

As natural gas production and usage continues to increase, displacing oil and coal, there is an escalating requirement to ensure that natural-gas burning equipment performs as cleanly and efficiently as possible to allow the environmental advantages of this fuel to be realized. Presented are encouraging results obtained from the three-dimensional, elliptic, CMC modeling of a low-swirl-stabilized nonpremixed flame of methane, which are representative of such technologies. Calculations were based upon the solution of the three-dimensional fluid-flow equations supplemented with a Reynolds stress and scalar flux second-moment closure, and the chemistry applied to represent mean production rates of species was a 16-step reduced mechanism. Predictions of species in both mixture fraction and real space display a level of conformity with experimental data that is encouraging for these methods in both a qualitative and a quantitative sense. Shortcomings of the modeling procedures are discussed in light of the results, and suggestions are made for future investigation.

Swirling turbulent non-premixed flames of methane: Flow field and compositional structure

This paper introduces a new swirl burner which has simple, well-defined boundary conditions and which stabilizes complex, turbulent, unconfined flames that are not unlike those found in practical combustors. Two flames with identical swirl numbers but differing bulk jet velocities Uj are selected for further inves- tigations. Flow field measurements reveal that a second recirculation zone may exist on the centerline of the flames further downstream of the primary recirculation zone. This is attributed to vortex breakdown. The measurements also show the presence of highly rotating collar-like flow features present between the primary and secondary recirculation zones. These regions of the flow are characterized by high tangential shear stresses uw� . The compositional structures of these methane flames are measured using the simultaneous Raman- Rayleigh laser-induced fluorescence (LIF) technique. The LIF technique is used to measure concentra- tions of OH, CO, and NO. Results are presented as scatter plots and radial Favre mean profiles of tem- perature, mixing, and composition fields. As the fuel jet velocity increases and the flame approaches blowoff, a higher proportion of fluid samples shifts away from fully burned conditions and closer to a mixing asymptote. An interesting feature of these flames is that these locally extinguished samples originate mostly from regions of the flow near the high-shear, collar-like region, which is not found in similar bluff- body flames.

Three-scalar imaging in turbulent non-premixed flames of methane

Raman scattering from nitrogen in laminar (Re=1600) and turbulent (Re=15,000) non-premixed flames of diluted methane has been investigated as a means of complementing the two-scalar mixture-fraction computation based on measurements of fuel mass fraction and temperature (ξ} FT ). Using a diluent consisting of argon and oxygen with an overall volumetric dilution ratio of 3/1 (diluent/fuel) allows sufficient variation in the measured nitrogen concentration for its use as a passive scalar. The present experimental setup requires only a single laser (532 nm) in a high-power intracavity configuration. Mixture-fraction profiles calculated using independent fuel-temperature and nitrogen-temperature two-scalar approaches show excellent agreement in the laminar flame. For the turbulent flame, a 100 single-shot average of mixture fraction shows reasonable agreement between the two approaches. Discrepancies are most evident in single-shot images in regions of large mixture fraction (ξ>ξ s ) where the nitrogen Raman signal is noise dominated. The location of the stoichiometric contour is consistently determined by both two-scalar approaches. A simple modification of the functional dependence for parameters appearing in the fuel-temperature mixture-fraction formulation is shown to correct deficiencies in the approach predicted by laminar flame calculations, most notably for values of ξ FT around stoichiometric and into lean regimes.

Computational and experimental investigation of a turbulent non-premixed methane flame

Mathematical modeling of turbulent combustion processes is a challenging task, mainly because of the problem of describing the coupling of the chemical kinetics with the turbulent flow. Existing models still have to be improved, which can only be achieved by comparison and validation with experimental results. We present experimental results and numerical simulations of a piloted non-premixed methane-airflame of semitechnical size (50 kW). The experimental data comprise measurements of velocities, temperature, and species concentrations. The model used for the simulation of the turbulent flame consists of a k-ε model for the turbulent flow field combined with a Monte Carlo method for the solution of the joint probability density function (PDF) of velocities and scalars that overcomes the closure problem for the chemical kinetics. Finite-rate chemistry is incorporated in the model using a two-step reduced chemistry model obtained by an automatic mechanism reduction procedure based on intrinsic low-dimensional manifolds, together with a table look-up procedure. Comparisons of experimental and computational results show that the model is able to predict the main characteristics of the flame well. Discrepancies can be mainly attributed to deficiencies of the micromixing model and inaccuracies in the description of the turbulent flow field by the k-ε model. However, it is clearly shown that finite chemistry effects play a major role in this flame.

Chemical Inhibition of Nonpremixed Flames of Hydrocarbon Fuels with CF3Br

Abstract Experimental data are presented for the extinction limits of pilot-stabilised turbulent nonpremixed flames of methane and propane fuels inhibited by Bromotrifluoromethane, CF3Br commonly known as halon-1301. Compared with thermal inhibitors like N2 and At it is shown that CF3Br is far more effective because it acts chemically as well. The halon is mixed either with the fuel or with the coflowing air stream. Results indicate that as the turbulent mixing rates increase, the mass fraction of CF3Br required to extinguish the flame decreases linearly with jet velocity regardless of whether the inhibitor is added to the fuel or to the air stream. Using a detailed chemical kinetic mechanism with 34 species and 91 reactions, the structures of laminar counterflow nonpremixed flames of methane inhibited with CF3Br are calculated. The flames are maintained at a constant, low stretch rate of a = 5s-1 and the inhibitor is mixed either with the fuel or with the counterflowing air. It is found that when halon i...

Raman-rayleigh measurements in bluff-body stabilised flames of hydrocarbon fuels

Bluff-body stabilised turbulent nonpremixed flames of methane and methanol are investigated. In the flame region downstream of the recirculation zone, turbulent mixing rates become very intense at high enough velocities and the interactions between turbulence and chemistry becomes significant. Simultaneous, space- and time-resolved, Raman-Rayleigh measurements of temperature and the mass fractions of fuel, CO, CO 2 , H 2 , H 2 O, O 2 and N 2 are presented for flames with low turbulent mixing rates as well as ones close to blowoff. The flames also range from being fuel-jet dominant, with just hot gases stabilising the flame to the nozzle, to coflow-air dominant with a fully visible recirculating zone. In methanol flames, the recirculation region is visibly soot-free, while in methane flames it is sooting and hence causes significant interferences on the laser Raman scattered light from all species especially CO 2 . The results are presented in the form of scatter plots as well as probability density functions conditioned with respect to mixture fraction. It is found that bluff-body stabilised flames close to extinction show similar characteristics to pilot-stabilised free jet-flames of the same fuel. Namely, a bimodal approach to blowoff with an increasing percentage of locally extinguished samples as the turbulent mixing rates incrase. In methane flames, the peak levels of CO and H 2 are higher than those calculated for laminar flamelets and this is consistent with measurements in pilot-stabilised flames. In methanol flames, the bimodality is more distinct and the peak levels of CO are lower than those measured in pilot-stabilised flames.

PDF calculations of piloted turbulent nonpremixed flames of methane

A velocity-composition joint pdf transport equation has been solved by the Monte Carlo method to calculate the structure of pilot-stabilized turbulent nonpremixed flames of methane. Three components of velocity and a conserved scalar, namely, mixture fraction, ξ are jointly represented in the pdf. A new model is used for turbulent frequency. Turbulent dissipation and the fluctuating pressure gradient terms are conditionally modeled. Two simple models for thermochemistry are used. In one, density is a piecewise function of ξ, and in the other, density is obtained from calculations of a laminar counterflow diffusion flame of methane with a stretch rate, a = 100 s −1. Calculations are compared with the corresponding experimental measurements performed on a number of flames ranging from flames with low mixing rates to ones close to extinction. The velocity, turbulence, and mixing fields are predicted with reasonable accuracy down to x D j ∼ 30 . The agreement with experiments is less satisfactory at x D j = 50 . This will improve with the use of better pdf models which have recently been developed [27, 30, 31]. The reactive scalars are very well predicted only for flames which are far from extinction. This is expected considering the simplicity of the thermochemical model. This paper is a first step towards implementing realistic chemistry in the code, and hence predicting the observed finite rate kinetic effects in these flames.

Conditional probability density functions measured in turbulent nonpremixed flames of methane near extinction

Spontaneous Raman-Rayleigh measurements of temperature and species concentrations have been made in turbulent nonpremixed flames of methane close to extinction. For each data point, five different values of reactedness are calculated from the measured temperature and the mass fractions of CH 4, O 2, H 2O and CO 2. These data are then sampled with respect to five ranges of mixture fraction and the conditional single and joint probability density functions (PDFs) of reactedness calculated from the various reactive scalars are presented for each range. Conditional means and root mean squares (rms) of fluctuations of reactedness, temperature, and species mass fractions are also presented. It is found that for mixture fraction ranges lying below the rich reaction limit, ξ < ξ R ( ξ R = 0.087 for methane) combustion is generally bimodal and fluid samples are either fully burned or unburned. For ranges with ξ > ξ R, the majority of mixtures are partially reacted and the conditional PDFs have centered distributions. The correlation between the conditional PDFs of various reactive scalars indicate that more than a single-step reaction mechanism is required to adequately describe the chemistry. Close to extinction, the flame was kept alight by fully burned pockets of fluid that provide a source of reignition for mixtures with slower mixing rates. The flame blows off when the frequency of occurrence of these pockets becomes small.

Turbulent nonpremixed flames of methane near extinction: Probability density functions

Space- and time-resolved measurements of major species concentrations and temperature have been made using the Raman-Rayleigh scattering technique in the blue (visibly soot free) regions of turbulent nonpremixed flames of methane close to extinction. The data are presented in this paper in the form of single variate probability density functions (pdfs) for the mixture fraction, temperature, and the mass fractions of CH 4, O 2, H 2O, H 2, CO 2, and CO. Representative instantaneous scatter plots and joint pdfs are also shown. When the mixing rates are low, the data show mostly fully burnt mixtures indicating that the chemistry is relatively fast. As the flame approaches extinction, most local mixtures become either partially burnt or simply mixed. The joint pdfs shown bimodality for mixture fractions less than ∼.1 and centered distributions for richer mixtures. When close to extinction, fully burnt pockets of fluid are still encountered and these may be responsible for keeping the flame alight. Questions relating to the local, instantaneous flame structure near extinction are discussed in light of existing theoretical models.

Turbulent nonpremixed flames of methane near extinction: Mean structure from Raman measurements

Spontaneous Raman-Rayleigh measurements of species concentration-temperature have been made in turbulent nonpremixed flames of methane near extinction. The means and rms of fluctuations of temperature, density, mixture fraction, and the mass fractions of CH 4, O 2, N 2, H 2O, H 2, CO, and CO 2 are presented in this paper. The mean Raman-Rayleigh results compare favorably with probe measurements of species concentrations and temperature made in similar flames. At a given axial location, unreactedness (finite rate kinetics) and not unmixedness [high ( ξ″ 2) 1 2 ] explains almost entirely the change in the flame structure. As the flame approaches extinction, the Favre averaged mass fractions of CH 4 and O 2 increase while those of the products H 2O, H 2, CO, and CO 2 decrease. The rms of fluctuations in the temperature and mass fractions O 2, H 2O, H 2, CO, and CO 2 decrease as well. Although these results favor the concept of a diffuse flame structure, the existence of partially premixed laminar diffusion flamelets cannot be excluded.

“Fluorescence” interference with Raman measurements in nonpremixed flames of methane

What appears to be fluorescence has been measured in the blue (visibly soot free) regions of laminar and turbulent nonpremixed flames of methane at atmospheric pressure. The measurements are part of an experiment to measure the spontaneous Raman scattering of eight species in the flame. The flames studied range from ones with low mixing rates to flames close to extinction. The “fluorescence” is nonresonantly excited by a pulsed dye laser and has been routinely monitored at λ = 516.5 nm which is a bandhead of diatomic carbon, C 2. It is Stokes “fluorescence” that interferes with the Raman signal detection. It is found that the “fluorescence” is broadband covering all the visible spectrum, and is observed only on the rich side of stoichiometry with noticeable intensities for fuel number densities in the range 0.1 to ⋍ 1.5 × 10 18 cm −3. The “fluorescence” intensity is of the order of the Raman signals, increases with the fourth power of the laser flux, is unaffected by flame stretch, and decreases with dilution of the unburned methane. No “fluorescence” is detected at temperatures below ⋍ 1000°C. Although the sources of this “fluorescence” are not fully known, it is likely that C 2 is only one of many contributing radicals and molecules. The possible contribution of incandescence from vaporizing soot particle is also discussed. Irrespective of the sources of the “fluorescence,” its interference with the Raman lines of stable species in the flame can be corrected for using Stokes and anti-Stokes “fluorescence” monitors, and the Raman scattering technique can therefore be used for measurements in the blue regions of methane flames and may be extendable to higher hydrocarbon flames.