Modeling of soot aggregate formation and size distribution in a laminar ethylene/air coflow diffusion flame with detailed PAH chemistry and an advanced sectional aerosol dynamics model

Soot aggregate formation and size distribution in a laminar ethylene/air coflow diffusion flame is modeled with a PAH-based soot model and an advanced sectional aerosol dynamics model. The mass range of solid soot phase is divided into 35 discrete sections and two variables are solved for in each section. The coagulation kernel of soot aggregates is calculated for the entire Knudsen number regime. Radiation from gaseous species and soot are calculated by a discrete-ordinate method with a statistical narrow-band correlated-k based band model. The discretized sectional soot equations are solved simultaneously to ensure convergence. Parallel computation with the domain decomposition method is used to save computational time. The flame temperature, soot volume fraction, primary particle size and number density are well reproduced. The number of primary particles per aggregate is overpredicted. This discrepancy is presumably associated with the unitary coagulation efficiency assumption in the current sectional model. Along the maximum soot volume fraction pathline, the number-based and mass-based aggregate size distribution functions are found to evolve from unimodal to bimodal and finally to unimodal again. The different shapes of these two aggregate size distribution functions indicate that the total number and mass of aggregates are dominated by aggregates of different sizes. The PAH-soot condensation efficiency γ is found to have a small effect on soot formation when γ is larger than 0.5. However, the soot level and primary particle number density are significantly overpredicted if the PAH-soot condensation process is neglected. Generally, larger γ predicts lower soot level and primary particle number density. Further study on soot aggregate coagulation efficiency should be pursued and more experimental data on soot aggregate structure and size distribution are needed for improving the current sectional soot model and for better understanding the complex soot aggregation phenomenon.     

A computational study of soot formation and flame structure of coflow laminar methane/air diffusion flames under microgravity and normal gravity

A numerical study is conducted of methane–air coflow diffusion flames at microgravity (μg) and normal gravity (1g), and comparisons are made with experimental data in the literature. The model employed uses a detailed gas phase chemical kinetic mechanism that includes PAH formation and growth, and is coupled to a sectional soot particle dynamics model. The model is able to accurately predict the trends observed experimentally with reduction of gravity without any tuning of the model for different flames. The microgravity sooting flames were found to have lower temperatures and higher volume fraction than their normal gravity counterparts. In the absence of gravity, the flame radii increase due to elimination of buoyance forces and reduction of flow velocity, which is consistent with experimental observations. Soot formation along the wings is seen to be surface growth dominated, while PAH condensation plays a more major role on centreline soot formation. Surface growth and PAH growth increase in microgravity primarily due to increases in the residence time inside the flame. The rate of increase of surface growth is more significant compared to PAH growth, which causes soot distribution to shift from the centreline of the flame to the wings in microgravity.     

Calculating the soot particle size distribution function in turbulent diffusion flames using a sectional method

Soot formation in a turbulent jet diffusion flame is modeled using an unsteady flamelet approach in post-process. In the present work, we apply a detailed kinetic soot model with a sectional method, and study the evolution of the particle size distribution. Detailed information on the evolution of the soot particle size distribution function is acquired. It is found that the particle size distribution function is bimodal throughout the flame. The transition from the small to large particle size distributions is strongly influenced by surface growth and oxidation reactions. We find that large particles are most likely to be emitted from the flame.     

Detailed investigation of soot formation from jet fuel in a diffusion flame with comprehensive and hybrid chemical mechanisms

A numerical and experimental study is performed to investigate soot formation from jet fuel in a laminar coflow diffusion flame. The combustion chemistry of the fuel is simulated using (1) the MURI jet fuel surrogate (Dooley et al. 2012) with a modestly reduced Ranzi mechanism (Ranzi et al. 2012), and (2) the recently proposed HyChem model (Xu et al. 2018) combined with the KAUST PAH mechanism 2 (Wang et al. 2013). The two reaction mechanisms are coupled with a sectional soot model to simulate a coflow diffusion flame of methane doped with the MURI jet fuel surrogate. The combined laser extinction and two-angle elastic light scattering method is used to perform non-intrusive in situ measurements of soot volume fraction, primary particle diameter and number density. The good agreement including soot particle size and number density between the experimental data and the simulation results computed with the reduced Ranzi mechanism demonstrate the robustness of the soot model to changes in fuel composition, as the model parameters are unchanged with a previous numerical study of soot formation of n-propylbenzene/n-dodecane mixtures (Zhang and Thomson, 2018). The computation with the combined HyChem/KAUST mechanism predicts similar results as the computation with the detailed chemistry of the reduced Ranzi mechanism for fuel breakdown, thus the basic premise of the HyChem model that the fuel decomposition process can be greatly simplified with the lumped reaction steps is supported. The results also show that by adding a PAH growth scheme to the HyChem model, the approach can be used to predict soot formation from jet fuel combustion in a laminar coflow diffusion flame. Finally, the dependency of the soot prediction on PAH chemistry is discussed and it is suggested that more experimental data is needed to validate the PAH mechanism and improve the predictive accuracy of the model.     

Numerical study of the effects of gravity on soot formation in laminar coflow methane/air diffusion flames under different air stream velocities

Numerical simulations of laminar coflow methane/air diffusion flames at atmospheric pressure and different gravity levels were conducted to gain a better understanding of the effects of gravity on soot formation by using relatively detailed gas-phase chemistry and complex thermal and transport properties coupled with a semi-empirical two-equation soot model. Thermal radiation was calculated using the discrete-ordinates method coupled with a non-grey model for the radiative properties of CO, CO₂, H₂O, and soot. Calculations were conducted for three coflow air velocities of 77.6, 30, and 5 cm/s to investigate how the coflowing air velocity affects the flame structure and soot formation at different levels of gravity. The coflow air velocity has a rather significant effect on the streamwise velocity and the fluid parcel residence time, especially at reduced gravity levels. The flame height and the visible flame height in general increase with decreasing the gravity level. The peak flame temperature decreases with decreasing either the coflow air stream velocity or the gravity level. The peak soot volume fraction of the flame at microgravity can either be greater or less than that of its normal gravity counterpart, depending on the coflow air velocity. At sufficiently high coflow air velocity, the peak soot volume fraction increases with decreasing the gravity level. When the coflow air velocity is low enough, soot formation is greatly suppressed at microgravity and extinguishment occurs in the upper portion of the flame with soot emission from the tip of the flame owing to incomplete oxidation. The numerical results provide further insights into the intimate coupling between flame size, residence time, thermal radiation, and soot formation at reduced gravity level. The importance of thermal radiation heat transfer and coflow air velocity to the flame structure and soot formation at microgravity is demonstrated for the first time.     

Analysis of the impact of agglomeration and surface chemistry models on soot formation and oxidation

Models for soot aggregation that account for the influence of soot surface chemistry on mass growth and oxidation are still at the formative stage. Past studies have considered techniques ranging from the method of moments to stochastic approaches and significantly different sensitivities to chemical processes such as mass growth and oxidation have been reported. The method of moments is computationally efficient and can yield encouraging results for laminar flames as well as for turbulent flames when combined with transported probability density function (PDF) methods. However, an assessment of the sensitivity to constituent model assumptions is not trivial and information regarding the soot size distribution is incomplete. In the current work, the ability of a sectional method to reproduce population dynamics data has been evaluated along with the sensitivity of predictions to closure elements associated with soot nucleation, agglomeration, surface growth and oxidation. A detailed chemistry model with 285 chemical species and 1520 reactions was used for the gas phase. It is shown that the approach to the fuel lean sooting limit can be reproduced with reasonable accuracy and that the inclusion of fractal aggregates and surface chemistry effects improve agreement with experimental data.     

Transported JPDF modelling and measurements of soot at elevated pressures

Accurate measurements and modelling of soot formation in turbulent flames at elevated pressures form a crucial step towards design methods that can support the development of practical combustion devices. A mass and number density preserving sectional model is here combined with a transported joint-scalar probability density function (JDPF) method that enables a fully coupled scalar space of soot, gas-phase species and enthalpy. The approach is extended to the KAUST turbulent non-premixed ethylene-nitrogen flames at pressures from 1 to 5 bar via an updated global bimolecular (second order) nucleation step from acetylene to pyrene. The latter accounts for pressure-induced density effects with the rate fitted using comparisons with full detailed chemistry up to 20 bar pressure and with experimental data from a WSR/PFR configuration and laminar premixed flames. Soot surface growth is treated via a PAH analogy and soot oxidation is considered via O, OH and O₂ using a Hertz-Knudsen approach. The impact of differential diffusion between soot and gas-phase particles is included by a gradual decline of diffusivity among soot sections. Comparisons with normalised experimental OH-PLIF and PAH-PLIF signals suggest good predictions of the evolution of the flame structure. Good agreement was also found for predicted soot volume statistics at all pressures. The importance of differential diffusion between soot and gas-phase species intensifies with pressure with the impact on PSDs more evident for larger particles which tend to be transported towards the fuel rich centreline leading to reduced soot oxidation.     

An aerosol model to predict size and structure of soot particles

An aerosol model to simulate soot formation and growth was developed using moving- and fixed-sectional methods. The new model is composed of a set of subroutines that can be easily combined with the Chemkin package. Using the model, we have simulated soot formation and growth in plug flow reactors.

Our model was compared with a previously published method of moments model for a simulation of the plasma pyrolysis of methane in a plug flow reactor. Inclusion of the transition correction factor for the condensation coefficient led to the prediction of a smaller condensation rate compared with the method of moments model. The average coagulation rate calculated by the sectional model was much higher than that by the method of moments model for a broad particle size distribution. The two models predicted significantly different soot precursor concentration and rates of aerosol processes, but substantially similar particle mass and number for the pyrolysis process.

We have also simulated soot formation and growth in a jet-stirred/plug flow reactor (JSR/PFR) system for which soot size distribution measurements are available in the literature. It is shown that the adjusted-point fixed-sectional method can provide comparable accuracy to the moving-sectional model in a simulation of soot formation and growth. It is also shown that the measured surface growth rate could be much higher than the value used in this study. Soot mass concentrations and size distributions for particles larger than 10 nm were well predicted with a surface reaction enhancement. The primary particle size was underpredicted by only about 30% compared with the measurements, without any model adjustments. As the new model can predict both the particle size distribution and structure, and is suitable for application in complex flows, its application to diverse soot formation conditions will enhance our knowledge on the evolution of soot structures.     

In-flame soot particle structure on the up- and down-swirl side of a wall-interacting jet in a small-bore diesel engine

This study shows how the structure of soot particles within the flame changes due to the relative direction of the swirl flow in a small-bore diesel engine in which significant flame–wall interactions cause about half of the flame travelling against the swirl flow while the other half penetrating in the same direction. The thermophoresis-based particle sampling method was used to collect soot from three different in-flame locations including the flame–wall impingement point near the jet axis and the two 60° off-axis locations on the up-swirl and down-swirl side of the wall-interacting jet. The sampled soot particle images were obtained using transmission electron microscopes and the image post-processing was conducted for statistical analysis of size distribution of soot primary particles and aggregates, fractal dimension, and sub-nanoscale parameters such as the carbon layer fringe length, tortuosity, and spacing. The results show that the jet-wall impingement region is dominated by many small immature particles with amorphous internal structure, which is very different to large, fractal-like soot aggregates sampled from 60° downstream location on the down-swirl side. This structure variation suggests that the small immature particles underwent surface growth, coagulation and aggregation as they travelled along the piston-bowl wall. During this soot growth, the particle internal structure exhibits the transformation from amorphous carbon segments to a typical core–shell structure. Compared to those on the down-swirl side, the soot particles sampled on the up-swirl side show much lower number counts and more compact aggregates composed of highly concentrated primary particles. This soot aggregate structure, together with much narrower carbon layer gap, indicates higher level of soot oxidation on the up-swirl side of the jet.     

Modeling and measurements of size distributions in premixed ethylene and benzene flames

This study demonstrates the major differences in the evolution of the particle size distributions (PSDs), both measured and modeled, of soot in premixed benzene and ethylene flat flames. In the experiments, soot concentration and PSDs were measured by using a scanning mobility particle sizer (SMPS, over the size range of 3-80 nm). The model employed calculations of gas phase species coupled with a discrete sectional approach for the gas-to-particle conversion. The model includes reaction pathways leading to the formation of nano-sized particles and their coagulation to larger soot particles. The particle size distribution, both experimental and modeled, evolved from a single particle mode (the nucleation mode) to a bimodal size distribution. An important distinction between the results for the ethylene and benzene flames is the behavior of the nucleation mode which persists at all heights above the burner (HAB) for ethylene whereas it was greatly suppressed at greater HAB for the benzene flames. The explanation for the decreased nucleation mode at higher elevations in the benzene flame is that the aromatics are consumed in the oxidation zone of the flame. Fair predictions of particle-phase concentrations and particle sizes in the two flames were obtained with no adjustments to the kinetic scheme. In agreement with experimental data, the model predicts a higher formation of particulate in the benzene flame as compared with the ethylene flame.     

Understanding soot particle size evolution in laminar ethylene/air diffusion flames using novel soot coalescence models

Two coalescence models based on different merging mechanisms are introduced. The effects of the soot coalescence process on soot particle diameter predictions are studied using a detailed sectional aerosol dynamic model. The models are applied to a laminar ethylene/air diffusion flame, and comparisons are made with experimental data to validate the models. The implementation of coalescence models significantly improves the agreement of prediction of particle diameters with the experimental data. Sensitivity of the soot prediction to the coalescence parameters is analysed. Finally, an update to the coalescence model based on experimental observations of soot particles in the flame oxidation regions has been introduced to improve its predicting capabilities.     

Constrained-temperature solutions of coflow laminar diffusion flames

We investigate the use of an experimental 2-D temperature profile to constrain detailed numerical solutions of a sooting coflow laminar diffusion flame. Experimentally, four optical diagnostic techniques are used to measure the two-dimensional temperature field in an ethylene-air coflow flame. This experimental temperature field is then used to impose the temperature in the solution process, thus obviating the need to solve the energy equation and, in particular, to incorporate costly models of radiative losses in the flame. Results are presented for a 40% ethylene-air flame on the Yale Coflow Burner. In the unconstrained solution of the complete set of governing equations, the location of maximum temperature is found along the flame wings, whereas the experimental temperature field has its maximum along the centerline. Similarly, the location of peak soot volume fraction migrates from along the flame wings in the unconstrained calculation, where soot surface growth processes dominate, to the centerline in the constrained case, where soot inception is the dominant condensed-phase formation mechanism. The distribution of soot in the constrained solution is much more consistent with experimental observations, and this fact illustrates how the validation of a soot sub-model may be complicated by the necessity of modeling distributed heat losses in the flame.     

Experimental and numerical study of the evolution of soot primary particles in a diffusion flame

The evolution of primary soot particles is studied experimentally and numerically along the centreline of a co-flow laminar diffusion flame. Soot samples from a flame fueled with C₂H₄ are taken thermophoretically at different heights above the burner (HAB), their size and nano-structure are analysed through TEM. The experimental results suggest that after inception, the nascent soot particles coagulate and coalesce to form larger primary particles (?～?5 to 15 nm). As these primary particles travel along the centreline, they grow mainly due coagulation and condensation and a layer of amorphous hydrocarbons (revealed by HRTEM) forms on their surface. This amorphous layer appears to promote the aggregation of primary particles to form fractal structures. Fast carbonisation of the amorphous layer leads to a graphitic-like shell around the particles. Further graphitization compacts the primary particles, resulting in a decrease of their size. Towards the flame tip the primary particles decrease in size due to rapid oxidation. A detailed population balance model is used to investigate the mechanisms that are important for prediction of primary particle size distributions. Suggestions are made regarding future model development efforts. Simulation results indicate that the primary particle size distributions are very sensitive to the parameterization of the coalescence and particle rounding processes. In contrast, the average primary particle size is less sensitive to these parameters. This demonstrates that achieving good predictions for the average primary particle size does not necessarily mean that the distribution has been accurately predicted.     

Investigation of soot sensitivity to strain rate in ethylene counterflow soot formation oxidation flames

Soot sensitivity to strain rate is mainly responsible for soot formation intermittence in practical combustion devices. This work provides a fundamental study on soot formation in Soot Formation Oxidation (SFO) counterflow flames at varying strain rates. While the problem has been extensively studied in Soot Formation (SF) configurations, where the dominant process is nucleation, investigations remain scarce in the corresponding SFO cases. In the latter, the high temperatures and strong oxidative environments make the surface reactions prevail over nucleation. The work provides a new dataset for ethylene SFO flames in a wide range of strain rates and sheds light on the main processes concurring in determining soot strain rate sensitivity in such conditions. In particular, the peak of soot volume fraction (SVF) is primarily controlled by surface growth and oxidation. The latter becomes progressively more dominant on the side of the SVF distribution toward the oxidizer nozzle, where the presence of oxidizing agents is significant. The soot mechanism adopted predicts a SVF distribution and sensitivity to strain rate in agreement with experimental data. The latter is found similar to corresponding SF cases, although soot loads in the two configurations differ by almost an order magnitude, and the SVF sensitivity is known to be more accentuated for lower soot loads. A deeper investigation revealed that the nucleation process through dimerizations primarily controls the SVF sensitivity, providing the onset of soot necessary for further growth. Then, the latter tends to reduce SVF sensitivity depending on its impact. PAH sensitivities mostly agree with theoretical observation even though further validations on the kinetic mechanism are needed to improve its predictions in lean conditions. The simplistic yet effective model based on the hybrid method of moments and the employment of a reduced kinetic mechanism makes the approach amenable for turbulent computational fluid dynamic (CFD) simulations.     

A computational framework for predicting laminar reactive flows with soot formation

Numerical modeling is an attractive option for cost-effective development of new high-efficiency, soot-free combustion devices. However, the inherent complexities of hydrocarbon combustion require that combustion models rely heavily on engineering approximations to remain computationally tractable. More efficient numerical algorithms for reacting flows are needed so that more realistic physics models can be used to provide quantitative soot predictions. A new, highly-scalable combustion modeling tool has been developed specifically for use on large multiprocessor computer architectures. The tool is capable of capturing complex processes such as detailed chemistry, molecular transport, radiation, and soot formation/destruction in laminar diffusion flames. The proposed algorithm represents the current state of the art in combustion modeling, making use of a second-order accurate finite-volume scheme and a parallel adaptive mesh refinement (AMR) algorithm on body-fitted, multiblock meshes. Radiation is modeled using the discrete ordinates method (DOM) to solve the radiative transfer equation and the statistical narrow-band correlated-k (SNBCK) method to quantify gas band absorption. At present, a semi-empirical model is used to predict the nucleation, growth, and oxidation of soot particles. The framework is applied to two laminar coflow diffusion flames which were previously studied numerically and experimentally. Both a weakly-sooting methane–air flame and a heavily-sooting ethylene–air flame are considered for validation purposes. Numerical predictions for these flames are verified with published experimental results and the parallel performance of the algorithm analyzed. The effects of grid resolution and gas-phase reaction mechanism on the overall flame solutions were also assessed. Reasonable agreement with experimental measurements was obtained for both flames for predictions of flame height, temperature and soot volume fraction. Overall, the algorithm displayed excellent strong scaling performance by achieving a parallel efficiency of 70% on 384 processors. The proposed algorithm proved to be a robust, highly-scalable solution method for sooting laminar flames.     

热电偶沉积法测量乙烯非预混火焰的试验研究

烟黑容积份额的测量是研究烟黑生成的反应机理的额的薪方法.本文详细描述了采用热电偶沉积法测量烟黑容积份额的理论基础和数据处理过程,并将此方法应用于层流乙烯非预混火焰的测量中.测量结果表明,该火焰中烟黑容积份额的分布同火焰结构和火焰温度都有关.     

Distributed-memory parallel computation of a forced,time-dependent,sooting, ethylene/air coflow diffusion flame

Forced, time-varying laminar flames help bridge the gap between laminar and turbulent combustion as they reside in an ever-changing flow environment. A distributed-memory parallel computation of a time-dependent sooting ethylene/air coflow diffusion flame, in which a periodic fluctuation (20 Hz) is imposed on the fuel velocity for four different amplitudes of modulation, is presented. The chemical mechanism involves 66 species, and a soot sectional model is employed with 20 soot sections. The governing equations are discretised using finite differences and solved implicitly using a damped modified Newton's method. The solution proceeds in parallel using strip domain decomposition over 40 central processing units (CPUs) until full periodicity is attained. For forcing amplitudes of 30%, 50%, 70% and 90%, a complete cycle of numerical predictions of the time-resolved soot volume fraction is presented. The 50%, 70% and 90% forcing cases display stretching and pinching off of the sooting region into an isolated oval shape. In the 90% forcing case, a well-defined hollow shell-like structure of the soot volume fraction contours occurs, in which the interior of the isolated sooty region has significantly lower soot concentrations than the shell. Preliminary comparisons are made with experimental measurements of the soot volume fraction for the 50% forcing case. The experimental results are qualitatively consistent with the model predictions.     

Numerical simulations of soot aggregation in premixed laminar flames

In this paper we make use of a detailed particle model and stochastic numerical methods to simulate the particle size distributions of soot particles formed in laminar premixed flames. The model is able to capture the evolution of mass and surface area along with the full structural detail of the particles. The model is validated against previous models for consistency and then used to simulate flames with bimodal and unimodal soot particle distributions. The change in morphology between the particles from these two types of flames provides further evidence of the interplay among nucleation, coagulation, and surface rates. The results confirm the previously proposed role of the strength of the particle nucleation source in defining the instant of transition from coalescent to fractal growth of soot particles.     

The Formation of Carbon in Combustion and how to Quantify the Impact on Human Health

We summarize our current research on combustion aerosols. First, sampling devices for the analyses of flame gases are described. The flame gas samples are investigated by mass spectroscopy and by standard aerosol techniques. Time-of-flight mass spectroscopy is well suited to study formation and growth of soot precursor molecules. Fullerenes can also be seen in some mass spectra of flame gases. Presumably, the fullerenes are evaporated from small soot particles in the mass spectrometer by the ionizing laser. Size spectra of soot particles from the flame are presented. The flame is optionally seeded with palladium aerosol to demonstrate that the particle size distribution is not altered during the sampling procedure. It is found that soot particles are already present low in the flame where large molecules are absent.Photoemission is applied to study surface properties of soot particles from the flame. It is shown that the surface of the particles is covered with polycyclic aromatic hydrocarbons (PAH). The PAH can be removed by heating and the properties of the carbon core are revealed. One can thereby distinguish a soot growth from a soot burnout region in the flame. Time-resolved desorption experiments of perylene (a PAH) from model aerosol particles are presented. It is shown that they follow a first order rate law. The photoelectric PAH sensor is introduced as a personal air quality monitor. The danger from inhaling combustion aerosol can be expressed in units of standard cigarettes.