Effect of Suspended Particles on the Laminar Burning Velocities and

The laminar burning velocities were measured for mixtures of CH4/O2/N2/carbon black particle with equivalence ratio values in the range of 0.8–1.2. ...
0 downloads 0 Views 3MB Size
Article pubs.acs.org/EF

Effect of Suspended Particles on the Laminar Burning Velocities and Markstein Lengths of CH4 Flames Jinou Song, Chonglin Song,* Gang Lv, Feng Bin, and Hao Li State Key Laboratory of Engines, Tianjin University, Tianjin 300072, People’s Republic of China ABSTRACT: The laminar burning velocities were measured for mixtures of CH4/O2/N2/carbon black particle with equivalence ratio values in the range of 0.8−1.2. The data were acquired at an initial pressure of 0.1 MPa and initial temperatures of 303, 353, and 403 K. High-speed schlieren visualization, used to monitor flame growth following ignition, provided a direct determination of the laminar flame velocity. The data were corrected for flame stretch, providing the unstretched laminar burning velocities and burned gas Markstein lengths. The values measured for CH4/O2/N2 flames were compared to those previously reported in the literature and computational prediction using the full mechanism. These comparisons revealed reasonable similarity in the data and demonstrated the reliability of the current experimental system and the accuracy of the full mechanism for CH4/O2/N2 flames. A decline in the Markstein lengths and burning velocities and an enhancement of the thermal diffusive instability upon the addition of carbon black particles were shown by the data. An early onset of cellularity preceded by the formation of toroidal cells for CH4/O2/N2/carbon black particle flames indicated particle mediated alteration of the cellular flame structure. The key reactions involved in the observed changes in the laminar burning velocities were identified on the basis of the sensitivity analysis. O2/N2/carbon black particle premixed flames were measured over a wide range of equivalence ratios, and the effect of carbon black particles on the Markstein lengths was investigated. This work, although performed at the atmospheric condition, provides a base for the further investigations at engine-like conditions.

1. INTRODUCTION The laminar burning velocity is a defining characteristic of the reactivity of combustible mixtures. The flames of fuel/air mixtures1−7 and fuel blend/air mixtures8−12 in this regard have been studied extensively. In a recent study, stoichiometric mixtures of hydrogen and oxygen with a steam content of up to 85 mol % have been tested in a spherical explosion chamber.13 Results indicated that steam dilution had a suppressing effect on flame propagation. Kian Eisazadeh-Far et al.14 investigated the effects of extra diluent gases on the laminar burning velocity and flame structure and developed a correlation between the laminar burning velocity and temperature, pressure, fuel air equivalence ratio (ER), and extra diluent gas. In contrast to the extensive number of studies outlined above, studies focused on fuel/air/particle mixtures are relatively scarce. Fuel/air/particle mixtures have recently received particular interest from the combustion community, because of an increasing demand for the post-injectionmediated enhancement of soot oxidation from the diesel engine industry.15−18 Post-injection of a small amount of fuel, which burns in the laminar premixed flames, produces little soot and provides higher temperatures late in the combustion cycle when soot oxidation is approaching a low level. This ultimately promotes oxidation by increasing the temperature and improving mixing.19,20 The efficiency of a post-injection strategy has been demonstrated in soot reduction without adverse impact upon NOx emission and fuel consumption.21,22 The effect of soot particles on the post-flame properties, however, remains unstudied. Among the fundamental flame properties required to better understand the combustion of fuel/air/particle mixtures, the flame velocity is probably one of the most important given that it provides detailed information about the combustion processes of post-injection fuel. In this paper, commercially available carbon black, dispersed in CH4/O2/N2 mixtures, was used to simulate particles in the laminar premixed flames. Laminar burning velocities of CH4/ © 2012 American Chemical Society

2. EXPERIMENTAL SECTION 2.1. General Configuration and Experimental Procedure. The experimental apparatus and diagnostic arrangement used in the current study are similar to those reported by Tang et al.23 As shown in Figure 1, the experimental system consists of a combustion vessel, heating section, ignition system, data acquisition section, and highspeed schlieren photography system. The constant volume combustion vessel is cylindrical with an inner diameter and length of 180 and 210 mm, respectively, allowing flame radii of up to 25 mm to be used in the data analysis without any significant effect of cylindrical confinement11 and deviation from constant pressure behavior.23 Two sides of the vessel are mounted with the quartz windows, allowing for observation of the combustion process inside the vessel. The flame propagation sequence was imaged with a schlieren photography system and recorded with a high-speed charge-coupled device (CCD) camera (HG-100 K, Redlake) operating at 10 000 pictures per second. A Kistler pressure transducer (model 4075A100) was used to record the combustion pressure with a resolution of 0.01 kPa. A high-speed data acquisition system (DL750, Yokogawa) was used to record the pressure−time data. Mixtures were made using the partial pressure method with a 0− 1000 Torr pressure transducer, and vacuum pressures were measured using a Varian 0−2000 mTorr transducer. Methane, oxygen, and nitrogen were introduced into the vessel sequentially according to their corresponding partial pressures for the specified equivalence ratio (the volumetric ratio of N2/O2 was 3.761 for each experimental run). The gas purities used in this study were grade 3.7 (99.97%) for CH4 Received: June 9, 2012 Revised: October 18, 2012 Published: October 24, 2012 6621

dx.doi.org/10.1021/ef3009822 | Energy Fuels 2012, 26, 6621−6626

Energy & Fuels

Article

Figure 1. Experimental setup.

Table 1. Specifications of Carbon Black diameter (nm)

surface area (m2/g)

density (g/cm3)

volatile content (%)

ash content (%)

heat loss (%)

325-mesh sieve residue (ppm)

11

143

2.0

1.5

0.5 maximum

0.5 maximum

50

0.8. The corresponding averaged number concentrations were 1.92 × 107, 2.00 × 107, and 1.95 × 107 cm−3 for the experimental runs conducted at 303, 353, and 403 K, respectively. The aerosol size distributions and number concentrations were analyzed by a scanning mobility particle sizer (SMPS) (model 3090, TSI, Inc.) via a bypass line. Mean relative standard deviation values were 2.3 and 1.0% for the number concentrations and particle diameters, respectively. Following preparation of the combustible mixture inside the vessel, a settlement period was required to obtain the quiescent mixture (generally 5−10 min after the gas motion inside the vessel was observed to settle with the aid of schlieren photography). An external electric heating circuit was mounted on the outside surface of the combustion vessel to facilitate the uniform heating of the mixture inside the vessel to a desired initial temperature. The gas temperature was measured using a NiCr−Ni thermocouple of 0.5 mm thickness, installed 45 mm from the inside vessel wall. The temperature deviation range of the temperature control system was ±0.5 K. The spatial non-uniformity of the temperature was less than ±1 K. The mixture was ignited by two centrally located electrodes with a standard capacitive discharge ignition system. The current and voltage

and UHP (99.999%) for both O2 and N2. For all experimental runs, the same mass of commercially available carbon black (Table 1) was introduced into the vessel with nitrogen through an aerosol generator (Generitor-JJL01) to form aerosols (aggregates) suspended in the gas phase, obtaining similar aerosol size distributions and number concentrations. Figure 2 shows the aerosol size distributions at ER =

Figure 2. Aggregate (aerosol) size distributions.

Figure 3. Schlieren images showing the cellularity transition (red circles illustrating the flame front edge detection). 6622

dx.doi.org/10.1021/ef3009822 | Energy Fuels 2012, 26, 6621−6626

Energy & Fuels

Article

across the electrodes were continuously adjusted to minimize the ignition energy of each run. In this study, the ignition energy was less than 45 mJ and the flame velocities became independent of the ignition energy when the flame radius was greater than 6 mm.9 Upon completion of the combustion, the combustion vessel was purged and flushed 3 times with dry N2 to avoid the influence of any residual gases on the next experimental run. Combustion experiments were conducted with CH4/O2 /N 2 mixtures (pure CH4 mixtures) and CH4/O2/N2/carbon black particle mixtures (particle CH4 mixtures) at an initial pressure of 0.1 MPa and initial temperatures of 303, 353, and 403 K. Equivalence ratios of 0.8, 0.9, 1.0, 1.1, and 1.2 were used. During the current investigation, each experimental run was repeated at least 3 times. 2.2. Processing Steps. The flame propagation observed on the recorded movies demonstrates a spherical evolution for all flame radii (Figure 3), revealing that buoyancy effects and disturbances because of the electrodes, spark gap, or initial spark energy could be neglected. The radius at each time step was determined using a best-fit algorithm to fit a circle to each recorded image.2 Figure 3 shows a spherical flame front with the corresponding best fit superimposed on the images (red circles). During the time when measurements were taken, the increase in the pressure was less than 1%, and it was therefore assumed that the experimental data were collected at a constant pressure. The classical outwardly propagating flame approach assumes that a linear relationship exists between the stretched propagation flame speed Sn and the applied stretch rate K24−26

Sn = Ss − L bK

chemistry in terms of 325 elementary reactions of 53 species. A converged solution was provided by the use of 1000 grid points. The temperature and species slopes at the boundaries were close to 0, and the gradient and curvature controls were both 0.01. The particle size distribution pre-generated in particle CH4 flames for computing laminar burning velocities is shown in Figure 2. The corresponding averaged number concentration was 1.78 × 107 cm−3. To consider the possible oxidation of particles, a kinetic subset of carbon black oxidation was added. The mechanism of carbon black oxidation was described using the Nagle and Strickland-Constable semi-empirical model33 to account for the oxidation caused by O2 attack. Furthermore, oxidation caused by OH attack was accounted for using the OH oxidation model developed by Neoh et al.34 The number density and size distribution of the particles in the vessel change as particles are oxidized. This change is related to the burned/ unburned gas density ratio, thus affecting the flame parameters.

4. RESULTS AND DISCUSSION 4.1. Comparison. Figure 4 shows measured flame velocities plotted against stretch at ER values of 0.8, 1.0, and 1.2 and

(1)

where Ss is the unstretched propagation flame velocity and Lb is the Markstein length for burned gases. The stretched propagation flame velocity, Sn, is related to the flame front radius as follows: Sn =

dr dt

(2)

The flame stretch, K, is expressed for a spherically symmetric flame as follows: K=

2 dr r dt

(3)

Temporal evolutions of the flame radii can be used to provide the (K; Sn) pairs and perform appropriate extrapolation procedures to yield both the unstretched flame propagation velocity (Ss) and the Markstein length of burned gases (Lb). In the present investigation, the nonlinear extrapolation methodology described by Kelley and Law27 was used to extract both Ss and L b.

⎛ Sn ⎞2 ⎛ Sn ⎞2 2L K ⎜ ⎟ ln⎜ ⎟ = − b Ss ⎝ Ss ⎠ ⎝ Ss ⎠

Figure 4. Flame propagation velocities (0.1 MPa): open symbols for Gu et al.35 and solid symbols for the present work.

initial temperatures of 303, 353, and 403 K. Experimental data reported by Gu et al.35 using closed vessels with ER values of 0.8, 1.0, and 1.2 and at initial temperatures of 300, 360, and 400 K are also included in the figure. The maximum flame radius in this work is smaller than that of Gu et al. Correspondingly, the range of the flame stretch is narrower. The smoothing for a narrow range of flame stretch results in the flat data relative to those by Gu et al. Methane, oxygen, and nitrogen introduced into the vessel hardly mix well without a stir, probably causing an error for the measurement results. On the whole, the measurements of the present study are similar to those reported by Gu et al.,35 especially for ER values of 0.8 and 1.0 and an initial temperature of 303 K. Figure 5 shows the burning velocities at 0.1 MPa as a function of ER. The experimental results of the present study are compared to the measurements reported by Gu et al.35 and Bosschaart et al.,1 together with the modeled data generated using GRI 3.0. The predictions using GRI 3.0 appear to be in better agreement with the present experimental results. The standard deviations based on three measurements ranged from 2.5 to 11% for the flame velocities

(4)

Fitting algorithms developed by Halter et al.28 were used in the present investigation. In the case of a flame expansion at a constant pressure, the laminar burning velocity, ul, was deduced from the unstretched propagation flame velocity using the burned/unburned gas density ratio ρ ul = b Ss ρu (5) where ρb and ρu are the burned and unburned gas densities, respectively.

3. NUMERICAL APPROACH The Sandia flame code PREMIX29 together with the CHEMKIN-PRO30 and the transport property subroutines31 were used to compute the unstretched laminar burning velocities of a freely propagating, one-dimensional, adiabatic premixed flame. The chemical reaction mechanism of GRIMech 3.032 was used to describe the methane oxidation 6623

dx.doi.org/10.1021/ef3009822 | Energy Fuels 2012, 26, 6621−6626

Energy & Fuels

Article

and decrease with the temperature. The addition of particles to pure CH4 flames results in a significant decrease in the Markstein length. Physically, this is manifested as a variation in the influence of the stretch rate on the flame propagation and burning velocities. Because of the high heat capacity of particles, the addition of particles will decrease the thermal diffusivity of the mixtures, leading to a low Le. It has been reported11 that the positive dependence of Le on ER suggests a positive dependence of Markstein lengths on ER and vice versa. By analogy, the addition of particles leading to a low Le should lead to a low Markstein length. This deduction is supported by the measurements of the Markstein length in Figure 6. Therefore, the addition of particles to the mixtures lowers the Le and enhances the diffusional thermal instability. 4.3. Laminar Burning Velocity. As shown in Figure 7, the laminar burning velocity is expected to decrease upon the addition of particles for three initial temperatures because of a decrease in the thermal diffusivity of unburned gas. The experimental data are compared to the results of the chemical kinetics model described above. The model appears to simulate the chemistry well and is in near-perfect agreement with the experimental data collected for pure CH4 flames. However, the burning velocities under lean and rich conditions for particle CH4 flames are overestimated, even though similar particle numbers and size distributions are computationally adopted. This observation indicates that there are chemistry interactions between CH4 oxidation and carbon black. The overestimation of burning velocities for the model under lean and rich conditions must be related to the chemistry interactions that the model fails to capture. By analogy with the adsorption of volatile organic compounds on porous carbon, it is supposed that there is adsorption of some radicals on the porous aerosols (aggregates), which induces the chemistry interactions. Figure 8 shows the sensitivity analyses performed for pure CH4 and particle CH4 mixtures with an ER value of 1.0 at an

Figure 5. Comparison of laminar burning velocities.

and from 0.5 to 1% for ER, which are plotted in the form of error bars in Figure 5. 4.2. Markstein Length. Figure 6 shows Markstein lengths as a function of ER at an initial pressure of 0.1 MPa. The

Figure 6. Markstein lengths.

standard deviations based on three measurements ranged from 5 to 14% for the Markstein lengths. In Figure 6 (also in Figure 7), a characteristic standard deviation is plotted in the form of error bars and the error bars are omitted for all other data for clarity. Figure 6 shows the trends of Markstein length for particle CH4 flames similar to those of pure CH4 flames; that is to say, the associated Markstein lengths increase with the ER

Figure 8. Flame velocity sensitivity analysis.

initial pressure of 0.1 MPa and an initial temperature of 303 K. Reactions with positive sensitivity coefficients are promoting reactivity, while those with negative coefficients are inhibiting reactivity. Some of the reactions exhibiting high sensitivity to particle CH4 mixtures (e.g., H + O2 ⇄ OH + O, HCO + H ⇄ H2 + CO, and HO2 + CH3 ⇄ OH + CH3O) are important for pure CH4 mixtures and other hydrocarbon fuel oxidations.12 Modifying these reactions to fit the experimental data of particle CH4 mixtures would significantly affect the oxidation mechanisms of pure CH4 mixtures. It is also found that some of

Figure 7. Laminar burning velocities. 6624

dx.doi.org/10.1021/ef3009822 | Energy Fuels 2012, 26, 6621−6626

Energy & Fuels

Article

Figure 9. Schlieren images showing the onset of toroidal cells (ER = 0.8).

values (0.8−1.2) at temperatures of 303, 353, and 403 K. A pressure of 0.1 MPa was employed throughout. It was concluded that addition of particles to the mixture enhanced the thermal diffusive instability and decreased the values of the Markstein lengths. An early onset of cellularity was preceded by the formation of toroidal cells for the particle CH4 mixtures, indicating that particles altered the cellular flame structure. The laminar burning velocities for pure CH4 mixtures were greater than those observed in particle CH4 mixtures, and the reactions showing special sensitivity to particle CH4 flames were identified on the basis of sensitivity analysis.

the reactions show large sensitivity to particle CH4 mixtures but little sensitivity to pure CH4 mixtures [e.g., CH2(S) + O2 ⇄ H + OH + CO and 2CH3 ⇄ H + C2H5]. Adjusting the rate constants of these reactions to fit the experimental data of particle CH4 flames would have less influence on the oxidation mechanisms of pure CH4 mixtures. The discrepancies between the measured and calculated burning velocities at the ER of 1.0 are less than those at lean and rich conditions (Figure 7), indicating that the burning velocities at the ER of 1.0 are less affected by the chemistry interactions between CH4 oxidation and carbon black. Thus, the sensitivity analysis made at the ER of 1.0 largely reflects the impact of the reduced temperature because of particle addition. 4.4. Flame Morphology. For all mixtures, the schlieren images show the flame surface to be smooth during the initial flame growth. The disturbances, initiated in the early flame by the spark or spark electrodes, grow in proportion to the flame radius and continue to do so until late in the combustion event, where disturbances become evident. Although the burning velocities for pure CH4 mixtures are greater than those for particle CH4 mixtures (Figure 7), a fast flame propagation velocity and an early transition to cellular flames were observed for particle CH4 mixtures because of the decrease in Marstein length (Figure 6), as shown in Figure 3. It is interesting to comment on the structure of the cells. As shown in Figure 3, pure CH4 flames, appearing initially smooth, later develop cracks and cross-cracks until a cellular structure appears. In contrast, the onset of cellularity within particle CH4 flames is preceded by the formation of toroidal cells. Similar toroids in fuel-rich flames were observed by Johnston and Farrell36 with a wide range of hydrocarbon fuels, including paraffins, olefins, and aromatic compounds, but no further explanation was given. Interpretation can now be offered by an extension of the current observations. Thus, soot particles may be formed in the fuel-rich flames, leading to the formation of the toroidal cells reported by Johnston and Farrell.36 The toroidal cells reflect the particles promoting the flame front perturbations and are supposed to represent the perturbationaffected zones. As shown in Figure 9, toroidal cells vary with test conditions. At 303 K, they appear similar to those observed by Johnston and Farrell,36 but they are concentric at 353 K and overlapped at 403 K. The current evidence indicates that the particle perturbation varies with test conditions. Because of the high density and heat capacity of particles compared to those of gas phase, there may exist a velocity slip and a temperature lag between the particles and the gas phase during flame propagation. Whether or not this slip could cause flame front perturbations, leading to an increase in the flame propagation velocity and an early cellularity transition for particle CH4 flames, could not be deduced from the present study.



AUTHOR INFORMATION

Corresponding Author

*Telephone: +86-22-27406840, ext. 8020. Fax: +86-2227403750. E-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was supported by the National Natural Science Foundation of China (51276126) and the National Key Basic Research and Development Program (2013CB228502). The authors gratefully acknowledge the assistance of Professor Zuohua Huang and Dr. Erjiang Hu at Xi’an Jiaotong University for performing some of the flame velocity experiments.



REFERENCES

(1) Bosschaart, K. J.; de Goey, L. P. H.; Burgers, J. M. Combust. Flame 2004, 136, 261−269. (2) Tahtouh, T.; Halter, F.; Mounaim-Rousselle, C. Combust. Flame 2009, 156, 1735−1743. (3) Bradley, D.; Lawes, M.; Mansour, M. S. Combust. Flame 2009, 156, 1462−1470. (4) Liao, S. Y.; Jiang, D. M.; Huang, Z. H.; Zeng, K. Fuel 2006, 85, 1346−1353. (5) Huang, Z. H.; Wang, Q.; Yu, J. R. Fuel 2007, 86, 2360−2366. (6) Chen, Z.; Wei, D.; Huang, Z. H.; Miao, H.; Wang, X.; Jiang, D. Energy Fuels 2009, 23, 735−739. (7) Wang, Y. L.; Holley, A. T.; Ji, C.; Egolfopoulos, F. N.; Tsotsis, T. T.; Curran, H. J. Proc. Combust. Inst. 2009, 32, 1035−1042. (8) Tang, C. L.; Huang, Z. H.; Law, C. K. Proc. Combust. Inst. 2011, 33, 921−928. (9) Huang, Z. H.; Zhang, Y.; Zeng, K.; Liu, B.; Wang, Q.; Jiang, D. M. Combust. Flame 2006, 146, 302−311. (10) Natarajan, J.; Kochar, Y.; Lieuwen, T.; Seitzman, J. Proc. Combust. Inst. 2009, 32, 1261−1268. (11) Burke, M. P.; Chen, Z.; Ju, Y.; Dryer, F. L. Combust. Flame 2009, 156, 771−779. (12) Lowry, W. B.; Serinyel, Z.; Krejci, M. C.; Curran, H. J.; Petersen, E. L.; Bourque, G. Proc. Combust. Inst. 2011, 33, 929−937. (13) Kuznetsov, M.; Redlinger, R.; Breitung, W.; Grune, J.; Friedrich, A.; Ichikawa, N. Proc. Combust. Inst. 2011, 33, 895−903. (14) Eisazadeh-Far, K.; Moghaddas, A.; Al-Mulki, J.; Metghalchi, H. Proc. Combust. Inst. 2011, 33, 1021−1027.

5. CONCLUSION Laminar burning velocities and Markstein lengths for particle CH4 and pure CH4 mixtures were measured over a range of ER 6625

dx.doi.org/10.1021/ef3009822 | Energy Fuels 2012, 26, 6621−6626

Energy & Fuels

Article

(15) Tree, D. R.; Svensson, K. I. Prog. Energ. Combust. Sci. 2007, 33, 272−309. (16) Yun, H.; Reitz, R. D. J. Eng. Gas Turbines Power 2007, 129, 279− 286. (17) Arrègle, J.; Pastor, J. V.; López, J. J.; García, A. Combust. Flame 2008, 154, 448−461. (18) Bobba, M.; Musculus, M.; Neel, W. SAE Int. J. Engines 2010, 3, 496−516. (19) Park, C.; Kook, S.; Bae, C. SAE [Tech. Pap.] 2004, DOI: 10.4271/2004-01-0127. (20) Mallamo, F.; Badami, M.; Millo, F. SAE [Tech. Pap.] 2002, DOI: 10.4271/2002-01-2672. (21) Desantes, J. M.; Arrègle, J.; López, J. J.; García, A. SAE [Tech. Pap.] 2007, DOI: 10.4271/2007-01-0915. (22) Hountalas, D. T.; Lamaris, V. T.; Pariotis, E. G.; Ofner, H. SAE [Tech. Pap.] 2008, DOI: 10.4271/2008-01-0641. (23) Tang, C. L.; He, J. J.; Huang, Z. H.; Jin, C.; Wang, J. H.; Wang, X. B.; Miao, H. Y. Int. J. Hydrogen Energy 2008, 33, 7274−7285. (24) Clavin, P. Prog. Energy Combust. Sci. 1985, 11, 1−59. (25) Dowdy, D. R.; Smith, D. B.; Taylor, S. C.; Williams, A. Proc. Combust. Inst. 1990, 23, 325−332. (26) Brown, J. M.; McLean, I. C.; Smith, D. B.; Taylor, S. C. Proc. Combust. Inst. 1996, 26, 875−881. (27) Kelley, A. P.; Law, C. K. Combust. Flame 2009, 156, 1844−1851. (28) Halter, F.; Tahtouh, T.; Mounaïm-Rousselle, C. Combust. Flame 2010, 157, 1825−1832. (29) Kee, R. J.; Grcar, J. F.; Smooke, M. D.; Miller, J. A. A Fortran Program for Modeling Steady Laminar One-Dimensional Premixed Flames; Sandia National Laboratories: Livermore, CA, 1985; Report SAND85-8240. (30) Reaction Design. CHEMKIN-PRO; Reaction Design: San Diego, CA, 2008. (31) Kee, R. J.; Dixon-Lewis, G.; Warnatz, J.; Coltrin, R. E.; Miller, J. A. A Fortran Computer Package for the Evaluation of Gas-Phase, Multicomponent Transport Properties; Sandia National Laboratories: Livermore, CA, 1986; Report SAND86-8246. (32) http://www.me.berkeley.edu/gri-mech/. (33) Nagle, J.; Strickland-Constable, R. F. Oxidation of carbon between 1000−2000 °C. Proceedings of the 5th Conference on Carbon; Pergamon Press: New York, 1962; Vol. 1, pp 154−164. (34) Neoh, K. G.; Howard, J. B.; Sarofim, A. F. Proc. Combust. Inst. 1974, 20, 951−957. (35) Gu, X. J.; Haq, M. Z.; Lawes, M.; Woolley, R. Combust. Flame 2000, 121, 41−58. (36) Johnston, R. J.; Farrell, J. T. Proc. Combust. Inst. 2005, 30, 217− 224.

6626

dx.doi.org/10.1021/ef3009822 | Energy Fuels 2012, 26, 6621−6626