air diffusion flames

Combustion and Flame 158 (2011) 547–563

Contents lists available at ScienceDirect

Combustion and Flame j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / c o m b u s t fl a m e

An experimental and numerical study of the effects of dimethyl ether addition to fuel on polycyclic aromatic hydrocarbon and soot formation in laminar coflow ethylene/air diffusion flames F. Liu a,⇑, X. He b, X. Ma b, Q. Zhang c, M.J. Thomson c, H. Guo a, G.J. Smallwood a, S. Shuai b, J. Wang b a b c

Institute for Chemical Process and Environmental Technology, National Research Council, Building M-9, 1200 Montreal Road, Ontario, Canada State Key Laboratory of Automotive Safety and Energy, Tsinghua University, Beijing, China Department of Mechanical and Industrial Engineering, University of Toronto, 5 King’s College Road, Toronto, Ontario, Canada

a r t i c l e

i n f o

Article history: Received 30 June 2010 Received in revised form 13 September 2010 Accepted 6 October 2010 Available online 28 October 2010 Keywords: Laminar diffusion flame Soot formation PAHs Synergistic effect Numerical modeling

a b s t r a c t The effects of dimethyl ether addition to fuel on the formation of polycyclic aromatic hydrocarbons and soot were investigated experimentally and numerically in a laminar coflow ethylene diffusion flame at atmospheric pressure. The relative concentrations of polycyclic aromatic hydrocarbon species and the relative soot volume fractions were measured using planar laser-induced fluorescence and two-dimensional laser-induced incandescence techniques, respectively. Experiments were conducted over the entire range of dimethyl ether addition from pure ethylene to pure dimethyl ether in the fuel stream. The total carbon mass flow rate was maintained constant when the fraction of DME in the fuel stream was varied. Numerical calculations of nine diffusion flames of different dimethyl ether fractions in the fuel stream were performed using a detailed reaction mechanism consisting of 151 species and 785 reactions and a sectional soot model including soot radiation, inception of nascent soot particle due to collision of two pyrene molecules, heterogeneous surface growth and oxidation following the hydrogen abstraction acetylene addition mechanism, soot particle coagulation, and PAH surface condensation. The addition of a relatively small amount of dimethyl ether to ethylene was found experimentally to increase the concentrations of both polycyclic aromatic hydrocarbons and soot. The synergistic effect on polycyclic aromatic hydrocarbons persists over a wider range of dimethyl ether addition. The numerical results reproduce the synergistic effects of dimethyl ether addition to ethylene on both polycyclic aromatic hydrocarbons and soot, though the magnitude of soot volume fraction overshoot and the range of dimethyl ether addition associated with the synergistic effect of soot are less than those observed in the experiment. The synergistic effects of dimethyl ether addition to ethylene on many hydrocarbon species, including polycyclic aromatic ones, and soot can be fundamentally traced to the enhanced methyl concentration with the addition of dimethyl ether to ethylene. Contrary to previous findings, the pathways responsible for the synergistic effects of benzene, polycyclic aromatic hydrocarbons, and soot in the ethylene/dimethyl ether system are found to be primarily due to the cyclization of l-C6H6 and n-C6H7 and to a much lesser degree due to the interaction between C2 and C4 species for benzene formation, rather than the propargyl self-combination reaction route, though it is indeed the most important reaction for the formation of benzene. Crown Copyright Ó 2010 Published by Elsevier Inc. on behalf of The Combustion Institute. All rights reserved.

1. Introduction Soot emitted from flames and various combustion devices have been identified to have an adverse health effect on humans [1] and also to be a major contributor to climate change [2]. Significant research attention has been paid in the last four decades to search for various strategies to control soot formation. One of such strategies is to blend the primary hydrocarbon fuel with an additive, which can be either chemically inert or active, to alter the sooting ⇑ Corresponding author. Fax: +1 613 957 7869. E-mail address: [email protected] (F. Liu).

characteristics of the primary fuel. To understand the effects of temperature and the C/H ratio on the tendency to soot in laminar diffusion flames, Schug et al. [3] systematically investigated the sooting heights of laminar coflow ethylene/air diffusion flames under the effect of various additives, of flames fueled by other hydrocarbons, and of flames fueled by acetylene–fuel mixtures. Work on the influence of additives on soot formation conducted up to 1980 was summarized in the review of Haynes and Wagner [4]. Besides the purpose of soot formation control, study of the effect of certain hydrocarbon additives, such as acetylene and benzene, on soot formation also serves as a powerful means to probe the reaction pathways leading to the incipient ring formation

0010-2180/$ - see front matter Crown Copyright Ó 2010 Published by Elsevier Inc. on behalf of The Combustion Institute. All rights reserved. doi:10.1016/j.combustflame.2010.10.005

548

F. Liu et al. / Combustion and Flame 158 (2011) 547–563

and nascent soot particles and/or to clarify the relative roles of different pathways to the formation of polycyclic aromatic hydrocarbon (PAH) and soot [5–7]. Soot formation is a very complex phenomenon. Although significant progress has been achieved in the last few decades towards the main chemical species and detailed reaction pathways responsible for the formation of PAHs, nascent soot particles, and their subsequent surface growth processes, certain key steps in the process from gaseous species to the formation of PAH and solid soot particles remain relatively poorly understood [8]. An excellent discussion of the gas-phase chemistry leading to soot formation has recently been provided by Miller et al. [9]. It is important to notice that the following major steps in the process of soot formation have been widely accepted [8]: (i) formation of molecular precursors of soot, (ii) inception of particles from heavy PAH molecules, (iii) mass growth of particles by addition of gas phase molecules, (iv) particle coagulation via collisions, (v) carbonization of particulate material, and (vi) oxidation. Correspondingly, continuous efforts have been made to develop generic soot models applicable to different flames, such as premixed, non-premixed, or fueled by different hydrocarbons, oxygenated fuels, or blended fuels. The drive towards gaining improved fundamental understanding of the effect of fuel molecular structure on soot formation and searching for effective additives for soot formation suppression in hydrocarbon combustion has generated considerable interest in recent years in the effects of adding oxygenated fuels to conventional hydrocarbon fuels on soot formation. Oxygenated fuels are in general very clean in terms of soot formation and their decomposition species can influence PAH and soot formation in a rather complex way. Two oxygenated fuels of particular interest are dimethyl ether (DME: CH3–O–CH3) and its isomer ethanol (EtOH: CH3–CH2–OH). These two fuels can be mass produced economically and have been proposed as possible soot reduction additives to conventional hydrocarbons and as alternative fuels (DME to diesel engines and EtOH to gasoline engines). A great deal of research has been conducted to investigate the effects of DME or EtOH addition to the combustion and emission characteristics of engines. The present study concerns the effects of DME addition to fuel on PAH and soot formation in laminar coflow ethylene/air diffusion flame. Ethylene was chosen as the fuel in this study for the following reasons. First, ethylene is an important intermediate with significant concentrations in the decomposition of larger hydrocarbons [10] and its detailed reaction mechanism is an essential sub-set for the development of combustion mechanisms of higher-order hydrocarbon fuels, such as propane, butane, heptanes, and jet-A. Secondly, ethylene is an interesting hydrocarbon fuel that has been shown to display the synergistic effect in terms of soot formation with the addition of a small amount of propane [11,12] or DME [13,14]. Thirdly, laminar coflow ethylene/air diffusion flames have been extensively investigated experimentally, e.g. [15–17], and there exists a quite comprehensive dataset of temperature and soot volume fraction distributions [15–17] that can be used to validate the numerical results for the base case (without DME addition) of this study. So far there have been only four studies published in the open literature to investigate the effects of DME addition to fuel on PAH and soot formation in laminar ethylene flames [10,13,14,18]. Wu et al. [10] conducted an experimental and computational study of the effects of DME addition on PAH and soot formation in rich premixed ethylene/air flat flames established in a McKenna burner. Experimentally, they observed a reduction in PAHs and soot concentrations when a small amount of DME (up to 10% oxygen by weight in the fuel) was added. They also conducted numerical calculations using the PREMIX flame code along the centerline of the flame with a specified temperature profile based on experimental measurements. The reaction mechanism used in their calculations consists

of 248 species and 1008 reactions. Soot was also calculated in a post-processing fashion using a model based on the studies of Wang and Frenklach [19] and Appel et al. [20]. Based on the numerical results they explained the decreased PAH and soot concentrations with the addition of DME mainly by the increased concentrations of O and OH radicals and reduced amount of carbon available for the formation of precursor species. McEnally and Pfefferle [14] conducted an experimental study of the effects of adding a relatively small amount of DME or EtOH to fuel (up to 10% as measured by the volumetric flowrate ratio of DME or EtOH to ethylene) on benzene and soot formation in laminar coflow ethylene/air nonpremixed flames. Their experimental results indicate that addition of both DME and EtOH increased the maximum benzene concentration and the soot volume fraction with the effect of DME being somewhat stronger, which is unexpected given the fact that both DME and EtOH produce very little soot. It is interesting to observe that McEnally and Pfefferle [14] (conducted in laminar diffusion flames) and Wu et al. [10] (conducted in laminar premixed flames) obtained opposite trends with regard to the effect of DME addition to ethylene on PAH and soot formation. The fundamental reason for the synergistic effects in ethylene non-premixed flames is that pyrolysis of ethylene produces virtually no odd-carbon species. However, oxidative pyrolysis of ethylene does produce odd-carbon species through reactions such as C2H4 + O = CH3 + HCO, where the stability of the carbonyl bond provides a driving force to break the C@C bond. This is demonstrated by the observation of a similar synergistic effect when air is added to the fuel in ethylene non-premixed flames [21,22]. Thus, it is not surprising that synergistic effects from DME addition are observed in non-premixed ethylene flames but not in premixed flames. McEnally and Pfefferle [14] explained the enhanced formation of PAH and soot by the addition of a small amount of DME or EtOH to ethylene in terms of the enhanced concentrations of methyl radical (CH3) from the decomposition of these oxygenates. The enhanced methyl radical concentrations then prompt the formation of propargyl radical (C3H3) through C1 + C2 addition reactions. The increased propargyl radical concentrations then lead to higher benzene concentrations through propargyl self-reaction, which has been identified to be the most important benzene formation reaction, e.g. [13,18]. The greater effect of DME addition on PAH and soot formation than EtOH can be attributed to the higher methyl radical concentrations from its decomposition [14,18]. Yoon et al. [13] experimentally investigated the effects of mixing DME with methane, ethane, propane, and ethylene on PAH and soot formation in laminar counterflow diffusion flames. The relative soot volume fractions and PAH concentrations were measured by planar laser-induced incandescence (LII) and laser-induced fluorescence (LIF) techniques, respectively. They also conducted numerical calculations of the flames using a reaction mechanism consisting of 115 species and 602 reactions steps. However, soot was not included in the numerical calculations. Their experimental results of both PAH concentration and soot volume fraction demonstrate that only the ethylene flames exhibit the synergistic effect, i.e., mixing a small amount of DME to ethylene causes the PAH concentration and soot volume fraction to increase beyond the pure ethylene case, but not the other three hydrocarbon flames. The findings of Yoon et al. with regard to the effects of DME addition to ethylene on PAH concentration and soot formation are in qualitative agreement with those of McEnally and Pfefferle [14]. It is important to notice that the synergistic effect for the PAH concentration persists over a much wider range of DME percentage in the fuel than that for the soot volume fraction (50% vs. 22% of DME in fuel). Yoon et al. [13] also explained the enhanced formation of PAH and soot with the addition of a relatively small amount of DME to ethylene by the enhanced methyl radical concentrations from the decomposition of DME, which then prompt propargyl radical formation and

549

F. Liu et al. / Combustion and Flame 158 (2011) 547–563

eventually lead to increased formation of PAH and soot, similar to the explanation proposed by McEnally and Pfefferle [14]. The most relevant study to the present one is that conducted by Bennett et al. [18], who reported a computational and experimental study of the effects of adding DME and EtOH to fuel (up to 10% as parameterized by the ratio of oxygenate volume flow rate to that of ethylene) on temperature and various hydrocarbon (up to benzene) concentrations in laminar coflow nitrogen-diluted ethylene/air non-premixed flames (69% N2 in fuel by volume). The experimental part of Bennett et al. [18] is that reported earlier [14] with detailed hydrocarbon mole fraction distributions along the flame centerline reported, which were not given in [14]. Computationally, they modeled the 2D axisymmetric laminar flames using three reactions mechanisms: (1) a 50-species, 257-reaction one for ethylene combustion, which included C1 to C6 hydrocarbons [23], (2) a 57-species, 311-reaction one for flames with DME addition, which was assembled from the base ethylene mechanism by adding a DME submechanism from [24], and (3) a 59species, 333-reaction mechanism for flames with EtOH addition, which was again based on the base ethylene mechanism and the EtOH submechanism from [24]. The numerical results of Bennett et al. [18] substantiate and provide numerical evidence to support the explanation of the enhanced PAH and soot formation by adding a relatively small amount of DME or EtOH to ethylene proposed by McEnally and Pfefferle based on their experimental study [14]. This explanation emphasizes the vital role of the increased methyl radical concentrations, which are very low in ethylene flames, from the decomposition of DME and EtOH. The increased methyl radical concentrations then lead to higher propargyl concentrations through the reactions of CH3 with C2 species, which are plenty in ethylene flames, and eventually higher benzene concentrations through the self-reaction of C3H3. It is noticed that the numerical calculations of Bennett et al. [18] did not take into account soot formation even though the maximum soot volume fractions are close to 2.5 ppm, which is believed to have some influence on the predicted flame temperature and species concentrations due to the influence of soot radiation on temperature and soot formation on species contributing to soot inception and surface growth. The objectives of this study are to further investigate the effects of DME addition to fuel on PAH and soot formation by conducting a combined experimental and numerical study in laminar coflow undiluted ethylene/air diffusion flames with various degrees of DME addition. The relative PAH concentrations and soot volume fractions were measured using the planar laser-induced fluorescence (PLIF) and 2D laser-induced incandescence (LII), respectively. Our numerical calculations took into account soot formation by employing a PAH based sectional soot model, which assumes that soot inception is due to PAH collision and soot surface growth and oxidation follows the hydrogen-abstraction C2H2 addition (HACA) mechanism. To our knowledge, no numerical studies of soot formation in laminar coflow hydrocarbon diffusion flames with DME addition have been reported in the open literature. The synergistic effect of PAH and soot formation by a small amount of DME addition to fuel in ethylene diffusion flames constitutes an interesting and challenging problem for a soot model to predict, since the chemical nature of such phenomenon requires the relative importance of various species involved in soot inception, surface growth, and oxidation processes to be correctly modeled. Our study also substantially extends the range of DME addition considered in previous coflow ethylene diffusion flame studies [14,18] by investigating flames from pure ethylene to pure DME. The motivation to investigate a wider range of DME addition is to establish the critical fractions of DME addition in coflow laminar ethylene/air diffusion flames where the PAH concentration and soot volume fraction peaks, since these critical fractions of DME addition serve as a useful quantity for the validation of

detailed reaction mechanism for ethylene/DME mixtures and soot formation model. Investigation of the entire range of DME addition also allows a better understanding of its effect on PAH and soot formation. 2. Experiment setup Axisymmetric coflow laminar diffusion flames fueled by ethylene/DME mixtures at atmospheric pressure were generated with a co-annular burner in which the fuel mixture was delivered at room temperature through a central fuel tube of 10.9 mm inner diameter (the fuel tube thickness is 0.94 mm) and air was supplied, also at room temperature, through the co-annular region between the fuel tube and the air tube of 88 mm inner diameter. This laminar diffusion flame burner has been described in detail in a previous publication [25]. Air supply to the burner was from a compressor and the air flow rate was maintained at 284 l/min, as metered with a rotameter. The fuels used in the experiments were research grade and supplied from cylinders. The flow rates of ethylene and DME were controlled using separate calibrated mass flow controllers. The volumetric flow rate of the fuel stream to the burner was kept at 194 ml/min. Under these flow conditions the visible flame height of the ethylene diffusion flame (without DME addition) was about 65 mm. When a certain amount of DME was added to the fuel stream, the same amount of ethylene was reduced to keep the total carbon flow rate constant. In total, nine diffusion flames with the fuel mixture covering the entire range of DME addition from pure ethylene to pure DME were investigated in this study. The composition of the fuel stream was parameterized using the ratio of the DME flow rate to the total fuel flow rate, i.e., b ¼ Q DME =ðQ DME þ Q C2 H4 Þ. The flow rates of DME and ethylene of the fuel stream for the nine flames are summarized in Table 1. 2D LII was employed to image the distribution of soot and the relative soot volume fractions under different levels of DME addition. A laser sheet of about 48 mm high and 1 mm wide formed from the second harmonic (k = 532 nm) of a Nd:YAG laser (Quantel, Brilliant B) operating at 10 Hz and 120 mJ per pulse energy was used to pass through the flame centerline to irradiate the soot particles. The temporal laser pulse has a 7 ns FWHM. The resultant incandescence signals were imaged using an ICCD at 90° from the laser sheet. The LII signals in the wavelength range from 400 to 450 nm were collected by using a bandpass filter in front of the ICCD to avoid signal noise from laser scattering (532 nm) and C2 swan band emissions (476 nm and 512 nm). The camera gate width was set at 20 ns with a 40 ns delay from the start of the laser pulse. The LII images presented below were the result of accumulation of 50 frames under each flame condition. The intensities of these 2D LII images reflect the relative soot volume fractions. The relative concentrations of aromatic species were measured using PLIF following the work of Sgro et al. [26]. In the PLIF measurements, the light source (48 mm high and 1 mm wide laser sheet) was from the same laser as that in the 2D LII measurements. However, the fourth harmonic, instead of the second one, of the pulsed Table 1 Ethylene and DME flow rates for the nine flames investigated. b

QDME (ml/min)

Q C2 H4 ðml=minÞ

0 0.0625 0.125 0.1875 0.25 0.375 0.5 0.75 1

0 12.1 24.3 36.4 48.5 72.8 97 145.5 194

194 181.9 169.7 157.6 145.5 121.1 97 48.5 0

550

F. Liu et al. / Combustion and Flame 158 (2011) 547–563

Nd:YAG laser (k = 266 nm) and a 50 mJ per pulse energy were used in the PLIF measurements for PAH concentrations. The same bandpass filter as that in the 2D LII experiments was used to collect LIF signals in the 400–450 nm spectral range, similar to the study of Yoon et al. [13]. The gatewidth for the camera was set at 80 ns without delay from the beginning of the laser pulse. Again, all the PLIF images obtained in this study were the result of accumulation of 50 frames and their intensities indicate the relative concentrations of PAHs. 3. Numerical model 3.1. Governing equations, soot and radiation models The steady-state fully-coupled elliptic conservation equations for mass, momentum, energy, and species mass fractions in axisymmetric cylindrical coordinates and in the low Mach number limit were solved. These conservation equations can be found in [27]. The effect of buoyancy was accounted for by retaining the gravity term in the momentum equation in the flow direction (z, vertically upwards). The method of correction diffusion velocity described in [28] was employed to ensure that the net diffusion flux of all species sums to zero in both r- and z-directions. The interaction between the soot chemistry and the gas-phase chemistry was accounted for through the reaction rates of the species involved in soot formation and oxidation. Only the thermal diffusion velocities of H2 and H are accounted for using the expression given in [28]. Such a treatment of the thermal diffusion is sufficiently accurate under the conditions of the present study based on the findings of Guo et al. [29], since only the thermal diffusion velocities of very light species, such as H, H2, or He, can be appreciable in comparison to their ordinary diffusion velocities. The thermal diffusion velocities of heavier species, however, are negligibly small. The source term in the energy equation due to radiation heat transfer was also included and calculated using the discrete-ordinates method along with the statistical narrow-band correlated-k method for the absorption coefficients of CO, CO2, and H2O. Details of thermal radiation calculations can be found in [30,31]. The spectral absorption coefficient of soot was assumed to be 5.5fvm with fv being the soot volume fraction and m the wavenumber of each spectral band. The radiation intensity at each spectral band was calculated using the 4-point Gauss–Legendre quadrature scheme [31]. The spectrally integrated radiation source term was evaluated by summing up contributions from all spectral bands covering the spectral range 150–9300 cm1. Besides the conservation equations mentioned above, additional transport equations were solved for sectional soot aggregate and primary soot particle number densities. In the current sectional model, the range of soot aggregate mass is divided into a number of discrete sections, each with a prescribed representative mass. According to their mass, soot aggregates are assigned to individual sections. The nucleation step connects the gaseous incipient species with the solid soot phase. Through coagulation and/or surface growth, aggregates in a lower section move to higher sections. Conversely, aggregates in a higher section move to lower sections or become gaseous products by oxidation and fragmentation. More details of the current sectional model can be found in [32]. The transport equations for sectional soot aggregate and primary particle number densities are given as [33,34] ðAÞ;ðPÞ

qv

@Nj

ðAÞ;ðPÞ

@Nj

þ qu @r @z ! ! ðAÞ;ðPÞ ðAÞ;ðPÞ @Nj @N 1 @ @ A þ ¼ qDAj j r qDj r @r @z @r @z 1 @ @ ðAÞ;ðPÞ ðAÞ;ðPÞ ðAÞ;ðPÞ ðr qNj V Tr;s Þ  ðqNj V Tz;s Þ þ xj r @r @z ðj ¼ 1; 2; . . . ; SNÞ 

ð1Þ

In Eq. (1),r, z, v, u, q are radial and axial coordinates, radial and axial ðAÞ;ðPÞ velocities and mixture density, respectively. N j is jth sectional number density of either aggregates (A) or primary particles (P) in unit mass of gaseous mixture; xj is jth sectional production rate; VTr,s and VTz.s are the radial and axial thermophoretic velocities of soot aggregates; SN is the total soot section number; DAj is the jth aggregate diffusivity, which is calculated by the method in [35]. Note that the same DAj appears in the jth primary particle number density equation. The system of equations given in Eq. (1) describes the evolution of soot particles under simultaneous nucleation, coagulation, surface growth, oxidation, and surface condensation processes. The source terms are obtained by the sectional aerosol dynamics model in conjunction with the PAH soot model. The contributions from nucleation (nu), coagulation (co), surface growth (sg), surface condensation (sc) and oxidation (ox) are included [33,34] 0      1 ðAÞ;ðPÞ  ðAÞ;ðPÞ  ðAÞ;ðPÞ  ðAÞ;ðPÞ  ðAÞ;ðPÞ  @Nj @Nj @Nj @Nj @Nj      A ðAÞ;ðPÞ xj ¼ q@  þ  þ  þ  þ  @t  @t  @t  @t  @t  nu

co

sg

sc

ox

ð2Þ

Nucleation rate is calculated by the collision rate of two A4 (pyrene) molecules in the free-molecular region with a van der Waals enhancement factor of 2.2 [19]. Coagulation terms are calculated using the same method given in [32]. The collision kernel of two aggregates in the entire Knudsen number regime is calculated based on [35]. It is noted that this method assumes unity coagulation efficiency, i.e., once two aggregates collide they stick together. Surface growth and oxidation terms are calculated by the HACA mechanism described in [20]. All parameters are kept as original except the parameter a, the fraction of the reactive soot surface. It is worth pointing out that the parameter a was introduced to simulate the so-called soot ageing phenomenon [36], i.e., the reactivity of soot particles weakens as they mature, albeit in an approximate manner. Here, the a recommended by Xu et al. [37] is used

a ¼ 0:004 expð10800=TÞ

ð3Þ

A similar correlation was used by Guo et al. [38] in the modeling of an ethylene/air diffusion flame. Using the original a of Appel et al. [20] the correct soot level could not be well reproduced. Unfortunately, there is currently no well established universal a in the literature that would work well under different flame conditions. The specific expression of a depends on the specific sooting conditions such as temperature. Systematic investigations on a should be made. PAH surface condensation accounts for the growth of soot particles due to the condensation of A4 molecules colliding with soot aggregates on their surface [20]. The condensation rate in the current model is calculated by the collision theory between A4 molecules and aggregates [32,39]. Since not all collisions lead to successful condensation [40], a PAH condensation efficiency c is introduced and assumed to be 0.5 to account for the probability of sticking in each collision following a recent study [33], which also provided a detailed discussion and sensitivity study of c. 3.2. Chemical kinetic mechanism To our knowledge a well validated detailed reaction mechanism for the combustion of ethylene and DME mixture that also includes the formation of PAHs is currently not available. For the purposes of this study a detailed reaction mechanism was assembled by adding the species and reactions related to the DME mechanism from [41], which contains 79 species, to the reaction mechanism of Appel et al. [20] for C2 fuels, which includes PAH formation and growth up to pryene (A4) and consists of 101 species and 544 reactions. The extended mechanism for ethylene and DME

F. Liu et al. / Combustion and Flame 158 (2011) 547–563

mixtures consists of 151 species and 785 reactions. This was achieved by simply appending reactions involving the 50-species that are in the DME mechanism [41] but not in the Appel et al. C2 mechanism [20] without validation or optimization, following the practice of a recent study [18]. This extended chemical kinetics mechanism was used throughout this study, even for the pure ethylene and the pure DME diffusion flames. It is recognized that the benzene formation and PAH growth sub-mechanisms in the Appel et al. C2 mechanism [20], which was published a decade ago, could be improved. This is particularly true for the propargyl recombination reaction in view of the recent advancement made by Miller and co-workers [42,43]. The propargyl recombination reaction was assumed to be a single-channeled step leading to benzene in the mechanism of Appel et al. [20]. However, the recent studies showed that the propargyl recombination is a multi-channeled and much more complicated reaction [42,43]. This multi-channeled propargyl reaction model was not incorporated into the gas-phase reaction mechanism employed in our study mainly for the reason that this multi-channeled reaction and its associated reactions have not been validated in modeling flames at atmospheric pressure, albeit some validation was made in low pressure premixed flames. 3.3. Numerical method Similar to several previous studies [33,34,38], the finite volume method was used to discretize the governing equations. The standard SIMPLE algorithm with the staggered mesh was used to handle the pressure and velocity coupling. The diffusive terms were discretized by the second order central difference scheme while the convective terms were discretized by the power law scheme. The gaseous species equations were solved simultaneously to effectively deal with the stiffness of the system and speedup the convergence process. The sectional soot equations were solved in the same manner as the species equations due to the stiffness of the system. The remaining governing equations were solved by the Tri-Diagonal Matrix Algorithm. The thermal and transport properties of gaseous species and the chemical reaction rates were obtained using Sandia’s Chemkin [44] and Transport [45] libraries and the database associated with the selected reaction mechanisms [20,41]. The atmospheric-pressure flame was modeled in a domain of 9.83 cm (z)  4.71 cm (r) with 210 (z)  88 (r) control volumes. The wall thickness of the central fuel tube was included in the numerical calculations. A non-uniform mesh was used to save computational time while resolving the large gradients. Very fine grids are placed in the r-direction (resolution 0.2 mm) near the burner exit in the z-direction (resolution 0.3 mm). It was checked that further refinement of the mesh had negligible effect on the results. Because the problem at hand is very computationally demanding, it would be intractable to obtain the solution with serial processing. Therefore, it is essential to employ parallel processing for such problems. In this study the distributed-memory parallelization with strip-domain decomposition was employed to speed up the calculations. The whole computational domain was divided uniformly (in terms of the number of control volumes) into 16 sub-domains along the axial direction and each subdomain was assigned to one CPU for the calculations [46]. The algorithm uses the message passing interface (MPI) library of Fortran to parallelize the code. A parabolic profile was assumed for the inlet fuel velocity, i.e., u = 2uF[1  (r/RI)2], where r is the radial position, RI is the inner radius of the fuel tube, and uF is the mean fuel stream velocity kept at 3.465 cm/s throughout this study. The resultant Reynolds numbers of the fuel stream (based on the fuel tube inner diameter) for the pure ethylene flame and the pure DME flame are about 43 and 68, respectively. For the air stream

551

a uniform velocity of 77 cm/s was assigned outside the boundary layer formed at the outer surface of the fuel pipe. Inside this boundary layer a boundary layer type velocity profile was assumed. The inlet temperatures for fuel and air were both assumed to be 300 K. Symmetry, free-slip and zero-gradient conditions were enforced at the centerline, outer radial boundary and the exit boundary, respectively. In total, 35 sections (SN = 35) were used for the sectional model with a sectional spacing factor of 2.35. The sectional representative sizes, as indicated by the volumeequivalent diameter, range from 0.9 nm to 13 lm. This sectional distribution gives adequate resolution of particle size and is large enough that the total mass fraction of soot particles in the largest section is zero. In other words, an empty largest section is secured to ensure that the selected sectional size range does not clip the formation of the largest particles. Further details can be found in [34]. Iterations in all the calculations conducted in this study were stopped after the maximum relative variation in temperature over 100 iterations was less than 1  104. To confirm that the computational domain in the streamwise direction (9.83 cm) is sufficiently long to model the present laminar diffusion flames, the pure ethylene flame was also computed on a larger computational domain of 16.37 cm (z)  4.71 cm (r) with 262 (z)  88 (r) control volumes. The results from the larger domain are virtually identical to those from the smaller domain, confirming that the smaller computational domain is adequate for modeling the laminar diffusion flames studied in this work.

4. Results and discussion 4.1. Experimental results The 2D LII images are shown in Fig. 1. The vertical distance measured from the burner exit (in mm) is marked in the plot. The images from the left to right in Fig. 1 are arranged in order of increasing b (increasing fraction of DME in the fuel stream), i.e., b = 0, 0.0625, 0.125, 0.1875, 0.25, 0.37, 0.5, 0.75, and 1, respectively. The color1 scales (image intensities) indicate the relative soot volume fractions in these flames. Due to the rapid decrease in the soot volume fraction with increasing amount of DME addition beyond b = 0.37, the image intensities decrease drastically. The image for b = 0.75 (the eighth image in Fig. 1) is barely visible and it becomes completely invisible for b = 1 (pure DME flame). The somewhat non-continuous image intensities observed at between z  43 and 48 mm is not physical. It is not clear presently what caused such unexpected image intensities. It is observed from Fig. 1 that the visible flame height, defined by the location on the flame centerline where soot is fully burnt out, first increases slightly as a relatively small amount of DME is added to the fuel stream up to about b = 0.125 and then starts to decrease as more DME is added. At b = 0.1875 the visible flame height is almost the same as that at b = 0 (the pure ethylene flame). The visible flame height decreases rather rapidly when more than 25% of DME is present in the fuel stream. It is noticed that the total carbon mass flow rate was maintained constant in our experiments. In laminar coflow diffusion flames fueled with conventional hydrocarbons the visible flame height remains more or less constant under the conditions of a constant carbon mass flow rate, e.g. [12]. However, the visible flame height of DME-ethylene mixed fuel overall decreases as more DME is added to the fuel stream due to the presence of oxygen in such fuel and reduced soot loading. The slight increase in the visible flame height at relatively small amount of DME addition (0 < b < 0.1875) may be attributed to 1 For interpretation of color in Figs. 1–19, the reader is referred to the web version of this article.

552

F. Liu et al. / Combustion and Flame 158 (2011) 547–563

Fig. 1. Images of soot volume fraction distribution obtained by 2D LII. The images from the left to right represent respectively the flames of b = 0 (pure ethylene), 0.0625, 0.125, 0.1875, 0.25, 0.375, 0.5, and 0.75. The 2D LII image of the pure DME flame (b = 1) is not visible due to the very low soot loading in this flame.

the increased fuel stream momentum, since the molar weight of DME is significantly higher than ethylene, and the enhanced soot volume fraction (see below), which could survive longer in the flame before it is fully oxidized. The mixture-strength flame height is the more fundamental quantity and it scales with carbon mass flowrate for conventional hydrocarbon fuels containing no oxygen. In laminar diffusion flames fueled by conventional hydrocarbon fuels, such as Ref. [12], the visible flame height correlates quite well with the mixture-strength flame height. However, this breaks down as more DME is added to ethylene, due to the addition of oxygen to the fuel stream. In other words, the mixture-strength flame height decreases with increasing the amount of DME in the fuel stream. It is also evident from the intensities of these images that the soot volume fractions first increase with relatively small amounts of DME addition and then start to decrease as more DME is added to the fuel stream. Such a synergistic effect of DME addition to ethylene on soot formation has been previously observed by McEnally and Pfefferle [14] and Yoon et al. [13] in laminar coflow nitrogendiluted ethylene and counterflow undiluted ethylene diffusion flames, respectively. Under the present experimental conditions, the soot volume fractions peak at b = 0.0625, the second image in Fig. 1, judged by the image intensities. This value is comparable to that observed by Yoon et al. [13] (b = 0.1) from the maximum LII signals in counterflow laminar ethylene diffusion flames. It is noticed that a finer interval of b was not attempted in our study. The peak soot volume fraction in the flame of b = 0.125 (the third image) still exceeds that of the pure ethylene flame. At b = 0.1875, the fourth image, the intensities of the 2D LII image exhibit levels very similar to, but perhaps slightly lower than, those of the pure ethylene case of b = 0 (the first image), suggesting that the soot volume fractions in the flames of b = 0.1875 and b = 0 are similar. This is consistent with the observation of the visible flame height discussed above. As more DME is added beyond b = 0.1875, the soot volume fractions decrease rapidly. Therefore, the synergistic effect of soot formation by adding DME to ethylene is found to occur in the range of 0 < b < 0.18 based on the present 2D LII investigation of laminar coflow undiluted ethylene/air diffusion

flames, which is in close agreement with the range of 0 < b < 0.22 observed by Yoon et al. [13] based on maximum LII signals in counterflow undiluted ethylene/air diffusion flames. It is important to clarify here that by synergistic effect we mean the maximum soot concentration exceeds that of the pure ethylene flame. Strictly speaking, the synergistic effect occurs when the maximum soot concentration is greater than the linear combination of those in the pure ethylene and DME flames. Figure 2 displays the PLIF images of the diffusion flames with various levels of DME addition to the fuel stream. As in Fig. 1, the images in Fig. 2 from the left to right are arranged in order of increasing b (b = 0, 0.0625, 0.125, 0.1875, 0.25, 0.37, 0.5, 0.75 and 1) and the color scales (image intensities) represent the relative concentrations of PAHs. The vertical distance from the burner exit (in mm) is shown in the figure. Once again, the PLIF image of b = 0.75 is barely visible and that of b = 1 (pure DME flame) is completely invisible, due to the very low PAH concentrations in these flames. Compared to the locations of soot shown in Fig. 1, PAHs appear much earlier starting almost right above the burner exit, which is consistent with the understanding that PAHs are soot precursors. As the fraction of DME in the fuel stream increases, the PAH concentrations first increase quite significantly and then start to decrease as more DME is added. It is clear from Fig. 2 that the synergistic effect of DME addition to ethylene on PAH formation is much stronger than that on soot formation shown in Fig. 1. Based on the image intensities as indicated by the color scales, the concentrations of PAHs peak at b = 0.125, the third image from the left in Fig. 2. The concentrations of PAHs at b = 0.375, the sixth image from the left, seem comparable to those in the pure ethylene flame (the first image), with the PAH concentrations in the centerline region between z = 13 mm to about 21 mm being somewhat lower. Our PLIF experiments indicate that the synergistic effect of DME addition to fuel on PAH formation in laminar coflow ethylene/air diffusion flame occurs when the fraction of DME addition in the fuel stream is in the range of 0 < b < 0.37. It is noticed that this range of DME addition is somewhat narrower than but comparable to that obtained by Yoon et al. [13] (0 < b < 0.5) from the maximum PAH LIF signals in counterflow undiluted ethylene/air diffusion

F. Liu et al. / Combustion and Flame 158 (2011) 547–563

553

Fig. 2. Images of PAHs obtained by PLIF. The images from the left to right represent respectively the flames of b = 0 (pure ethylene), 0.0625, 0.125, 0.1875, 0.25, 0.375, 0.5, and 0.75. The PLIF image of the pure DME flame (b = 1) is not visible due to the very low concentrations of PAHs in this flame.

flames. Both our present experiments and those conducted by Yoon et al. [13] establish that the synergistic effect of DME addition to ethylene on PAH formation persists over a much wider range of DME fraction in the fuel stream than that on soot formation. These results may be explained by our current understanding of soot inception and surface growth. Soot inception is mainly due to the collision and coagulation of PAHs. Although soot inception is the bottleneck of the soot formation process, it contributes only a small portion of the total soot loading. The majority of the soot mass comes from the surface growth process, which is primarily dependent on the concentration of C2H2, but not PAHs, according to the HACA mechanism [19,20]. Therefore, the synergistic effect of DME addition to ethylene on PAH formation does not necessarily corresponds to the synergistic effect on soot formation. 4.2. Numerical results 4.2.1. Temperature and axial velocity distributions The predicted temperature distributions in the nine flames of different DME fractions in the fuel stream are compared in Fig. 3 with the peak temperature indicated in each plot. In the pure ethylene flame, the first plot, the maximum temperature is 2056.5 K and occurs in the annular region low in the flame around z = 1.5 cm and r = 0.5 cm, instead in the centerline region around the flame tip due to radiation heat loss from soot, CO2, and H2O in the upper portion of the flame. As more DME is added to the fuel stream up to b = 0.5, Fig. 3g, the peak flame temperature continuously decreases slightly, while the temperatures in the centerline region in the upper portion of the flame gradually increase. It is interesting and also surprising to observe that the peak flame temperature actually increases from 2043.2 K to 2062.8 K when the fraction of DME in the fuel stream increases from b = 0.5 to 0.75, Fig. 3g and h. As b is further increased from 0.75 to 1, the peak flame temperature decreases from 2062.8 K to 2018.4 K (about 44 K), which is the lowest peak flame temperature among the nine flames, Fig. 3h and i. It is even more surprising to realize that the peak temperature of 2062.8 K at b = 0.75 is even higher than that of the pure ethylene flame (2056.5 K). The initial decrease in the peak flame temperature with relatively small amount of DME addition (up to b = 0.0625) can be attributed to the increased radiation loss through enhanced soot formation (shown later) and the fact that DME has a lower adiabatic flame temperature than ethylene in air (2296 K vs. 2385 K). The latter is responsible for the continuous decrease in the peak flame temperature as more DME is added. The sudden increase in the peak flame temperature at b = 0.75 is not expected. Therefore, the flame of b = 0.75 behaves differently from the other flames in terms of the value of the peak flame temperature. Moreover, it is the only flame among the nine flame studies where the peak flame temperature occurs in the flame centerline around z = 4.9 cm. In all other flames, however, the peak flame temperature occurs in the annular region low in the flame around z = 2 cm.

Examination of the calculated heat release rate distributions in these flames, especially in the flames of b = 0.5, 0.75, and 1, does not provide an explanation for the sudden increase in the peak flame temperature for the flame of b = 0.75, though the heat release rates in the centerline region decrease with increasing the amount of DME in the fuel stream. It is plausible that the highest flame temperature reached in the centerline region in the flame of b = 0.75 is mainly the result of two competing processes between heat release, which increases temperature, and radiation heat loss from soot and radiating species (CO2 and H2O), which decreases temperature. To confirm the role of thermal radiation in the occurrence of the unexpected higher peak flame temperature in the flame of b = 0.75 additional calculations were conducted for the flames of the three largest DME fractions in the fuel stream considered, i.e., b = 0.5, 0.75, and 1, with thermal radiation term removed. These results indicate that the peak flame temperature (2317.7 K, 2240 K, and 2134.5 K for b = 0.5, 0.75, and 1, respectively) decreases monotonically with increasing the fraction of DME in the fuel stream. Therefore, it can be concluded that thermal radiation is indeed responsible for the unexpected higher peak temperature in the flame of b = 0.75. Although the numerical results of these additional calculations reveal that thermal radiation is responsible for the occurrence of the highest flame temperature in the flame of b = 0.75, they do not answer the question how this happens. To understand how thermal radiation affects the flame temperature at different levels of DME addition, the distributions of CO2 and H2O mole fraction in the flames of b = 0.5, 0.75, and 1 were examined, since these two species are responsible for gas radiation. The mole fractions of CO2 in the centerline region decrease slightly as more DME is added to the fuel stream. However, the mole fractions of H2O in the centerline region around z = 5 cm increase somewhat significantly with increasing amount of DME in the fuel stream. Hence, the occurrence of the highest flame temperature in the flame of b = 0.75 can be understood as the result of competition between soot and gas radiation. In flames of lower values of b (<0.75) where soot concentrations are relatively high, see Fig. 5 shown later, thermal radiation from soot dominates gas radiation and lowers the flame temperature. In flames of higher values of b (>0.75), where soot concentrations are very low but water vapor concentrations are higher, thermal radiation loss due to water vapor is enhanced to lower the flame temperature. This is why the highest flame temperature occurs in the flame of b = 0.75. It is also noticed that the location of the peak temperature on the flame centerline, which correlates to the flame height defined by the location of the stoichiometric mixture fraction on the flame centerline, shifts upstream closer to the burner as more DME is added to the fuel stream. This is actually expected due to the presence of oxygen in DME. Distributions of axial velocity along the flame centerline are compared in Fig. 4. Overall the axial velocity distribution is only weakly dependent on the fraction of DME in the fuel stream, which

554

F. Liu et al. / Combustion and Flame 158 (2011) 547–563 (a) Pure C2H4 2056.5 K

8 7 6

z, cm

5 4 3 2 1 0 0

0.5

1

T 2100 2055 2010 1965 1920 1875 1830 1785 1740 1695 1650 1605 1560 1515 1470 1425 1380 1335 1290 1245 1200 1155 1110 1065 1020 975 930 885 840 795 750 705 660 615 570 525 480 435 390 345 300

r, cm

(b) 6% DME 2051.4 K

(c) 12% DME 2050.0 K

(d) 18% DME 2049.0 K

(e) 25% DME 2048.1 K

(f) 37% DME 2045.9 K

(g) 50% DME 2043.2 K

(h) 75% DME 2062.8 K

(i) 100% DME 2018.4 K

8

8

8

8

8

8

8

8

7

7

7

7

7

7

7

7

6

6

6

6

6

6

6

6

5

5

5

5

5

5

5

5

4

4

4

4

4

4

4

4

3

3

3

3

3

3

3

3

2

2

2

2

2

2

2

2

1

1

1

1

1

1

1

1

0

0 0

0.5

1

0 0

r, cm

0.5

1

0 0

r, cm

0.5

1

r, cm

0 0

0.5

r, cm

1

0 0

0.5

r, cm

1

0 0

0.5

r, cm

1

0 0

0.5

r, cm

1

0

0.5

1

r, cm

Fig. 3. Comparison of the calculated temperature distributions in the nine laminar coflow diffusion flames.

Axial velocity, cm/s

250

200 β=0 β = 0.0625 β = 0.125 β = 0.1875 β = 0.25 β = 0.375 β = 0.5 β = 0.75 β=1

150

100

50

0 0

1

2

3

4

5

6

7

8

9

z, cm Fig. 4. Distributions of the axial velocity along the flame centerline.

is expected given the fact that these laminar diffusion flames are buoyancy controlled. It is worth pointing out that the velocity field in the flame varies fairly little with the amount of DME in the fuel stream. The degree of axial velocity variation with the amount of DME addition along the flame centerline displayed in Fig. 4 is representative of the overall axial velocity variation from one flame condition to another (2D plots are not shown). Consequently, it can be concluded that the residence times in the sooting regions of these diffusion flames are very similar. With increasing amounts of DME in the fuel stream, the axial velocity first decreases slightly and then increases more rapidly to reach slightly higher velocities. The initial decrease in the axial velocity with increasing the amount of DME in the fuel stream is likely caused by the increasingly smaller heat release rates in the centerline region between about z = 4–10 mm (not shown) as more DME is added to the fuel stream. The more rapid increase in the axial velocity at higher DME fractions in the fuel stream is due to stronger flow expansion associated with higher temperatures in the flame centerline region. 4.2.2. Soot volume fraction The predicted soot volume fraction distributions under different fractions of DME addition to the fuel stream are compared in Fig. 5.

In the case of pure ethylene flame, Fig. 5a, the predicted peak soot volume fraction (8.37 ppm) and its location (around z = 3.5 cm) are in good quantitative agreement with the measured values (8 ppm and z = 3.1 cm) [17]. The absence of soot in the flame centerline region in Fig. 5 is a known deficiency of the current soot models in ethylene flames, e.g., [33,38]. Overall the predicted soot volume fraction distributions are in qualitative agreement with the 2D LII images shown in Fig. 1, i.e., soot concentrations decrease rapidly when the fraction of DME in the fuel stream is greater than 0.25. Similar to the LII images at the two highest values of b (0.75 and 1), the color plots of the predicted soot volume fraction are barely visible at b = 0.75, Fig. 5h, and become invisible at b = 1. As the amount of DME addition to ethylene increases, the peak soot volume fraction first increases from 8.37 ppm to 8.55 ppm, Fig. 5a and b, and then start to decrease rather rapidly. The predicted degree of soot volume fraction enhancement by addition of a small amount of DME (b = 0.0625) seems less than that revealed by the LII image intensities shown in Fig. 1, albeit both the experiment and the modeling indicate that the maximum soot volume fraction occurs at b = 0.0625, Figs. 1 and 5. Although the synergistic effect of DME addition on soot formation is predicted by the numerical model, in agreement with the experimental observation shown in Fig. 1, the range of DME fraction in fuel exhibiting the synergistic effect predicted by the model (0 < b < 0.0625) is much narrower than the experimental one (0 < b < 0.18) shown in Fig. 1. The failure of the numerical model in this aspect may be attributed to the deficiency in the soot inception model, which was assumed to be due to the collision of two pyrene molecules. On the other hand, the soot inception rate is very sensitive to the concentration of pyrene, which in turn is determined by the gas-phase chemistry, especially the chemistry leading to the first ring structure (benzene) and the subsequent growth to larger PAHs. Therefore, the soot inception model is intimately coupled with the gas-phase chemistry through the PAH formation and growth and it is difficult to make a conclusive remark about the adequacy of the soot inception model before the PAH chemistry is properly validated.

4.2.3. Benzene and PAHs In this section the calculated mole fraction distributions of A1 (benzene) and A4 (pyrene) are examined. Benzene and pyrene are of particular importance since A1 is the first ring structure

555

F. Liu et al. / Combustion and Flame 158 (2011) 547–563

8

(a) Pure C2H4 8.374 ppm vol rac f

7 6

z, cm

5 4 3 2 1 0 0

0.5

1

8.00 7.80 7.60 7.40 7.20 7.00 6.80 6.60 6.40 6.20 6.00 5.80 5.60 5.40 5.20 5.00 4.80 4.60 4.40 4.20 4.00 3.80 3.60 3.40 3.20 3.00 2.80 2.60 2.40 2.20 2.00 1.80 1.60 1.40 1.20 1.00 0.80 0.60 0.40 0.20 0.00

(b) 6% DME (c) 12% DME (d) 18% DME 8.552 ppm 7.554 ppm 6.465 ppm

(e) 25% DME 5.459 ppm

(f) 37% DME 3.682 ppm

(g) 50% DME 2.26 ppm

(h) 75% DME (i) 100% DME 0.4636 ppm 0.00149 ppm

8

8

8

8

8

8

8

8

7

7

7

7

7

7

7

7

6

6

6

6

6

6

6

6

5

5

5

5

5

5

5

5

4

4

4

4

4

4

4

4

3

3

3

3

3

3

3

3

2

2

2

2

2

2

2

2

1

1

1

1

1

1

1

1

0

0 0

r, cm

0.5

1

0 0

r, cm

0.5

1

0 0

r, cm

0.5

1

0 0

r, cm

0.5

1

0 0

r, cm

0.5

1

0 0

r, cm

0.5

1

0 0

r, cm

0.5

1

0

0.5

r, cm

1

r, cm

Fig. 5. Comparison of the calculated soot volume fraction distributions in the nine flames of different DME fractions in the fuel stream.

(a) Pure C2H4 Peak: 2.7304E-4

8

A1 3 2.9 2.8 2.7 2.6 2.5 2.4 2.3 2.2 2.1 2 1.9 1.8 1.7 1.6 1.5 1.4 1.3 1.2 1.1 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

7 6

z, cm

5 4 3 2 1 0

(b) 6% DME 3.9584E-4

0.5

r, cm

1

(d) 18% DME 3.6365E-4

(e) 25% DME 3.4069E-4

(f) 37% DME 3.0496E-4

(g) 50% DME 2.6949E-4

(h) 75% DME 1.9031E-4

(i) 100% DME 1.0545E-4

8

8

8

8

8

8

8

8

7

7

7

7

7

7

7

7

6

6

6

6

6

6

6

6

5

5

5

5

5

5

5

5

4

4

4

4

4

4

4

4

3

3

3

3

3

3

3

3

2

2

2

2

2

2

2

2

1

1

1

1

1

1

1

1

0 0

(c) 12% DME 3.8647E-4

0 0

0.5

r, cm

1

0 0

0.5

r, cm

1

0 0

0.5

1

r, cm

0 0

0.5

r, cm

1

0 0

0.5

r, cm

1

0 0

0.5

r, cm

1

0 0

0.5

r, cm

1

0

0.5

1

r, cm

Fig. 6. Comparison of the calculated benzene mole fraction distributions in the nine flames of different DME fractions in the fuel stream.

and A4 is the assumed soot inception species. Figures 6 and 7 display the 2D and centerline mole fraction distributions of A1, respectively. The same color scale is used for all the plots in Fig. 6. It is evident that the addition of DME to the fuel stream results in higher A1 mole fractions than those in the pure ethylene flame over a wide range of DME fraction in the fuel stream, up to at least b = 0.375, Fig. 6f. When more DME is added to the fuel stream, i.e., b > 0.375, the mole fractions of A1 continue to decrease to levels well below those in the pure ethylene flame, up to a factor of about 2.6 in the pure DME flame. It is clear from the results shown in Fig. 6 that addition of DME to ethylene leads to a strong synergistic effect on benzene formation over a wide range of DME fraction in the fuel stream up to about b = 0.38. The centerline distributions of A1 mole fraction shown in Fig. 7 clearly illustrates the synergistic effect of DME addition on A1 formation, since the centerline peak mole fractions for b between 0.0625 and 0.375 all exceed that for b = 0. Addition of a small amount of DME to ethylene is very effective in prompting A1 forma-

tion. This is because addition of 6.25% DME to the fuel stream results in a 45% increase in the peak A1 mole fraction compared to that in the pure ethylene flame. The addition of relatively low amounts of DME, up to b = 0.375, first suppresses the formation of A1 in the centerline region close to the burner exit but then accelerates its formation to peak values above that in the pure ethylene flame. Compared to the numerical results of Bennett et al. [18], who showed that the peak A1 mole fractions in a nitrogen diluted (69% volume basis) ethylene laminar coflow diffusion flame with up to about 9% DME addition in the fuel stream are between about 110 (pure ethylene flame) to 140 ppm (9% DME addition), the A1 mole fractions in our present study are much higher and exhibit a stronger sensitivity to DME addition. Such differences in the A1 mole fraction can be attributed to the following two reasons. First, Bennett et al. [18] studied nitrogen diluted flames while undiluted ethylene flames are investigated in our study. Secondly, different reaction mechanisms were used in these numerical studies, which are likely to result in different species concentrations, especially for

F. Liu et al. / Combustion and Flame 158 (2011) 547–563

0.0004 β=0 β = 0.0625 β = 0.125 β = 0.1875 β = 0.25 β = 0.375 β = 0.5 β = 0.75 β=1

A1 Mole Fraction

0.0003

0.0002

0.0001

0.0000 0

1

2

3

4

5

6

7

z, cm Fig. 7. Distributions of the calculated mole fraction of A1 along the flame centerline in the nine flame studied.

species of low concentrations such as PAHs. It is also noticed that benzene (A1) was the largest species in the three mechanisms used in [18]; therefore, there were fewer pathways for A1 destruction than the gas-phase mechanism used in this study. As more DME is added to the fuel stream beyond b = 0.375, the mole fractions of A1 continuously decrease. However, even in the pure DME flame the peak A1 mole fraction is still about 100 ppm, which is about 37% of that in the pure ethylene flame. To explore the relative importance of the propargyl selfreaction and reactions between C4 and C2 species in the formation of A1, reaction rates of the 14 reactions involving A1 in the Appel et al. [20] reaction mechanism, based on which the present reaction mechanism was built, were analyzed. It was found that the formation of A1 under the present flame conditions is primarily due to the following reactions

C3 H3 þ C3 H3 ! A1

ðR234Þ

n-C4 H5 þ C2 H2 $ A1 þ H

ðR296Þ

l-C6 H6 þ H $ A1 þ H

ðR348Þ

n-C6 H7 $ A1 þ H

ðR354Þ

A1- þ H þ ðMÞ $ A1 þ ðMÞ

ðR371Þ

7 6

z, cm

5 4 3 2 1 0 0 0.25 0.5 0.75 1

(b) 6% DME Peak: 1.7937E-6

(c) 12% DME Peak: 1.6924E-6

(d) 18% DME Peak: 1.6717E-6

8

8

8

7

7

7

6

6

6

5

5

5

z, cm

8

R234 1.5E-06 1.45E-06 1.4E-06 1.35E-06 1.3E-06 1.25E-06 1.2E-06 1.15E-06 1.1E-06 1.05E-06 1E-06 9.5E-07 9E-07 8.5E-07 8E-07 7.5E-07 7E-07 6.5E-07 6E-07 5.5E-07 5E-07 4.5E-07 4E-07 3.5E-07 3E-07 2.5E-07 2E-07 1.5E-07 1E-07 5E-08 0

z, cm

(a) Pure C2H4 Peak: 1.8656E-6

The rate of the last reaction (R371) peaks around z = 4 cm, where the benzene is rapidly consumed as shown in Figs. 6 and 7. This reaction will not be further discussed here. All other four reactions contributing to A1 formation have also been identified by McNesby et al. [47] in their study of fuel side addition of 8% ethanol in counterflow ethylene/air diffusion flames. The self-combination reaction of propargyl (R234) for A1 formation has received considerable attention in recent years and was believed to be the leading reaction for benzene formation in diffusion flames, e.g., [13,18]. To demonstrate how DME addition affects the propargyl route and the C2 and C4 species route for A1 formation, Figs. 8 and 9 display the distributions of reaction rates of C3H3 + C3H3 ? A1 (R234) and nC4H5 + C2H2 M A1 + H (R296), respectively, for b = 0, 0.0625, 0.125, and 0.1875. The effect of DME addition on the reaction rates of lC6H6 + H M A1 + H (R348) and n-C6H7 M A1 + H (R354) is not shown, since it is very similar to that on n-C4H5 + C2H2 M A1 + H (R296) shown in Fig. 9, i.e., addition of DME enhances the reaction rates of all these three reactions (R296), (R348), and (R354). Although the peak values of the reaction rate of (R234), Fig. 8, are about a factor of 25 higher than those of (R296), Fig. 9, they appear in the region just above the burner rim, rather than in the centerline region. On the other hand, addition of DME significantly enhances the reaction rate of (R296) (as well as (R348) and (R354), though not shown) in the centerline region, Fig. 9. Therefore, in the region around z = 1.5 cm along the flame centerline the differences between the reaction rates of (R234) and (R296) are less pronounced with the addition of a relatively small amount of DME. To demonstrate this Fig. 10 compares the reaction rates of (R234) and (R296) along the flame centerline for b = 0, 0.0625, 0.125, and 0.1875. It is evident that addition of DME only somewhat enhances the reaction rate of (R234) around z = 2 cm, which is directly related to the C3H3 concentrations shown in Fig. 17 below, but has very little influence on (R234) at higher axial locations. However, addition of DME significantly increases the reaction rate of (R296) in the region of 1 cm < z < 2 cm. Nonetheless, the self-combination reaction of propargyl (R234) is still much more important than (R296) to A1 formation. Figure 11 compares the centerline distributions of reaction rates of the other two important A1 forming reactions, i.e., (R348) and (R354), for b = 0, 0.0625, 0.125, and 0.1875. It is first observed from the results shown in Figs. 10 and 11 that the self-combination reaction of propargyl (R234) is indeed the most important reaction leading to A1 formation in the pure ethylene diffusion flame (b = 0). However, addition of DME to the fuel stream somewhat

4

z, cm

556

4

4

3

3

3

2

2

2

1

1

1

0

0 0 0.25 0.5 0.75 1

0 0.25 0.5 0.75 1

r, cm

r, cm

r, cm

r, cm 3

Fig. 8. Distributions of reaction rate (in mol cm

0

0 0.25 0.5 0.75 1

s

1

) of reaction C3H3 + C3H3 ? A1 (R234) for b = 0, 0.0625, 0.125, and 0.1875.

557

F. Liu et al. / Combustion and Flame 158 (2011) 547–563

7 6

z, cm

5 4 3 2 1 0 0 0.25 0.5 0.75 1

8

8

8

7

7

7

6

6

6

5

5

5

4

4

4

3

3

3

2

2

2

1

1

1

0

r, cm

(d) 18% DME Peak: 8.051E-8

z, cm

8

(c) 12% DME Peak: 8.78E-8

(b) 6% DME Peak: 8.61E-8

z, cm

R296 7.2E-08 6.96E-08 6.72E-08 6.48E-08 6.24E-08 6E-08 5.76E-08 5.52E-08 5.28E-08 5.04E-08 4.8E-08 4.56E-08 4.32E-08 4.08E-08 3.84E-08 3.6E-08 3.36E-08 3.12E-08 2.88E-08 2.64E-08 2.4E-08 2.16E-08 1.92E-08 1.68E-08 1.44E-08 1.2E-08 9.6E-09 7.2E-09 4.8E-09 2.4E-09 0

z, cm

(a) Pure C2H4 Peak: 6.74E-8

0

0

0 0.25 0.5 0.75 1

0 0.25 0.5 0.75 1

0 0.25 0.5 0.75 1

r, cm

r, cm

r, cm

Fig. 9. Distributions of reaction rate (in mol cm3 s1) of reaction n-C4H5 + C2H2 M A1 + H (R296) for b = 0, 0.0625, 0.125, and 0.1875.

β=0 β = 0.0625 β = 0.125 β = 0.1875

1.0e-8

8.0e-9

Reaction Rate, mol cm-3 s-1

Reaction Rate, mol cm-3 s-1

1.2e-8

6.0e-9

4.0e-9 R234 2.0e-9

β=0 β = 0.0625 β = 0.125 β = 0.1875

2.0e-8

1.5e-8

1.0e-8

5.0e-9

R348

R296

R354

0.0

0.0 0

1

2

3

4

5

6

z, cm Fig. 10. Distributions of reaction rates of C3H3 + C3H3 ? A1 (R234) and nC4H5 + C2H2 M A1 + H (R296) along the flame centerline for b = 0, 0.0625, 0.125, and 0.1875.

alters the relative importance of the propargyl self-combination reaction, the C2 and C4 species reaction, and the two cyclization reactions for A1 formation. Addition of DME drastically enhances the reaction rates of (R348) and (R354), with the rate of lC6H6 + H M A1 + H (R348) by more than threefold and that of (R354) more than doubled. It is also evident from Figs. 10 and 11 that the cyclization reactions, (R348) and (R354), are much more important than the C2 and C4 species reaction (R296) for A1 formation. The increased rates of the cyclization reactions (R348) and (R354) are the result of significant increases in the concentrations of, H radical, l-C6H6, and n-C6H7 in the centerline region around z = 1.75 cm when a relatively small amount of DME is added to ethylene. Pathways leading to higher concentrations of l-C6H6 and n-C6H7 are discussed later following the presentation of methyl formation. It is also noticed from Fig. 11 that the reaction rates of R354 at b = 0.0625, 0.125, and 0.1875 are comparable to those of the selfcombination reaction of R234 shown in Fig. 10, while the reaction rates of (R348) at b = 0.0625, 0.125, and 0.1875 become much higher than those of (R234), albeit the reaction rates of (R234) persist over a larger region along the centerline. It can therefore be concluded based on the results shown in Figs. 10 and 11 that compared to the pure ethylene flame case the additional formation of

0

1

2

3

4

5

6

z, cm Fig. 11. Distributions of reaction rates of l-C6H6 + H M A1 + H (R348) and nC6H7 M A1 + H (R354) along the flame centerline for b = 0, 0.0625, 0.125, and 0.1875.

A1 under a relatively small amount of DME addition are due to, in order of decreasing importance, l-C6H6 + H M A1 + H (R348), nC6H7 M A1 + H (R354), and n-C4H5 + C2H2 M A1 + H (R296), but not due to the self-combination reaction of propargyl (R234). This is because the addition of DME greatly enhances the reactions rates of these three A1 forming reactions. It is noticed that the contribution of the two cyclization reactions, (R348) and (R354), is much greater than that of reaction (R296). Although addition of DME also alters the distributions of reaction rate of (R234), Fig. 10, the overall effect is quite small when b is less than about 0.2. The importance of propargyl self-combination reaction (R234) to A1 formation in flames with DME addition should still be recognized, since it still makes the most contribution to A1 formation (the overshoot of A1 mole fraction above the level in the pure ethylene flame, Figs. 6 and 7, is only about 45%). These findings are in qualitative agreement with those of by McNesby et al. [47] in the study of ethanol addition to ethylene in counterflow diffusion flames. In this regard, it is also worth pointing out that the same C2 reaction mechanism [20], including PAH formation up to A4, was used in [47] and the present study. Figure 12 compares the 2D distributions of A4 (pyrene) mole fraction in the nine diffusion flames of different DME fractions in the fuel stream, while the corresponding A4 mole fraction

558

F. Liu et al. / Combustion and Flame 158 (2011) 547–563 (a) Pure C2H4 Peak: 1.9353E-7

8 7 6

z, cm

5 4 3 2 1 0 0

0.5

r, cm

1

A4 2 1.95 1.9 1.85 1.8 1.75 1.7 1.65 1.6 1.55 1.5 1.45 1.4 1.35 1.3 1.25 1.2 1.15 1.1 1.05 1 0.95 0.9 0.85 0.8 0.75 0.7 0.65 0.6 0.55 0.5 0.45 0.4 0.35 0.3 0.25 0.2 0.15 0.1 0.05 0

(b) 6% DME 2.3181E-7

(c) 12% DME 2.2891E-7

(d) 18% DME 2.2215E-7

(e) 25% DME 2.1364E-7

(f) 37% DME 1.9693E-7

(g) 50% DME 1.7369E-7

(h) 75% DME 1.0596E-7

(i) 100% DME 3.8865E-8

8

8

8

8

8

8

8

8

7

7

7

7

7

7

7

7

6

6

6

6

6

6

6

6

5

5

5

5

5

5

5

5

4

4

4

4

4

4

4

4

3

3

3

3

3

3

3

3

2

2

2

2

2

2

2

2

1

1

1

1

1

1

1

1

0

0 0

0.5

1

0 0

r, cm

0.5

1

0 0

r, cm

0.5

1

r, cm

0 0

0.5

r, cm

1

0 0

0.5

r, cm

1

0 0

0.5

r, cm

1

0 0

0.5

r, cm

1

0

0.5

1

r, cm

Fig. 12. Comparison of the calculated pyrene mole fraction distributions in the nine flames of different DME fractions in the fuel stream.

2.5e-7 β=0 β = 0.0625 β = 0.125 β = 0.1875 β = 0.25 β = 0.375 β = 0.5 β = 0.75 β=1

A4 Mole Fraction

2.0e-7

1.5e-7

1.0e-7

5.0e-8

0.0 0

1

2

3

4

5

6

7

z, cm Fig. 13. Distributions of the calculated mole fraction of pyrene along the flame centerline in the nine flames studied.

distributions along the flame centerline are displayed in Fig. 13. Compared to the mole fraction distributions of A1, pyrene appears downstream at higher centerline locations, which is expected from the PAH growth mechanism. Addition of DME to the fuel stream also gives rise to a synergistic effect on A4 formation over a wide range of DME addition at least up to b = 0.375, Fig. 12f, very similar to that observed in Fig. 6 for A1. This is also evident in Fig. 13, which also shows that the decrease in the peak A4 mole fraction from the pure ethylene flame to the pure DME flame is quite large (about a factor of 4.5). However, it is noticed that the synergistic effect on A4 formation is somewhat weaker than that on A1 formation, judged by the overshoot of the peak mole fractions above the corresponding values in the pure ethylene flame, Figs. 7 and 13. Examination of the mole fraction distributions of A2 and A3 (not shown) reveals that they have the same trend as that observed for A1 and A4 in terms of the synergistic effect of DME addition on their mole fractions. Therefore, the synergistic effect of DME addition on PAH formation observed experimentally is also well reproduced numerically, including the range of DME addition to the fuel stream over which the PAH concentrations are higher than

those in the pure ethylene flame, i.e., 0 < b < 0.38 from both the PLIF images and the numerical modeling. A closer examination of Fig. 2 indicates that the PAH concentrations peak at b = 0.125 from the PLIF images, the third plot in Fig. 2; however, the mole fractions of both A1 and A4 peak at b = 0.0625 in the numerical results, Figs. 6 and 12. It is noticed that the maximum PAH LIF signal obtained by Yoon et al. [13] in counterflow diffusion flame occur at about b = 0.1, which is in close agreement with the present result of b = 0.125 based on the PLIF signals of laminar coflow diffusion flame shown in Fig. 2. It is expected that the enhanced A4 mole fractions in the flame centerline region for 0.0625 6 b 6 0.375 leads to higher soot inception rates, given the fact that the temperature in the centerline regions between z = 3–4 cm in these flames varies only slightly (the addition of DME lowers the temperature in this region), Fig. 3, and the weak dependence of the soot inception rate on temperature (square root). However, such enhanced soot inception rates do not translate into higher soot volume fractions in the flame centerline region for the reasons that the concentrations of acetylene, which is the soot surface growth species in the HACA mechanism, decrease rapidly in the centerline region above z = 3 cm (shown later in Figs. 14 and 15) and the residence times for soot surface growth are quite short. This is why the predicted soot volume fractions are very low in the centerline region in these flames. On the other hand, the annular sooting region predicted by the model, Fig. 5, is along the path where A4 and acetylene coexist up to about z = 4 cm, see Figs. 12 and 14 shown below. Soot is then transported by the flow field further downstream, while experiencing oxidation. 4.2.4. Acetylene Because acetylene is the only species responsible for soot surface growth process (the contribution of A4 condensation is small), it is of importance to examine how DME addition affects acetylene. The calculated 2D distributions of C2H2 mole fraction in the nine flames are compared in Fig. 14, while the centerline distributions of C2H2 mole fraction are shown in Fig. 15. It is seen that with a small amount of DME addition, i.e., b = 0.0625 (the second plot in Fig. 14), the peak mole fraction of acetylene actually increases slightly above that in the pure ethylene flame, which is also indicated in Fig. 15. As more DME is added to the fuel stream beyond b = 0.0625, the peak mole fraction of acetylene continues to decrease. Under the conditions of the present study, DME addition

559

F. Liu et al. / Combustion and Flame 158 (2011) 547–563 (a) Pure C2H4 Peak: 4.8961E-2

8 7 6

z, cm

5 4 3 2 1 0 0

0.5

r, cm

1

C2H2 4.8 4.68 4.56 4.44 4.32 4.2 4.08 3.96 3.84 3.72 3.6 3.48 3.36 3.24 3.12 3 2.88 2.76 2.64 2.52 2.4 2.28 2.16 2.04 1.92 1.8 1.68 1.56 1.44 1.32 1.2 1.08 0.96 0.84 0.72 0.6 0.48 0.36 0.24 0.12 0

(b) 6% DME 4.9294E-2

(c) 12% DME 4.8162E-2

(d) 18% DME 4.6488E-2

(e) 25% DME 4.4511E-2

(f) 37% DME 4.0032E-2

(g) 50% DME 3.5101E-2

(h) 75% DME 2.4613E-2

(i) 100% DME 1.4119E-2

8

8

8

8

8

8

8

8

7

7

7

7

7

7

7

7

6

6

6

6

6

6

6

6

5

5

5

5

5

5

5

5

4

4

4

4

4

4

4

4

3

3

3

3

3

3

3

3

2

2

2

2

2

2

2

2

1

1

1

1

1

1

1

1

0

0 0

0.5

1

r, cm

0 0

0.5

1

0 0

r, cm

0.5

1

r, cm

0 0

0.5

r, cm

1

0 0

0.5

r, cm

1

0 0

0.5

r, cm

1

0 0

0.5

r, cm

1

0

0.5

1

r, cm

Fig. 14. Comparison of the calculated acetylene mole fraction distributions in the nine flames of different DME fractions in the fuel stream.

0.06 β=0 β = 0.0625 β = 0.125 β = 0.1875 β = 0.25 β = 0.375 β = 0.5 β = 0.75 β=1

C2H2 Mole Fraction

0.05

0.04

0.03

0.02

0.01

0.00 0

1

2

3

4

5

6

7

z, cm Fig. 15. Comparison of the calculated acetylene mole fraction distributions along the flame centerline in the nine diffusion flames studied.

to ethylene leads to a very weak synergistic effect on the concentration of acetylene only at b = 0.0625, which correlates well to that observed for the synergistic effect of DME addition on soot formation shown in Fig. 5. This is not surprising given the important role of acetylene in soot surface growth, which is the process responsible for almost all the soot mass. Therefore, the combined effect of enhanced A4 and C2H2 mole fractions along the annular soot growth path, as a result of DME addition, is the cause of the predicted synergistic effect of DME addition on soot formation. This is why the synergistic effect on soot formation occurs only for a narrow range of DME addition, i.e., b is around 0.0625, though the synergistic effect on PAH formation persists over a much wider range of DME addition. Relative to the location where C2H2 peaks in the pure C2H4 flame (b = 0), the addition of DME to fuel first shifts the location of the maximum C2H2 slightly upstream closer to the burner rim and then gradually somewhat downstream as more DME is added, Fig. 15. There is a significant decrease in the peak mole fraction of acetylene from the pure ethylene flame (b = 0) to the pure DME flame (b = 1), about a factor of 3.5. The significant decreases in the mole fractions of A4 and C2H2, Figs. 12–15, as large amounts

of DME are added to the fuel stream lead to greatly reduced soot inception and surface growth rates and eventually to the very low soot volume fractions in the DME flame, Fig. 5i. An examination of the reaction rates for reactions associated with C2H2 formation in flames of b = 0 and 0.0625 indicates that C2H2 is primarily formed through C2H3 + H M C2H2 + H2 (R158) and the decomposition of ethylene, i.e., C2H4 + (M) M C2H2 + H2 + (M) (R164), in both the flame without DME addition (b = 0) and the one with DME addition (b = 0.0625). Vinyl (C2H3) is primarily formed by the attack of H and OH radicals on ethylene, i.e., C2H4 + H M C2H3 + H2 (R166) and C2H4 + OH M C2H3 + H2O (R168) in both the pure ethylene flame (b = 0) and the flame with a small amount of DME addition (b = 0.0625). In the flame with DME addition, the importance of methyl attack on ethylene to form vinyl is significantly enhanced via C2H4 + CH3 M C2H3 + CH4 (R169) in the flame centerline region around z = 1.5 cm due to enhanced concentration of methyl from the decomposition of DME. Hence R169 is responsible for the synergistic effect of vinyl (not shown). The peak mole fraction of vinyl in the b = 0.0625 flame is more than doubled compared to that in the pure ethylene flame. 4.2.5. Propargyl In view of the recent emphasis on the potential importance of propargyl in benzene formation through its self-combination reaction (R234), e.g., [13,18], it is of interest to examine how DME addition to ethylene affects the concentration of propargyl. The calculated 2D mole fraction distributions of C3H3 in the nine diffusion flames studied are compared in Fig. 16. The concentration of propargyl peaks in the annular region just above the burner rim in all the nine flames calculated. The peak mole fraction of C3H3 decreases monotonically with increasing the fraction of DME in the fuel stream. The peak mole fraction of C3H3 decreases just over a factor of 2 from the pure ethylene flame to the pure DME flame. To further illustrate the effect of DME addition on the propargyl concentration, the centerline distributions of C3H3 mole fraction in the nine flames are compared in Fig. 17. The peak mole fraction of C3H3 along the centerline, at about z = 3.1 cm, remains almost constant over a wide range of DME addition up to b = 0.5. Addition of DME delays the occurrence of C3H3 along the centerline. For a moderate amount of DME addition up to b = 0.25 the centerline mole fraction of C3H3 is slightly enhanced around z = 1.8 cm. The

560

F. Liu et al. / Combustion and Flame 158 (2011) 547–563 (a) Pure C2H4 Peak: 7.8349E-5

8 7 6

z, cm

5 4 3 2 1 0 0

0.5

1

C3H3 6 5.85 5.7 5.55 5.4 5.25 5.1 4.95 4.8 4.65 4.5 4.35 4.2 4.05 3.9 3.75 3.6 3.45 3.3 3.15 3 2.85 2.7 2.55 2.4 2.25 2.1 1.95 1.8 1.65 1.5 1.35 1.2 1.05 0.9 0.75 0.6 0.45 0.3 0.15 0

r, cm

(b) 6% DME 7.7346E-5

(c) 12% DME 7.7506E-5

(d) 18% DME 7.6846E-5

(e) 25% DME 7.5436E-5

(f) 37% DME 7.2696E-5

(g) 50% DME 6.7891E-5

(h) 75% DME 5.5E-5

(i) 100% DME 3.6139E-5

8

8

8

8

8

8

8

8

7

7

7

7

7

7

7

7

6

6

6

6

6

6

6

6

5

5

5

5

5

5

5

5

4

4

4

4

4

4

4

4

3

3

3

3

3

3

3

3

2

2

2

2

2

2

2

2

1

1

1

1

1

1

1

1

0

0 0

0.5

r, cm

1

0 0

0.5

r, cm

1

0 0

0.5

1

r, cm

0 0

0.5

1

0 0

r, cm

0.5

1

r, cm

0 0

0.5

1

0 0

r, cm

0.5

1

0

r, cm

0.5

1

r, cm

Fig. 16. Comparison of the calculated propargyl mole fraction distributions in the nine flames of different DME fractions in the fuel stream.

C2 H2 þ CH2 $ C3 H3 þ H

ðR193Þ

C2 H2 þ CH2 $ C3 H3 þ H

ðR194Þ

followed by

AC3 H4 þ H $ C3 H3 þ H2

ðR235Þ

and

C2 H2 þ HCCO $ C3 H3 þ CO

ðR205Þ

It is estimated that reactions (R193) and (R194) contribute to more than 90% of C3H3 production, with (R193) more than 65% while (R194) more than 25%. Moreover, these C3H3 forming pathways remain the same with or without DME addition. Formation of C3H3 in the centerline region close to the burner is mainly attributed to reaction (R205). The delayed appearance of C3H3 with the addition

5e-5 β=0 β = 0.0625 β = 0.125 β = 0.1875 β = 0.25 β = 0.375 β = 0.5 β = 0.75 β=1

4e-5

C3H3 Mole Fraction

peak mole fraction of C3H3 along the centerline decreases moderately, about 47%, between the pure ethylene flame and the pure DME flame. It is interesting to compare our calculated centerline distributions of propargyl shown in Fig. 17 with those reported by Bennett et al. [18]. The results of Bennett et al. showed that the mole fraction of C3H3 along the flame centerline increases monotonically, albeit slightly, with DME addition up to about 9% in the fuel stream. In addition, in their results there was no delay in the occurrence of C3H3 when DME was added. It is somewhat surprising to notice that their calculated peak mole fractions of C3H3 along the centerline are even slightly higher than the values shown in Fig. 17 (45 ppm vs. 38 ppm) at relative small amount of DME addition, since Bennett et al. [18] investigated the effects of DME addition to nitrogen-diluted ethylene flames. Despite these differences, the present results and those of Bennett et al. [18] are qualitatively similar in the sense that the peak mole fraction on the centerline remains approximately constant (for b as large as 0.5) and there is a slight enhancement in the mole fraction of C3H3 lower in the flame centerline. Although two different C2 reaction mechanisms were used in [18] and the present study, we may still speculate that the concentrations of C3H3 stay fairly constant in ethylene diffusion flames without or with dilution by nitrogen or DME. Analyses of the reaction rates of reactions responsible for the formation of C3H3 in the pure ethylene flame (b = 0) and the flame of b = 0.0625 indicate that C3H3 is primarily formed through

3e-5

2e-5

1e-5

0 0

1

2

3

4

5

6

7

z, cm Fig. 17. Comparison of the calculated propargyl mole fraction distributions along the flame centerline in the nine diffusion flames studied.

of relatively small amount of DME in the centerline region is associated with the decrease of temperature. While the somewhat small enhancement of the mole fraction of C3H3 with the addition of DME around z = 1.75 cm is due to the increase of the HCCO mole fraction in this region. 4.2.6. Methyl Methyl radical is another very important species in the present context. As mentioned above, the decomposition of ethylene produces primarily C2 species, such as C2H3, C2H2. Therefore, it is expected that the methyl radical concentration is very low in the pure ethylene diffusion flame. On the other hand, the dissociation of DME produces methyl radical and methoxy radical, which further dissociates primarily to formaldehyde [13,48], i.e.,

CH3 OCH3 $ CH3 þ CH3 O CH3 O $ CH2 O þ H

ðR705Þ ðR78Þ

It is expected that addition of DME enhances the concentration of methyl through its direct dissociation. The calculated 2D methyl mole fraction distributions in the nine diffusion flames are

561

F. Liu et al. / Combustion and Flame 158 (2011) 547–563

compared in Fig. 18. It is evident that the mole fractions of methyl continue to increase with increasing the fraction of DME in the fuel stream. It is interesting to observe from Fig. 18 that addition of DME not only increases the peak methyl mole fraction, but also leads to another branch of relatively high methyl concentration at lower axial locations within the branch in the pure ethylene flame, Fig. 18b–i. The concentrations of methyl in both branches are enhanced as more DME is added to the fuel stream, which is clearly shown in Fig. 19, which compares the methyl mole fraction distributions along the flame centerline. Without DME addition the centerline distribution of methyl mole fraction is single peaked. The distribution of methyl mole fraction becomes double peaked when DME is added over the entire range of DME fraction in the fuel stream considered in this study, i.e., b P 0.0625 up to 1. The methyl mole fractions increase significantly as more DME is added. At relatively small amount of DME addition up to about b = 0.5, the addition of DME increases the methyl mole fractions of the inner branch more than the outer branch, which is clearly shown in Fig. 19 along the centerline, especially at b = 0.0625. When more than 37% of DME is added, methyl disappears earlier at a lower centerline location, which is associated with the decrease in the mixture-strength flame height as more DME is added to the fuel stream. Analyses of the reaction rate for reactions leading to methyl formation in the pure ethylene flame indicate that reactions, in order of decreasing importance,

0.00025 β=0 β = 0.0625 β = 0.125 β = 0.1875 β = 0.25 β = 0.375 β = 0.5 β = 0.75 β=1

CH3 Mole Fraction

0.00020

0.00015

0.00010

0.00005

0.00000 0

1

2

CH4 þ OH $ CH3 þ H2 O

CH2 CO þ H $ CH3 þ CO

ðR152Þ

CH3 þ CH3 $ C2 H5 þ H

C2 H2 þ OH $ CH3 þ CO

ðR150Þ

and

ðR714Þ

8 7 6

z, cm

5 4 3 2 1 0 0

0.5

r, cm

1

CH3 5 4.875 4.75 4.625 4.5 4.375 4.25 4.125 4 3.875 3.75 3.625 3.5 3.375 3.25 3.125 3 2.875 2.75 2.625 2.5 2.375 2.25 2.125 2 1.875 1.75 1.625 1.5 1.375 1.25 1.125 1 0.875 0.75 0.625 0.5 0.375 0.25 0.125 0

(d) 18% DME 6.2425E-4

ðR120Þ

ðR102Þ

4.2.7. Other hydrocarbon species The calculated mole fractions of several other hydrocarbon species in the nine flames of different DME fractions in the fuel stream were also analyzed. Although these results are not shown here, it was found that addition of DME to ethylene has three different effects on various hydrocarbon species. These effects can be classified into (i) synergistic effect, (ii) monotonic increasing effect, and (iii) monotonic decreasing effect. More hydrocarbon species experience the synergistic effect in the ethylene/DME system.

ðR705Þ

(c) 12% DME 5.687E-4

(e) 25% DME 6.8145E-4

(f) 37% DME 7.9767E-4

(g) 50% DME 9.127E-4

(h) 75% DME 1.1239E-3

(i) 100% DME 1.2848E-3

8

8

8

8

8

8

8

8

7

7

7

7

7

7

7

7

6

6

6

6

6

6

6

6

5

5

5

5

5

5

5

5

4

4

4

4

4

4

4

4

3

3

3

3

3

3

3

3

2

2

2

2

2

2

2

2

1

1

1

1

1

1

1

1

0

0 0

0.5

r, cm

1

0 0

0.5

r, cm

1

0 0

0.5

r, cm

1

7

also make enhanced contributions to methyl formation in the outer branch of high methyl concentrations, corresponding to the second peak in Fig. 19.

are primarily responsible for the production of methyl in the pure ethylene flame shown in Figs. 18a and 19. Similar analyses conducted in the pure DME flame reveal that in addition to the three reactions identified in the pure ethylene flame, i.e., (R118), (R152), and (R150), for methyl production, several more reactions contribute to methyl formation. Reactions

(b) 6% DME 5.3596E-4

6

are responsible for the inner branch of high methyl concentrations shown in Fig. 18i and the first peak in Fig. 19 in the pure DME flame, while reactions (R118), (R152), (R150) and

and the reverse of

(a) Pure C2H4 Peak: 5.1913E-4

5

Fig. 19. Comparison of the calculated methyl mole fraction distributions along the flame centerline in the nine diffusion flames studied.

ðR118Þ

CH3 OCH2 $ CH2 O þ CH3

4

z, cm

CH4 þ H $ CH3 þ H2

CH3 OCH3 $ CH3 þ CH3 O

3

0 0

0.5

r, cm

1

0 0

0.5

r, cm

1

0 0

0.5

r, cm

1

0 0

0.5

r, cm

1

0

0.5

r, cm

Fig. 18. Comparison of the calculated methyl mole fraction distributions in the nine flames of different DME fractions in the fuel stream.

1

562

F. Liu et al. / Combustion and Flame 158 (2011) 547–563

Species experiencing the synergistic effect include at least C2H3, AC3H4, n-C4H3, C4H4, i-C4H5, n-C4H5, C4H6, l-C6H6, A1, A2, A3, and A4. As more DME is added to ethylene, from the pure ethylene flame to the pure DME flame, the mole fractions of CH3, CH4, and C2H6 increase monotonically, while the mole fractions of C2H2, C2H5, C3H3, and C4H2 decrease monotonically. It is noticed that the very slight increase in the peak mole fraction of C2H2 shown in Fig. 14 or in the centerline peak mole fraction of C3H3 shown in Fig. 17 is neglected in this classification of the DME effect. The synergistic effects of DME addition on many of the hydrocarbon species (including benzene), PAHs, and soot can be attributed to the interaction between C1 and C2 chemistry in the ethylene/DME system. The former (ethylene flame) has low concentrations of C1 species but high concentrations of C2 species, while the latter (the DME flame) is abundant in C1 species, in particular methyl. This is why the synergistic effect of a small amount of DME addition is quite significant, since it significantly boosts the concentrations of methyl in the fuel pyrolysis region. Under the addition of a large amount of DME to ethylene, the system has a lack of C2 species, which in turn suppresses the growth of large hydrocarbons and PAHs. 4.2.8. Pathways from methyl to benzene It was identified in Section 4.2.3 that the synergistic effect of DME addition to ethylene on benzene formation is primarily caused by the cyclization of l-C6H6 and n-C6H7. Therefore, it is important to understand how DME addition to ethylene leads to enhanced formation of l-C6H6 and n-C6H7. For this purpose, the distributions of reaction rates associated with the formation of l-C6H6 and n-C6H7 were analyzed for the pure ethylene flame (b = 0) and the flame of the smallest amount of DME addition (b = 0.0625). The findings are summarized here. As shown in Section 4.2.6, addition of DME significantly enhances the concentration of methyl radical, especially in the centerline region between about z = 1– 2 cm. The enhanced formation of methyl radical from the direct dissociation of DME interacts with ethylene to form additional vinyl through reaction C2H4 + CH3 M C2H3 + CH4 (R169), as pointed out earlier at the end of Section 4.2.4. The extra vinyl radicals then prompt the concentrations of C4H4 and n-C4H5 through their interactions with acetylene via reactions

C2 H2 þ C2 H3 $ C4 H4 þ H

ðR200Þ

and

C2 H2 þ C2 H3 $ n-C4 H5

ðR201Þ

Finally higher concentrations of l-C6H6 and n-C6H7 are produced from C4H4 and n-C4H5 through the following reactions

C4 H4 þ C2 H3 $ l-C6 H6 þ H

ðR282Þ

and

n-C4 H5 þ C2 H2 $ n-C6 H7

ðR293Þ

l-C6 H6 þ H $ n-C6 H7

ðR346Þ

It is therefore clear from the above discussions that C1 (methyl) and C2 (ethylene, vinyl, and acetylene) interaction plays a vital role in the synergistic effects of DME addition to ethylene on benzene formation. The interaction of C2 (C2H2 and C2H3) and C4 (C4H4 and n-C4H5) chemistry is important in this system to the formation of l-C6H6 and n-C6H7, but not directly to the formation of benzene. The direct pathways for the synergistic effect of benzene are the cyclization of l-C6H6 and n-C6H7 through reactions l-C6H6 + H M A1 + H (R348) and n-C6H7 M A1 + H (R354).

5. Conclusions An experimental and numerical study was conducted to investigate the effects of DME addition to fuel on PAH and soot formation in laminar coflow ethylene/air diffusion flames at atmospheric pressure. The entire range of DME addition from pure ethylene to pure DME was studied. Experimentally the relative concentrations of PAHs and the relative soot volume fractions were measured using the PLIF and 2D LII techniques, respectively. The experimentally investigated diffusion flames were also numerically simulated using a detailed gas-phase reaction mechanism and a PAH based sectional soot model. The PLIF and 2D LII images indicate that addition of DME to ethylene leads to synergistic effects on both PAHs and soot, with the synergistic effect on polycyclic aromatic hydrocarbons persists over a wider range of DME addition. The numerical results reproduce the synergistic effects of DME addition to ethylene on both PAHs and soot, though the magnitude of soot volume fraction overshoot and the range of DME addition associated with the synergistic effect of soot are less than those observed in the experiment. The synergistic effects of DME addition to ethylene on many hydrocarbon species, including PAHs, and soot can be fundamentally traced to the enhanced methyl radical concentrations from the direct dissociation of DME. The interaction of C2 and C4 chemistry is important in the ethylene/DME system to the formation of l-C6H6 and n-C6H7, but not directly to the formation of benzene. The self-combination reaction of propargyl is the most important reaction for benzene formation in the ethylene/DME system studied; however, the reaction rate of this reaction remains more or less unchanged over the range of DME addition where the synergistic effect of benzene formation is observed. Therefore, the additional amount of benzene, PAHs, and soot formed at a small amount of DME addition associated with the synergistic effect is not due to the self-combination reaction of propargyl. The pathways responsible for the synergistic effects of benzene formation with DME addition are primarily the cyclization reactions l-C6H6 + H M A1 + H and n-C6H7 M A1 + H and to a much less degree the interaction of C2 and C4 reaction n-C4H5 + C2H2 M A1 + H. These reaction rates are significantly enhanced when a relatively small amount of DME is added. Although the numerical results demonstrate the synergistic effect of DME addition on both PAHs and soot, discrepancies exist between the numerical results and the experimental observations with regard to the magnitude of these synergistic effects and at what fraction of DME in the fuel stream the synergistic effects peak. Such discrepancies may be attributed to deficiencies in the gas-phase reaction mechanism, in particular the benzene and PAH formation sub-set, as well as the soot inception model. More detailed and quantitative laser-induced incandescence measurements are required to provide both soot volume fraction and primary soot particle size distributions to gain further understanding on how DME addition affects the soot inception and surface growth processes. Acknowledgments This work was supported by the Science Fund of the State Key Laboratory of Automotive Safety of China (No. KF09031) and the National Natural Science Foundation of China (No. 50806039). References [1] C.A. Pope III, D.W. Dockery, Air Waste Manage. Assoc. 56 (2006) 709–742. [2] M.Z. Jacobson, J. Geophys. Rev. 107 (2002). ACH 16-1/16-22. [3] K.P. Schug, Y. Manheimer-Timnat, P. Yaccarino, I. Glassman, Combust. Sci. Technol. 22 (1980) 235–250. [4] B.S. Haynes, H.G. Wagner, Prog. Energy Combust. Sci. 7 (1981) 229–273. [5] C.S. McEnally, L.D. Pfefferle, Combust. Flame 112 (1998) 545–558.

F. Liu et al. / Combustion and Flame 158 (2011) 547–563 [6] S.M. Lee, S.S. Yoon, S.H. Chung, Combust. Flame 136 (2004) 493–500. [7] S.A. Zelepouga, A.V. Saveliev, L.A. Kennedy, A.A. Fridman, Combust. Flame 122 (2000) 76–89. [8] H. Richter, J.B. Howard, Prog. Energy Combust. Sci. 26 (2000) 565–608. [9] J.A. Miller, M.J. Pilling, J. Troe, Proc. Combust. Inst. 30 (2005) 43–88. [10] J. Wu, K.H. Song, T. Litzinger, S.-Y. Lee, R. Santoro, M. Linevsky, Combust. Sci. Technol. 178 (2006) 837–863. [11] J.Y. Hwang, W. Lee, H.G. Kang, S.H. Chung, Combust. Flame 114 (1998) 370– 380. [12] S. Trottier, H. Guo, G.J. Smallwood, M.R. Johnson, Proc. Combust. Inst. 31 (2007) 611–619. [13] S.S. Yoon, D.H. Anh, S.H. Chung, Combust. Flame 154 (2008) 368–377. [14] C.S. McEnally, L.D. Pfefferle, Proc. Combust. Inst. 31 (2007) 603–610. [15] R.J. Santoro, T.T. Yeh, J.J. Horvath, H.G. Semerjian, Combust. Sci. Technol. 53 (1987) 89–115. [16] Ü.Ö. Köylü, C.S. McEnally, D.E. Rosner, L.D. Pfefferle, Combust. Flame 110 (1997) 494–507. [17] Ö.L. Gülder, D.R. Snelling, R.A. Sawchuk, Proc. Combust. Inst. 26 (1996) 2351–2358. [18] B.A.V. Bennett, C.S. McEnally, L.D. Pfefferle, M.D. Smooke, M.B. Colket, Combust. Flame 156 (2009) 1289–1302. [19] H. Wang, M. Frenklach, Detailed mechanism and modeling of soot particle formation, in: H. Bockhorn (Ed.), Soot Formation in Combustion, Mechanism and Models, Series in Chemical Physics, vol. 59, Springer, 1994, p. 165. [20] J. Appel, H. Bockhorn, M. Frenklach, Combust. Flame 121 (2000) 122. [21] A. Mitrovic, T.-W. Lee, Combust. Flame 115 (1998) 437–442. [22] C.S. McEnally, L.D. Pfefferle, Combust. Flame 121 (2000) 575–592. [23] B.A.V. Bennett, C.S. McEnally, L.D. Pfefferle, M.D. Smooke, M.B. Colket, Combust. Flame 127 (2001) 2004–2022. [24] K.H. Song, P. Nag, T.A. Litzinger, D.C. Haworth, Combust. Flame 135 (2003) 341–349. [25] D.R. Snelling, K.A. Thomson, G.J. Smallwood, Ö.L. Gülder, Appl. Opt. 38 (1999) 2478–2485. [26] L.A. Sgro, P. Minutolo, G. Basile, A. D’Alessio, Chemosphere 42 (2001) 671–680. [27] K.K. Kuo, Principle of Combustion, Wiley, New York, 1986 (Chapter 3). [28] R.J. Kee, J.F. Grcar, M.D. Smooke, J.A. Miller, A Fortran Program for Modeling Steady Laminar One-Dimensional Premixed Flames, SANDIA Report, SAND858240, 1985.

563

[29] H. Guo, F. Liu, G.J. Smallwood, Ö.L. Gülder, Int. J. Comput. Fluid Dyn. 18 (2004) 139–151. [30] F. Liu, H. Guo, G.J. Smallwood, Combust. Flame 138 (2004) 136–154. [31] F. Liu, G.J. Smallwood, J. Quant. Spectrosc. Radiat. Transfer 84 (2004) 465–475. [32] J.Z. Wen, M.J. Thomson, M.F. Lightstone, S.N. Rogak, Energy Fuels 20 (2006) 547–559. [33] Q. Zhang, H. Guo, F. Liu, G.J. Smallwood, M.J. Thomson, Proc. Combust. Inst. 32 (2009) 761–768. [34] Q. Zhang, M.J. Thomson, H. Guo, F. Liu, G.J. Smallwood, Combust. Flame 156 (2009) 697–705. [35] S.N. Rogak, R.C. Flagan, J. Colloid Interface Sci. 151 (1992) 203–224. [36] M. Frenklach, H. Wang, Proc. Combust. Inst. 23 (1990) 1559–1566. [37] F. Xu, P.B. Sunderland, G.M. Faeth, Combust. Flame 108 (1997) 471–493. [38] H. Guo, F. Liu, G.J. Smallwood, Ö.L. Gülder, Combust. Flame 145 (2006) 324– 338. [39] S.H. Park, S.N. Rogak, W.K. Bushe, J.Z. Wen, M.J. Thomson, Combust. Theory Model. 9 (2005) 499–513. [40] D.F. Kronholm, J.B. Howard, Proc. Combust. Inst. 28 (2000) 2555–2561. [41] E.W. Kaiser, T.J. Wallington, M.D. Hurley, J. Platz, H.J. Curran, W.J. Pitz, C.K. Westbrook, J. Phys. Chem. A 104 (2000) 8194–8206. [42] J.A. Miller, S.J. Klippenstein, J. Phys. Chem. A 107 (2003) 7783–7799. [43] N. Hansen, J.A. Miller, T. Kasper, K. Kohse-Höinghaus, P.R. Westmoreland, J. Wang, T.A. Cool, Proc. Combust. Inst. 32 (2009) 623–630. [44] R.J. Kee, F.M. Rupley, J.A. Miller, Chemkin-II: A Fortran Chemical Kinetics Package for the Analysis of Gas-Phase Chemical Kinetics, Sandia Report SAND89-8009, Sandia National Laboratories, 1989. [45] R.J. Kee, G. Dixon-Lewis, J. Warnatz, M.E. Coltrin, J.A. Miller, A Fortran Computer Code Package for the Evaluation of Gas-Phase Multicomponent Transport Properties, Sandia Report SAND86-8246, Sandia National Laboratories, 1986. [46] Q. Zhang, H. Guo, F. Liu, G.J. Smallwood, M.J. Thomson, Combust. Theory Model. 12 (2008) 621–641. [47] K.L. McNesby, A.W. Miziolek, T. Nguyen, F.C. Delucia, R.R. Skaggs, T.A. Litzinger, Combust. Flame 142 (2005) 413–427. [48] H.J. Curran, W.J. Pitz, C.K. Westbrook, P. Dagaut, J.-C. Boettner, M. Cathonnet, Int. J. Chem. Kinet. 30 (1998) 229–241.