Effect O2 Co2 On Soot Formation Non-premixed Flame

Fuel 85 (2006) 615–624 www.fuelfirst.com

The effect of oxygen and carbon dioxide concentration on soot formation in non-premixed flames Kwang Chul Oh a,*, Hyun Dong Shin b b

a Enviromental Parts R&D Center, Korea Automotive Technology Institute, 74 yongjung-Ri, Pungse-Myun, Chonan, Chungnam 330-912, South Korea Department of Mechanical Engineering, Korea Advanced Institute of Science and Technology, 373-1 Guseong-dong, Yuseong-gu, Daejon 305-701, South Korea

Received 11 May 2005; accepted 16 August 2005 Available online 12 September 2005

Abstract The influence of oxygen concentration and carbon dioxide as diluents in the oxidizer side on soot formation was studied by Time Resolved Laser Induced Incandescence (TIRE-LII) and TEM photography in non-premixed co-flowing flames. TIRE-LII method was used to measure the distribution of two-dimensional soot volume fraction and primary particle size. The soot was directly sampled by the thermophoretic method, and its diameter was examined by TEM photography. Two suitable delay times of the TIRE-LII method affecting measurable range and sensitivity were determined by comparing TEM photographs with the TIRE-LII signal. The effects of oxygen concentration and carbon dioxide as diluents in the oxidizer side on soot formation were investigated with these calibrated techniques. An O2C(CO2, N2, and [ArCCO2]) mixtures in co-flow were used to isolate carbon dioxide effects systematically. The primary particle number concentration and soot volume fraction were abruptly decreased by the addition of carbon dioxide to co-flow. This suppression was resulted from the short residence time in inception region because of the late nucleation and the decrease of surface growth distance by the low flame temperature due to the higher thermal capacity and the chemical change of carbon dioxide. The increase of oxygen concentration in the co-flow caused an enhancement of soot nucleation and thus the residence time increase, but the specific growth rate showed almost the same value regardless of the co-flow mixture in the growth region. This result suggests that the specific growth rate has a weak dependence on the relative change of co-flow conditions in non-premixed co-flowing flames. q 2005 Elsevier Ltd. All rights reserved. Keywords: Soot; LII method; TEM; Carbon dioxide; Non-premixed flame

1. Introduction Recently many combustion systems such as EGR (Exhaust Gas Recirculation) systems and oxygen combustors with CO2 recycling have been proposed to reduce pollutant emissions and to enhance thermal efficiency. However, the proposed systems are mostly accompanied by drawbacks; namely, nitric oxide (NOX) and soot particles emissions [1– 3]. Even though there have been developments in the mechanisms of NOX formation and its reduction methods, there are still many ambiguities in the explanations of soot emission due to various changes in the features of soot aggregates (particle diameter, aggregate size, the number concentration of primary particles, soot volume fraction, etc.) in flames.

* Corresponding author. Tel.: C82 41 559 3089; fax: C82 41 559 3242. E-mail address: [email protected] (K.C. Oh).

0016-2361/$ - see front matter q 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.fuel.2005.08.018

A number of investigations have been focused on reducing soot emission by introducing several techniques, including the use of different fuels and addition of diluents [3–11] on the oxidizer or fuel side. However, it is still difficult to elucidate the effects of diluents on soot formation in flames because soot particle diameter, volume fraction, and the number density of particles are strongly dependent on location within a flame. [6, 11], Gu¨lder observed the effect of oxygen addition to the fuel side on soot volume fraction in methane, propane and n-butane diffusion flames and showed that the addition of oxygen to methane suppresses soot formation chemically, while oxygen addition enhances the soot formation by direct chemical interaction for propane and n-butane. Through the measurement of the soot inception limit, [8] reported that CO2, whether added to the fuel or oxidizer side, could suppress soot formation chemically in a counter diffusion flame. But because parameters that previous researchers had used to examine CO2 effects generally, those are, maximum soot volume fraction [1,11], soot inception limit [8] and smoke height [5], are affected by all the processes of soot formation which are inception, surface growth and oxidation, we cannot judge

616

K.C. Oh, H.D. Shin / Fuel 85 (2006) 615–624

neither where, nor how the effects independently act on apparently. Therefore, despite these advances in the understanding of soot formation, the work done to date does not provide a clear picture on the effect of addition of diluents due to the deficient information of the features of soot in flames. Several experimental techniques have been developed to identify the features of soot aggregates in flames. Optical techniques are generally considered to be the best one suitable for this purpose because of their non-intrusiveness; for example, light scattering [12] for aggregates size, various scattering and extinction methods [13,14] for measuring of both soot volume fraction and cluster structure, and multi-wavelength analysis [15] for soot volume fraction. Unfortunately, it is impossible to identify soot characteristics on the whole combustion field simultaneously with these methods because of their pointwise or line-of-sight character. Alternatively, laser induced incandescence (LII) techniques have been used successfully for twodimensional soot volume fraction measurements in a wide range of combustion processes [9,16]. Recently, there have been some attempts to combine LII, elastic scattering, and the TIRE-LII (time-resolved LII) method to measure soot characteristics simultaneously [1,17]. In this study, we tried to improve the measuring technique for primary particle diameter by proper determination of two delay times through a calibration procedure of the TIRE-LII signal with TEM photographs. With this calibrated technique, we obtained characteristic value of soot, such as soot volume fraction (fv), particle size (dp) and the number concentration of primary soot particles (Np), in laminar non-premixed flames. The effects of oxygen concentration (25–45%) in the oxidizer side, and diluents (N2, CO2, ArCCO2) were investigated in the soot inception and growth regions.

2. Methodology 2.1. LII and soot volume fraction The Laser Induced Incandescence (LII) method has been applied to the measurement of soot volume fraction (fv) in many combustion systems. LII involves the heating of soot particles to temperatures above the surrounding gas temperature via the absorption of laser energy, and subsequent detection of blackbody radiation corresponding to the elevated temperature of the soot particles. If one assumes a sufficiently loose structure of soot aggregates, which may be regarded as justified with the typical fractal dimension of soot aggregates in the range 1.6–1.8, and ignores gas radiation, the energy balance of soot is governed by the size of the primary pdp2 DHv dm Qabs Ei KLðT KTo Þpdp2 C 4 M dt ð pdp3 dT Z0 rC Kpdp2 3ðdp ; lÞMlb ðT; lÞdlK dt 6 where

(1)

Qabs, absorption efficiency Ei, irradiance L, heat transfer coefficient T, particle temperature To, surrounding gas temperature M, molar mass of solid carbon 3, emission coefficient Mlb , blackbody spectral radiant existence l, wavelength r, density of solid carbon C, specific heat of solid carbon. the primary particles to the following equation [18,19]. In order to improve soot measurement techniques such as LII and TIRE-LII methods, there have been many detailed examinations about the heat transfer paths in heated soot particles by many researchers [18–20]; the absorption of laser irradiance, the heat transfer to ambient gas, vaporization, blackbody radiation, and the change of particle temperature. Melton [18] showed that the intensity of the LII signal, SLII, for a group of soot particles has a dependence on primary particle diameter as follows SLII f Np dpx ; x Z 3 C 154=lem

(2)

where Np is the number concentration of primary particles, dp is the diameter of the primary particles, and lem is the measured wavelength. For lem between 700 and 400 nm, for example, the LII signal is proportional to the primary soot particle diameter raised to the power of 3.22–3.38, or approximately to the soot volume fraction. Ni and Quay [16,21] reported that initially the LII signal increases rapidly as laser power increases (linear region) and once laser power reaches a saturation threshold, the LII signal shows a small increase (saturation region). Fig. 1 shows the intensity of the LII signal to the laser fluence at different positions in a flame. The intensity of the LII signal shows the same trend as the result of Ni and Quay. The laser fluence in a saturated region was used to compensate a slightly inhomogeneous laser profile. In this experiment, a planar laser sheet was formed using a cylindrical lens and convex lenses, and the edge of the laser sheet, which is weaker in intensity, was cut off by a slit to obtain a sheet having a more uniform profile (I/ImaxO0.75 at test section). The fluence distribution of the laser sheet was confirmed with the Laser Rayleigh Scattering (LRS) method. Extinction method [13] was used to quantify the relative soot volume fraction obtained by the LII method. 2.2. TIRE-LII and primary particle diameter The TIRE-LII technique is based on the fact that after a laser pulse, smaller particles cool down faster than larger ones due to their larger specific surface [19,20,22]. There are three paths for the energy loss, which can be elucidated from the energy balance equation (Eq. (1)): vaporization, heat conduction to the surrounding gas, and radiation. Will et al. [19] showed that the temperature of soot particles abruptly changes within 100 ns

K.C. Oh, H.D. Shin / Fuel 85 (2006) 615–624

617

Fig. 1. Influence of laser fluence on LII signal at three points in annular or enveloped region.

after irradiance by a laser pulse due to the vaporization of soot and the temperature decay rates were governed by a conduction loss. Therefore the initial LII signal, within 100 ns, is similar regardless of the soot particle diameter, but the decay trend of the LII signal depends on the primary particle size in the region is governed by the conduction term due to its specific surface. The ratio of the LII signals, measured at two different delay times after the laser pulse irradiation, is known to be the function of soot diameter. STIRELII Z SLII2 =SLII1 wf ðdp Þ

(3)

Here, SLII1 is the LII signal at the first delay time and SLII2 is the LII signal at the second delay. The decay curves of the LII signal after the absorption of laser energy are shown in Fig. 2. The temporal profile of LII signal changes to a laser fluence; that is the parameter determining the initial particle temperature. As the laser fluence decreases, the cooling rate decreases due to the small temperature difference between soot particles and ambient, but there is little difference of the decay rate of LII signal above (a) 0.265 J/cm2 0.465 J/cm2 0.737 J/cm2 1.08 J/cm2

SLII/SLII, 100ns

0.8

position: 2 position: 3

1.0 0.8

0.6

0.6

0.4

0.4

0.2

0.2

0.0 0

200 400 600 800 1000 0

0.0 200 400 600 800 1000

time(ns) Fig. 2. Effects of laser fluence on the decay rate of LII signal.

SLII/SLII, 100ns

1.0

(b)

saturation region. In the present study, the error in measurement of soot particle size resulted from inhomogeneous heating profile is estimated withinG3 nm from the calibration curve (Fig. 6a). 3. Experimental detail The experimental setup consists of three parts as shown in Fig. 3; a co-flowing burner to make stable flames, a measurement systems for LII and TIRE-LII techniques and a thermophoretic soot sampling system to measure quantitative soot particle diameter. The inner diameter of the fuel nozzle is 5 mm and the contraction nozzle with a cut-area ratio of 1/6.5 around the fuel nozzle was used to obtain the uniform co-flowing velocity and to block the ambient air. CO2CO2, [0.61ArC0.39CO2]CO2, and N2CO2 mixtures in the oxidizer side (co-flow) were used to discriminate the CO2 effects. Blow-off of flame occurs at 19% of oxygen concentration and 27% of oxygen concentration in the case of N2 and CO2 dilution, respectively. Therefore, experiments carried out in the stable flame region, those are, 25–45% of oxygen concentration for the N2 dilution and 30–45% of oxygen concentration for the CO2 dilution. The adiabatic flame temperature (Tad) was used as the representative flame temperature. The experimental conditions and flame lengths are shown in Table 1. A second harmonic Nd:Yag laser (lZ532 nm, ImaxZ 500 mJ/pulse) was used as a laser source and its duration of pulse (w7 ns) was very short. A laser sheet of 40 mm width and 500 mm thickness was formed across the flame axis by using a cylindrical lens and a convex lens. The detection part consisted of a filter to obtain the monochromatic signal, 450 nm (G10 nm) in LII measurement and 532 nm (G10 nm) in Rayleigh scattering in the air-flow to confirm the uniformity of the laser sheet, an ICCD camera for two-dimensional measurement and a pulse generator for synchronization between the laser and the ICCD camera. The camera has

618

K.C. Oh, H.D. Shin / Fuel 85 (2006) 615–624

Fig. 3. Experimental apparatus: (a) thermophoretic sampling apparatus, (b) measuring system for LII and TIRE-LII techniques, (c) co-flowing burner.

a 1024*256 CCD array and its spatial resolution of this experiment was about 66 mm/pixel, which was sufficient to measure soot characteristics (dp, Np, fv) in detail. Laser power is one of the most important parameters in LII measurement. If excessive laser power is used, the LII signal at the laser incident side is smaller than the other side due to the vaporization of soot. On the contrary, when insufficient laser power is used, the opposite tendency occurs because of the attenuation of laser power by soot on the laser path [9,16]. A 0.737 J/cm2, mean laser fluence was used in this experiment to ensure the symmetry of soot volume fraction. 50 frames with 50 ns gate were accumulated in order to increase S/N ratio and flame luminosity was subtracted from the LII images. The soot-sampling device was composed of three parts, a component that fixes a TEM grid of 3 mm (200 mesh) diameter, a potentiometer to measure the location of the TEM grid, and a compressed air piston to move the grid. The moving speed of the TEM grid was controlled by the control of air pressure at the piston inlet. The soot was sampled by inserting a cold grid into the flame by the following procedures: rapid insertion (w1 m/s)/soot sampling (w0.2 s in the residence time)/rapid withdrawal (w1 m/s). The soot

morphologies were examined by TEM (Transmission Electron Microscopy, Philips Tecnal F20) with a magnification of 115 K, and the soot particle sizes were measured by image processing technique with 0.73 nm/pixel resolution. 4. Result and discussion 4.1. Soot particle size by TEM photography and TIRE-LII signal The soot particles were sampled using thermophoresis with the TEM grid in flames of 35% oxygen concentration with N2 or CO2 dilution in the co-flow to calibrate the relative distribution of soot particle size obtained by the TIRE-LII method. Figs. 4 and 5 shows TEM photographs for the case of N2 and CO2 dilution at the position of maximum soot volume fraction along the axial direction. Each soot aggregate consists of individual particles and the particle diameters at a given specified location within the flame show a statistical distribution with narrow band. As shown in Fig. 4, the diameters of the primary particles are strongly dependent on

Table 1 Experimental conditions Fuel Co-flow N2CO2

[0.61ArC0.39CO2]C O2 CO2CO2

Xo2 Tad (K) Hf (mm) Tad (K) Hf (mm) Tad (K) Hf (mm)

Propane (Z1.32 cc/s), nozzle exit velocityZ6.7 cm/s Dilluents (N2, CO2, Ar)CO2; (Z20 l/min) Coflow exit velocityZ48.5 cm/s 0.25 0.30 0.35 2424 2563 2663 32.4 26.2 21.1 2346 2479 2581 31 24.04 20.1 2197 2334 28.95 23.3

0.40 2739 17.9 2662 18 2447 20.6

Hf, visible flame length; Tad, adiabatic flame temperature; Xo2 , oxygen concentration in oxidizer side. [0.61ArC0.39CO2] equivalent to N2 in terms of rCp (298–1100 K).

K.C. Oh, H.D. Shin / Fuel 85 (2006) 615–624

619

Fig. 4. TEM photographs of soot aggregates and particle size distribution at the position of maximum soot volume fraction along the axial direction in the N2CO2 (XO2 Z0.35) co-flow flame with magnification of 115,000.

the location within the flame. At 4 mm from the nozzle, small aggregates start to appear and numerous small (w2 nm) soot particles, which are not yet agglomerated, surround the aggregates. As a distance far from the nozzle, the primary particle diameter becomes large and the aggregate becomes larger one through the agglomeration process. The diameter of the primary particles reaches its maximum value (w35 nm) at about 14 mm from the nozzle and decreases by oxidation of soot. When CO2 was used as diluents in oxidizer side, a similar tendency with that observed in the N2 case was observed

except the different maximum soot particle diameter, that was, 26 nm. The TIRE-LII method was applied to flames under the same conditions as those with TEM photography. As explained previously, the intensity of the TIRE-LII signal is a function of the primary particle size governed by heat transfer to the surrounding gas. Therefore, a planar distribution of soot particle size can be obtained if the calibration between the diameter and the signal intensity is properly performed.

Fig. 5. TEM photographs of soot aggregates and particle size distribution at the position of maximum soot volume fraction along the axial direction in the CO2CO2 (XO2 Z0.35) co-flow flame with magnification of 115,000.

K.C. Oh, H.D. Shin / Fuel 85 (2006) 615–624

from Will et. al.'result(1998, applied optics) t2=400ns t2=600ns t2=550ns, interpolation 400ns and 600ns

signal ratio(LIIt2/LII100ns)

0.4

dp,TEM=7.4 + 84.03Ssignal ratio present study(t2=550ns) linear curve fitting by least square method (σ: 2.77 nm) experimental result by TEM and TIRE-LII

0.3

0.2

0.1

50

Primary soot particle size(nm)

620

40

N2+O2 case : TEM CO2+O2 case : TEM

:TIRE-LII :TIRE-LII

30

20

10

0.0 10

15

20

25

30

35

soot particle diameter(nm) (a)

40

3

6

9

12

15

18

21

distance from the nozzle(mm) (b)

Fig. 6. Calibration curve between TEM photography and TIRE-LII signal in the CO2, N2CO2 co-flowing condition (XO2 Z0.35). (a) Comparison of calibration curve between present work and Will’s numerical result, (b) Primary particle size along the axial direction from TEM and TIRE-LII methods.

The calibration curve of the present study and previous result [19] are shown in Fig. 6a. At t2Z400 ns of Will’s results the slope of the signal ratio gradually decreases in large diameter. That was resulted from the small temperature difference in large particles due to the relatively short duration of heat transfer against the heat capacity of particles and same trend is shown in small particle region at t2Z600 ns due to long duration. This reveals that the calibration curve has a linear relation according to selection of second delay time and the measuring range of particle size. From Fig. 6a, the relation between the particle diameter and TIRE-LII signal has a linear tendency with a slope of Ddp/DSTIRE-LII and an intercept that is the minimum measurable diameter in the moderate particle size range (8– 40 nm) at t2Z550 ns (sZ2.77 nm). In this experiment if t2 is smaller than 550 ns, the calibration curve deviated from linear profile in large particle region and when t2 is larger than 550 ns, the deviation occurred in small particle region. The primary soot particle diameter in the axial direction is shown in Fig. 6b. The soot particle size measured by the laser method is in good agreement with that measured by TEM photographs, though some deviations are observed in oxidation regions, likely resulting from the aerosol process of cluster– cluster aggregation (CCA), as shown in previous TEM photographs. Through the CCA process, the soot aggregates become dense and form a compact structure, and thus the cooling rate of soot aggregates can decrease because of the decrease of specific surface. This result is consistent with Vander Wal et al.’s work [22]. 4.2. Distribution of soot volume fraction (fv), primary particle size (dp), and number concentration of primary particle (Np) Soot volume fraction and primary soot particle size were measured by using the calibrated LII and TIRE-LII methods, respectively, and the number concentration of primary particles was calculated from soot volume fraction and primary particle size according to the following relation, Np Z ð6fv =pdp3 Þ103 , particles/mm3.

Fig. 7 shows the radial distributions of soot volume fraction, primary particle size and the number concentration of primary particles in the N2CO2 case at (a) zZ7, (b) 11, (c) 16 mm and a simultaneous OH LIF and LII image. At zZ7 mm, the distribution of soot is symmetrical at the maximum point ðjrjZ rfv ;max Þ of soot volume fraction. The distribution of primary particle size is almost flat, and thus the number density seems to affect the distribution of soot volume fraction. At zZ11 mm, fv, dp and Np abruptly decrease in the outer region ðjrjO rfv ;max Þ of the maximum point of soot volume fraction due to the oxidation of soot by OH radicals, but the number density of primary particles increases in the inner region ðjrj! rfv ;max Þ through the nucleation process. The soot volume fraction reveals a maximum value at the maximum position of the primary particle size. This result was shown that the increase of soot volume fraction was caused by the growth of primary particle not nucleation at the maximum position of soot volume fraction. At zZ16 mm, the maximum of the primary particle size still lies on the side of the sooting zone, but the size is smaller than that of zZ11 mm. And the soot volume fraction increases in the enveloped region (at center) through active nucleation due to the increase of temperature. Santoro et al. [12] argued that the particle trajectory coincides with the line showing the maximum of the soot volume fraction. Examining changes of fv, dp and Np along the line showing the maximum of soot volume fraction, as shown in Fig. 7, the characteristics of soot distribution can be qualitatively categorized into three parts. The first region where the primary particle number density increases while the primary particle size is kept small is called the inception region. In this region, the soot volume fraction increases due to soot inception. The surface growth region follows, where the primary particle number density is nearly constant while primary particle size and soot volume fraction increase due to the condensation of growth species, C2H2 and/or PAH (Poly Aromatic Hydrocarbon), on the soot particle surface. The last is the oxidation region, where soot volume fraction and soot

K.C. Oh, H.D. Shin / Fuel 85 (2006) 615–624

621

Fig. 7. Radial distribution of soot characteristics in the N2CO2 (XO2 Z35%) case (a) soot inception region, (b) surface growth region, (c) soot oxidation region.

particle size decrease due to the attack of soot particles by OH and O radicals [23]. 4.3. Inception and growth region of soot Fig. 8 shows the distributions of primary soot particles (dp), soot volume fraction (fv), and the number concentration of

primary particles (Np) at 30% oxygen concentration in the coflow. In the case of using CO2 as the diluents instead of N2, the primary particle size and soot volume fraction decrease abruptly while the primary particle number concentration increases at the enveloped region, as shown in Fig. 8a and b. These changes result from the effect of CO2 addition, which are temperature reduction, dilution of reactive species and direct

Fig. 8. Soot volume fraction, primary particle size, and the number concentration of primary particles distribution in 30% oxygen volume fraction (a) N2CO2, (b) CO2CO2, (c) (ArCCO2)CO2 case in co-flow (*residence timeZDz/Uz, where Uz is the instant velocity at Z).

622

K.C. Oh, H.D. Shin / Fuel 85 (2006) 615–624

chemical effect. But it is impossible to completely isolate the CO2 effects experimentally because these effects occur simultaneously. Therefore, we examined the CO2 effects by dividing two parts, the change of physical property (rCp) and the chemical change of CO2. In order to compensate the thermal capacity variation of CO2, N2 was substituted with CO2 and Ar to be equivalent in terms of specific heat capacity (rCp) at 298–1100 K. In this ArCCO2 diluents case although the inception point in this experiment, which is determined by the measurable minimum soot particle size of TIRE-LII method (w8 nm), is slightly retarded about 12% in distance from nozzle, this case is similar to the N2 diluents case. Therefore the difference of Fig. 8b from c results from the higher heat capacity (rCp) of CO2 and the 12% retardation of a nucleation point (Fig. 8a and c) results from the chemical change of CO2. The flames in this experiment are controlled by a buoyancy force (Fr!!1). It is known that the axial velocity of a buoyancy-controlled flame has a constant acceleration regardless of the location of the path lines, even though there is a slight difference in the magnitude. Roper [24] and Santoro [12] showed through velocity measurements in various flames that the velocity acceleration in flame is somewhat constant about 25 m/s2. Therefore, in this study the acceleration due to buoyancy, a, is assumed to be constant (25 m/s2) and the axial velocity, Uz, is given as qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 C 2az; U Uz Z Uz;o (4) z;o : fuel exit velocity

fZ

soot volume fraction(ppm) soot particle diameter(nm) the number concentration of primary particle*5(#/µm3)

35 30 25 20 15 6 4 2 0 0.020

0.025

0.030 t(sec)

0.035

ddp rsoot Uz ddp r Z soot 2 dz 2 dt

(5)

where rsoot is soot density (Z1.86 g/cm3 (Megardis, 1988)). The specific growth rate is calculated from Fig. 9a data and plotted in Fig. 9b with previous results; Harris and Weiner [25] for a ethylene–air premixed flame with a C/O ratio of 0.76, Santoro [12] for an ethylene–air diffusion flame (flame length: 88 mm) with optical technique and Megaridis and Dobbins [26] with thermophoretic sampling in the same flame as Santoro’s experiment. The present flame, a propane flame with 30% oxygen concentration in the co-flow, has a short surface growth region and the drastic change of the specific growth rate as compared to other previous works because of a short flame length and an enriched oxygen fraction in co-flow. As shown in Fig. 9b the specific growth rate is larger than the premixed case and the value of specific growth rate is comparable to the results of diffusion flame. Fig. 10 shows the primary particle size and the specific growth rate along the path-line exhibiting the maximum soot volume fraction to the change of oxygen concentration and diluents in the co-flow. The growths of primary particle size to the elapsed time have a similar trend in all cases though growth distances vary due to temperature field [5] or flame length. And the specific growth rate shows almost the same value regardless of the co-flow mixture at any position in growth region, even though it had a slight deviation of about 20% according to the increase of the oxygen concentration in the co-flow. This result suggests that the specific growth rate has a relatively weak dependence on the change of co-flow conditions in diffusion flame. Maximum primary particle diameter (dp,max), maximum soot volume fraction (fv,max), and maximum integrated soot volume fraction (fIv,max) according to the oxygen concentration

Specific Growth Rate(g/cm2 sec)

fv (ppm), dp (nm), Np (number/µm3)

An axial velocity and a residence time are shown in Fig. 8. The retardation in ArCCO2 case causes the number concentration of the primary particles to decrease in the soot inception region in Fig. 8a and c (Np distribution) for a relative short residence time during nucleation process. And these changes also decrease the growth of soot particle size in the initial stage through the coagulation process, where Np decreases and thereby decreases the total surface area of the soot particles. Fig. 9shows fv, dp and Np along the path-line and the specific growth rate in the N2 dilution case (XO2 Z30%). In this region, fv and dp increase gradually while Np is nearly constant. This

suggests that the increase of soot volume fraction along the path-line is mainly affected by surface growth rather than by nucleation. The specific growth rate, f, was defined as the mass deposition rate per unit of particle surface area to compare soot growth rate in flames under different nucleation conditions, as referred to in the previous paragraph.

10–3

0.02

0.03

0.04

0.05

Present work(in N2+O2:30% in coflow)

1x10–4

1x10–5

Harris & Weiner Megaridis & Dobbins Santoro

10–6 0.02

0.03

0.04

0.05

t(sec)

Fig. 9. Soot characteristics and specific growth rate along the path-line (a) fv, dp and Np at N2CO2: 30% in the co-flow, (b) specific soot surface growth rate.

K.C. Oh, H.D. Shin / Fuel 85 (2006) 615–624

623

10–3

Increase 1x10–4

Specific Growth Rate(g/cm2 sec)

Soot Particle Diameter(nm)

35 30 25 20 15 10 0.020

0.025

0.030

0.035

t (sec)

1x10–5 10–6 10–3 1x10–4 1x10–5

[Ar+CO2]+O2 : 35% [Ar+CO2]+O2 : 30% [Ar+CO2+O2 : 25%

10–6 10–3 1x10–4 1x10–5 10–6

Specific growth rate ~ ∆dp/ ∆t=slope

N2+O2 : 35% N2+O2 : 30% N2+O2 : 25%

CO2+O2 : 35% CO2+O2 : 30% 0.020

0.025 0.030 t (sec)

(a)

0.035

(b)

Fig. 10. Primary particle size and the specific growth rate to the change of oxygen concentration and diluents in co-flow, (a) change of soot particle diameter (b) the specific soot surface growth rate.

40

soot particle diameter[nm]

90

N2+O2 [Ar+CO2]+O2 CO2+O2

80

35 30

70

25 maximum soot volume fraction[ppm] N2+O2 20 [Ar+CO2]+O2 CO2+O2 15 maximum 6

60 50 40

integrated soot volume fraction

4 2 0.10

N2+O2 [Ar+CO2]+O2 CO2+O2

30

maximum integrated soot volume fraction[ppm*mm2]

maximum soot volume fraction[ppm] primary particle size(nm)

and diluents are shown in Fig. 11. As the oxygen concentration in the co-flow increases, the local maximum soot volume fraction (fv,max) increases due to the increase of nucleation rate and in CO2 diluents case the maximum soot volume fraction is smaller than N2 case because of the retardation of nucleation region by the low flame temperature from the higher thermal capacity (comparison between CO2 and ArCCO2 cases) and the chemical change of CO2 (comparison between ArCCO2 and N2 cases). But the maximum of the integrated soot volume fraction (fIv,max) decreases because the flame volume containing soot is small due to the enhanced oxidation rate by OH radical oxygen. The maximum primary particle diameter increases in a certain oxygen fraction in the co-flow, 25–40% in the N2 dilution case, and then decreases in high oxygen concentration as shown in Fig. 11. This tendency is resulted from the competition between early nucleation (nucleation) and growth

20 0.15

0.20

0.25

0.30

0.35

0.40

0.45

oxygen concentration in coflow (XO2) Fig. 11. Maximum soot volume fraction, maximum integrated soot volume fraction and maximum primary particle diameter to the oxygen concentration in ÐR co-flow (* Integrated soot volume fraction, fIv ðzÞZ 2p fv ðr; zÞr dr, which is a measure of total amount of soot present at a particular0height in flame).

distance (surface growth). As the oxygen concentration increases, the distance of growth region is getting short but the residence time increase because of early nucleation. 5. Concluding remarks The influences of oxygen concentration and diluents in the oxidizer side on soot inception and growth were examined by a combined LII and TIRE-LII method in propane non-premixed flames. (1) We measured soot particle diameter with both thermophoretic sampling and TIRE-LII techniques to refine the TIRE-LII technique. With these calibrated techniques, the soot volume fraction and the primary soot particle size were measured for N2 and CO2 cases. The soot particle size measured by the laser method is in good agreement with that measured by TEM photography, though some deviations observed in oxidation regions likely resulted from the dense structure and the aerosol process of cluster – cluster aggregation (CCA). (2) In the case of using CO2 as a diluents instead of N2, the primary particle size and soot volume fraction abruptly decrease. The suppression of soot volume fraction and the reduction of soot particle size in the CO2 diluted case was resulted from the processes of the short residence time in inception region because of late nucleation and the decrease of surface growth distance by the low flame temperature resulted from the higher thermal capacity and the chemical change of CO2. (3) The specific growth rates were calculated to compare soot growth rates in flames of different nucleation condition. The specific growth rate shows almost the same value regardless of the co-flow mixture even though it had a slight deviation of about 20% according to the increase of oxygen

624

K.C. Oh, H.D. Shin / Fuel 85 (2006) 615–624

concentration in the co-flow. This result suggests that the specific growth rate has a relatively weak dependence on the change of co-flow conditions. Although there are uncertainties regarding flame temperature and chemical composition, the present experimental investigations supply detailed information on soot characteristics in flames. Acknowledgements This research was supported by the Korea Science and Technology Foundation (KOSEF) through the Combustion Engineering Research Center (CERC). References [1] Angrill O, Geitlinger H, Streibel T, Suntz R, Bockhorn H. Influence of exhaust gas recirculation on soot formation in diffusion flames. Proc Combust Instit 2000;28:2643. [2] Lapuerta M, Salavert JM, Domenech C. Modelling and experimental study about the effect of exhaust gas recirculation on diesel engine combustion and emission, SAE paper, 95-0216. [3] Shimazaki, N., Hatanak, H., Yokota, K., and Nakahira, T. A study of diesel combustion process under the condition of EGR and high-pressure fuel injection with gas sampling method. SAE paper 96-0030. [4] Zhang C, Atreya A, Lee L. Sooting structure of methane counterflow diffusion flames with preheated reactants and dilution by products of combustion. Proc Combust Instit 1992;24:1049. [5] Glassman I. Sooting laminar diffusion flames: effect of dilution, additives, pressure, and microgravity. Proc Combust Instit 1998;27:1589. ¨ L. Effects of oxygen on soot formation in methane, propane, and [6] Gu¨lder O n-butane diffusion flames. Combust Flame 1995;101:302. ¨ , L. The chemical effects of carbon [7] Liu F, Guo H, Smallwood GJ, Gu¨lder O dioxide as an additive in an ethylene diffusion flame: implications for soot and NOx formation. Combust Flame 2001;125:778. [8] Du DX, Axelbaum RL, Law CK. The influence of carbon dioxide and oxygen as additives on soot formation in diffusion flames. Proc Combust Instit 1990;23:1501. [9] Ni T, Gupta SB, Santoro RJ. Suppression of soot formation in ethane laminar diffusion flames by chemical additives. Proc Combust Instit 1994; 25:585.

[10] Hainsworth Dougalas, Pourkashanian M, Richadson Andrew P, Rupp Joanne L, Williams Alan. The influence of carbon dioxide on smoke formation and stability in methane–oxygen–carbon dioxide flames. Fuel 1996;75(3):393. ¨ L, Snelling DR, Sawchuk RA. Influence of hydrogen addition to [11] Gu¨lder O fuel on temperature field and soot formation in diffusion flames. Proc Combust Instit 1996;26:2351. [12] Santoro RJ, Yen TT, Horvath JJ, Semerjian HG. The transport and growth of soot particles in laminar diffusion flames. Combust Sci Tech 1987;53:89. [13] Santoro RJ, Semerjian HG, Dobbins RA. Soot particle measurements in diffusion flames. Combust Flame 1983;51:203. [14] Santoro RJ, Semerjian HG. Soot formation in diffusion flames: flow rate, fuel species and temperature effects. Proc Combust Instit 1984;20:997. [15] Iuliis S, De Barbini M, Benecchi S, Cignoli F, Zizak G. Determination of the soot volume fraction in an ethylene diffusion flame by multiwavelength analysis of soot radiation. Combust Flame 1998;115:253. [16] Ni T, Pinson JA, Gupta S, Santoro RJ. Two-dimensional imaging of soot volume fraction by the use of laser-induced incandescence. Appl Opt 1995;34:7083. [17] Will S, Schraml S, Leipertz A. Comprehensive two dimensional soot diagnostics based on laser-induced incandescence (LII). Proc Combust Instit 1996;26:2277. [18] Melton LA. Soot diagnostics based on laser heating. Appl Opt 1984;23: 2201. [19] Will S, Schraml S, Bader K, Leipertz A. Performance characteristics of Soot primary particle size measurements by time-resolved laser-induced incandescence. Appl Opt 1998;37:5647. [20] Schraml S, Dankers S, Bader K, Will S, Leipertz A. Soot temperature measurements and implications for time-resolved laser-induced incandescence (TIRE-LII). Combust Flame 2000;120:439. [21] Quay B, Lee TW, Ni T, Santoro RJ. Spatially resolved measurements of soot volume fraction using laser-induced incandescence. Combust Flame 1994;97:384. [22] Vander Wal RL, Ticich TM, Stephens B. Can soot primary particle size be determined using laser-induced incadescence? Combust Flame 1999;116:291. [23] Kennedy Ian M, C. Yam, Rapp DC, Santoro RJ. Modeling and measurements of soot and species in laminar diffusion flame. Combust Flame 1996;107:368. [24] Roper FG. Prediction of laminar jet diffusion flame size: Part I—theoretical model. Combust Flame 1977;29:219. [25] Harris SJ, Weiner AM. Surface growth of soot particles in premixed ethylene/air flames. Combust Sci Tech 1983;31:155. [26] Megaridis CM, Dobbins RA. Soot aerosol dynamics in laminar ethylene diffusion flame. Proc Combust Instit 1988;22:353.