MEASUREMENTS OF LAMINAR BURNING VELOCITIES AND FLAME STABILITY ANALYSIS OF BIOMASS DERIVED GAS AIR PREMIXED FLAMES

Transcription

1 MCS 7 Chia Lagna, Cagliari, Sardinia, Italy, September 11-15, 211 MEASUREMENTS OF LAMINAR BURNING VELOCITIES AND FLAME STABILITY ANALYSIS OF BIOMASS DERIVED GAS AIR PREMIXED FLAMES Won Sik Song*, Tran Manh V*, Jeong Park*, Oh Boong Kwon*, Hyn Seok Yo**, Jin Han Yn*** jeongpark@pkn.ac.kr *School of Mechanical Engineering, Pkyong National University, San 1, Yongdang-dong, Nam-g, Bsan , Korea **New Energy & Environmental Team, R&D Center, Korea Gas Corporation, 638-1, Il-dong, Sangrock-g, Ansan , Korea ***Environment & Energy Research Division, Korea Institte of Machinery and Materials, 171 Jang-dong, Yseong-g, Daejeon , Korea Abstract Three biomass derived gases (BDGs, named GG-H, GG-L and GG-V), which are derived from indstry facilities and can be sefl for redcing CO 2 and the application to combstors, are stdied and examined for some basic flame characteristics sch as nstretched laminar brning velocity, Markstein length, and cell formation over the entire flame srface. Experiments were condcted in a constant pressre combstion chamber sing a schlieren system. A better agreement between the measred and predicted nstretched laminar brning velocities is obtained sing a sggested reaction mechanism modified from the GRI-Mech 3. mechanism. Additionally, cell formations on flame srfaces of the three mixtres were also analyzed and compared sing high-speed schlieren images. It is shown that the GG-H air flames and the GG-L air flames have similar flame wrinkled srfaces, while the GG-V air flames shows a stronger celllarity behavior. The effects of each fel component in mixtres to celllarity are also evalated by varying the concentration of each fel in the reactant mixtres. The celllar instability is promoted (diminished) with hydrogen enrichment (methane addition); meanwhile the similar behavior is obtained for carbon monoxide addition. Introdction In recent years, for the prposes of redcing of CO 2 emissions dring combstion of fossil fels and removing of wastes de to environmental and health concerns, biomass derived gas (BDG) has been widely sed to replace fossil fels in combstion engines. The most serios barrier in designing combstion engines sing BDGs arises from a large variety of compositions prodced from varios biomass sorces and different techniqes. BDGs mainly compose of H 2, CO, CH 4, CO 2, N 2, and can occasionally contain a small amont of higher hydrocarbons [1]. Becase of the small lower heating vale (LHV) of BDGs, they can ordinary be brned together with natral gas to achieve a higher efficiency of internal combstion engine [2]. In literatre, there have been a lot of reports on the fndamental characteristics of pre fel premixed flames. So far, most stdies related to biomass have been rather limited to processes technologies sch as gasification and pyrolysis [3 5]. It is rather srprising that the flame characteristics of BDGs flames have not been yet nderstood thoroghly. The deep nderstanding in combstion behaviors of BDGs air flames is very important in designing of combstion devices, validating and developing the chemical kinetic mechanisms, as well as predicting the performance and emission of combstion systems. In this work, three BDGs namely GG-H, GG-L and GG-V which are derived from indstry facilities and can be sefl

2 Table 1. Compositions of biomass derived gases stdied in the measrements. BDGs CO CO 2 H 2 CH 4 N 2 Biomass Ref. GG-H Celllose [3] GG-L Pine wood [4] GG-V Crde glycerol [5] for redcing CO 2 and the application to combstion devices [2], are stdied and examined for some basic flame characteristics sch as nstretched laminar brning velocity, Markstein length, and cell formation over the entire flame srface. The GG-H represents the gasification gas reported by Hanaoka et al. [3], the GG-L denotes the gasification gas given by Lv et al. [4], and the GG-V is the gasification gas stdied by Valliyappan et al. [5]. The compositions of the three BDGs are listed in Table 1. The most important parameter in designing premixed combstion engines is laminar brning velocity, in that it determines the strctre and the flame stability. The behavior of cell formation in premixed flames is another important factor in the side of increasing the flame speed and leading to the engine knock. The celllar instabilities in premixed flames can reslt from body-force, diffsive-thermal and hydrodynamic effects [6]. The diffsive-thermal effect is cased by the non-similar diffsion of mass verss heat and has a destabilizing inflence for Lewis nmbers of the deficient reactant lower than nity. The hydrodynamic effect is cased by the thermal expansion throgh the flame front and is the most essential factor in the flame instability [7]. To clarify the flame characteristics of the BDGs air mixtres, this stdy focses on measrements of the laminar brning velocities and Markstein lengths for a wide range of eqivalence ratios in the three BDGs air (i.e. GG-H, GG-L, GG- V) premixed flames. A revised mechanism of the GRI-Mech 3. is presented throgh modifying rate coefficients of some key reactions. The behaviors of cell formation of the three BDGs air flames with and withot adding 1%, 2% by volme of each fel component (i.e. H 2, CO, CH 4 ) are also discssed. All the experiments were done at room temperatre and elevated pressres sing the centrally ignited, otwardly propagating spherical flames method. This method yields highly accrate reslts for both laminar flame speeds and celllar flame instabilities and can easily accont for high initial pressres and high initial temperatres [8]. Experimental facility and data analysis Experimental method A fll, detailed description of the experimental facility is available in V et al. [9 11], so only a brief overview is provided here. The total layot of the facility can be seen in Fig. 1. The constant-pressre, otwardly propagating spherical flame was sed in this stdy to collect the experimental data. The internal cylindrical chamber has 22-mm diameter and 22-mm length, with two 1-mm diameter, 4-mm thickness qartz windows monted on both flat opposite sides of the chamber for flame observation. The reactant mixtres were spplied by adding individal component gases at corresponding partial pressres sch that a desired initial chamber pressre cold be obtained. The mixtres were centrally ignited by creating a spark across two.5-mm diameter electrodes. The propagating spherical flames were imaged sing a schlieren system with a 3-W halogen light sorce and a pair of 15-mm diameter spherical concave mirrors and were recorded sing a high-speed digital camera (Phantom v7.2) operated at 1, fps. The measrements were restricted to flames with radii larger than 6 mm and smaller than 3 mm to avoid both ignition distrbances and pressre increasing more than 1% of the initial pressre.

3 Figre 1. Schematic representation of the experimental setp. Nmerical modeling In the present stdy the steady, one-dimensional, laminar premixed flame code PREMIX [12] was sed to predict the nstretched laminar brning velocity, which was then compared to experimental data. The GRI-Mech 3. mechanism [13], which consists of 53 species and 325 elementary reaction steps, was adopted for the nmerical simlations. This model was chosen becase it incldes all of the species reqired in this stdy. However, the GRI-Mech 3. is primarily designed to describe methane combstion, therefore it is reqired to be revised and modified to get a better agreement with the experimental reslts by replacing rate coefficients in Arrhenis form of some important reactions with ones from several referential reaction mechanisms in literatre. The mechanisms referred here are the optimized H 2 /CO mechanism of Davis et al. [14], the comprehensive C 1 mechanism of Li et al. [15], the C 1 C 3 model socalled the San Diego mechanism [16], the CO/H 2 mechanism developed by Sn et al. [17], the H 2 /CO/C 1 C 4 model, USC Mech Version II, developed by Wang et al. [18], and the mechanism developed by Konnov for small hydrocarbons flames [19]. Laminar brning velocity and Markstein length of biomass derived gases air calclation In the present stdy the constant pressre method was sed for the otwardly propagating spherical flame. The detailed calclated process can be fond elsewhere [1,11]. The flame front has a propagation velocity dr dt, where R is the instantaneos radis of the flame in the schlieren photograph and t is the elapsed time from spark ignition. Flame stretch rate of a spherical flame srface can be expressed as [1,11] 1 da 2 dr = = (1) 1 K A 1 d t R d t where A 1 is the srface area of the flame front. For weakly stretched flames, a linear relationship between the stretched flame speed and the stretch rate exists that is qantified by a brned gas Markstein length, L b [2,21] S1 Sn = Lb K (2) where S 1 is the nstretched flame speed with respect to the brned mixtre. Base on the plot

4 of S n K from Eq. (2), S 1 can be obtained as the intercept vale at K = and L b is the negative vale of the slope of the S n K line. The nstretched laminar brning velocity with respect to the nbrned mixtre, S, is given throgh mass conservation S ρ = S b 1 ρ (3) where ρ and ρ b are densities of the nbrned and brned mixtres, respectively. Important factors for celllar instabilities The celllar instabilities of the three BDGs air premixed flames were identified and evalated with respect to hydrodynamic and diffsional-thermal instabilities. In the present stdy bodyforce effect was not significant and cold be neglected becase the laminar brning velocities of the flames mentioned in this stdy are large enogh sch that the flames can overcome the impact of the body-force factor. Initially, celllar instabilities are sppressed by the strong crvatre associated with a small flame radis. However, as the flame expands and flame stretch decreases, a state is reached in which the cell development can no longer be sppressed, and cells will appear almost instantaneosly over the entire flame srface. The inflences of the hydrodynamic and diffsive-thermal effects were mentioned in detail in Ref. [11]. The hydrodynamic effect is cased by the thermal expansion ratio throgh the flame front, σ = ρ ρb, and laminar flame thickness, l f. The laminar flame thickness is a characteristic length scale which is given by Law et al. [22] as lf = ( λ cp) / ( ρs ), where λ and c P are the thermal condctivity and specific heat at 12 K, respectively, which is an approximate average of the free stream and flame temperatres [22]. Diffsive-thermal effect is cased by the non-similar diffsion of mass verss heat and are represented by the Lewis nmber, Le, which is defined as a ratio of the heat diffsivity of the mixtre to the mass diffsivity of the limiting reactant relative to the abndant inert [2]. The effective Lewis nmber is defined as the combination of the fel and oxidizer Lewis nmbers [9 11] Le eff ( Le 1) + ( Le 1) A = A E D 2 2 (4) where Le E, Le D are the Lewis nmbers of excessive and deficient reactants, and A 2 is a measrement of the mixtre strength, which were defined in Ref. [9 11]. In case of lean mixtre, the Lewis nmber of the deficient reactant is named fel Lewis nmber defined as a weighted average of the Lewis nmbers of the three fels (CO, H 2, CH 4 ) [9 11,22] Le F qco ( LeCO 1) + qh ( Le 2 H 1) + q 2 CH ( Le 4 CH 1) 4 = 1+ (5) q where LeCO, LeH, Le 2 CH are the fel Lewis nmbers of pre CO air mixtre at φ 4 CO, pre H 2 air mixtre at φ H 2, pre CH 4 air mixtre at φ CH 4, respectively [9 11]; q = qco + qh + q 2 CH is 4 the total heat release, where q j (j = CO, H 2, CH 4 ) is the nondimensional heat release associated with the consmption of species j [22]. Reslts and discssion Unstretched laminar brning velocities

5 The measred S of the three BDGs air mixtres at P =.1 MPa as a fnction of eqivalence ratio are plotted in Fig. 2. Together with the reslts from the present work, the S of the 5H 2 :5CO air flames and the 5H 2 :95CO air flames by McLean et al. [23] are also shown in the same figre. It is clear from Fig. 2 that the GG-V air flames have the highest brning velocity compared to those of the two other mixtres becase of the highest H 2 content in the GG-V mixtre. Next, becase of the higher ratio of H 2 /CO concentrations in the mixtre of GG-L compared to one of GG-H, it demonstrates a higher S of the GG-L air flames compared to those of the GG-H air flames, (the ratios of H 2 /CO/CO 2 are 1.18/.93/1 and 1.6/1.31/1 for GG-L and GG-H respectively). The maximm S of all three BDGs air premixed flames were fond at φ = 1.4 (i.e cm/s, 8.64 cm/s, and cm/s for the GG-H air, the GG-L air, and the GG-V air flames, respectively). In discssing the location of the peak brning velocity, reference to the two dashed line in Fig. 2 wold indicate that the S of the 5H 2 :5CO air flames and the 5H 2 :95CO air flames respectively have the peaks at φ = 2.5 and φ = 2, also literatre shows that H 2 air flames have the maximm S at φ = 1.4 [24] and CH 4 air flames have the peak S at φ = 1.1 [25]. The GG-V mixtre has H 2 :CO = 3:1, hence the peak brning velocity may be at φ between 1.4 and 2.; with the existence of CH 4 in the mixtre, the peak has to shift to lower φ (i.e. φ = 1.4 here). The GG-H has H 2 :CO =.8:1 casing the peak brning velocity at φ a little higher than 2., while the GG-L has H 2 :CO = 1.27:1 indcing the peak brning velocity at φ a little lower than 2., bt the high percentages in CO 2 diltion make the decrease of thermal diffsivity in the reactant mixtres [26], therefore the brning peaks tend to shift to lower eqivalence ratio (i.e. φ = 1.4 here). Sensitivity analysis As mentioned above, the PREMIX code [12] and the GRI-Mech 3. mechanism [13] were sed for the prediction of brning velocities. As shown in Fig. 3, the experimental measrements and predictions for S agree well at lean and stoichiometric flames, bt they diverge mch at rich flames. As an effort to identify the dominant reactions in the GRI-Mech 3. that lead to the observed differences at rich flames, a sensitivity analysis with respect to the reaction rate coefficients was performed for the three BDGs air premixed flames. In this stdy, the first-order normalized sensitivity coefficient of a given species affected by a certain reaction is defined as a relative vale eqaling the qotient of the sensitivity coefficient of (cm/s) S P =.1 MPa 5H 2 :5CO [23] 5H 2 :95CO [23] GG-H GG-L GG-V Eqivalence Ratio Figre 2. Unstretched laminar brning velocities of the three BDGs air premixed flames at P =.1 MPa, together with data of 5H 2 :5CO air and 5H 2 :95CO air by McLean et al. [23].

6 (cm/s) S (cm/s) S (cm/s) S GG-H P =.1 MPa GG-L P =.1 MPa GG-V P =.1 MPa GRI-Mech 3. Modified mechanism This stdy (experiment) GRI-Mech 3. Modified mechanism This stdy (experiment) GRI-Mech 3. Modified mechanism This stdy (experiment) Eqivalence Ratio Figre 3. Experimental (points), calclated by GRI-Mech 3. (dashed lines), and calclated by revised mechanism (solid lines) S of the three BDGs air flames at P =.1 MPa. that reaction and the sensitivity coefficient of the most sensitive reaction, which holds the largest sensitivity coefficient. This definition can be applied to compte the normalized sensitivity coefficients of the three fel species (i.e. CH 4, H 2, CO) of the BDGs sed in this stdy and only the relative vales larger than +.1 or smaller than.1 are reported here. Fig. 4 demonstrates the first-order normalized sensitivity coefficient of CH 4 species at P =.1 MPa, φ = 1.5 for the 21 elementary reactions exhibiting the largest sensitivities based on the GRI-Mech 3. mechanism, which emphasizes reslts at fel-rich conditions where the discrepancies between measrements and predictions are large. Similarly, the normalized sensitivity coefficients of H 2 and CO species of the GG-H air flames at P =.1 MPa, φ = 1.5 for the seven reactions exhibiting the largest sensitivities are presented in Fig. 5. A positive sensitivity coefficient indicates that increasing the reaction rate coefficient increases brning velocity; conversely, a negative sensitivity coefficient means that increasing the reaction rate coefficient decreases brning velocity. Therefore, in order to get a better agreement between measred and predicted S at fel-rich conditions, the GRI-Mech 3. mechanism were revised by replacing the reaction rate coefficients of these sensitive reactions by ones from referential reaction mechanisms ntil reaching a best fit between predicted S and or experimental data. The modified parameters of pre-exponential factor A, power of n, activation energy E a were collected from the kinetic data for combstion modeling in literatre [14 19] and are listed in Table 2. As can be seen in Fig. 3, the experimentally measred S of the three BDGs air premixed flames get better agreement with the prediction vales obtained by the revised

7 R3: O+H2<=>H+OH R1: O+CH3<=>H+CH2O P =.1 MPa φ = 1.5 R12: O+CO(+M)<=>CO2(+M) R33: H+O2+M<=>HO2+M R35: H+O2+H2O<=>HO2+H2O R36: H+O2+N2<=>HO2+N2 R38: H+O2<=>O+OH R41: 2H+H2O<=>H2+H2O R43: H+OH+M<=>H2O+M R45: H+HO2<=>O2+H2 R46: H+HO2<=>2OH R53: H+CH4<=>CH3+H2 R55: H+HCO<=>H2+CO R84: OH+H2<=>H+H2O R85: 2OH(+M)<=>H2O2(+M) R97: OH+CH3<=>CH2(S)+H2O R98: OH+CH4<=>CH3+H2O R52: H+CH3(+M)<=>CH4(+M) GG-H GG-L GG-V R3: O+H2<=>H+OH R52: H+CH3(+M)<=>CH4(+M) R53: H+CH4<=>CH3+H2 R84: OH+H2<=>H+H2O H 2 species CO species R38: H+O2<=>O+OH GG-H P =.1 MPa φ = 1.5 R99: OH+CO<=>H+CO2 R166: HCO+H2O<=>H+CO+H2O R284: O+CH3<=>H+H2+CO R98: OH+CH4<=>CH3+H2O R99: OH+CO<=>H+CO Normalized sensitivity coefficient of CH 4 species Normalized sensitivity coefficients of H 2 and CO species Figre 4. First-order normalized sensitivity coefficient of CH 4 species of the three BDGs air premixed flames at φ =1.5, P =.1 MPa. Figre 5. First-order normalized sensitivity coefficient of H 2 and CO species of the three BDGs air premixed flames at φ =1.5, P =.1 MPa. mechanism, even at fel-rich flames. It is shown that the largest discrepancy between them is abot 3 cm/s. The maximm S of the three BDGs air premixed flames by predicting sing the revised mechanism occr at eqivalence ratio of 1.4, matching with the peak brning velocities of experimental data. Markstein lengths The negative vale of the slope of the straight-line fit between stretched flame speed and flame stretch rate is defined as the brned gas Markstein length, L b, which represents the inflence of the flame speed on the flame stretch rate. The L b varies with the difference eqivalence ratios, mixtre compositions and initial pressres. For the case of a negative L b, the flame speed increases along with an increase in the flame stretch rate. In this case, if any protberance occrs on the flame front, then the flame speed increases, leading to the enhanced flame instability. On the other hand, if L b >, the flame-front instabilities are sppressed, thereby contribting to the flame stabilization [21,25]. Fig. 6 plots the measred L b of the three BDGs air premixed flames as a fnction of eqivalence ratio at the initial pressre of.1 MPa. This figre indicates that the L b of all mixtres are smaller than zero; therefore sch mixtres are nstable to flame stretch effects. For all measred BDGs air mixtres, L b increase along with eqivalence ratio as shown in Fig. 6. This tendency is similar to the tendencies of L b of the 5H 2 :5CO air flames [26] and the CH 4 air flames [25]. The L b of the GG-H air flames and the GG-L air flames are lower than those of the GG-V air flames, becase of the high CO 2 concentrations in the GG-H and the GG-L mixtres, casing the significant decrease in the L b of the mixtres [1,26].

8 n Table 2. Reaction rate coefficients in Arrhenis form k = AT exp( Ea R T ). Units are mole-cm-s-cal-k. No. Name Reaction A n E a Ref. 1 R3 O+H2<=>H+OH 3.82E [17] 2 R1 O+CH3<=>H+CH2O 8.43E+13.. [15,16,18,19] 3 R38 H+O2<=>O+OH 3.547E [15] 4 R41 2H+H2O<=>H2+H2O 5.624E [14,18] 5 R45 H+HO2<=>O2+H2 1.66E [15,16] 6 R46 H+HO2<=>2OH 1.7E [19] 7 R52 H+CH3(+M)<=>CH4(+M) 1.27E [15,16,18] 8 R53 H+CH4<=>CH3+H2 5.47E [15] 9 R55 H+HCO<=>H2+CO 5.E+13.. [16] 1 R84 OH+H2<=>H+H2O 1.17E [16] 11 R98 OH+CH4<=>CH3+H2O 5.72E [15] 12 R166 HCO+H2O<=>H+CO+H2O 2.244E [14,18] Flame stability and celllar strctre In premixed flames, a corrgated flame front de to the formation of celllar instabilities cold indce the trblence of nbrned mixtre and sbseqently the rapid increase of flame propagation velocity, hence cold be one of the main reasons for gas explosion. As previosly mentioned, two instabilities of premixed flame were observed in this stdy: diffsionalthermal instability and hydrodynamic instability. In order to compare the celllarity behavior of the three BDGs air premixed flames, schlieren images of different seqences of the GG- H air flames, the GG-L air flames, and the GG-V air flames for eqivalence ratio of.8 are shown at P =.2 MPa in Fig. 7a and at P =.3 MPa in Fig. 7b for the same Karlovitz nmber, which is the non-dimensional measre of stretched rate and can be defined as a ratio of chemical time scale, which is represented by the ratio of flame thickness to flame speed, and physical time scale, which is given by the inverse of flame stretch, Ka = (2 R)(dR d t) ( S l ) [8]. Three important parameters (, σ, l f ), which affect the f. -.5 P =.1 MPa L b (mm) GG-H GG-L GG-V Eqivalence Ratio Figre 6. Markstein lengths of the three BDGs air premixed flames at P =.1 MPa for a wide range of eqivalence ratios.

9 celllarity of the flames, are tablated at the bottom of the figre. This figre exhibits that the flame fronts behavior of the GG-H air flames and the GG-L air flames are qite similar, while the GG-V air flames are more wrinkled on the flame srfaces. For the GG-H air flames and the GG-L air flames, the are eqal, casing no different effect on the diffsional-thermal instability. Concerning the hydrodynamic instability, the σ and the l f change a little, hence casing a slight inflence on the hydrodynamic effect. For the GG-V air flames, compared to the GG-H air flames and the GG-L air flames, the σ increases, and the l f significantly decreases; ths the effect of the hydrodynamic instability wold be promoted. In addition, the is mch lower than those of the GG-H air flames and the GG- L air flames; therefore the flame front instabilities of the GG-V air flames can be enhanced. It is clear from Fig. 7a and b that different initial pressres case different seqences of the flames, cells form earlier and the size of cells is smaller at higher P. The is insensitive to initial pressre changes; therefore the propensity to destabilize the flames was not affected by the diffsive-thermal effect. The enhancement of celllarity at higher P cold be explained by the significant decrease in the l f, whereas the σ maintains nearly the same vale. In discssing the effects of each fel component in mixtres, flame seqences are evalated by increasing the concentration of each fel in the reactant mixtres. For hydrogen enrichment, Fig. 8 indicates that the propensity of destabilization tends to be progressively promoted. It is noteworthy to mention that the σ does not change so mch while the l f decreases significantly casing the enhancement in hydrodynamic effect. Additionally, the of the flame decreases with adding of H 2 content hence intensifies the diffsive-thermal effect. It can be conclded that both the hydrodynamic instability and diffsional-thermal instability are promoted when the amont of hydrogen increases in the reactant mixtres. a GG-H GG-L GG-V P =.2 MPa φ =.8 Ka =.59 P =.2 MPa φ =.8 Ka =.53 GG-H GG-H+1%H 2 GG-H+2%H 2 P =.2 MPa φ =.8 Ka = , , ,.83 b , , ,.75 GG-H GG-L GG-V P =.2 MPa φ =.8 Ka =.73 GG-L GG-L+1%H 2 GG-L+2%H 2 P =.3 MPa φ =.8 Ka = , , ,.78 GG-V GG-V+1%H 2 GG-V+2%H 2 P =.3 MPa φ =.8 Ka =.5 P =.2 MPa φ =.8 Ka = , , , , , ,.54 Figre 7. Schlieren images of the three BDGs air premixed flames at φ =.8, (a) P =.2 MPa and (b) P =.3 MPa. Figre 8. Schlieren images of the three BDGs air premixed flames with H 2 enrichment at φ =.8, P =.2 MPa.

10 Regarding to effects of CO addition, Fig. 9 shows that there are no differences in seqences of the flame front srfaces between the BDGs air premixed flames with and withot CO additions. As increasing content of CO in the reactant mixtres, the slightly increases, thereby the diffsional-thermal instability wold be slightly diminished. Meanwhile, the hydrodynamic instability wold be a little promoted becase of a small increase in the σ and a slight decrease in the l f. Hence the net effects of the two instabilities can be negligible. The combination of the diffsional-thermal instability and the hydrodynamic instability indicates that increasing CO concentration in the fel blends cold not sppress the celllar instabilities of the BDGs air flames. The effects of methane additions on celllarity of the BDGs air flames are illstrated in Fig. 1. It is different from the cases in the 5H 2 :5CO syngas air premixed flames that cold not be sppressed by methane additions [9,11], herein the flame front instabilities are diminished as the amont of methane addition in the fel blend increases. The most sitable parameter that denotes the stabilizations in the BDGs air flames with CH 4 additions is the significant increase in the l f. For methane addition, the decreases, casing an enhancement in diffsional-thermal instability. The σ increases, however the l f progressively increases, and these ths sppress hydrodynamic instability. It indicates that l f can be the dominant factor as the sppressant or in the enhancement flame front instabilities. It also demonstrates that the hydrodynamic effect is the main factor when the flame has a large size, while the diffsive-thermal effect is only dominant at the time when flame is ignited and has a small size [8]. Compared to the GG-H air flames and the GG-L air flames, the GG-V air flames is not so mch sppressed by CH 4 addition, and this tendency is similar to the behaviors of the CH 4 -added 5H 2 :5CO syngas air flames [9,11]. Becase both flames have high brning velocities and thereby small flame thicknesses, therefore the increase in flame thickness cold not significantly affect the celllarity of the flames. GG-H GG-H+1%CO GG-H+2%CO GG-H GG-H+1%CH 4 GG-H+2%CH 4 P =.3 MPa φ =.8 Ka =.49 P =.3 MPa φ = 1.2 Ka = , , , , , ,.74 GG-L GG-L+1%CO GG-L+2%CO GG-L GG-L+1%CH 4 GG-L+2%CH 4 P =.3 MPa φ =.8 Ka =.47 P =.3 MPa φ = 1.2 Ka = , , , , , ,.77 GG-V GG-V+1%CO GG-V+2%CO GG-V GG-V+1%CH 4 GG-V+2%CH 4 P =.3 MPa φ =.8 Ka =.48 P =.3 MPa φ = 1.2 Ka = , , , , , ,.54 Figre 9. Schlieren images of the three BDGs air premixed flames with CO addition at φ =.8, P =.3 MPa. Figre 1. Schlieren images of the three BDGs air premixed flames with CH 4 addition at φ = 1.2, P =.3 MPa.

11 Conclsions The major conclsions of the stdy are as follows: The S of the GG-V air flames are higher than those of the GG-L air flames and the GG-H air flames and the peak brning velocities of the three BDGs air premixed flames were fond at eqivalence ratio of 1.4. The experimental measrements and predictions sing GRI-Mech 3. of S agree well at lean and stoichiometric flames, bt they diverge mch at rich flames. For getting a better agreement between the measred and predicted brning velocities at fel-rich conditions, sensitivity analysis was performed to identify the dominant reactions. Sbseqently the rate coefficients of these important reactions were replaced by ones from referential reaction mechanisms in literatre. The new calclated brning velocities from the revised mechanism agree rather well with the experimental data. The L b of the three BDGs air premixed flames are smaller than zero, thereby the flames are sensitive to flame stretch effects and the L b increase with eqivalence ratio. The flame front behaviors of the GG-H air flames and the GG-L air flames are qite similar, while the GG-V air flames are more wrinkled on the flame srfaces becase of the combined inflences of the hydrodynamic and the diffsional-thermal instabilities. For hydrogen enrichment, the propensity of destabilization tends to be progressively promoted becase of the enhancement on both effects. For carbon monoxide addition, there are no differences in seqences of the flame front srfaces between the BDGs air premixed flames with and withot carbon monoxide additions. For methane additions, the flame front instabilities are diminished becase of the significant increase in the flame thickness. Acknowledgements This work was spported by the cooperation bsiness between Korea Gas Corporation and Pkyong National University in References [1] Yan, B., W, Y., Li, C., Y, J.F., Li, B., Li, Z.S., et al., Experimental and modeling stdy of laminar brning velocity of biomass derived gases/air mixtres, Int. J. Hydrogen Energy 36: (211). [2] Li, C., Yan, B., Chen, G., Bai, X.S., Strctres and brning velocity of biomass derived gas flames, Int. J. Hydrogen Energy 35: (21). [3] Hanaoka, T., Inoe, S., Uno, S., Ogi, T., Minowa, T., Effect of woody biomass components on air steam gasification, Biomass Bioenergy 28:69 76 (25). [4] Lv, P., Yan, Z., Ma, L., W, C., Chen, Y., Zh, J., Hydrogen-rich gas prodction from biomass air and oxygen/steam gasification in a downdraft gasifier, Renew. Energy 32: (27). [5] Valliyappan, T., Ferdos, D., Bakhshi, N.N., Dalai, A.K., Prodction of hydrogen and syngas via steam gasification of glycerol in a fixed-bed reactor, Top. Catal. 49:59 67 (28). [6] Williams, F.A., Combstion theory, 2 nd ed. Redwood City, CA:Addison-Wesley, p [7] Kadowaki, S., Szki, H., Kobayashi, H., The nstable behavior of celllar premixed flames indced by intrinsic instability, Proc. Combst. Inst. 3: (25). [8] Law, C.K., Kwon, O.C., Effects of hydrocarbon sbstittion on atmospheric hydrogen air flame propagation, Int. J. Hydrogen Energy 29: (24). [9] V, T.M., Park, J., Kwon, O.B., Kim, J.S., Effects of hydrocarbon addition on celllar instabilities in expanding syngas air spherical premixed flames, Int. J. Hydrogen Energy 34: (29).

12 [1] V, T.M., Park, J., Kwon, O.B., Bae, D.S., Yn, J.H., Keel, S.I., Effects of dilents on celllar instabilities in otwardly propagating spherical syngas air premixed flames, Int. J. Hydrogen Energy 35: (21). [11] V, T.M., Park, J., Kim, J.S., Kwon, O.B., Yn, J.H., Keel, S.I., Experimental stdy on celllar instabilities in hydrocarbon/hydrogen/carbon monoxide air premixed flames, Int. J. Hydrogen Energy 36: (211). [12] Kee, R.J., Grcar, J.F., Smooke, M.D., Miller, J.A., A Fortran program for modeling steady laminar one-dimensional premixed flames, Report No. SAND85-824, Sandia National Laboratories, [13] Smith, G.P., Golden, D.M., Frenklach, M., Moriarty, N.W., Eiteneer, B., Goldenberg, M., et al., GRI-Mech 3., [14] Davis, S.G., Joshi, A.V., Wang, H., Egolfopolos, F., An optimized kinetic model of H 2 /CO combstion, Proc. Combst. Inst. 3: (25). [15] Li, J., Zhao, Z., Kazakov, A., Chaos, M., Dryer, F.L., Scire, J.J., A comprehensive kinetic mechanism for CO, CH 2 O, and CH 3 OH combstion, Int. J. Chem. Kinet. 39: (27). [16] Saxena, P., Williams, F.A., Testing a small detailed chemical-kinetic mechanism for the combstion of hydrogen and carbon monoxide, Combst. Flame 145: (26). [17] Sn, H., Yang, S.I., Jomaas, G., Law, C.K., High-pressre laminar flame speeds and kinetic modeling of carbon monoxide/hydrogen combstion, Proc. Combst. Inst. 31: (27). [18] Wang, H., Yo, X., Joshi, A.V., Davis, S.G., Laskin, A., Egolfopolos, F., Law, C.K., USC Mech Version II. High temperatre combstion reaction model of H 2 /CO/C 1 C 4 componds. [19] Konnov, A.A., Detailed reaction mechanism for small hydrocarbons combstion, Release.5, 2. [2] Clavin, P., Dynamic behavior of premixed flame fronts in laminar and trblent flows, Prog. Energy Combst. Sci. 11:1 59 (1985). [21] Bradley, D., Gaskell, P.H., G, X.J., Brning velocities, Markstein lengths, and flame qenching for spherical methane air flames: a comptational stdy, Combst. Flame 14: (1996). [22] Law, C.K., Jomaas, G., Bechtold, J.K., Celllar instabilities of expanding hydrogen/propane spherical flames at elevated pressres: theory and experiment, Proc. Combst. Inst. 3: (25). [23] McLean, I., Smith, D.B., Taylor, S.C., The se of carbon monoxide/hydrogen brning velocities to examine the rate of the CO+OH reaction, Proc. Combst. Inst. 25: (1994). [24] Dowdy, D.R., Smith, D.B., Taylor, S.C., The se of expanding spherical flames to determine brning velocities and stretch effects in hydrogen/air mixtres, Proc. Combst. Inst. 23: (199). [25] G, X.J., Haq, M.Z., Lawes, M., Woolley, R., Laminar brning velocity and Markstein lengths of methane air mixtres, Combst. Flame 121:41 58 (2). [26] Prathap, C., Ray, A., Ravi, M.R., Investigation of nitrogen diltion effects on the laminar brning velocity and flame stability of syngas fel at atmospheric condition, Combst. Flame 155: (28).