Experimental Investigation of Lean Methane-Air Laminar Premixed Flames at Engine-Relevant Temperatures.

Lean premixed combustion is one of the most effective methods to constrain pollutant emissions for modern industrial gas turbines. An experimental study was performed on its propagation speed and internal structure at engine-relevant temperatures. A Bunsen burner was employed for the measurement with an optical schlieren system. The results show that the increase of preheating temperature dramatically accelerates the propagation of methane flames. The numerical results predicted by GRI-Mech 3.0, FFCM-1, and USC Mech II were also compared. The GRI-Mech 3.0 seems to overestimate the laminar flame speed at high operating conditions, while FFCM-1 underestimates the laminar flame speed compared to the present experimental data. The prediction by FFCM-1 shows good agreement with the overall existing data. The USC Mech II seems to overestimate the laminar flame speed at fuel-lean conditions while shows good agreement with present experimental measurements at stoichiometric conditions when the inlet temperature increases. It is also indicated that the flame is thinned at high-temperature conditions and the importance of CO production to the propagation speed increases. Finally, based on the experimental data, an empirical correlation of the laminar flame speed was developed in the range of T u = 300-800 K and ϕ = 0.7-1.0, the maximum deviation of which was less than 8%. The results of this study may contribute to the optimization of advanced gas turbine combustors.

Introduction

Methane is one of the most important energy sources nowadays and the main fuel for modern industrial gas turbines. To avoid excessive NOx emissions, lean premixed flames have been widely adopted in combustors under elevated temperature and pressure conditions. For a typical H class gas turbine, the pressure ratio of the compressor is 19–23,[1−3] and therefore the inlet temperature to the combustor is about 690–750 K when the air compression process is regarded as an isentropic process. Much attention has been paid to the topic of the laminar flame speed of methane–air mixtures since it is an important factor in the combustor design. For an ideal low emission combustion process, the flame propagation speed must match the local flow speed. Fail to do so, the combustor may experience some problems, e.g., dynamic instability, flashback, blowoff,[4,5] or high CO emissions.[6−8]

The laminar flame speed (SL) is defined as the propagation speed of a planar, unstretched, adiabatic, premixed flame.[9,10] It is a key parameter and of great importance for a combustible mixture because it governs many properties of combustion such as the shape and stabilization of a premixed flame. Extensive experiments have been conducted to measure SL of methane–air mixtures using various apparatuses. Gu et al.[11] measured SL of a methane–air mixture between 300 and 400 K using spherically expanding flames. Correlations between SL and equivalence ratio were empirically given as an exponential function within the range of ϕ = 0.8–1.2. Hu et al.[12] measured SL of methane–air mixtures up to 443 K. Ogami and Kobayashi[13] measured SL of stoichiometric methane–air mixtures diluted by helium up to 600 K. More recently, SL of methane–air mixtures has been measured at 300–573 K by Mohammad et al.,[14] 489–573 K by Ferris et al.,[15] and 300–650 K by Varghese et al.[16] Egolfopoulos and co-workers[17] measured SL for C1–C4 hydrocarbons at 400–520 K and 0.8–3.0 MPa. Along with theoretical and numerical studies, it is unequivocal that the flame propagation speed and its gradient to temperature increase with the preheating temperature. However, this dependence has not been quantificationally measured at engine-relevant conditions, which leaves a major shortcoming in the fundamental database. As summarized by Konnov,[18] previous data were performed over a wide range of pressure (up to 6 MPa), while the preheating temperature was mostly below 500 K. Furthermore, large discrepancies still lie between existing literature for elevated temperature conditions.

The main experimental approaches to determine the flame speed are Bunsen burner,[6−8,10,13] spherically expanding bomb,[10,11,19−25] heat flux,[26−30] and stagnation flame.[31−33] Detailed advantages and limitations of these methods were reviewed by Egolfopoulos et al.[34] The Bunsen burner is deemed as a suitable choice to investigate the flame speed under elevated temperatures and ambient pressure. This approach is feasible to control the preheating process and form an unstretched, quasi-adiabatic premixed flame. The shortcoming of this method is that the flame shape is strongly affected by the heat loss at the nozzle rim, curvature on the flame tip, and the equivalence ratio discrepancy due to fresh air entrainment around the flame. To avoid these effects, the Angle method[13,35−37] is usually employed, which measures the slope of the flame front and excludes the nonlinear sections, i.e., the tip and base of the flame.

In addition, numerical simulations also play an important role in combustion modeling and engineering design. For methane, many detailed chemical mechanisms have been developed. GRI-Mech 3.0[38] is a well-developed mechanism that consists of 53 species and 325 reactions. It has been optimized using experimental laminar flame speeds of methane up to 600 K. FFCM-1[39] is one of the most advanced mechanisms that consists of 37 species and 291 reactions. USC Mech II[40] is a high-temperature combustion model of H2/CO/C1–C4 compounds, which consists of 111 species and 784 reactions and was subject to validation tests against reliable combustion data. Mechanisms mentioned above have been used to model the laminar methane–air flames.

The objective of this work is to investigate the premixed laminar flame speed of lean methane–air mixtures at preheating temperatures up to 800 K, which covers the inlet temperature range of modern industrial gas turbine combustors. The remainder of this paper is organized as follows. Section presents the experimental and numerical details. Section presents the results and discusses the response of the laminar flame structure to different preheating temperatures. Finally, in Section , some important conclusions are summarized.

Experimental and Numerical Details

The experimental setup is shown in Figure , including a Bunsen burner and an optical schlieren system.

Figure 1

Experimental setup.

Bunsen Burner

An axisymmetric premixed burner was designed and developed to generate a steady conical laminar premixed flame stabilized on the outlet of a contracting nozzle, and the diameter (d) and the contraction ratio of which are 10 mm and 2.25, respectively. The height of the Bunsen burner is 700 mm, which is long enough to mix the fuel with an air stream. Two K-type thermocouples were placed, one downstream of the gas inlet, while the other upstream of the contracting nozzle, to monitor temperatures throughout the Bunsen burner. The preheating temperature of the unburned mixture (Tu) was determined by the latter. During each measurement, the fluctuation of temperature was controlled within ±5 K.

Optical Diagnostics and Determination of the Laminar Flame Speed

As shown in Figure , the optical system consists of a tungsten–halogen lamp, a pinhole, a set of lenses, and a camera. In the experiments, the light through the pinhole was collimated by a convex lens. The collimated light passed through the Bunsen flame before it was focused on a vertically installed knife edge. The schlieren images were taken using a digital CMOS camera (FE 1.8/50, 2768 × 1560).

As shown in Figure a, the brightness of the schlieren image represents the corresponding temperature gradient in the Bunsen flame field. The line formed by the brightest pixel in the direction perpendicular to the knife edge is determined as the flame front. The Angle method[13,35−37] was chosen to calculate the flame speed from schlieren images. Other than the area-average method,[6−8,10] this approach only uses data in the middle section of the flame front, where it is free from the adverse effects such as flame stretch and curvature.

Figure 2

(a) Flame front obtained by the optical method. (b) Angle method for determination of SL.

As shown in Figure b, the flame propagates toward the unburned mixture at an angle θ/2. The laminar flame speed (SL) is determined to be equal to the velocity component of the unburned mixture, which is normal to the flame front, and therefore, SL is calculated as followswhere U is the average velocity of the unburned mixture jet, and QV is the volumetric flow rate of the unburned methane–air flow.

The uncertainty of our measurements was estimated from two main sources: the uncertainty of the volumetric flow rate of the unburned methane–air flow (U) and the uncertainty of the calculated slope of the flame front (Usin θ). U came from the mass flow controller uncertainty, which was estimated to be ∼1.4%, and Usin θ came from the postprocess of the schlieren method. Then, the overall uncertainties were estimated from .

Numerical Setup

The simulation of the one-dimensional premixed flame was carried out using the PREMIX code[41] of the CHEMKIN package[42] with the GRI-Mech 3.0,[38] FFCM-1,[39] and USC Mech II[40] mechanisms.

Results and Discussion

System Validation

To validate the present measurement system, the methane–air flame speed at 300 and 373 K were measured and compared with existing literature,[6−8,14,16,20,27−29,43−45] as shown in Figure . Each data point was obtained by averaging the results of 24 schlieren images, and the error bars represent the scale of one standard deviation. Error bars of the equivalence ratio are also plotted in Figure , which represent the standard deviation of equivalence ratio fluctuation during measurements. Furthermore, the simulation results were plotted for comparison. The present measurements agree well with the existing data. SL increases with the equivalence ratio and the preheating temperature. The laminar flame speeds of the stoichiometric mixture at 300 and 373 K are 35.9 and 52.3 cm/s, respectively. The standard deviation of data at 373 K is larger than that at 300 K. This may be due to the noises on schlieren images, which were caused by the natural convection between the heated experiment rig and the ambient environment. It is also seen that the numerical results by GRI-Mech 3.0 quantificationally agree with present measurements, while FFCM-1 slightly underestimates the flame speed. Predictions for 373 K by GRI-Mech 3.0 seem to be a little overshooting near fuel stoichiometric conditions. As for USC Mech II, at 300 and 373 K, it seems to slightly overestimate the flame speed at lean conditions compared to the overall existing data. The maximum overall uncertainties of the flame speed are estimated to be ∼4.7% at 300 K and ∼4.2% at 373 K, respectively.

Figure 3

Laminar flame speed of methane–air flames at 300 and 373 K.

Experimental Results

The measurements of laminar flame speeds were performed at ambient pressure in the temperature range of 350–800 K. The present experimental data can be seen in Table in Appendix A. Figure shows the comparison of the present experimental data with existing SL data[14,16,31,46−48] at elevated preheating temperatures. The results indicate that GRI-Mech 3.0 overestimates the SL while FFCM-1 shows a good agreement with existing data at Tu ≤ 550 K. The USC Mech II mechanism shows a good agreement with present experimental measurements at stoichiometric conditions when the inlet temperature elevates but still slightly overestimates the flame speed at lean conditions as previously mentioned. GRI-Mech 3.0 and FFCM-1 have a tendency to overestimate when the preheating temperature increases (Tu > 550 K).

Table A1

Experimental Data of the Laminar Flame Speed of the Methane–Air Mixture

Tu (K)	ϕ	SL (cm/s)	√σ (cm/s)
299	0.697	19.3	0.174
302	0.792	26.1	0.668
303	0.898	33.4	1.00
303	0.951	34.0	0.867
303	1.00	34.8	1.13
304	0.896	32.0	1.03
305	0.998	37.0	0.363
305	0.850	29.3	1.01
306	0.794	25.7	0.924
307	0.753	23.0	0.932
308	0.694	18.7	0.850
345	1.00	46.8	1.39
346	0.998	46.8	0.495
346	1.00	47.5	1.45
346	0.951	44.9	1.89
347	0.951	47.4	1.84
347	0.897	42.3	1.10
348	0.950	46.0	0.858
348	0.897	45.2	1.04
348	0.850	38.1	0.711
350	0.896	43.0	1.26
350	0.850	39.5	2.44
350	0.794	34.2	0.780
352	0.850	39.8	1.16
352	0.794	36.7	1.85
352	0.753	30.8	0.757
354	0.796	36.0	1.12
354	0.750	32.6	0.919
354	0.695	26.9	0.779
357	0.753	30.8	1.15
357	0.662	23.2	0.561
358	0.695	28.7	0.963
359	0.695	26.6	1.15
362	0.666	23.1	1.17
362	0.644	22.7	2.34
366	1.00	53.8	1.20
367	1.00	52.1	1.40
367	0.951	53.7	1.54
368	0.950	50.1	1.51
368	0.896	50.4	1.70
368	1.00	51.0	1.09
369	0.895	50.0	0.851
369	0.850	47.0	1.07
369	0.950	49.3	1.73
370	0.849	45.9	1.13
370	0.794	42.9	0.787
370	0.897	46.2	0.752
371	0.794	41.7	1.47
371	0.751	37.6	1.35
371	0.850	42.1	1.05
372	0.750	36.7	0.815
372	0.795	37.9	1.19
373	0.693	30.3	1.23
373	0.695	32.4	1.03
373	0.751	34.3	0.895
374	0.695	29.3	1.03
375	0.661	26.1	0.631
375	0.656	28.6	0.679
376	0.660	25.2	0.986
395	1.00	56.9	1.11
396	1.00	59.1	1.28
396	1.00	58.0	1.83
396	0.951	54.9	1.48
397	0.950	56.6	1.99
397	0.896	50.9	0.931
398	0.950	57.3	1.00
398	0.896	55.3	2.04
398	0.851	49.0	1.18
399	0.850	51.0	1.36
400	0.895	55.2	1.53
400	0.794	47.8	1.12
400	0.794	43.0	1.44
401	0.751	38.2	0.857
402	0.849	50.8	2.03
402	0.751	41.3	1.67
403	0.695	32.2	1.13
404	0.794	45.6	1.82
404	0.695	36.5	1.42
405	0.661	29.5	0.703
406	0.750	40.2	0.659
406	0.656	29.9	0.738
407	0.693	34.5	1.27
409	0.656	28.1	0.970
445	1.00	70.2	0.945
445	1.00	69.7	1.71
446	1.00	72.0	2.69
446	0.950	70.1	2.07
447	0.896	69.4	2.46
447	0.950	67.3	2.57
448	0.896	67.4	1.48
448	0.950	71.1	1.91
448	0.850	64.3	2.52
448	0.896	64.4	1.33
449	0.850	65.2	1.08
450	0.795	58.4	1.69
450	0.795	58.6	3.29
450	0.850	59.9	0.656
451	0.751	52.4	2.61
452	0.751	50.2	1.84
452	0.694	45.9	1.88
452	0.795	55.0	1.79
453	0.656	38.7	1.08
454	0.751	50.2	1.49
455	0.694	43.4	1.73
455	0.694	42.7	0.758
457	0.656	36.3	0.615
458	0.656	36.5	0.841
468	1.00	81.5	2.51
468	1.00	77.4	2.77
469	0.950	80.3	2.88
469	0.950	73.9	1.16
470	0.850	72.9	1.84
470	0.896	78.5	2.49
470	0.897	72.2	2.05
471	0.795	68.2	1.17
472	0.950	78.5	2.49
472	1.00	82.2	2.20
472	0.751	61.6	1.88
472	0.849	66.6	1.59
473	0.897	76.5	1.96
473	0.694	52.7	1.86
473	0.795	62.1	1.22
474	0.850	72.8	2.44
474	0.656	45.7	1.61
474	0.751	54.0	0.897
475	0.795	65.5	2.76
475	0.694	47.2	0.984
476	0.751	61.0	1.79
477	0.696	51.9	2.05
477	0.661	40.8	1.25
478	0.656	44.3	1.81
495	1.00	86.0	1.91
495	1.00	83.0	2.40
496	0.950	85.1	2.87
496	0.950	81.0	1.33
497	0.896	83.2	2.98
497	0.896	79.4	1.78
498	1.00	86.7	2.04
498	0.850	72.8	2.18
499	0.950	84.5	1.82
499	0.850	78.2	3.32
500	0.897	83.8	2.84
500	0.795	71.7	3.50
500	0.795	66.7	3.13
501	0.850	77.7	3.47
501	0.751	64.6	1.72
501	0.751	62.2	1.51
502	0.795	71.1	2.86
502	0.694	56.7	1.62
502	0.694	53.4	1.60
503	0.751	64.6	1.42
503	0.656	47.5	1.48
503	0.657	45.9	1.20
504	0.696	55.3	2.09
505	0.656	47.1	1.97
545	1.00	103	3.97
545	0.998	102	3.56
546	0.950	102	3.64
546	0.950	98.0	4.10
547	0.896	93.8	3.45
548	0.896	97.8	3.02
548	0.794	82.6	3.22
548	0.848	90.6	3.78
549	0.848	96.6	4.20
549	0.695	64.7	2.12
549	0.751	76.1	3.85
550	0.796	87.5	3.75
550	0.664	52.3	1.72
550	0.677	59.3	3.02
552	0.751	78.6	1.80
553	0.695	70.4	2.11
554	0.655	59.4	2.05
555	0.596	50.6	1.33
568	1.00	117	2.70
568	1.00	109	3.77
569	0.949	114	3.16
569	0.949	106	3.47
570	0.896	113	3.97
570	0.895	102	3.24
571	0.847	104	3.05
571	0.848	97.0	3.04
572	0.795	87.5	3.68
573	0.793	99.8	2.12
573	0.750	78.8	2.94
574	0.750	91.6	2.36
575	0.695	81.9	3.16
575	0.698	72.0	2.66
576	0.652	71.8	1.76
576	0.674	64.4	3.00
578	0.600	57.5	1.77
578	0.650	57.2	2.80
595	1.00	122	2.68
595	1.00	121	4.40
596	0.949	121	3.50
596	0.949	117	4.78
597	0.896	121	3.83
597	0.847	105	3.92
597	0.897	111	3.30
598	0.847	114	4.05
598	0.750	89.0	3.33
598	0.793	98.3	3.40
599	0.695	78.9	3.07
600	0.793	105	3.71
600	0.654	65.1	2.86
600	0.670	72.0	2.77
601	0.750	95.7	4.52
602	0.695	85.4	2.00
603	0.652	76.6	2.48
604	0.600	63.5	3.36
644	1.00	147	5.36
645	0.949	145	6.76
645	1.00	142	4.83
646	0.949	138	5.45
647	0.895	141	6.20
647	0.895	135	3.64
648	0.848	136	5.86
648	0.848	126	5.30
649	0.794	116	5.35
650	0.794	131	3.00
650	0.749	105	4.22
651	0.695	99.4	4.16
652	0.749	117	3.65
652	0.678	85.6	3.79
653	0.695	105	3.13
654	0.654	94.4	3.75
654	0.655	79.3	4.05
656	0.596	81.3	1.62
695	1.00	167	7.01
695	1.00	164	3.26
696	0.948	165	5.40
696	0.948	162	4.07
697	0.895	163	7.71
698	0.847	157	4.13
698	0.895	156	5.19
699	0.795	149	6.62
699	0.847	152	6.46
700	0.749	142	5.01
700	0.749	135	6.41
700	0.795	140	6.63
701	0.695	128	4.01
701	0.704	117	4.76
702	0.681	105	4.98
703	0.651	112	4.21
703	0.657	99.6	3.72
705	0.597	96.8	3.02
745	1.00	196	7.86
745	1.00	191	8.75
746	0.948	192	7.24
746	0.950	188	6.09
747	0.897	188	8.92
747	0.897	187	8.13
748	0.848	186	7.07
748	0.848	177	7.11
749	0.795	176	6.64
749	0.797	164	8.29
750	0.749	164	5.90
750	0.749	154	10.5
751	0.705	140	5.89
752	0.696	149	4.27
753	0.684	130	7.48
754	0.652	133	4.97
755	0.653	122	8.03
757	0.598	119	6.79
795	1.00	226	5.47
795	1.00	221	12.0
796	0.949	225	4.44
796	0.949	216	10.2
797	0.896	224	7.42
797	0.896	211	14.6
798	0.848	212	7.23
798	0.848	204	13.4
799	0.795	206	7.21
799	0.795	193	14.1
800	0.749	192	5.83
800	0.749	180	12.0
801	0.696	182	7.69
801	0.707	169	9.79
802	0.651	162	9.08
802	0.688	147	8.04
803	0.598	142	4.30
803	0.667	139	6.03

Figure 4

Comparisons between present experimental and existing SL data at 0.1 MPa and elevated preheating temperatures.

Figure shows the present experimental SL data with numerical predictions of GRI-Mech 3.0 at 350–800 K. In addition, the uncertainties of the mass flow rate and temperature measurements were also considered in numerical simulations. Each pair of dashed lines in Figure represents the flame speed corresponding to the upper and lower error limits of equivalence ratio (left column) and preheating temperature (right column) measurements, respectively. It is indicated that the impact of temperature uncertainty on the measurement is about 3 times larger than that of the equivalence ratio. The propagation speed of the lean premixed flame monotonously increases with the preheating temperature and equivalence ratio. At 800 K, the flame speed of the stoichiometric mixture is 223.8 cm/s, which is about 6 times larger compared to that at 300 K. Interference between the preheated mixture jet and the ambient environment was encountered. There were noises in the schlieren images due to the vortices caused by the natural convection. As the preheating temperature increases from 350 to 800 K, those noises escalated the measuring error of SL. Nevertheless, the measuring error in this study was still under ±7%. The overall uncertainty of the flame speed at 800 K was estimated to be ∼7.1%, which is about 2 times larger compared to that at 300 K.

Figure 5

Laminar flame speeds from simulations and experiments at different preheating temperatures. Solid lines: predictions at Tu and ϕ. Dashed lines in (a)–(c): predictions at ϕ ± Δϕ. Dashed lines in (d)–(f): predictions at Tu ± ΔTu.

The numerical results predicted by GRI-Mech 3.0 agree well with the measurements at low preheating temperatures (below 400 K). Between 450 and 500 K, quantitative prediction is made at fuel-lean conditions (ϕ ≤ 0.90). For cases with a higher temperature or equivalence ratio, the flame speed is constantly overpredicted by the numerical method. At 800 K, this discrepancy is 6% for the stoichiometric mixture. One of the reasons might be the thermal radiation in the experiments.

To evaluate the impact of thermal radiation on the experiments, an Optical Thin (OPT) model was used with the PREMIX code to predict the laminar flame speed of methane. As shown in Figure , the numerical results were plotted with the present experimental results. The results indicate that the effect of thermal radiation is negligible.

Figure 6

Numerical predictions of SL at 350–800 K using Optical Thin model.

Numerical Simulation

Since the GRI-Mech 3.0 mechanism has been optimized using experimental data up to 600 K, it was chosen to simulate the methane–air flames structure at elevated preheating temperatures. Sensitivity analysis was performed to determine the importance of each specific elementary reaction to the flame propagation speed at preheating conditions. The sensitivity coefficients are defined as , where k is the reaction rate for the ith elementary reaction.[12] Nine elementary reactions with the largest magnitude of the sensitivity coefficient are shown in Figure . It is shown that the flame speed is most sensitive to the chain-branching reaction R38, and the coefficient of which is much greater than the others. For most reactions, the magnitude of the coefficient decreases with increasing temperature. However, for R166, R167, and R284, the sensitivity coefficient changes otherwise, which may suggest that the importance of CO production elevates. For the fuel-lean scenario, the importance of R35 and R99 increases since the oxygen is sufficient in the reaction system, which can promote the oxidation of CO.

Figure 7

Effects of different preheating temperatures on sensitivity coefficients.

Figure shows the computed methane premixed flame structure by GRI-Mech 3.0 under the flame coordinate normalized by the characteristic flame thickness (λ/CP)/f0. For comparison, a set of nondimensional variables is defined as followswhere subscripts u and b represent the unburned mixture and the burned mixture, respectively. λ is the thermal conductivity, CP is the specific heat, ρu is the density, and XB is the mole fraction of species B. The flame front (x̃ = 0) is defined as the position with the maximum temperature gradient.

Figure 8

Computed flame structure. Solid lines: Tu = 300 K; dashed lines: Tu = 500 K; and short dashed lines: Tu = 800 K.

The numerical results show that the nondimensional temperature gradient increases with the preheating temperature. This indicates that the thickness of the flame reduces due to the change of chemistry and radical distribution. For the conditions ϕ = 0.8 and 1.0, the flame thicknesses (δT) at Tu = 800 K are 58% and 63% of that at 300 K, respectively. δT is calculated based on the maximum gradient of the temperature profile (dT/dx)max as follows

Figure also shows that the peak concentration of the minor species increases with the preheating temperature, which suggests that the reactivity of the mixture is amplified. In the typical preheating zone of the flame (x̃ ∼ −5), the concentrations of CO and CH2O decrease with the inlet temperature, while the concentration of HO2 notably increases. R119 (HO2 + CH3 = OH + CH3 O) is a dominant step for methyl oxidation in the autoignition process.[4] The high concentration of HO2 in front of the flame indicates that the ignition process is activated ahead of the reaction layer at high preheating temperatures.

Hence, it is suggested that with increasing inlet temperature, the flame is accelerated and thinned not only due to the reactivity elevation in the reaction layer but also due to that in the preheating zone.

Empirical Correlation

An empirical correlation was developed based on the present measurements, the form of which was chosen as followswhere a, b, c, n1, and n2 are independent constants. This function is an extension of the previous analysis conducted by Yu[49] (eq 24). Two exponents, n1 and n2, were added to mimic the nonlinear effects and improve the accuracy in data fitting. In the range of Tu = 300–800 K and ϕ = 0.7–1.0, the flame speed is

The comparison between these correlations and experimental results is shown in Figure . It is suggested that the results increase monotonously with the preheating temperature and the equivalence ratio, the uncertainty of which is within ±8%.

Figure 9

Comparison between the empirical correlations and the present experimental results.

In the existing literatures,[12,48,50−56] power-law correlations were used to fit the stoichiometric methane–air laminar flame speed at 0.1 MPa. However, as shown in Figure , the discrepancy is quite large when the preheating temperature is high (Tu > 550 K). Correlations by Iijima and Takeno, Hill and Hung, Akram, and Amirante et al. show underestimation, while the correlation by Wang et al. shows overestimation at Tu > 550 K. Correlations by Elia et al., Rahim, Hu et al. and Hinton et al. show relatively good agreement with present experiments.

Figure 10

Correlations of stoichiometric methane–air laminar flame speed at 0.1 MPa.

Conclusions

The laminar flame speed of a methane–air mixture was measured using a Bunsen burner at Tu = 300–800 K, ϕ = 0.7–1.0, and ambient pressure. A numerical investigation was also conducted to gain an understanding of the impact of preheating temperature on the propagation and structure of premixed flames. The main conclusions are as follows.

The increase of preheating temperature dramatically accelerates the flame propagation, while the reduction of the flame thickness is relatively moderate. At 800 K, the flame speed of the stoichiometric mixture is about 6 times larger compared to that at 300 K, and the flame thickness is 63% of that at 300 K. It indicates that the importance of CO production to the flame propagation elevates with the preheating temperature.

Predictions made by GRI-Mech 3.0 generally agree well with the experimental measurements. At high operating conditions, i.e., Tu > 550 K or ϕ > 0.90, the numerical results by GRI-Mech 3.0 incline to overshooting. FFCM-1 seems to underestimate the flame speed compared to the present experiments; however, it is suggested that FFCM-1 agrees well with the overall existing data. Predictions of USC Mech II also agree well with present experiments at stoichiometric conditions; however, USC Mech II seems to overestimate the flame speed at fuel-lean conditions.

An empirical correlation of the laminar flame speed was developed in the range of Tu = 300–800 K and ϕ = 0.7–1.0, as shown in eq , the maximum deviation of which was less than 8%.