Effects of Equivalence Ratio on the Combustion Performance of Staged. Swirl Flame

Transcription

1 Effects of Equivalence Ratio on the Combustion Performance of Staged Swirl Flame Bing Ge*, Yinshen Tian, Yongbin Ji, Shusheng Zang, Jianhua Xin Institute of Turbomachinery, School of Mechanical Engineering, Shanghai Jiao Tong University Dongchuan Road 800, Shanghai , China Mingmin Chen, Guangyun Jiao, Dongfang Zhang Shanghai Electric Gas Turbine CO.,Ltd, Shanghai , China ABSTRACT The effect of equivalence ratio on the combustion performance was experimentally investigated at atmospheric pressure conditions using a radical-staged DLE burner. The overall combustion equivalence ratio in experiment is 0.50~0.75. Planar laser-induced fluorescence (PLIF) of OH radical measurement is adopted to identify main reaction zones and burnt gas regions. And gas analyzer was used to obtain NOx/CO emissions. Moreover, dynamic pressure sensors and OH planar laser-induced fluorescence was employed to identify combustion instability and reaction zone distribution in the staged swirl combustor. The experimental results show that at equivalence ratio lower than 0.65, flame center is close to the burner and the pressure fluctuation inside the combustor is limited, which means no periodic pressure oscillation is observed. The emission of CO is high while the NOx is low. However, at equivalence ratio higher than 0.65, transition of flame characteristics, pressure as well as emission occurs. To be specific, the center of the flame presents away from the burner with periodic pressure oscillation coming out. Also at this condition, CO emission is low while the NOx emission is high. So equivalence ratio of 0.65 performs as a critical value for this burner and test. 1. INTRODUCTION There have been demands for lowering NOx emissions from gas turbines to meet increasingly stringent emissions regulations. Since NOx formation decreases drastically with decreasing flame temperature, efforts have been devoted to preparing more homogeneous fuel-air mixtures and to burning at leaner conditions in gas-turbine dry low emission (DLE) burner [1]. The DLE burner premix air and fuel prior to injection into the combustor and tune the mixture equivalence ratio to optimize operations, match power level to ambient conditions or achieve a desired power output (staged systems). The understanding of combustion in gas turbines has been strongly improved by the supporting effort of laser-based experimental studies in GT model combustors [2]. In particular, GT model combustors with optically accessible combustion chambers which facilitate the application of laser and optical measurement techniques enabled a deeper insight into the complex interaction between flow field, combustion and acoustic modes of the system [3]. Among the main advantages of these techniques is their ability to non-intrusively measure instantaneous two-dimensional distributions of various quantities, to image short-lived combustion radicals and heat release rates, to measure several quantities simultaneously and to capture the temporal development using high-speed imaging techniques [4]. Main research topics in GT combustion have been thermo-acoustic instabilities, coherent flow structures, flame stabilization, pollutant emissions, effects of mixing and turbulence chemistry interaction, fuel flexibility, flashback and the potential of new combustion systems. Many laser-based studies in GT (model) combustors addressed these topics, for example, the dynamics of swirl flames by using particle image velocimetry (PIV) for the characterization of the flow field or planar laser-induced fluorescence (PLIF) for the measurement of the flame front, often in combination with chemiluminescence imaging [5]. PLIF was also applied to investigate the mechanisms of flame stabilization [6]. The role of equivalence ratio fluctuations and effects of unmixedness in lean premixed flames were investigated by Schürmans et al. [7] using chemiluminescence and infrared absorption measurements and by Meier et al. [8] using laser Raman scattering. The primary objective of this study is to experimentally investigate the effect of equivalence ratio on flame structure and combustion dynamics in radially-staged DLE burner. The burner was investigated in an atmospheric combustion test rig with optical access for application of optical and laser diagnostic techniques. NOx and CO emissions are measured by an infrared gas analyzer. OH PLIF is used to examine the flame structure, and used to identify the reaction zone and post flame regions from a location defined by a laser sheet.

2 2. EXPERIMENTAL SETUP 2.1 Experimental Apparatus The experiment was conducted on the advanced high temperature rise combustion test bench in Institute of Turbomachinery (TMI) of Shanghai Jiaotong University. Fig.1 depicts the schematic view of the experimental system. The setup is mainly made up of four parts, mainstream air system, pilot and main fuel system, PLIF system and test section. An 130 kw screw compressor with maximum air capability of 1440 m3/h is used to supply the main flow going through the combustor at atmospheric pressure. And 72KW electric heater is used in air supply system. The optically accessible combustor is made of quartz with an inner diameter of 250 mm and a length of 300 mm, providing full optical access to the flame. After burning in the combustor, the burned gas is exhausted into the water-cooled exhaust section. Fig.1 Sketch of the combustor and test rig OH-PLIF Measurement zone 1.4D 1.4D Main flow Quartz combustor Pilot swirler main swirler Fuel orifice Pilot fuel Main premixed fuel Fig.2 Configuration of DLE Burner and OH PLIF measurement zone The structure of the radially-staged DLE burner is shown in Fig.2. The flame from pilot nozzle is non-premixed, and the flame from main nozzle is lean premixed. Pilot and main fuels are controlled by two mass flow controllers respectively. The outlet diameter of DLE burner (D) is 70mm. 2.2 Pressure and OH-PLIF Measurement There are six high frequency dynamic pressure sensor used in experiment. The pressure fluctuations in the combustor and air inlet chamber are measured using Kulite XTL-190(M) pressure transducers (P1, P2, P3) and WCTV SG pressure transducers (P4, P5, P6), with a sampling rate of 2500 Hz. Planar laser induced fluorescence (PLIF) of OH radical is employed to measure flame structure and heat release rate. The excitation laser derives from a pulsed Nd: YAG laser pumping a tunable dye laser with Rhodamine 6G as dye solution before going through a frequency double crystal. The output ultraviolet laser beam has the wavelength of nm with pulse duration of

3 20ns at the power of 70 mw, which is used to excite OH radicals. The ultraviolet laser beam is expanded by a set of spherical and cylindrical lenses, forming a laser sheet with the thickness less than 500μm. The laser sheet is guided horizontally through over the center of three nozzles. The excited fluorescence is then collected by an ICCD camera placed perpendicular to the laser sheet plane with Nikon UV lens, in front of which a combined UG11 and WG305 interference filter set is installed to suppress scattered laser light and background flame radiation. Timing delay of PLIF system is controlled by a pulse delay generator DG535. The exposure time of ICCD camera is set to 50ns to include a complete OH fluorescence for each instantaneous laser shot. Fig.3 shows the corresponding PLIF imaging region (98mm 98 mm). 2.3 Operating Conditions Experiments were carried out with natural gas from high pressure gas cylinder, which consists of almost pure methane. The experimental conditions were set as follow: P=1.0atm, T air,in=600k, T fuel,in=288k, Q air,in=120 Nm 3 /h. Table.1 lists the specific testing conditions for different equivalence ratio cases. Strictly speaking, only fuel amount is adjusted to reach set points for different conditions without changing air amount, which will cause the velocity out from the nozzle to be varied. Thanks to that the fuel amount is quite small compared to air amount, the highest and lowest velocity at the nozzle outlet only differ in the range of 2.5%, which can be neglected. Table.1 Test matrix No. Equivalence ratio, Air inlet temperature, T air,in (K) Air amount, Q air,in (Nm3/h) Fuel ratio Premixed Pilot fuel fuel % 11% RESULTS AND DISCUSSION 3.1 Flame structure Fig.3 shows the results of flame photos as well as time-averaged PLIF images for different testing conditions, which gives direct impression how the flame response to equivalence ratio variation. It can be seen that flame locates in the almost same range inside the combustor as equivalence ratio is adjusted to increase in the wide range of 0.5 to During increase of ϕ, flame stretches to the combustor liner wall. It is very obvious that flame at the near wall region can be observed at ϕ 0.75 case from right column PLIF results. Also the phenomenon that infrared signal gets more and more intense as equivalence ratio is elevated to be higher and higher clarifies that temperature inside the combustor is rising. Detail flame topology can be told from OH-PLIF results. When the equivalence ratio is low ( 0.50), chemical reaction is mainly distributed in the spindle reaction zone. The shape of reaction zone is similar to the diffusion flame, but the ratio of region with high OH-PLIF fluorescence signal to the overall measurement region is larger than diffusion combustion, which proves that premixed reaction distributes in a wider area. Besides, OH intensity is distributed in a more smooth pattern in the reaction zone, representing that concentrated high temperature spot is not severe than diffusion reaction. With increase of equivalence ratio, root of the flame starts to gather to the combustor centerline gradually, while two tips of the flame tend to depart from each other to the liner wall. Especially at the highest equivalence ratio case in current study ( 0.75), reaction zone near the axis (marked with black dash dot line) and near the liner wall seems to be connected and form a reaction band or strip across the whole combustor. Center of reaction region is calculated as centroid of OH distribution from PLIF images for different conditions, as shown in Fig.4 and Fig.5. Fig.4 shows that although flame shape are different at different cases, the centroid of reaction region stays consistent,. It is 0.73D away from the center axis of the combustor in the lateral direction and 0.75D from the combustor front endwall in the streamwise direction. This is because ratio of premixed fuel to pilot fuel is kept same although equivalence ratio varies. Opening angle of flame is also shown in Fig.6 pointing out that the angle decreases as equivalence ratio increases. It can be concluded that center of reaction area is closer to the combustor axis and front endewall at three lower equivalence ratio conditions. While at three higher equivalence ratio cases, increment of the fuel amount causes it take more time finish

4 chemical reaction. That s why the flame tends to reach out to the combustor liner wall as well as move downwards. In addition, the difference between 0.60 and 0.64 is notable, which is also can be seen from OH-PLIF results. That is, as equivalence ratio increases from 0.60 to 0.64, high temperature reaction region (red and white area) at two sides of the combustor axis starts to get connected from detached status. At the same time, area near the liner wall presents to be at high temperature also. So there is a significant flame structure transition from 0.60 to 0.64 for the current studied nozzle. (a) 0.50 (b) 0.55 (c) 0.60 (d) 0.64 (e) 0.70

5 (f) 0.75 Fig.3 Flame pictures (left) and OH PLIF average images (right) at different conditions Fig.4 Axial distance of centroid of reaction zone Fig.5 Radial distance of centroid of reaction zone Fig.6 Flame opening angle (in degree)at different conditions Fig.7 OH signal intensity at different conditions The maximum and average OH signal intensity in the investigated area for different cases are plotted in Fig.7. The transition from the condition 0.60 to 0.64 is also evident here for both maximum and average results. According to the correlation between OH signal to temperature, it means temperature inside the combustor undergoes a significant rise from 0.60 to However, it doesn t change too much when equivalence ratio is lower than 0.60 or higher than Dynamic pressure characteristics

6 Fig.8 Dynamic pressure fluctuation at different conditions Dynamic pressure inside the combustor is monitored and illustrated in Fig.8. Dynamic pressure is nearly flat without obvious fluctuation for low equivalence ratio cases ( 0.50 and0.55), and periodic pressure pulsation is distinguished for higher equivalence ratio ( 0.64, 0.70 and 0.75). The fluctuation amplitude for 0.60 is relatively less. Maximum dynamic pressure fluctuating amplitude is displayed in Fig.9. Again the transition shows up as equivalence ratio is increased from 0.60 to 0.64, the maximum amplitude increases from bar to 0.037bar (95%). at conditions <0.6,the maximum amplitude improves gradually as equivalence ratio is raised. at conditions 0.64,maximum amplitude almost keeps at the level of bar not to increase with the equivalence any more. Fig.9 Maximum amplitude of dynamic pressure fluctuation at different conditions Fig.10 FFT of dynamic pressure signal at different conditions Fast Fourier Transform (FFT) is carried out on the dynamic pressure signal at different conditions, as shown in Fig.10. As discussed above, OH signal results (Fig.7) doesn t show any trend of increasing or decreasing as the equivalence ratio is raised. It just tells two types that the lower equivalence ratio group ( 0.50, 0.55 and 0.60) has lower OH signal intensity and higher equivalence ratio group ( 0.64, 0.70 and 0.75) gets higher OH signal intensity. For the FFT result of dynamic pressure, the

7 conclusion is also similar as OH intensity. GPPS NA Fig.11 Comparison of FFT results at different conditions Fig.11 summarizes FFT results for difference equivalence ratio cases. Only a relatively low characteristic frequency near 75Hz is discernable for the lower equivalence ratio group ( 0.50, 0.55 and 0.60) without any other clear characteristic frequency. However, at the higher equivalence ratio group ( 0.64, 0.70 and 0.75), amplitudes at four characteristics frequencies are high, they are 40Hz, 80Hz, 120Hz and 160Hz. Although the amplitude is high, the dynamic pressure fluctuation amplitude is still under 5% of the average pressure inside the combustor. So, all the cases are viewed as reacting in stable status. 3.3 Outlet temperature and emissions Fig.12 Outlet temperature at different conditions Fig.13 NOx and CO emissions at different conditions Average temperature at the model combustor outlet is presented in Fig.12. Outlet temperature is not as sensitive as flame and dynamic pressure to the transition mentioned above. It goes up as equivalence ratio is elevated. Fig.13 shows NOx and CO at the model combustor outlet during experiments. The overall trend is that NOx increases with equivalence ratio, while CO goes down when equivalence ratio is increased. Bounded by the condition of 0.64, CO emission is very high when equivalence ratio is below 0.64 and it is all lower than 3ppm when equivalence ratio exceeds There isn t any separation point for NOx emission. It can be explained by OH-PLIF signal intensity as shown in Fig.7. Average OH intensity is low at lower equivalence ratio conditions means average temperature is also relatively low. At higher equivalence ratio, it is opposite. Chemical reaction will be not enough at low temperature, which means unburnt CO can t be further consumed, leading to high level CO emission at 0.50, 0.55 and Maximum OH intensity doesn t show the same transition process, that is to say, the highest temperature inside the combustor doesn t undergo the transition. NOx emission is mainly influenced by the highest temperature, that s the reason why NOx emission is not experiencing transition. Concluded from the emission characteristics, both NOx and CO obtain low level emission at Although pressure fluctuation amplitude is a little greater at 0.64 than that at conditions with lower (0.50, 0.55, 0.60), the combustion doesn t reach to oscillation status. Compared with higher (0.70, 0.75), dynamic pressure and emission characteristics are similar at 0.64, but chemical reaction near the liner wall proves to be more intense at higher conditions, which will cause higher heat load on the combustor liner and thus make the liner wall suffer damage risk. To summarize, operation condition for the

8 real gas turbine is chosen according to flame structure, dynamic pressure, outlet temperature and emission characteristics, which verifies the reasonability of current scaled combustor and nozzle study based on the real operation condition parameters. Fig.14 Relation between overall equivalence ratio and fuel percentage of CH 4 Overall equivalence ratio of the burner and CH 4 concentration in the main pre-mixed nozzle can be obtained by calculating the air flow distribution through main injector and pilot injector, as shown in Fig. 14. As can be seen from the figure, when the equivalence ratio is 0.50, 0.55 and 0.60, CH 4 concentration in the main burner is lower than the ignition limit. The premixed gas at the exit of the main burner is heated and consumed to chemical reaction when passing through the high-temperature recirculation zone downstream of the burner, so effective flame front can t be formed. Flame front is mainly concentrated in the reaction zone of pilot nozzle, as shown in Figure 3(a), 3(b), 3(c). When the equivalence ratio is 0.64, 0.70 and 0.75, the concentration of CH 4 in the main burner is higher than the ignition limit, so premixed combustion occurs at the outlet of the main burner after the premixed gas is discharged, forming a conspicuous flame front between the pilot flame and the combustor wall as shown in Figures 3(d), 3(e), 3(f). In the above conditions, the main mode is premixed combustion, so it is easy to generate combustion oscillation. CONCLUSIONS The combustion characteristics of the current radially-staged DLE burner fueled with natural gas have been investigated with varying equivalence ratio from 0.5 to The experimental results led to the following conclusions: 1) When equivalence ratio is lower than 0.65, flame center is close to the burner and the pressure fluctuation inside the combustor is limited, which means no periodic pressure oscillation is observed. The emission of CO is high while the NOx is low. 2) When equivalence ratio higher than 0.65, transition of flame characteristics, pressure as well as emission occurs. To be specific, the center of the flame presents away from the burner with periodic pressure oscillation coming out. Also at this condition, CO emission is low while the NOx emission is high. So equivalence ratio of 0.65 performs as a critical value for this burner and test. 3) When the fuel concentration is lower than ignition limit, non-premixed combustion is the main combustion mode in the combustor, leading to high CO emission but combustion instability avoided. When the fuel concentration is higher than ignition limit, premixed mode is dominant and reaction is sufficient resulting in obvious combustion oscillation. REFERENCES [1] Griebel, P., Siewert, P., and Jansohn, P., Flame characteristics of turbulent lean premixed methane/air flames at high pressure: Turbulent flame speed and flame brush thickness. Proceedings of the Combustion Institute, 31(2), pp [2] T.C. Lieuwen, V. Yang, Combustion instabilities in gas turbine engines, Progress in Astronautics and Aeronautics, vol. 210, AIAA, USA, [3] Kim, K.T., Combustion instability feedback mechanisms in a lean-premixed swirl-stabilized combustor. Combustion and Flame, : p [4] Duan, X.R., Meier, W., Weigand, P., and Lehmann, B Phase-resolved laser Raman scattering and laser Doppler velocimetry applied to periodic instabilities in a gas turbine model combustor. Appl. Phys. B, 80, [5] Sadanandan, R., Stöhr, M., and Meier, W Simultaneous OH-PLIF and PIV measurements in a gas turbine model combustor. Appl. Phys. B, 90,

9 [6] M. Stöhr, I. Boxx, C. Carter, W. Meier, Dynamics of lean blowout of a swirl-stabilized flame in a gas turbine model combustor, Proc. Combust. Inst. 33 (2011) [7] B. Schürmans, V. Bellucci, F. Güthe, F. Meili, P. Flohr, C.O. Paschereit, a detailed analysis of thermoacoustic interaction mechanisms in a turbulent premixed flame, Proc.ASME Turbo Expo 2004, GT [8] W. Meier, P. Weigand, X.R. Duan, R. Giezendanner-Thoben, Detailed characterization of the dynamics of thermoacoustic pulsations in a lean premixed swirl flame, Combust. Flame150 (2007) 2 26.