| OCR Text |
Show Figure 13 displays the average of 100 O H emission intensity images obtained with a camera positioned at the furnace exhaust end and facing an A L G L A S S ™ burner operating at 1.5 MMBtu/hr. The image provides information on flame geometry, stability, and a qualitative indication of temperature which can be correlated with N O x production. Real-time variations in the O H emission intensity are shown in Figure 14 which illustrates the change in the O H emission as the N G flow is being oscillated from fuel-rich to fuel-lean conditions [19]. In the sequence of images shown in Figure 14. frames 1 and 2 are in the fuel-lean portion of the oscillating cycle while frames 3-6 represent the fuel-rich section. For the fuel-lean cycle the area of the O H emission is smaller compared to the fuel-rich due to the reduced flow of fuel. Increasing the fuel flow expands the area of O H emission and creates a small area void of any emission due to the center core of fuel being injected. A correlation between the OH emission images and NOx production can be performed for the oscillating example by averaging hundreds of images and integrating the image to obtain the total O H emission. Normalizing the total O H emission by the total average O H emission observed for the case with no oscillation can be conducted for various oscillating conditions and correlated with N O x flue gas measurements. Results using this procedure are shown in Figure 15 for different oscillating frequencies and percent of the total N G being oscillated. From the results on Figure 15 a strong relationship between N O x and O H emission is observed. It can be noted that as the frequency of oscillation increases or the amount of N G being oscillated decreases the N O x level approaches the case of no oscillation. i.e.. standard A L G L A S S ™ operating conditions. The technique demonstrates the sensitivity of the O H emission to the flame performance. 4.3 Planar Mie scattering for flow mixing characterization Mie scattering has been carried out in the Air Liquide Countryside facilities and is currently being implemented on the 2 M W furnace in France. Experience gathered over the past years has shown that planar Mie scattering visualization is a powerful, yet simple method to characterize flow mixing in industrial natural gas flames [3,21.22]. The technique has proven especially useful to burner designers to visualize the interactions between the fuel and oxidant streams and to optimize the fuel and oxidant injection characteristics. This laser sheet technique is also well suited to characterize the soot spatial envelope in natural gas flames, which supplements the local soot concentration data obtained from techniques described above. The main advantage of the Mie scattering technique over other imaging techniques is its applicability to 2.0 M W scale pilot furnaces. The Mie scattering visualization technique consists in introducing marker particles in a burner oxidant or fuel stream and illuminating a flow section with a laser sheet. The laser light scattered by the particles as they cross the laser sheet is recorded by a monochrome camera. The Mie scattering image yields a quantitative or semi-quantitative measurement of the particle number density field. In the present work [11], the laser sheet was produced by a pulsed N d : Y A G laser. It was introduced horizontally from the furnace slot and crossed the burner axis. Either the natural gas or oxygen stream could be seeded with zirconium oxide particles. Images of the instantaneous particle concentration field were recorded with a black and white C C D camera which was externally triggered by the laser pulses. Continuous flame and wall radiation could be totally eliminated by using a camera gating time of 1 (as. A computer software developed by the IFRF [20.21] was used to automatically analyze ensembles of 250 instantaneous images. In order to relate the image gray level to a stream particle number density, the software first rescales the images gray levels so that the maximum signal intensity measured in the jet potential core corresponds to 1 0 0 % and the minimum signal intensity7 corresponds to 0. The analysis yields several time-averaged parameters that quantitatively or semi-quantitatively characterize mixing. They include the mean, standard deviation, intermittency. spatial gradient, and normalized radial profiles images. Figures 16 and 17 display analyses from visualization of the oxygen stream of a coaxial oxygen/natural gas burner. The camera field of view covers an area of 650*430 m m. Figure 16 was obtained by averaging the norm of the spatial gradients of 250 particle number density images. It highlights the location of the oxygen stream mixing layers, the jet entrainment boundary, and the rapid reduction in mixing intensity along the burner axis. Figure 17 displays the normalized mean concentration field obtained by normalizing the radial mean concentration profiles by its m a x i m u m value on the radial profile. The 5 0 % normalized mean profiles can be used conveniently to define the beginning of the self-similar region, as well as the jet half angle and apparent origin. A s from Figure 16, it can be seen that the oxygen stream is very slow to expand towards the central combusting natural jet, while its outer expansion angle is larger than for a free jet. Both images confirm previous observations that mixing of a central natural jet with an annular oxygen jet occurs at a much reduced rate when combustion takes place. 7 |
