| OCR Text |
Show 4.2. Validation of modeling To validate the C F D analysis pre-processed by if-Diss, a parametric study of FDI regenerative burner was conducted. Figure 9 shows schematics and design parameters of FDI regenerative bumer[4]. The burner injects fuel gas and the air directly into the furnace. Fuel and air jets induce a large amount of combustion products before being mixed and ignited, generating a low temperature relatively lazy flame, which leads to substantially low N O x emission. Design parameters of the burner are exit velocities of gas and air, the angle and P C D (Pitch Circle Diameter) of gas and air nozzles. Simulated results were compared with the experimental result in terms of flame length and N O x emission. To predict flame length and NOx emission of the FDI regenerative burner, the mixing process between fuel and the air, the behavior of self induced flue gas recirculation and the distribution of temperature and gas species in the FDI flame must be simulated properly. The two-step reaction eddy dissipation model[5] was employed to simulate turbulent combustion process consequent to the fuel and air mixing, which can not be simulated by the heat release distribution model, the default setting of if-Diss. Thermal N O was calculated by using a thermal N O x post-processor[6]. The experimental furnace had a pair of FDI burners on both of the front and the rear end with one burner in firing mode and the other in the flue mode. With the ignorance of buoyancy, a quarter of furnace domain was simulated as shown in Figure 10. C F X code was employed as a solver. Due to the parametric study, similar pre-processing was needed for many times. Use of if-Diss substantially reduced the time needed for pre-processing. It took three quarters shorter time than otherwise using the conventional method of pre-processing. The calculation results are shown in Figures 11 to 13. Figures 11 and 12 show comparison between calculated and measured flame length and N O x concentration at the flue, with the parameter of nozzle angle. Flame length was determined by measuring visible flame in the experiment. Calculated flame length were determined by applying the assumption that the contour line of C O concentration being 0.1 % was equivalent to visible flame shape. Calculated flame length when nozzle angle was 15 degree was 2.4m, while measured value was 3.0m. This result was acceptable if considering practical measurement of fluctuating flame. Figure 12 shows that with the increase in nozzle angle, N O x emission increased. Both calculated and measured result showed similar trend. Figure 13 shows the distributions of vector, temperature and mass fraction of N O at the symmetric plane in the center of nozzles. Figure 13(a) and (b) shows the results of the nozzle angle 0 and 30 degree respectively. W h en nozzle angle was 0 degree, fuel and air mixed slowly inducing sufficient amount of combustion products in the furnace, creating relatively low temperature flame. W h e n nozzle angle was 30 degree, fuel and air mixed relatively quick creating rather high temperature region, thus higher N O x concentration. The effect of other burner parameters showed the similar trend with the experiment as well. From the comparison between measured and predicted NOx emission, it was clearly demonstrated that the simulated results pre-processed by if-Diss showed reasonable agreement with measured values, suggesting that the accuracy was acceptable in practice. if- Diss can be used for parametric study to optimize furnaces and burners configurations. 6 |
