| OCR Text |
Show flow of the fuel air mixture divided by the burner throat area. For the 5 c m LSB, w e found it to operate between 2.7 < U_ < 16.6 m/s (18 k W to 106 k W ) corresponding to a turndown of 6:1. The minimum Sg requirement is about 0.03 and is consistent with the results obtained using the L B N L water heater simulator. Our previous study showed the upper Sg limit to be greater than 0.15. The 10 c m L S B was successfully tested at 6.2 < U.. < 24.8 m/s (146 k W to 585 k W ) corresponding to a 4:1 turndown. However, the 10cm L S B requires a higher swirl rate as the minimum Sg is about 0.06. Under the condition of low U_, w e found an upper limit of Sg for the 10cm L S B which created an operating range that is quite narrow. However, as shown by the data at U_ = 12 m/s, the operating range increases substantially with increasing input power. The increase in swirl requirement for the large LSB is a direct consequence of linear scaling. This scaling method is essentially a constant velocity approach that ratios the input power to R2 with R being the throat radius. The swirl processes, however, does not scale with R. Instead, the swirl requirement is proportional to the residence time of the swirl flow. At a given U„, Sg is a function of the length of the burner tube, /, above the swirler. Previous study [11] had shown that the swirl requirement increases with /. This is because the longer residence time weakens the swirl interactions and therefore a higher swirl flow is required to generate the necessary flow divergence for flame stabilization. With an exit tube of £ = 140 m m compared to 70 m m in the smaller LSB, the higher swirl requirement of the 10cm L S B is consistent with this explanation. These test results are extremely encouraging, They demonstrate that the LSB concept is easily scalable to different input powers using the constant velocity scaling approach, and that the swirl requirement can be predicted using a constant residence time analysis. Moreover, the relatively constant swirl number found at different input power suggests that the control needs for the L S B should be quite straightforward. It should also be noted that the maximum input powers shown here are not the highest attainable in either burner. In both cases, the test matrix was limited by the capacities of the peripheral components of these facilities. The water cooling circuit of the generic burner evaluation facility limited our tests to less than 106 k W . At the furnace simulator, the pressurized air supply for our air-swirler could not delivery the high flow rates needed for testing beyond 585 k W . Therefore, the potential for firing well beyond 1 M W with a 10 c m L S B should be quite good. Of course, the most important attribute of the LSB is its low NOx emission. Figure 7 compares the N O x emissions measured in the L B N L water heater simulator (5cm L SB Ac / A b = 15), the U C I C L burner evaluation facility (5 c m LSB, A c / A b = 142) and the U C I C L furnace simulator (10cm LSB, A c / Ab = 733). All the test conditions were maintained at (j) = 0.80. The data of Figure 7 are plotted against U „ but upper and lower axes are relabeled to give a direct reference to input power. Therefore, the horizontal scale for the small (upper axis) and the larger L S B (lower axis) differ by approximately a factor of four. The results are again very promising and clearly show that N O x emissions of the L S B are not sensitive to burner size, enclosure size and input power. The N O x level 7 |
