OCR Text |
Show Figure 3a and 3b show schematics of the 10 c m L S B (10.16 c m ID exact) and the furnace simulator at UCICL. Because this is the first attempt to design a large LSB, we have chosen to use a linear scaling method based on the dimension of the 5 c m LSB. By increasing the diameter from 5.28 c m to 10.16 cm, the non-dimensional parameters R /R (=0.06) and £IK (=2.7) are held nearly constant. The swirl injector radius R^ is then 3.2 m m and an exit tube length, £, is 140 m m . The exit tube is also tapered to 45 . The fuel injection point is also scaled linearly to 80 c m upstream of the exit rim. The natural gas injector is a small pipe with holes pointing upstream against the incoming reactant air. The two perforated plates enhance the mixing of the fuel and air. Immediately upstream of the swirler section, a 7 c m thick section of honeycomb material is used to remove large scale turbulence structures due to flow oscillations created in the premixing zone (Figure 3 a). The large L S B is mounted to fire horizontally. The interior dimensions of the furnace simulator are 240 c m square by 300 c m long (Ac / Ab = 733) with the exhaust at the wall opposite to the burner. Viewports give visual access from the front, rear, and side of the furnace. More details of the U C I C L test chambers can be found in [13, 14]. RESULTS AND DISCUSSIONS Optimizing the emissions of N O x and C O is an important criterion for setting the operating conditions of lean premixed burners. This is because C O emissions increase with excess air while N O x emissions decrease. To set the test conditions for the large LSB, w e used, as a guide, the tests results obtained in the L B N L water heater simulator. Figures 4 and 5 show, respectively, the N O x and C O emissions of this system over a range of equivalence ratios, (J), between 0.70 and 0.90 with input powers varied from 12 to 18 k W . With only a weak dependence on input power, N O x is primarily a function of <]) and increases from 5 p p m at <f> = 0.70, to about 35 p p m at (j) = 0.90. This dependence of N O x on § is a direct consequence of the increase in flame temperatures. For lean premixed methane/air flames, the production of N O x is primarily attributed to the so-called "thermal" N O x . A s described by the Zeldovich mechanism, this production is an exponentially increasing function with temperature. As the L S B is a premixed system, input power has little effect on flame temperature. Therefore N O x emissions can be controlled through the fuel-air ratio <{>. O n the other hand, the C O emissions of Figure 5 show a strong inverse dependence on (j) as well as on the input power. For <)) < 0.8, the C O emissions are consistently higher than 250 ppm. At <}) = 0.8 these results show N O x of about 15 p p m and C O less than 100 ppm (at 18 k W ) . Therefore, (j) = 0.8 was chosen as the condition to test the large LSB. As the 10 cm LSB is a linearly scaled-up version of the 5 cm LSB, the throat area is almost four times larger. W h e n operating with the same reference velocity, U_ and (j), it has close to four times the input power. Figure 4 compares the operating regimes of the 5cm L S B in the burner evaluation facility and the 10cm L S B in the furnace simulator. All test conditions are with (J) = 0.8. The operating regime of the L S B is defined by the range of swirl numbers, Sg,, that support stable flame propagation. It is shown here as a function of the reference velocity U^. This is the velocity computed based on the total volumetric 6 |