OCR Text |
Show temperature (K) 8 0 20 60 WOan mass froctloo of oxygen (%) .0 5 10 15 0 5 10 15 0 5 10 15 0 5 W 15 0 5 10 15 20 W0 cm FIGURE 13: TYPE-1 FLAME; SENSITIVITY OF FLAME PREDICTIONS TO THE MIXING RATE CONSTANT ; A=0.6;- - - A=1.2;-~A=0.3 12 18 axial distance (m) The sensitivity to the high temperature volatile yield ( V M ^ , ^ is examined with the yield ( V M m a x ) reduced from 6 5 % to 47%, while the char fraction is simultaneously increased from 3 5 % to 53%. The decrease in V M m a x corresponds to a change in the Q factor from 1.6 to 1.2. For high volatile bituminous coals, the Q factor is often estimated as 1.2. The calculated profiles of temperature and oxygen are relatively insensitive to V M m a x . A similar insensitivity was reported by Lockwood and Mahmud (1989). The insensitivity of the flame properties to V M m a x indicates that volatile combustion rale in the considered swirling flame is not limited by the availability of volatiles, but by the turbulent diffusion of oxygen from the combustion air into the IRZ. Some effects of the value of V M m a x on the predictions may be noted downstream of the IRZ. At this position, most of the volatilcs are burnt and the volatile combustion rate becomes dependent on the amount of volatiles existent The effect is however limited in the present cases, since the char combustion rate is relatively high for the applied Coal Valley coal. Thus char reaction compensates for volatile combustion. A larger sensitivity of the flame properties downstream of the IRZ to high temperature volatile yield is expected for coals with a less reactive char. Type-1 Flames Of Pulverized Coal The comparison between the measured and computed temperature and oxygen fields of the lype-1 flame is shown in Figure 9. Clearly, the differences between the predicted and measured properties arc substantial. The type-1 flame and the lypc-2 flame considered previously, were computed with an identical set of model parameters determining the devolalilization rale, the rate of volatile matter and char combustion. However, the quality of predictions for the type-1 flame is worse than that for ihe type-2 flame. It is believed that the imperfections in fluid flow calculations, in particular with respect to the depth of primary jet penetration are responsible for the lesser quality. A detailed sensitivity study of the predictions to a number of model parameters for type-1 flames of 0.9, 2.2 and 3.4 M W thdrmal FIGURE 14: MEASURED AND COMPUTED TEMPERATURE. OXYGEN (TOP) AND NITRIC OXIDE (BOTTOM) IN 2 M W NATURAL GAS FLAME inputs can be found in Visser and Weber (1990) and Visser (1991). Figure 13 depicts the sensitivity to the mixing rate constant A. However, until full confidence in calculating the fluid flow pattern for type-1 flames is gained and a methodology for verifying the fluid-flow predictions is derived (including hysteresis effects), any sensitivity study is of lower value. P R E D I C T I O N S O F NITRIC O X I DE Nitric Oxide In A Natural Gas Flame The N O x post-processor has been first tested for predictability of thermal and prompt N O . T o this end a swirling natural gas flame of 2 M W thermal input has been selected (flame 258 in the experiments of Dugue* et al. (1991)) for which the inlet conditions to the burner and thermal boundary conditions in the furnace have been measured. The flame is a high-NOx flame of 134 p p m (al 3 % 02 ) flue emissions;, neither air staging nor fuel gas staging is applied to reduce NOx. The in-flame measurements of velocities, temperature, oxygen, carbon oxides and N O x allow for validation of the thermal- and prompt-NO models. Any thermal NO calculations are dependent on both temperature and oxygen fields. Figure 14 shows the measured and predicted temperature and oxygen distribution along the symmetry axis of the furnace. The temperature in the vicinity of the burner, where most N O is formed, is predicted with accuracy of around 100 K. In the region 1.8-2m downstream of the furnace the calculated gas temperature is 200-250 K too high. It is believed that the flux-model used for radiation calculations is responsible for this discrepancy. The measured and calculated temperature and oxygen at the furnace exit are 1196K, 1218K; and 2.7%(dry), 12 11-11 |