OCR Text |
Show A=0.6; A=1.2; A=0.3. O 0.189. L C V 28.46 MJ/kg; proximate analysis (as fired) fixed carbon 0.518, volatilcs 0.356, ash 0.106 and moisture 0.02. Prior to the flame computing, the hvBb coal was characterized in a vertical furnace. The devolalilization data are shown in Figure 3 while Ihc cxpcrimcnlaly derived char combustion model parameters take values 1^=1.8 kg n r 2 s Pa-0-5 and Ec=92.1 kJ/mol. Tvpe-2 Flames Of Pulverized Coal. A comparison between computed and measured temperature and oxygen fields is given in Figure 7. A n important quantity to verify is the oxygen concentration in the flame. The path of the air through the flame can be followed via a sequence of maxima in the radial profiles of the oxygen concentration. The agreement between measured and predicted locations of these maxima is very good. This indicates that the path of the air in the near burner zone is correctly predicted and hence, that the predicted IRZ size is in agreement with the actual one. There is also good agreement between the measured and predicted values of the oxygen mass fraction in the combustion gases. This is a strong indication that the axial consumption rate of oxygen and hence the combustion rate is correctly predicted by the model. The predicted temperatures agree well with those measured; only in the secondary air stream the measured values are higher than the predicted ones. The oxygen concentration inside the IRZ is well predicted. The agreement is better than expected since, it has been experienced that the chemical composition inside the IRZ is sensitive to the setting of a number of model parameters. The reason for this sensitivity is that the concentration is determined by two large terms; the turbulent diffusion of oxygen into Ihe IRZ and the combustion rale within the recirculation zone. Small changes of these terms may have a significant effect on the chemical composition of the IRZ. The carbon monoxide concentrations inside the IRZ are substantially over predicted. At the quarl outlet, only a few hundred ppm were measured while around 5 % of C O were predicted. -W 0 20 60 WOcrn FIGURE 12: TYPE-2 FLAME; SENSITIVITY OF FLAME PREDICTIONS TO DE VOLATILIZATION; BADZIOCH-HAWSKLEY MODEL(1970), KNILL et al. (1989) Model sensitivity for Tvpe-2 flames An extensive sensitivity study of the predictions to many model parameters can be found in Visser (1991). Herewith only the most important effects are described; namely sensitivity to the mixing rate parameter in the eddy-break-up model, to the rate of coal devolalilization and to the high temperature volatile yield. Generally, a value of 4 is used for the mixing rate constant of eddy-break-up model (see Eg.(5)). Indications that the value of A should be in the range 0.5 to 0.7 appeared while modeling of 2.5 M W swirling pulverized coal flames (see Visser and Weber, 1989, Visser et al., 1990). Predicted profiles of oxygen mass fraction and temperature for A values of 0.3, 0.6 and 1.2 arc presented in Figure 11. A relatively high sensitivity of the chemical composition and temperature to the value of Ihc mixing rate constant is found. In particular IRZ properties arc strongly dependent on the local combustion rate. Outside the IRZ, the predicted flame properties are less dependent on the A value Simulations with A equal to 4 lead to temperatures along the IRZ boundary in excess of 2000 K and Ihe oxygen concentration inside the I R Z lower than 0.1%. The sensitivity of the predicted flame properties to the rate of devolatilization is investigated using the (B-H) Badzioch-Hawskey (1970) model and the model of Knill et al. (1989). The first model, when applied using the originally proposed set of constants, substantially underpredicts the rale of devolalilization. For example, according to the B-H model a 80 micron meter particle in 1270 K environment would give off 6 5 % volatilcs within 80-90 ms while Knill's calculations predict less than 40 m s (sec Figure 3). Figure 12 shows the effect of slow and fast dcvolatilizalion on the flame predictions. The slow devolalilization results in a substantial underprediction of temperature and ovcrprediction of oxygen concentrations. The flame properties inside and just downstream of the burner quarl are determined by the availability of volatiles. W h e n the instantaneous devolalilization model is used, it is predicted that the combustion rate inside the burner quarl is determined by oxygen diffusion. n 1 1 - 11 |