OCR Text |
Show temperoture (K) 8 meosured flow pot tern 06 b 1 -^Ov \N^ - \ \- ^^""^Nw*.* * \X \; \ \*' J. (. / ^^v^/ i \ -' V i \ V \ \ \ \\ \\ "™ I \ 5\x VI' 1 '.\ 1 1 -w 0 20 60 WO an mass fraction of oxygen (%1 .0 5 10 6 0 5 10 15 0 5 1 0 1 5 0 5 10 6 0 510 15 20 WO an FIGURE 7: TYPE-2 FLAME; COMPUTED AND MEASURED TEMPERATURE AND OXYGEN INSIDE THE AAS BURNER; ASM COMPUTED, K-e COMPUTED (-10cm, 0cm 100cm - Distance From The Furnace Front Wall) and cither RSM or ASM is applied. The k-e computations failed dready in the burner quarl where flow expansion is inviscid (see Figure 6). Neither the (anisotropic) generation of turbulence not the distribution of the tangential momentum in the downstream flow could be correctly predicted by the k-e model. N o w , the question arises whether second-order turbulence modeling is also required for computing type-2 flames of pulverized coal. Rigorous assessment of performance of the k-e, ASM and RSM in the near burner zone of pulverized coal flames is not possible for the absence of measured data containing velocity and turbulence. Thus, it is usually assumed that the predicted flow patterns arc in agreement with the existing velocities when the predicted and measured distributions of temperature and chemical composition agree. It has been experienced that the oxygen distribution in a flame provides a reliable indication of the accuracy of the predicted flow pattern since small changes in the predicted flow patterns have generally a very significant effect on the predicted oxygen distribution. Figure 7 shows the computed temperature and oxygen fields inside and just downstream of the quarl of the A A S Burner. The compulation were performed using a fine 75 x 56 grid with 20 x 33 nodes located inside the burner quarl. Only small differences can be observed between the k-e and the A S M predictions. Both turbulence models perform very well. The computed flame of 3.4 M W thermal input is a typical type-2 flame of 0.96 inlet swirl (sec Smart et al., 1989). The ignition front is located in the close vicinity of the fuel injector and the flow pattern is very similar to the flows of typc-2 flames of coke-oven gas which were considered by Weber and Dugud (1992). Thus, the 'effects of combustion on the swirling flows must be similar in both cases. The detailed laser vclocimctry and flow-field analysis of the coke-oven gas flames have proven that in type-2 flames the basic effect of combustion on swirling flows is to reduce the importance Q2 0.4 0.6 0.8 E (ml ASM 02 0.4 0.6 0.8 * (m) JL±. 02 0/ 0J6 08 z (ml FIGURE 8: MEASUREMENTS AND PREDICTIONS OF ISOTHERMAL SWIRLING FLOW WITH A CENTRAL NON-SWIRLING JET, So=1.5. w of the centrifugal forces with respect to the inertia forces by increasing the latter substantially. Despite 0.7 or 1.4 inlet swirl level, the effective swirl number of the type-2 coke-oven gas flames is low being in the range 0.1 to 0.2. Thus, the flow pattern inside the A A S Burner resembles strongly a swirling ambient air flow of 0.15-0.2 inlet swirl and therefore the k-e predictions are of the same quality as the A S M results in the type-2 pulverized coal. The work of Weber and Dugue* (1992) has demonstrated that the combustion induced effects on swirling flows depend on the combustion front location. When the original cold vortex is combustion accelerated in the vicinity of the quarl inlet, the IRZ strength and size are substantially reduced. W h e n the same vortex is combustion accelerated downstream of the burner quarl. the IRZ is increased in size and strength. The above observations have important implication for mathematical modeling of pulverized coal flames. In the 3.4 M W pulverized coal flame, the volatilcs arc given off rapidly and in large quantities. This results in the ignition front located at the coal injector. Combustion of an anthracite coal in the same A A S Burner would result in the ignition front located downstream of the burner quart. The flow-field inside the burner would not be so drastically affected by combustion *and it is expected that the R S M (or A S M ) calculations would offer improvements over the k-e predictions. The above considerations on the pulverized coal predictions could be carried out without referring to the coke-oven gas flames if there were velocity and turbulence data available for the 3.4 M W type-2 flame considered. However, it was not until very recently that Laser Doppler Technique has been adapted to allow for velocity measurements in a 0.5 M W (Brodbek, 1991) and 2.5 M W flames of pulverized coal (Dugue* and Weber, 1992). 1 1 - 11 |