| OCR Text |
Show Ilt = pl~ IVii I C #Apk2 Jit = ,CIJ = 0.09 l (1) (2) \'elocity predictions for the constant eddy diff usivity su bmodel are shown in Figure 1 A. This figure shows a strong vortex centered below the burners and attached to the burner wall. The shaded areas in the figures represent reactor walls and the arrows signify velocity vectors constructed from the two components o'f velocity parallel to the designated plane. The length and direction of each vector represents predicted velocity for the location specified by the vector tail. In order to reduce congestion, less than half of the computational nodes are represented with vectors. There are two sizes of unfilled arrowheads and their ratio along with the scale for the smal1 vectors is given Dear the t.op of each figure. Figure 1 B provides the flow field prediction for the central width plane using a Prandt rs mixing length turbulence model. This figure shows a highly viscous type flow, especially in the near- burner region and the absence of major vortices in this plane. The viscous flow is generated because the mixing length model predicts high eddy diffusivities produced b) high velocity gradients in the near-burner regions. The K-( turbulence submodel option yielded the most complex flow field prediction. Figure 1 C reveals numerous swirling patterns in this single plane. Separate vortices in the ash bin, above the burners, and behind each burner, as well as strong burner centerline recirculations are all predicted. The k-{ turbulence model predicted a central recirculation zone with gases flowing downward in the center of the reactor, but flowing upward near the east, west, north and south walls. Figure 1 illustrates the vast differences in both magnitude and direction of the velocities predicted using the three turbulence submodels. \ elocity measurements in the Consol furnace were made "'ith a ~ inch pitot tube connected to an electro-manometer. Experimental velocity data for the x and z components of velocity were obtained for 50 locations grouped in four horizontal plane5 all situated above the burners. The filled arrowheads in Figure 1 represent these predictions . The k-{ simulation is the only model that predicted downward flow in the reactor center and higher velocities on the burner wall than on the opposite wall. The majority of experimental velocity vectors in Figure 1 C correlate with the predictions in both direction and n1agnitude. The obvious exceptions are the cent.er data ,-ectors at a normalized height of 0.58 and 0.70. The most probable cause for thi5 discrepancy is the inability of the k-{ n10del to correctly simulate the size of large scale vortices. An earlier study on two-dimensional swirling flows faulted the k-{ model with o\'erpredicting recirculation zone lengths [8]. These conlparisons illustrate the importance of sophisticated turbulence closure when simulating complex flo\\' phenomena. 4 |
