OCR Text |
Show For the results reported here, flow and combustion models were completed using both 2-D axisymmetric and 3-D cylindrical structured orthogonal models and 2-D axisymmetric and 3-D unstructured models. Grid sensitivity studies were completed over a range of grid sizes from 7130 nodes to 47,600 nodes for 2-D models and 65,500 nodes to 153,500 nodes for 3-D models. Further discussion of grid sensitivity is reported in Kaufman and Fiveland (1995). Non-reacting Flow Predictions for the non-reacting flow case were completed using 27,000 nodes on both structured and unstructured 2-D axisymmetric meshes. As noted before, this non-reacting flow produces the primary central recirculation zone immediately behind the burner center body, but also creates a large secondary recirculation zone along the burner centerline. Figure 9 shows the comparison between measurements and predictions of axial and tangential velocity at five traverse locations. Both predictions show very good agreement with the data with several exceptions. At the 27mm location, all features of the flow are captured by the model except for the magnitude of the recirculation zone, which is under predicted, and the pronounced double peak on the tangential velocity component. These are both known artifacts of the k-£ turbulence model, which under predicts the strength of the recirculation zone. In Figure 10, predicted axial, radial, and tangential velocity components are plotted at the quarl exit and compared with measured LDV profiles for the mean velocity components. In these figures, the profiles are bounded by the envelope of measured RMS fluctuations . The predictions show good agreement with the measurements, and fall well within the RMS envelope, again with the exception of the IRZ strength. Simple Chemistry Models For the bulk of industrial and utility combustion applications, gas phase chemistry is modeled in two steps: (1) oxidation of fuel producing a pool of carbon monoxide and other products; and (2) oxidation of carbon monoxide to carbon dioxide. In COMO, this step is accomplished using a rate limiting step between the turbulent mixing (Magnussen and Hjertager, 1976) and the kinetics. This eddy dissipation model (EDM) provides reasonable results for many large applications and is time efficient when large models are required to resolve the geometry and flow, temperature, and species gradients. Results from combustion predictions of the cold-wall case with the 2-step mechanism are shown in Figures 10-12. Figures 10 and 11 show 2-D predictions of axial and tangential velocity, temperature, O2 and CO at the 27mm and 208mm traverses compared with the measurements. These profiles are 0.3 and 2.4 burner diameters (Do) downstream of the quarl exit. These figures indicate that the 2-D model does well in capturing the trends near the burner. Downstream, however, the 2-D model does not compare as well, indicating a longer flame, with slower mixing near the burner centerline. This is the result of the axisymmetric assumption requiring the use an annular natural gas inlet stream, as opposed to discrete gas jets that can be correctly modeled in 3-D. This idealization for 2-D models has a significant impact on mixing patterns near the burner, which in turn affect downstream predictions. Understanding these impacts is important, since simplified 2-D models are often used since they provide faster turnaround time for parametric studies. Even with the noted limitation, however, the 2-D model still provides a reasonable representation of the overall flame features and trends. In Figure 12, a Mie scattering image obtained for the baseline flame is compared with the velocity field, gas temperature, and local stoichiometric ratio contours from an axisymmetric prediction. Although the Mie scattering image portrays an instantaneous image of the flame, while the predictions are steady, time-averaged representations of the flame, obvious similarities in shape and structure are apparent. This comparison, though not quantitative, does provide additional verification of the model's ability to correctly simulate the characteristics of the flame. 7 |