| OCR Text |
Show between predicted and measured NOx emissions is 95%. Figure 14 shows the variation in NOx emissions with combustion air preheat as obtained from the measurements (open symbols) over all burner scales, and as predicted by the scaling model under the assumption of perfect similarity (solid symbols). The correlation between measured and predicted values in the bottom left panel is 89%. Figure 15 shows corresponding comparisons between the scaling model predictions and measurements for the dependence of N Ox emissions on fuel staging ratio. 5.3 BurlIer Scaling Principles The results in Figs. 11-15 indicate that the simple burner scaling model presented in §4 appears to largely account for most major trends seen in the NOx emissions performance of this class of burners and furnaces. Moreover, at least some of the discrepancies between model predictions and measurements are due to departures from the perfect similarity assumed in §5.1 and §5.2. These similarity departures result from the constant-velocity principle used to design the tests and departures from true geometric and thermal similarity among the various furnaces used in the tests. Corrections to account for these in the scaling model are examined in §6. However, since the model appears to correctly reflect the major physical mechanisms responsible for NOx formation, it should be possible to use the model to obtain insights into the proper scaling laws for design of small-scale burner performance tests. In this regard it is noteworthy that, except at the smallest scale, the NOx emissions under full-load conditions appear to be largely controlled by the near-burner region. Achieving similarity between differing burner scales would then hinge principally on achieving the proper scaling in the volume, temperature, and oxygen concentration in the near-burner region. In addition to maintaining geometric similarity in the burners, this would depend strongly on matching the heat extraction in the near-burner zone, and on similarity of the flow and mixing properties in the near-burner zone. The latter require aerodynamic similarity in the near-burner zone between different burner scales, which apparently was not fully achieved by the constant-velocity principle used in the SCALIN G 400 tests. In place of the constant-velocity principle (or the constant furnace residence time principle, as was evident in Fig. 8), the present scaling model suggests a rather different scaling principle for burners of this type. The emphasis on achieving fuel-air mixing similarity in the near-burner region reinforces the traditional principles of keeping the ratio of burner and furnace diameters the same, and to the extent possible keeping the product of combustion air velocity and burner size Do constant (i. e. classical Reynolds number scaling for the near-burner region). More important than the latter, however, is the requirement that at the same time the fuel injector hole sizes should be scaled to keep the ratio of combustion air velocity and fuel injection velocity constant. This principle is needed to insure similarity in the injection of fuel into the near-burner zone. Under the constant-velocity principle, the combustion air velocity and fuel injection velocity are kept the same at all burner scales, while the combustion air annulus into which the fuel is .inje~ted increases _ip _direct proportion to Do. As a con~equence, in the 30 kW burner 16 |
