OCR Text |
Show the fuel issued at 140 m/s into a 28 m/s combustion air crossflow within an annular gap smaller than 5 mm. The result is that the injected fuel essentially penetrated completely across the combustion air duct and entered the near-burner recirculation zone at the outer edge of the annulus. In the 12 MW burner, on the other hand, the fuel also issued at 140 m/s into a 28 m/s combustion air crossflow, but into an annular gap that was nearly 10 em large. The injected fuel thus did not penetrate far across the annulus, and instead entered the recirculation zone much closer to the inner wall of the annulus. Consequently, depending on the burner scale Do, the fuel entered the near-burner recirculation zone at different points and at different levels of premixing with the combustion air. The effect of this on fuel-air mixing in this NOx-controlling region will have been significant, and is believed to have been a major source of departures from perfect similarity in the near-burner region among the burners in the SCALING 400 tests. The scaling laws to meet the similarity requirements outlined above can be readily derived. Denoting the combustion air velocity Uo and annulus diameter Do as in Fig. 1, and the fuel injection velocity Uj and injector hole size D j, the requirement for constant Reynolds number relevant to the near-burner region aerodynamics dictates UoDo "'-J constant. Since the thermal input to the burner scales as Qo "'-J UoD6, the burner size then must be scaled as Do "'-J Qo, rather than the Do "'-J Q~/2 scaling obtained under the constant-velocity principle. The air velocity then must scale as Uo "'-J Q01 rather than the Uo "'-J constant scaling under the constant-velocity principle. The simultaneous requirement that Uj / Uo "'-J constant for similarity of the fuel injection then requires that Dj "'-J Do, which preserves geometric similarity as well. A burner designed for small-scale testing under these scaling principles must of course also satisfy additional requirements for flame stability, compressibility (Mach number) effects, and possibly other constraints as well. As is typical in scaling problems governed by more than one par~meter, these additional constraints may conflict with the scalings dictated by Reynolds number and fuel injection similarity. Even maintaining Re and fuel injection similarity alone may not always be practical. In that case, the requirement for strict matching of Reynolds numbers should be relaxed first, as is often done in fluid dynamics, since in practice Re generally needs only to be in the same viscous flow regime for aerodynamic similarity in the near-burner region to be achieved. By comparison, the requirement for strict similarity in the fuel injection process appears to be of controlling importance to mixing and combustion in the near-burner region. 6. CORRECTIONS FOR IMPERFECT SIMILARITY The results in §5 were obtained from the scaling model under the assumption that perfect geometric, aerodynamic, and thermal similarity were achieved in the measurements at all scales. Consequently, a single value for each parameter in the scaling model (i. e. T1•2 , T), [02h.2' [02]), and L1,2/ Do), obtained as described in §4, was used at all scales to obtain the performance predictions shown in Figs. 11-15. However, inspection of the inflame data from all scales shows clear departures from perfect similari ty. Although the basic features of the NOx-production regions in Fig. 10 remain unchanged, some of the pq.ramgters characterjz.ing the near:-burner region showed clear variations with burner 17 |