OCR Text |
Show 4.2.3 [02]eq in FUrnace Region The oxygen concentration in the furnace region is determined by the overall air-fuel equivalence ratio ,,\ as 2(A - 1) [02]eq,jurnace = 1 + 9.52A (30) This is the same as the theoretical O2 level in the flue gas. 4.2.4 [02]eq in Staging-jet Region The underlying physical picture is that the staged fuel is injected directly into the primary air issuing from the burner, as suggested by the in-flame data and as shown schematically in Fig. lOa. As a consequence, the excess O2 level relevant to NO formation in the staging jet region in Fig. IDe depends only on the overall air-fuel ratio, and is independent of the staging ratio. The in-flame measurements indicate that the O2 levels in this region are generally slightly above the flue gas O2 concentration, and are thus taken to be given by (30). 4.3 [N2]eq Concentration Scaling The nitrogen mole fraction is taken to be constant in all regions under all conditions, and given by at the level in air as (31 ) 4.4 Tenlperature Scaling The temperature dependence of EINOx formation in (6) is highly non-linear, and as a consequence using mean temperatures alone is inadequate to predict the NOx emissions scaling. It is essential to account for temperature fluctuations. Based on available temperature data from turbulent diffusion flames, the probability density function (3(T) for temperature is assumed to be of clipped-Gaussian form as (3(T) = C 1 exp (-(T - (T)m)2) V2i(7 2(72 (32) where (T)c is the characteristic mean temperature for the given region, (7 is the Tms temperature fluctuation level, and C is the normalization factor to assure uni ty total probability. For any two clipping temperatures tl and t2 , the temperature (T)m corresponding to the maximum in (3(T) is set to give the correct mean temperature (T)c, namely (T)c = ;t2 T C ~ exp (-(T ~ \T)m)2) (33) tl 27r(7 ~(7 For the results given here, the pdf (3(T) is clipped at tl = 400 K and at t2 given by the adiabatic stoichiometric equilibrium temperature for the given operating conditions. The scaling predictions are largely insensitive to the choice of t l , provided comparisons between emission lev~l~ are made with consistent limits. 12 |