OCR Text |
Show 3. MODEL DESCRIPTION The computation of the full finite-rate chemistry for all the intermediates in a turbulent coal-fired furnace or boiler is beyond the capability of current numerical combustion simulators. To meet most practical engineering objectives, however, a model of this complexity is not required. Our engineering objectives are to determine the effects of design and operating variables in a large-scale coal-fired furnace on outlet NO concentrations, fuel burnout, and furnace outlet temperatures. Furthermore, it is a goal to meet this objective with a reduced chemical kinetic mechanism and the use of 2- dimensional CFD models - with the stipulation that physical mechanisms of first-order importance are included in the analysis. For the low level of NO concentrations required by current environmental regulations, a sub model is required that can reliably calculate the relative effects of fuel, thermal, and prompt NO as a function of local stoichiometry and temperature. Such a model has been developed and is described elsewhere (Smith, et ai., 1982 and Eddings, et aI., 1994), however, the principle features of the model are: 1. The concentration of each pollutant species is computed from: where p is the Reynolds average density u is the Favre-averaged velocity vector, Yi is the Favre-mean mass fraction of the species of interest (i), De is the eddy diffusivity, and Wi is the time mean volumetric net formation rate. 2. The devolatilization process is modeled using a mechanism comprised of two competing reaction steps developed by Kobayashi, et al. (1977) with the rate information from Ubhayakar, et al. (1977) and using maximum devolatilization constants obtained from the ABB Power Plant Laboratories drop tube furnace (at pyrolysis conditions of 2700 oF). 3. The thermal NO formation is based on the extended Zeldovich mechanism with three kinetic steps (Miller and Bowman, 1989). The 0 and OH atom concentrations are determined from instantaneous local partial equilibrium. In the turbulent flow field, the mean values for the reaction rates are computed by integrating over the statistical distribution or probability density function (pdf) of the process variables characterizing the local stoichiometry and energy loss (temperature). The pdf is computed from an f-g formulation with a form of the k-e turbulence model to determine the variance of each of the process variables (Smoot and Smith, 1985). 4. Prompt NO is calculated via a kinetic mechanism proposed by deSoete (1975) for reaction of N2 with fuel species to form eN compounds which then decay through competing paths to N2 or NO. The fluctuations due to fluid turbulence are incorporated with multiple pdf's. 8 |