Numerical simulation of combustion and nitrogen pollutants in furnaces:Dissertation

This publication considers the numerical simulation of flow, combustion, heat transfer and nitrogen pollutants in furnaces. The main emphasis is on the presentation and application of a computational environment for pulverized fuel (and spray) combustion in multi-burner furnaces. The submodels of the computational environment include the (k,epsilon)-model and a multiple-time-scale model of turbulence, the eddy breakup and eddy dissipation concepts for gaseous combustion, a stochastic Lagrangian description of particles, a flux method and a variant of the discrete transfer method for radiative heat transfer, and reduced schemes for the formation and destruction of nitric oxide. The flow equations are solved with the Phoenics-program of CHAM Ltd, to which a technique to refine the computational grid locally has been added by the author. Three applications are presented. In the simulation of the combustion of light fuel oil in a 300 kW furnace and of pulverized peat in a 5 MW furnace only some of the submodels have been used. The computed results of these two applications agree well with the experiments in the far field region of the furnace and for the overall features. The agreement in the near burner regime is only satisfactory. The qualitatively different behaviour of the combustion of pulverized peat under various firing conditions could also be simulated. The computation of the combustion of pulverized coal in a 250 MW boiler furnace under various firing conditions showed the applicability of the submodels to a multi-burner furnace. The computed results are in reasonable agreement with the available measurements at the exit of the boiler furnace for temperature, char burnout and the concentration of nitric oxide. The principal factors related to the submodels, to the boundary conditions and to the discretizations, which affect the accuracy of the computed results, are discussed. Expected development of submodels and computing techniques are briefly considered. The results obtained indicate that, properly used, the present submodels and computing power are already capable of providing a useful insight into the local conditions in the furnace. Such an insight is needed e.g. for a successful application of in-furnace methods to reduce nitrogen pollutants