OCR Text |
Show chemical reaction processes are separated by working in terms of the elemental mixture fraction field £(x,f). which is a conserved scalar quantity. Modeling is done only at the very smallest scales, where experimental studies have clearly established that conserved scalar quantities in turbulent flows form universal, self-similar, layer-like structures. The LIM model incorporates this self-similar structure at the fine scales through a local parabolization of the governing transport equations across a time-evolving material surface on which gradients in the scalar field are initially centered. This transforms the original governing partial differential equations, which must be solved throughout the three-dimensional volume, to a set of ordinary differential equations that can be solved on this time-evolving surface. The smallest scales in the flow field are treated analogously, with the resulting vorticity field setting the velocity field, which deforms the computational vorticity and scalar gradient surfaces with time. The time-evolution of the scalar field is then determined by solving the resulting set of ordinary differential equations (the LIM equations) at each time step on this time-evolving material surface. The scalar dissipation rate field V£-V£(x,0 is reconstructed from the integral moments on this surface, and from this the scalar field C,(\j) is obtained. The resulting joint scalar and scalar dissipation rate fields allow highly complex reaction chemistry, in this case the complete GR1 Mech 2.11 chemical kinetics, to be coupled to the underlying time-varying molecular mixing process. The resulting LIM model is well within the computational capabilities of current desktop computing workstations and accurately incorporates the physics of the large-scale structure, turbulent transport, molecular mixing, and chemical reaction processes that are inaccessible to traditional turbulent reacting flow models. In this paper, we examine the applicability of LIM simulations for modeling flow. mixing, and combustion processes in coal-fired utility boilers, focusing in particular on gas reburning technologies for such units. Results are presented from LIM simulations of fuel-lean gas reburning (FLGR) in a 112 M W e roof-fired unit (Duquesne Light Company's Elrama Unit 2) and a 327 M W e cyclone-fired unit (Commonwealth Edison's Joliet Unit 6). In the former, the objective was to identify the origins of several unexpected factors that limited the NOx reduction attainable with the reburn system originally designed for that unit, and to identify recommendations for modifying the gas reburn system to achieve higher NOx reduction at lower gas input. In the second unit, the simulations were conducted prior to any field data being obtained. Predictions for NOx and CO emissions obtained from the simulations are compared with subsequent fuel-lean gas reburn performance measured from field testing at that facility. Conclusions are drawn as to the utility of LIM simulations as an accurate tool for supporting the development of combustion systems in general and gas reburning systems in particular. 2. OVERALL FUEL-LEAN GAS REBURNING AT ELRAMA UNIT 2 Elrama Unit 2. operated by Duquesne Light Company (DLC). is a 112 M W e roof-fired split-furnace coal-burning utility boiler manufactured by Babcock & Wilcox and placed in service in 1953. A schematic of the facility is shown in Fig. 1. The roof-fired burner is a unique design that is unlike conventional wall-fired burners. Pulverized coal is introduced through eight intertube burner blocks, four in each half of the furnace. As shown in Fig. 2. each burner block is comprised of twelve |