| OCR Text |
Show simulations for supporting the design of gas reburning systems. The first of these units, Elrama Unit 2, was a roof-fired boiler that had been significantly modified for low-NOx operation prior to installation of the gas injection system. In that case, the LIM simulations were conducted after field testing of the reburn system had been completed in order to gain insights into the origins of several peculiarities that appeared to limit the reburn performance. The simulation results showed that the SOFA system installed on the boiler for low-NOx operation led to reburn gas being injected into a fuel-rich zone near the furnace front wall, which provided conditions for potentially high NOx removal. However the SOFA system also produced a very large region of very high oxygen concentrations in the back part of the furnace, which is detrimental to effective NOx removal. The LIM simulations showed that at high SOFA settings, the gas jets penetrated the mixing layer separating these two parts of the furnace, and that this limited the NOx removal achieved. In this manner the simulations were able to explain why higher gas input often did not produce higher NOx removal, why the downward-oriented injector nozzles provided better reburn performance, and why the sidewall injectors generally gave poorer NOx reduction. At zero SOFA setting the modeling correctly predicted that these performance limitations would not occur, but that the level of NOx reduction achieved for each percent gas heat input to the boiler "would be lower than at high SOFA settings. Based on these insights into the reburn process at Elrama Unit 2. the simulations suggested specific recommendations for achieving higher levels of NOx reduction at lower levels of gas heat input, which are currently being considered for implementation. At Joliet Unit 6. the simulations predicted that very different performance characteristics would be found for the fuel-lean gas reburn system. These differences were directly traceable to the more homogeneous nature of the boiler. In particular. the simulations predicted that the level of NOx reduction achieved would be the same for all injector groupings and elevations in the proposed test matrix, at just slightly higher than 5 % NOx reduction for each 1% gas heat input to the boiler. Moreover, the simulations predicted that the limit in NOx reduction found at Elrama Unit 2 would not occur at Joliet Unit 6. and that the unit would show a nearly linear variation in % NOx reduction with % gas heat input. Lastly, the simulations predicted 188-214 ppm of C O at 7 % gas heat input to the boiler. All of these predictions were subsequently borne out by field testing of the reburn system on that unit, and the quantitative predictions of the simulations were found to be in excellent agreement with the field test results. Collectively, the experience gained from the LIM simulations of these two very different boilers shows, firstly, that this comparatively new large-eddy simulation approach to modeling of combustion flows is capable of producing quantitative accuracy that far exceeds that of traditional turbulent reacting flow models. Moreover, these two examples show how modeling of this type can be used in a traditional predictive role, such as was done in the simulations of Joliet Unit 6. as well as in a probative role such as was done at Elrama Unit 2 to gain insights into origins of unexpected performance and to develop recommendations for achieving improved performance. Based on the successful application of LIM simulations in these two fundamentally different boilers, this new simulation approach appears to be a valuable tool that can be applied to support the design of effective reburning systems for a wide range of utility boilers. |
