The efficient use of pulverized coal is crucial to the utility industries. The use of computational fluid dynamics (CFD)-based numerical models has an important role in the design of new boiler furnaces or in retrofitting situations. The results of CFD simulations can be used to better understand the complex processes occurring within the boiler furnace. The use of these results to support boiler operation and training of operators requires that the CFD models can be easily accessed and the results are easily analysed. This paper discusses two ways to simulate the heat transfer process in boiler furnaces. The method directly applying CFD results is employed, in which the grid for solving the energy equation is the same as the flow grid in the CFD simulation while radiation heat transfer is solved in another relatively coarse grid. Comparison of the prediction results between CFD and Heat Transfer code (Simple model) is performed under boiler full load (100%) with one side wall fouling, as well as for different boiler loads (100, 98 and 95 per cent boiler full load, respectively). Finally, the flexible use of the results of CFD and the simple model for pulverized coal-fired boilers is presented. To facilitate the use of the system, a user-friendly interface was developed which enables the user to manipulate new calculations and to view results, namely performing'what-if' analysis. Copyright ~ 2000 John Wiley & Sons, Ltd.

mechanism including Hg reactions involving HgO was proposed. Among those reactions involving HgO, the progress of reaction HgO +HCl→HgCl+OH is HgO+HCl→TS1(HgClOH)→M(HgClOH)→TS2(HgClOH)→HgCl+OH, in which the controlling step is HgO+HCl→TS1(HgClOH)→M(HgClOH). The progress of reaction HgO+HOCl→HgCl+HO2 is HgO+HOCl→M(HgClOOH)→TS(HgClOOH)→HgCl+ HO2, in which the controlling step is M(HgClOOH)→TS(HgClOOH)→HgCl+HO2. Four other reactions are one-step, with no intermediates formed. The performance of the model was assessed through comparisons with experimental data conducted by three different groups. The comparison shows that model calculations were in agreement with only one set of all the three groups experimental data. The deviation occurs due to the absence of accurate rate constants of existing echanism, the adding of reactions involving HgO, as well as the exclusion of heterogeneous Hg oxidation mechanism. Analyses by quantum chemistry and sensitivity simulations illustrated that the pathway Hg+ClO=HgO+Cl is more significant than some of the key reactions in the kinetic mechanism proposed in the literature, which indicates the necessity of including reactions involving HgO in the mercury speciation kinetic mechanism. Studies on the effects of oxygen show that O2 weakly promotes homogeneous Hg oxidation, especially under the condition of low Cl2 concentration. In all cases, 1.5–6.0% of the mercury is predicted to be present as HgO.

The text of a paper should be preceded by a summary in English. This should be short, but should mention all essential points of the paper.

The mechanism of detailed chemical reactions in combustion is very complicated, especially when the formation of pollutant such as NOx is considered. Based on Glarborg's and Bilger's full model (Glarborg P. et al., Combustion and Flame, 65, 177; (1986), Bilger R. W., Starner S.H., Kee, R.J., Combustion and Flame, 80, 135 (1990)), a simplified model of the mechanism of NOx formation in a CH4/Air system is presented. Sensitivity calculation and eigenvalue-eigenvector analysis fprincipal component analysis (Vajda S., Valko P., Turanyi T., Int. J. Chemicaf Kinetics, 17: 55 (1985)) are used in the model. Compared with the conventional methods of simplification of chemical reaction mechanisms such as quasi. steady-state approximation, the eigenvalue-eigenvector analysis can indicate the important reactions efficiently without providing any balance hypothesis. The new model consists of only 15 elementary reactions, which is simpler than the full model composing 39 reactions. The deviations in NOx concentration and other major species between the simplified and the full model are rather small over a wide range of fuel-air ratio．Meanwhile, the calculated NOx concentrations by the simplified model are in quite good agreement with Bartok's experimental data (Bartok W. et al., AICHE Symp. Ser. 126 (1972)), especially in the fuel-lean region.

This paper presents numerical simulation of the flow and combustion process in the furnace of a pulverized coal fired utility boiler of 350 MWe with 24 swirl burners installed at the furnace front wall. Five different cases with 100, 95, 85, 70 and 50% boiler full load are simulated. The comparison between the simulation and the plant data is stressed in this study. The heat flux to furnace walls between the measured values and the calculation results is compared. It is found that increasing the load leads to consistent variations in the properties presented and the exception is observed for the full load case where the predicted exit gas temperature is lower than the 95% one and the total heat to the boiler walls is smaller. This might be due to the fact of considering a linear scaling of the input parameters between the 70% and 100% load. The increase of the air flow rate led as expected to a reduction of the furnace outlet temperature and to a small decrease in NOx emissions. It shows that the NOx model used shows a higher sensitivity to temperature than to oxygen level in the furnace. The model used considered the De Soete mechanism for the nitrogen from volatiles and the contribution of char was considered in a similar way. The agreement for all cases except the one of 50% boiler load between the calculation results with the plant data validates the models and algorithm employed in the computation. The furnace performance under different boiler loads is predicted and compared in order to meet the requirements of NOx abatement and avoiding some negative side effects on the furnace. q2000 Elsevier Science Ltd. All rights reserved.