Formulación matricial de problemas con subregiones

This study used a series of published kinetic rates to construct a model for mercury oxidation by chlorine and bromine species under the influence of NO and SO₂pollut- ants. A genetic algorithm was used to optimise the model to the experimental results of Preciado et al. [117] in order to analyse the important mechanisms in the chemical process. Without optimisation, the presence of NO significantly mitigated mercury oxidation by consuming most of the reactive halogen species, and the model was in- capable of predicting any of the experimentally measured oxidation rates.

While the optimised model demonstrated reasonable agreement for the baseline air and oxyfuel cases, as well as promising agreement to an alternative validation data set, the models predicted considerable sensitivity to the presence of pollutant species that was not present in the measured data. This sensitivity in the model is in spite of the optimisation process that included these cases. In particular, it was not possible for the optimisation process to mitigate the detrimental effects of NO on the consumption of reactive halogen species. Therefore, it must be concluded that there are mechanisms that are not present in the current model that need to be accounted for. This may include surface-catalysed reactions, or reactions between mercury species and nitrogen species, such as ONCl or ONBr.

4.3 Summary

This chapter presented a thermodynamic and chemical kinetic study on mercury ox- idation in both air-fired and oxy-fired combustion environments. While the thermo- dynamic study yielded no significance in speciation between air-fired and oxy-fired conditions, the chemical kinetic study was able to capture the differences between the combustion environments, although with a heightened sensitivity to the NO concen- tration.

The thermodynamic study in Section 4.1 illustrates that for both air-fired and oxy- fired cases, Hg0is the dominant gas-phase species of mercury at higher temperatures. The equilibrium distributions also show that there is a tipping point, determined by the total halogen concentrations, at which Hg2+ becomes dominant. Both mercury and halogen species’ equilibrium distributions are unaffected by changing the combustion environment.

The chemical kinetic investigation in Section 4.2 highlights the importance of mod- elling several aspects of the combustion chemistry. A genetic algorithm was employed to optimise a chemical kinetic model based on published mechanisms and kinetic rates for mercury and halogen species. The optimisation process was successful in achiev- ing agreement across several measured values for mercury oxidation rates, however the model failed to capture some of the measured chemical behaviour.

The chemical kinetic study has demonstrated that the observed variations in mer- cury oxidation rates under air-fired and oxy-fired combustion can be somewhat ex- plained by variations in NO concentrations. However, the study has also demonstrated that the currently published mechanisms for mercury oxidation are incapable of rep- licating the insensitivity to the NO concentration in air-fired combustion, even with a fairly liberal optimisation process. Further work should investigate the importance of additional routes to mercury oxidation, which may include surface catalysed reactions or interactions with other halogen species, such as the nitrosyl halides.

5 Radiation model validation for

oxyfuel conditions

This chapter presents calculations of idealised environments that represent potential oxyfuel conditions to provide validation of modelling approaches for the spectral ab- sorption coefficient. Firstly, in Section 5.1, the LBL and narrow band models are validated against the high resolution measurements of gas transmissivity by Alberti et al. [256] for CO₂/H₂O/N₂mixtures. In Section 5.2, the LBL and narrow band mod- els are used to calculate radiative transfer through a one-dimensional inhomogeneous medium to test the inherent approximations introduced by the narrow band models against LBL calculations. In Section 5.3, the global models that were introduced in Subsection 3.1.2 are compared against narrow band calculations of radiative transfer within a three-dimensional enclosure, which represents conditions that are relevant to CFD calculations of oxy-coal combustion environments. Conclusions from these studies are presented in Section 5.4.

The methods that are considered in this chapter are the LBL method using the HITEMP-2010 [224] and CDSD-4000 [253] spectral databases, the SNB and CK nar- row band models with the parameters provided by Rivi`ere and Soufiani [263], the SLW model using the ALBDF provided by Pearson et al. [309], the WSGG model with the coefficients suggested by Smith et al. [220], Yin et al. [286], Johansson et al. [291], Krishnamoorthy [293] and Kangwanpongpan et al. [292], and the FSK models using the narrow band k-g distributions from Cai and Modest [441]. The SLW method uses the multiplicative approach to generate the ALBDF from the single-specie lookup tables, and the CK and FSK models both employ the mixing scheme of Modest and Riazzi [282] to calculate the mixture k-g distributions. This study aims to validate these various approaches and suggest a method for modelling the spectral component of radiation that is suitable for CFD calculations of oxyfuel combustion.

5.1 Validation of spectral models

Measurements by Alberti et al. [256] provide high resolution transmissivity values for mixtures of CO₂, H₂O and N₂at elevated temperatures, and at concentrations that are relevant for oxyfuel conditions. These fundamental measurements of homogeneous paths allow for the validation of the computationally expensive LBL and narrow band models, which are used as benchmark solutions to validate more expensive methods in complex cases, such as the studies in Sections 5.2 and 5.3.