A multi-scale approach to assessing the influence of gas phase combustion mechanisms on the prediction of pulverised coal flames
Computational fluid dynamics simulations were conducted to assess the effect of using different gas phase combustion mechanisms (one, two and five-step) in predicting the flame structure and combustion species distribution during the pulverised coal combustion process. The simulations were conducted at several scales i.e., Laboratory, Pilot Scale and Full Scale. To enable this a comprehensive model of the Eskom Research, drop tube furnace and pilot scale combustion test facility was developed. The full-scale assessment was done on a 200 MWe pulverised coal boiler operating in the South African power generation industry. It was shown that it is essential to develop detailed characterisations of the devolatilisation and char combustion characteristics of the reference coal using advanced techniques. The predictive capability using an iterative approach using ANSYS Fluent® and PC-Coal Lab® was extended from previous work and demonstrated that the reference coal exhibits an atypically slow burn out characteristic. This behaviour was modelled successfully by introducing modifications to the mass diffusion coefficient in the char combustion model. These submodels were applied to the pilot scale simulations, and it was demonstrated that the choice of gas phase combustion mechanism has a profound effect on the predicted structure of the pulverised coal flame. The use of the more advanced five-step mechanism improved the temperature predictions within the pilot scale combustion test facility’s furnace. The latter part of the study considered the application of the techniques developed at the lower scales, to a full-scale boiler. Hence, a comprehensive model of a full-scale boiler was developed, which included a faithful representation of the boiler geometry and particularly the behaviour of the burners. A simplified burner was developed for use in the full-scale model and its behaviour was verified against the results predicted by a detailed model of the installed low NOx burner. The simplified burner demonstrated that it could mimic the flow structure emanating from the detailed model, with specific emphasis on the velocity and turbulence profiles. The full-scale model simulations further emphasised the role of the gas phase combustion mechanisms on the prediction of pulverised coal flames. This was made starkly evident as the five-step mechanism provided better correlation with the measured data and moreover predicted a fundamentally different combustion flow and temperature field, that was considered realistic as it confirmed the observed flame behaviour. The five-step mechanism demonstrated an extended ability over the one and two-step mechanisms, to address the strong coupling between localised flame aerodynamics, thermodynamics, chemistry, and particle dynamics. This whilst also showing a sensitivity to the char burn out characteristics. The five-step mechanism provided the best prediction of temperature, O2, and importantly CO at the furnace exit. Furthermore, it also demonstrated good correlations with measured NOx values when used in conjunction with the standard NOx post-processing algorithm. It was concluded that the one and two step mechanisms were appropriate for predicting bulk quantities such as furnace and boiler exit temperature (to an extent) but provided erroneous predictions in the near burner and overall burner belt region. The five-step model also showed that adverse phenomena such as particle roping and its effect on ignition can be predicted. The research study also showed that the particle burnout characteristics were consistent across all scales, and this gives credence to the established testing and evaluation criteria used to specify a coal for large scale adoption in the production environment. However, a subtlety was noted in terms of the size basis for the drop tube tests and recommendations for improvement were proposed. In summary, the research work also confirmed that the choice of gas phase combustion model, and its underlying assumptions has a profound effect on the prediction of the structure of a pulverised coal flame. An ancillary observation also played a key role i.e., that the peculiarities of South African coals must be assessed using advanced CFD and phenomenological models for char and devolatilisation. This is to effectively model a realistic combustion characteristic. The work demonstrates that the appropriate use of the tools developed could be used to assess the performance of existing coal fired furnaces using South African coals and could provide key insight in optimising their thermal performance and reducing their emissions footprint as the global energy landscape changes to a more environmentally friendly future.
A dissertation submitted in partial fulfilment of the requirements for a Master of Science in Engineering in the Faculty of Engineering and Built Environment, School of Mechanical, Industrial and Aeronautical Engineering, University of the Witwatersrand, Johannesburg, 2022