(Georgia Institute of Technology, 2020-04-24)
Goh, Edwin
The ground power industry is targeting combined cycle plant efficiencies of 65% and above, which can be achieved primarily through higher combustor firing temperatures. Because conventional combustors fail to meet NOx regulations at such temperatures, there is a pressing need for high-temperature, low-emissions combustors. In this regard, the staged combustion architecture is one such concept that shows promise due to its enhanced emissions performance and operational flexibility. The prohibitive cost of building prototypes relegates full-scale testing to the final stages of the product design cycle, while accurate models with turbulence and detailed chemistry cannot be used to efficiently explore the design space. Therefore, an efficient computational model is necessary to study a broad range of architectures. Despite extensive research on staged combustion and the related jet-in-crossflow (JICF) problem, there is little published research regarding the minimum NOx levels achievable by staged combustion architectures. The first contribution of this thesis presents a set of fundamental minimum NOx levels that are obtained by wrapping a constrained optimization routine around a reduced-order staged combustor model. For a firing temperature of 1975 K which corresponds to 65% efficiency, the minimum NO levels are determined to be roughly 1 ppm. Sensitivities of these minimum NOx levels to operational, geometric and computational parameters are identified and discussed. Recognizing that a turbulent flow field affects NOx chemistry primarily through mixing, the second contribution presents a reduced-order Limited Mixing and Entrainment (LiME) reactor model to predict emissions based on mixing and entrainment time scales. Molecular mixing is simulated based on the Interaction by Exchange with the Mean (IEM) model through interacting Lagrangian particles. The consensus is that better JICF mixing leads to lower NOx emissions, but little work has been done to characterize the effects of large-scale entrainment and small-scale mixing on NOx in isolation. The third contribution elucidates the sensitivity of NOx to mixing and entrainment time scales using reduced-order models and demonstrates a potential use case of this model in a constrained design optimization problem to identify minimum NOx levels under fixed entrainment rates. The impact of fuel and air staging on NOx under mixing and entrainment-limited scenarios is elucidated.