A Liquid-Fueled Reactor Network for NOx Prediction in Gas Turbine Combustors

Aviation NOx emissions from fossil fuel combustion significantly impact the environment by contributing to climate change and atmospheric chemistry alteration. As the aviation industry continues to grow, accurate and computationally efficient methods for predicting NOx emissions become crucial for combustor design and optimization. While detailed three-dimensional CFD simulations can provide NOx estimates, their high computational costs make them impractical for rapid screening during design phases. This work presents a novel reactor network approach specifically designed for liquid-fueled combustion systems to predict NOx emissions in gas turbine combustors. The key innovation lies in incorporating detailed spray dynamics, including droplet breakup and evaporation, into the reactor network framework, significantly improving prediction accuracy compared to traditional gaseous fuel reactor networks. The reactor network uses k-means clustering to automatically partition the combustor domain into distinct zones, each representing specific physical and chemical processes. Validation against detailed 3D CFD simulations of a swirl-stabilized combustor operating with Jet-A fuel demonstrates superior performance of the liquid fuel reactor network over gaseous fuel approaches. The liquid fuel reactor network achieves computational speed-ups of 3200× while maintaining NOx prediction errors below 10\% across varying inlet air temperatures (300K-500K) and fuel flow rates (2.2-2.6 g/s). Sensitivity analysis reveals that increasing cluster resolution from 5 to 9 clusters improves prediction accuracy. The validated reactor network provides a robust, computationally efficient tool for combustor design optimization and represents a significant advancement in reduced-order modeling capabilities for liquid-fueled combustion systems.