Large eddy simulations of NH₃-H₂ jet flame at elevated pressure using PCA with inclusion of NH₃/H₂ ratio variation

Ammonia-hydrogen blends are receiving significant attention as a viable alternative to hydrocarbon fuels, and improved fundamental understanding of their characteristics is essential for their application. To investigate the fundamental characteristics of turbulent non-premixed ammonia-hydrogen flames at practically-relevant pressure conditions, large eddy simulation (LES) computations are conducted using the PC-transport model, which is based on Principal Component Analysis (PCA). To enhance the size-reduction potential of PCA, it is coupled with nonlinear regression that employs deep neural networks (DNN). This work aims to advance the PC-transport approach by extending the training data set to include variations in local NH3/H2 ratios due to chemical and transport effects. LES results from the PC-DNN approach with a training data set based on fixed (baseline manifold) and varied (extended manifold) NH3/H2 ratios are compared with the recent experimental measurements obtained at the KAUST high-pressure combustion duct (HPCD). The results show that the PC-DNN approach with the extended manifold provides improved predictions, compared to the baseline manifold, and is able to capture key flame characteristics with reasonable accuracy using two principal components only.