Assessment of combustion models in hydrogen engine simulations using optical measurements

Hydrogen internal combustion (IC) engines have attracted significant attention as a viable alternative for carbon-free transportation, with promising advancements towards practical implementation. Predictive computational fluid dynamics (CFD) simulations play a crucial role in rapid turnaround in the design process, but there are fundamental challenges concerning combustion submodels, specifically for hydrogen combustion. This study aims to assess different combustion submodels that are commonly used in IC engine modeling through validation against experimental data obtained from a heavy-duty optical engine employing port fuel injection with varying spark timing and lean air–fuel ratios (λ). First, a multi-zone well-stirred reactor model, known as SAGE, was evaluated with different chemical kinetic models. Predictions of laminar flame speed using various kinetic models remained consistent under normal temperature and pressure conditions for a different range of evaluated equivalence ratios. Furthermore, at lean conditions (λ≈2.5), good agreement between CFD results and experiments was observed for different chemical kinetic models, based on the pressure traces and flame propagation analysis in natural combustion luminosity images at engine conditions. Nonetheless, at ultra-lean conditions (λ≈3.0), significant differences were encountered. As an alternative combustion submodel, the G-equation model with the standard turbulent flame speed correlation was assessed, revealing discrepancies against experimental data. One hypothesis behind the observed underpredictions concerns the absence of current turbulent flame speed models to account for the additional flame wrinkling resulting from diffusive-thermal instabilities for lean hydrogen–air mixtures, potentially enhancing flame propagation. To represent the effect, the study explored the impact of the “b₁” constant in Peters's turbulent flame speed correlation, in order to account for the large-scale wrinkling effect. The comparative analysis between two main combustion submodels and experimental validations provide crucial insights for future developments in high-fidelity predictive modeling of hydrogen engines.