Accelerating Detailed Chemistry Simulations of Hydrogen Combustion in IC Engines Using Reynolds-Averaged Navier-Stokes (RANS) Turbulence Modeling and a Thickened Flame Model

Abstract

Abstract Green hydrogen, with its potential to significantly reduce transportation CO2 emissions, emerges as a promising solution to combat human-induced climate change. However, challenges such as flame instability and pre-ignition pose significant obstacles to its integration into internal combustion engines (ICEs). Computational fluid dynamics (CFD) is an important tool available for engineers to overcome these challenges towards enhancing engine efficiency, combustion stability and reducing emissions. Hence, novel CFD approaches to simulate hydrogen ICEs in an economical way is of importance. In the current study, hydrogen combustion is simulated in a turbocharged port fuel injection (PFI) spark ignited (SI) engine utilizing detailed chemistry along with a thickened flame model (TFM). The primary objective of employing a TFM is to increase the turbulent flame thickness, thus enabling more economical simulations while maintaining solution accuracy. The turbulence model utilized in this investigation is the RNG k-ϵ model, as this model also facilitates economical simulations. While TFMs are conventionally used with Large Eddy Simulation (LES) turbulence models, the current implementation extends this approach to Reynolds-Averaged Navier-Stokes (RANS) models. RANS models inherently introduce a thickening effect on the flame due to their higher turbulent viscosity. However, additional thickening can be beneficial for hydrogen flames, which tend to be thinner compared to traditional hydrocarbon fuel flames. This model effectively alleviates the computational burden associated with resolving thin hydrogen flames, rendering them computationally efficient for CFD simulations of H2 ICEs. In this study, a mixture-averaged diffusion model combined with detailed chemistry is employed to account for both preferential diffusion and H2 combustion. Several engine operating points are simulated and compared against measured cylinder pressure data, demonstrating good agreement. The utilization of TFM enables the simulation of multiple engine cycles with a significantly reduced wall clock time. The paper compares the cylinder pressure predictions and the run times from RANS simulations with and without the use of TFM with that of LES simulations with TFM to show how the use of TFMs with RANS in hydrogen ICE simulations offer a promising avenue for advancing research in sustainable transportation technologies.