Super-knock that occurs in spark ignition (SI) engines is investigated using two-dimensional (2D) numerical simulations. The temperature, pressure, velocity, and mixture distributions are obtained and mapped from a top dead center slice of full cycle three-dimensional (3D) engine simulations. Ignition is triggered at one end of the cylinder and a hot spot of known temperature was used to initiate a pre-ignition front to study super-knock. The computational fluid dynamics code CONVERGE was used for the simulations. A minimum grid size of 25 μm was employed to capture the shock wave and detonation inside the domain. The Reynolds averaged Navier-Stokes (RANS) method was employed to represent the turbulent flow and gas phase combustion chemistry was represented using a reduced chemical kinetic mechanism for primary reference fuels. A multi-zone model, based on a well-stirred reactor assumption, was used to solve the reaction terms. Hot spots introduced inside the domain at various initial temperatures initiated a pre-ignition front, which resulted in super-knock due to detonation of the end-gas. The detonation speed was around 2000 m/s. The detonation was induced for temperatures greater than 1000 K during the start of pre-ignition flame propagation. For temperatures between 800 K to 1000 K detonation was exhibited when almost all the fresh gases are consumed by the propagating pre-ignition front. Multiple auto-ignition sites in the end-gas region were observed at higher temperatures. High peak pressures were generated during the detonation onset. The low temperature case, 700 K, exhibited a deflagration mode of flame propagation without detonation development. The results were analyzed and reported by comparison with Bradley diagram which predicted a deflagration propagation mode for the lowest temperature case and developing detonation mode for all other cases considered in this study.
Bibliographical noteKAUST Repository Item: Exported on 2020-10-01
Acknowledgements: This work was sponsored by Saudi Aramco and King Abdullah University of Science and Technology. The simulations made use of the high performance computing resources at KAUST Supercomputing Laboratory. We thank Convergent Science for providing the licenses.