Direct Numerical Simulation of Autoignition in Turbulent Non-premixed Combustion

Abstract

Combustion in a CI engine is initiated by self-ignition of fuel–air mixture caused by high pressure and high temperature; a process known as autoignition. Autoignition is a challenging problem to simulate as the temperature increases from initial temperature to the adiabatic flame temperature in a very short duration. Numerical study of a turbulent flow using RANS/LES encounters a closure problem. Accuracy of closure models can be improved through experimental results, theoretical reasoning and direct numerical simulation (DNS) data. A review of DNS of autoignition in a turbulent non-premixed medium is presented in this chapter. As observed from DNS study, autoignition sites in a turbulent non-premixed medium are not randomly distributed but follow a pattern in the mixture fraction-scalar dissipation rate space. Turbulent flow is always three-dimensional in nature. 2D DNS of autoignition shows that ignition delay time increases with increase in initial turbulence intensity, which contradicts with the experimental observation. 3D DNS of autoignition resolves this conflict. The conflict is mainly due to the absence of vortex-stretching phenomenon in 2D DNS. Homogeneous charge compression ignition (HCCI) is being considered as one of the strategies toward improving performance of conventional CI engines. However, HCCI engines suffer from drawbacks like lack of control over combustion and limited operating regime. One of the modifications suggested in the HCCI technology to overcome these drawbacks is the use of stratification in the fuel–air mixture. Therefore, a few DNS studies on autoignition in the stratified medium have been discussed here. Further, discussion on the conditional moment closure (CMC) model and its validation using DNS data has been presented. Ignition delay time predicted by CMC was found to be in good agreement with DNS predictions.