Computational Studies of Autoignition and Combustion in Low Temperature Combustion Engine Environments.

Abstract

Computational studies are performed on the autoignition and combustion characteristics encountered in modern internal combustion (IC) engines in which combustion is achieved primarily by autoignition of the reactant mixture. High-fidelity computational tools with varying levels of complexity are employed in order to systematically investigate the phenomena under consideration.

As a first baseline study, the effects of unsteady temperature fluctuations on the ignition of homogeneous hydrogen-air mixture in a constant-volume reactor is studied both computationally and theoretically using asymptotic analysis. It is found that ignition delay shows a harmonic response to the frequency of imposed temperature fluctuation and the response monotonically attenuates as frequency increases.

The effects of spatial transport on the autoignition characteristics are next investigated using a one-dimensional counterflow configuration, in which unsteady scalar dissipation rate represents the effects of turbulent flow field. A newly defined ignitability parameter is proposed which systematically accounts for all the unsteady effects. n-Heptane, which exhibits a two-stage ignition behavior is studied next using similar configuration. Interestingly, two-stage ignition is observed even at significantly high initial temperatures when the ignition kernel is subjected to unsteady scalar dissipation rate. Mechanism for the appearance of two-stage ignition in unsteady conditions is found to be not chemical but is attributed to the spatial broadening of the ignition kernel and subsequent radical losses.

Guided by the above findings, multi-dimensional simulations are conducted to investigate the effects of spatial fluctuations in temperature and composition. Non-reacting 3D RANS engine simulations are first conducted to investigate different mixture formation scenarios that might exist in LTC engines prior to autoignition. Small-scale effects of these different mixture formation scenarios on the autoignition and subsequent front propagation are then studied using high-fidelity direct numerical simulation (DNS).

In the last part of dissertation, a novel principal component analysis (PCA) based approach is used to identify intrinsic low-dimensional manifolds in a complex autoigniting environment. A small number of principal components (PCs) are found to very well represent the complex reacting system. The approach thus provides a promising modeling strategy to reduce the computational complexity in solving realistic detailed chemistry in mixed-mode combustion systems.