Injection and autoignition of methane in diesel engines, a numerical study

Abstract

This thesis presents a numerical study on the autoignition process of methane in diesel engines. Traditionally, the method used by similar studies is based on numerical studies in shock tube geometries. This work represents the first computational work on the autoignition of natural gas in geometric environments which closely mimic those encountered in diesel engines.

In the present work, a numerical study that models the injection process up to the time of autoignition uses an in-house code developed by the author which solves the Reynolds averaged Navier-Stokes equations coupled with a simplified reaction (ignition) model. The equations are implicitly solved in fully coupled format. Turbulence viscosity is obtained by solving a specialized k - e turbulence model which has been specially modified for application to the engine flow. The transient flow computation stops as soon as local ignition occurs.

As part of this research, a specialized conjugate gradient algorithm with block correction algorithm, B-CGSTAB, has been developed. The scheme greatly improves upon the inherent disadvantage of standard conjugate gradient methods, i.e., it improved convergence rate for solving large numbers of grids and improved capabilities for handling large aspect ratio grids.

This work establishes that the injector/cylinder orifice velocity and temperature profiles are characterised by a double-hump shape, significantly different from the conventional box-hat or parabolic profile usually assumed for unsteady high speed injection processes. The double-hump profile results from supersonic injection causing radical expansion when the flow is transient.

The present study formulates a new kinetic model for modelling natural gas autoignition. Numerical regression is used to modify Arrhenius constants and activation energies so as to make the computational results conform to the experimental data. In addition, this analysis has lead to a better understanding of how to determine possible ignition locations. Ignition generally occurs at locations where the fuel-air mixture is near: stoichiometric and the fluid velocities are low. This latter condition is intepreted to coincide with regions where the convective heat losses are minimal.