Predicting the Onset of End-Gas Autoignition with a Quasi-Dimensional Spark Ignition Engine Model

Abstract

A predictive, quasi-dimensional simulation of combustion in a spark-ignition engine has been coupled with a chemical kinetic model for the low temperature, preflame reactions of hydrocarbon fuel and air mixtures. The simulation is capable of predicting the onset of autoignition without prior knowledge of the cylinder pressure history. Near-wall temperature gradients were computed within the framework of the engine cycle simulation by dividing the region into a number of thin mass slices which were assumed to remain adjacent to the combustion chamber surfaces in both the burned and unburned gas. The influence of the near-wall turbulence on the temperature field was accounted for by means of a boundary layer turbulence model developed by the authors. Fluid motion in the bulk gases has been considered by the inclusion of the turbulence model based on k-\ge theory while the flame propagation rate was predicted using a fractal flame model. Validation of the engine-cycle simulation is afforded by comparison of predicted turbulence intensities, gas temperatures, gas-wall interface heat fluxes and pressure histories with appropriate measured values. All showed good agreement. The heat release from the preflame reactions was estimated using the reduced kinetic model and has been accounted for in the overall engine simulation energy balance. The chemical kinetic and engine cycle models were solved simultaneously enabling the simulation of the coupled effects of the heat release by the preflame reactions on the end-gas temperatures and the subsequent influence of these temperatures on the rates of reaction in the end-gas region as combustion proceeds. Validation of the model is afforded by comparison of cylinder pressure predictions with those measured in a knocking engine. After modifying the Arrhenius parameter for the chain branching reaction in the chemical kinetic model the knock onset time was accurately predicted. In this context, the model was then employed to demonstrate the influence of the flame propagation rate on the knock onset time in a simulation of cyclically varying knock.