A Tabulated-Chemistry Approach Applied to a Quasi-Dimensional Combustion Model for a Fast and Accurate Knock Prediction in Spark-Ignition Engines

摘要

The description of knock phenomenon is a critical issue in a combustion model for Spark-Ignition (SI) engines. The most known theory to explain this phenomenon is based on the Auto-Ignition (AI) of the end-gas, ahead the flame front. The accurate description of this process requires the handling of various aspects, such as the impact of the fuel composition, the presence of residual gas or water in the burning mixture, the influence of cool flame heat release, etc. This concern can be faced by the solution of proper chemistry schemes for gasoline blends. Whichever is the modeling environment, either 3D or 0D, the on-line solution of a chemical kinetic scheme drastically affects the computational time. In this paper, a procedure for an accurate and fast prediction of the hydrocarbons auto-ignition, applied to phenomenological SI engine combustion models, is proposed. It is based on a tabulated approach, operated on both ignition delay times and reaction rates. This technique, widely used in 3D calculations, is extended to 0D models to overcome the inaccuracies typical of the most common ignition delay approaches, based on the Livengood-Wu integral solution. The aim is to combine the predictability of a detailed chemistry with an acceptable computational effort. First, the tabulated technique is verified through comparisons with a chemical solver for a semi-detailed kinetic scheme in constant-pressure and constant-volume configurations. Then a phenomenological model, based on the end-gas AI computation, is utilized to predict the knock occurrence in different SI engines, including both naturally-aspirated and turbocharged architectures. 0D/1D simulations are performed both with an online solution of the chemistry and employing the tabulated approach. Assessment with reference KLSA values shows that the knock model, based on the tabulated chemistry, is able to well reproduce the essential features of the auto-ignition process in the analyzed engines, with a limited impact on the computational time.