0D/1D modeling methods of in-cylinder combustion

Abstract

This thesis was a literature review for 0D/1D modeling methods for in-cylinder combustion in internal combustion engines, focusing on their theoretical foundations, practical applications, and comparative performance.

The study categorized combustion models into non-predictive, semi-predictive, and predic-tive approaches. Non-predictive models, such as the widely used Wiebe function, offered simplicity and computational efficiency but lacked adaptability to complex combustion phe-nomena. Semi-predictive models integrate empirical data and machine learning techniques, such as artificial neural networks (ANN), to estimate burn rate parameters, which improved accuracy within the training range but limited generalization. Predictive models, including entrainment and fractal approaches, relied on phenomenological principles to simulate tur-bulence and flame propagation, providing physical fidelity and predictive relevance at the cost of increased calibration complexity and computational demand.

A comprehensive literature review highlighted advancements in modeling strategies for spark-ignition engines under diverse conditions, including alternative fuels and lean-burn operation. Results indicated that while non-predictive models remain suitable for preliminary analysis and control-oriented applications, predictive models offered greater physical fidelity for performance optimization and emissions prediction. Semi-predictive methods bridged the gap between simplicity and precision but required extensive datasets for robust imple-mentation.

The findings emphasized the trade-offs between model complexity, accuracy, and computa-tional efficiency, offering guidance for selecting appropriate modeling techniques in engine design and simulation.