A deterministic model for wear of piston ring and liner and a machine learning-based model for engine oil emissions

Abstract

Nowadays, more constraints are required for design of internal combustion engines, to meet the energy saving and the emissions standards in the new era. Engine emissions and engine durability are two of the most important factors in the development of IC engines.

Engine particulate emissions are strongly correlated with the lubricant oil consumption. On the other hand, the carbon soot particles mixed in the lubricant from the combustion are the major source for long term wear of the piston, piston ring, and cylinder liner. Costly engine tests are required to develop the new system to meet emission and durability requirements. More advanced data analytics and models connecting critical design and operating parameters to performance will help shorten the development lead time for more efficient and cleaner engines.

This thesis work aims to model the engine wear during break-in and steady-state stages, capture oil emission correlations with engine operating parameters, and provide engine design guidance. This work is the first time to build deterministic physics-based wear models to perform systematic level engine wear simulations, including the effect of the liner topography. The wear simulation results are compared to experimental outcomes for both engine stages. It is also the first try to model the oil emission based on machine learning and connect the data-driven results with different engine ring-pack designs. The results suggest a good consistency of the machine learning analyzation and the underlying oil emission physics. The entire defined data-driven procedures show a promising future to accelerate engine development cycle, reduce engine testing cost, and help understand oil transport mechanisms and design influences.