A Fundamental Study of the Factors Affecting Adverse Autoignition (Knock) in Internal Combustion Engines

Project Description

To comply with the increasingly stringent emissions standards and to improve engine efficiencies, engine downsizing and turbo-charging are becoming a standard adoption in the automotive industry. However, further downsizing, boosting, or increases of compression ratio face the trade-off of increased levels of abnormal combustion, one example being knock. Engine knock must be addressed by engine designers to meet performance and emission requirements. Even after a century of addressing knock via measures such as leaded fuels, fuel octane adjustments and combustion phasing, the engine-research community still does not have a comprehensive understanding of knock, partly because of the limitations of experiments and partly because computational simulations have only recently achieved a high-enough level of fidelity.

With the development of increasingly more powerful computational capabilities, studies of the knock phenomena can be shifted from a time-consuming and expensive experimental space to the virtual space. This can offer a new perspective into understanding adverse, abnormal combustion at a much higher level of detail and with greater control of the test space when compared to experiments. In the industry, Reynolds-Averaged Navier-Stokes (RANS) turbulence models are commonly incorporated for engine combustion simulations due to it being computationally inexpensive. However, it is known that knocking behavior cannot be captured well using RANS turbulence models due to the lack of detail, as this model tends to average out the turbulent fluctuations in CFD calculations. Thus, this study shall employ LES to provide greater detail into the flow field. LES models only are executed for small periods of time and length scales of turbulence. Therefore, more of the flow is computationally resolved to get more accurate results in terms of knocking prediction to form a better understanding of knocking behavior. Additionally, the final simulations will apply a state-of-the-art methodology to resolve high-speed pressure oscillations in the cylinder; this level of detail is typically not realized in standard simulations because of computational costs, but it does provide critical information regarding the onset and propagation of knock.

It has also been well studied in the literature that wall temperatures play a significant part in influencing knock. Thus, this study shall also incorporate Conjugate Heat Transfer (CHT) modelling to fully capture the transient and spatially varying effects that combustion has on the solid surfaces that, in turn, influence knock; this is an advance over current standard practice in which surface temperatures are assumed and often set constant per surface. Finally, a SAGE detailed chemical mechanism will be employed to model combustion. This will include detailed NOx chemistry so that the effects of knock on emissions can be analyzed.

By utilizing the greatly increased computing capability at OLCF, the aim is to simulate an internal combustion engine cylinder in high fidelity, in conjunction with CHT and LES turbulence modelling, to better understand the conditions that lead to engine knock. The level of detail required to include these components has rarely been captured with the limited resources available in industry. An improved understanding of knocking behavior can influence future engine designs that can reduce knocking, and to better predict knocking onset from a controls perspective.