Search Results

The US Department of Energy has formulated different gasoline fuels called ''Fuels for Advanced Combustion Engines (FACE)'' to standardize their compositions. FACE I is a low octane number gasoline fuel with research octane number (RON) of approximately 70. The detailed hydrocarbon analysis (DHA) of FACE I shows that it contains 33 components. This large number of components cannot be handled in fuel spray simulation where thousands of droplets are directly injected in combustion chamber. These droplets are to be heated, broken-up, collided and evaporated simultaneously. Heating and evaporation of single droplet FACE I fuel was investigated. The heating and evaporation model accounts for the effects of finite thermal conductivity, finite liquid diffusivity and recirculation inside the droplet, referred to as the effective thermal conductivity/effective diffusivity (ETC/ED) model.

An earlier paper has shown the ability to predict the phasing of knock onset in a gasoline PFI engine using a simple ignition delay equation for an appropriate surrogate fuel made up of toluene and PRF (TPRF). The applicability of this approach is confirmed in this paper in a different engine using five different fuels of differing RON, sensitivity, and composition - including ethanol blends. An Arrhenius type equation with a pressure correction for ignition delay can be found from interpolation of previously published data for any gasoline if its RON and sensitivity are known. Then, if the pressure and temperature in the unburned gas can be estimated or measured, the Livengood-Wu integral can be estimated as a function of crank angle to predict the occurrence of knock. Experiments in a single cylinder DISI engine over a wide operating range confirm that this simple approach can predict knock very accurately.

The accurate prediction of engine performance maps can guide data-driven optimization of engine technologies to control fuel use and associated emissions. However, engine operational maps are scarcely reported in literature and often have missing data. Assessment of missing-data resilient algorithms in the context of engine data prediction could enable better processing of real-world driving cycles, where missing data is a more pervasive phenomenon. The goal of this study is, therefore, to determine the most effective technique to deal with missing data and employ it in prediction of engine performance characteristics. We assess the performance of two machine learning approaches, namely Artificial Neural Networks (ANNs) and the extreme tree boosting algorithm (XGBoost), in handling missing data.