New approaches to design of experiment modeling based on engine control levers and fuel properties

Abstract

In this paper, prediction of performance and emission of a single cylinder heavy duty diesel engine has been studied by considering “Partly Homogeneous Charge Compression Ignition” combustion. The fuel matrix consisted of 10 different diesel and diesel like fuels which were utilized for the entire investigation. First, with the help of a Principal Component Analysis (PCA) approach, the least correlated fuel properties were detected. The most independent fuel properties, as well as four engine control parameters were considered as independent DoE (Design of Experiment) parameters for engine response models. The EGR rate, boost pressure, rail pressure and SOI of early pilot injection for premixed combustion were varied as engine control levers. Separate DoE models including engine control levers plus 2–3 independent fuel properties were setup and analyzed. DoE tests have been performed at the test bench on a 2.1 L single cylinder Heavy Duty Diesel Engine for each fuel. The DoE test results were carefully modeled and analyzed by a DoE program called ProCal. Afterwards, series of combined DoE models were virtually created. A V-Optimal regression model was considered for the modeling and separate strategies used for engine control parameters and fuel properties to derive reasonable models. Results show excellent predictions for a combination of all engine and fuel variables. Furthermore, prepared models can predict engine response to the defined premixed combustion at different operating points and impact of fuel properties at the certain areas of engine map.