Control-Oriented Combustion Modeling for Compression-Ignition Engines Operating With Low Cetane Fuels

Abstract

To support the transition toward sustainable alternative fuels in the transportation sector, advanced statistical modeling techniques are needed to characterize and control combustion at low cetane fueling conditions. In this work, a novel control-oriented combustion phasing model is presented that simulates asymmetrical phasing distributions as a function of fuel cetane, fuel injection timing, and electrical power supplied to a thermal ignition assist device. The model was regressed using experimental data from a commercial compression-ignition (CI) engine operating with four fuel blends of cetane number ranging from 25 to 48. The model estimates the mode and median of phasing distributions within 2.4 and 2.6 crank angle degrees (CAD), respectively, over a range of 52 CAD. Additionally, kernel density estimation (KDE) is proposed for nonparametric modeling of torque as a function of phasing. A set of KDE-driven extremum seeking algorithms for learning torque-optimal phasing ranges and probabilistic misfire limits is validated across the multi-fuel data. By using a probabilistic criteria, this approach can permit an acceptable proportion of undesired events to occur before classifying a limit. This feature better suits it for engine operation with low cetane fuels, where late burns and misfires may not be entirely avoidable. This work will facilitate online learning-based control strategies that can enable CI engines to adapt to fuels with a wide range of ignition behaviors.