Spark Assisted Compression Ignition (SACI) aims to increase the load limit of homogeneous charge compression ignition (HCCI) engines, enabling the benefits of dilute combustion over a larger engine operation range. Compared to HCCI, SACI exhibits higher cyclic variation of several combustion features. Due to the necessity of control of the timing of the auto-ignition event during SACI operation, a suitable characterization of the combustion at a given set of actuator inputs is required to enable robust model-based controls of combustion.
This paper investigates statistical approaches to analyze in-cylinder pressure data of SACI in order to find a real or reconstructed cycle that will represent the important characteristics of combustion. To determine the representativeness of such a cycle, several combustion characteristics were compared that could serve as operational limits. The statistical approaches contain both artificial and physical cycles, including mean pressure, median pressure and cycles that were determined by the median value of a certain characteristic.
The analysis first compares the combustion parameters of the selected cycles to the average value of the operating point. Secondly, a linear regression analysis was conducted, allowing the coefficient of determination to be used as a basis for representativeness. The results show that although reconstructing cycles with statistical approaches might adequately represent certain characteristics, averaging may result in significant information loss. Therefore, finding a real cycle representing all of the unique behaviors of an operating point has the possibility of being a better approach for identifying combustion parameters in SACI.