Increased residual levels in homogeneous charge compression ignition (HCCI) engines employing valve strategies such as recompression or negative valve overlap (NVO) imply that accurate estimation of residual gas fraction (RGF) is critical for cylinder pressure heat release analysis. The objective of the present work was to evaluate three residual estimation methods and assess their suitability under naturally aspirated and boosted HCCI operating conditions: (i) the simple state equation method employs the ideal gas law at exhaust valve closing (EVC); (ii) the Mirstcy method assumes isentropic exhaust process; and (Hi) the Fitzgerald method models in-cylinder temperature from exhaust valve opening (EVO) to EVC by accounting for heat loss during the exhaust process and uses measured exhaust temperature for calibration. Simulations with a calibrated and validated "virtual engine" were performed for representative HCCI operating conditions of engine speed, fuel-air equivalence ratio, NVO and intake pressure (boosting). The state equation method always overestimated RGF by more than 10%. The Mirsky method was most robust, with average errors between 3-5%. The Fitzgerald method performed consistently better, ranging ffom no error to 5%, where increased boosting caused the largest discrepancies. A sensitivity study was also performed and determined that the Mirsky method was most robust to possible pressure and temperature measurement errors.
Virtual Engine ApproachIn-depth analysis of residual estimation methods requires engine data over a wide range of operating conditions. Moreover, validation of these methods requires benchmark results of incylinder content, temperature, and properties that are difficult to measure directly. For these reasons, it was opted to perform the analysis using a virtual engine, i.e., a model of the actual engine. The virtual engine is calibrated and validated against experimental results to ensure representative behavior and then employed as the main data source for the investigation. As with any model-based approach, discrepancies are expected, but given the conceptual level of the present study, proper trend-wise performance of the model is considered of greater interest than obtaining an exact
Journal of Engineering for Gas Turbines and Power