A real-time control-oriented discrete nonlinear model development for in-cylinder air charge, residual gas and temperature prediction of a Gasoline Direct Injection engine using cylinder, intake and exhaust pressures. (February 2022)

• Record Type: Journal Article
• Authors: Khameneian, Amir; Wang, Xin; Dice, Paul; Naber, Jeffrey D.; Shahbakhti, Mahdi; Archer, Chad; Moilanen, Peter; Glugla, Chris; Huberts, Garlan
• Is Part Of: Control engineering practice. Volume 119(2022)
• Publication Date: 2022-02
• Subjects: SI -- Real-time -- Air charge -- Residual gas -- Cylinder temperature; Automatic control -- Periodicals; 629.89
• DOI: 10.1016/j.conengprac.2021.104978
• Languages: English
• ISSNs: 0967-0661

Abstract

Abstract: The in-cylinder trapped air, residual gas, and temperature are critical dynamic parameters in Gasoline Direct Injection (GDI) Spark Ignition (SI) engines for fuel and combustion control. However, their real-time prediction for transient engine operations is complicated, especially when concerning variable valve timing. A dynamic cycle-by-cycle control-oriented discrete nonlinear model is proposed in this study to estimate the in-cylinder mixture temperature and the mass of trapped air and residual gas at the point of Intake Valve Closing (IVC). The developed model uses in-cylinder, intake, and exhaust pressures as the primary inputs. The model consists of two major sub-models for air charge and residual gas. The air charge sub-model estimates the mass of trapped air and total residual gas, and in-cylinder gas temperature. The residual gas sub-model calculates the exhaust gas backflow during valve overlap, capturing gas exchange dynamics. The exhaust gas backflow into the cylinder is estimated using a compressible ideal gas model designed for engines equipped with Variable Valve Timing (VVT). The integrated model into a rapid-prototype control system for real-time operation runs 2800 times faster than the GT-Power TPA model. The model's dynamic behavior is validated using an engine dynamometer transient test cycle under real-time conditions. The model's cycle-based output parameters are in good agreement with dynamic experimental data with minimal delay and overshoot. The experimental validation results show that the input dynamics propagated in the in-cylinder pressure trace are followed by the estimated outputs, with 2.2 mg, 0.16%, and 4.6 C average steady-state error for estimated air mass, residual fraction, and in-cylinder temperature at IVC, respectively. Highlights: A physics-based model to estimate the mass of trapped air charge and residual gas. Dual-Pressure Pegging method using intake and exhaust absolute pressures. Dual-Scale Amplification technique for in-cylinder pressure transducer signal. Real-time validation of the model on the rapid-prototype engine control system. DOE analysis for minimum calibration effort.