A methodology for boundary condition optimization in 3D-CFD of internal combustion engines

This study presents a novel computationally efficient methodology for tuning boundary conditions in 3D CFD simulations of internal combustion engines. The objective is to achieve errors below 1% for in-cylinder pressure at spark timing and below 5% for intake mass. The approach employs an iterative procedure which is grounded on thermodynamic principles, applying the laws of thermodynamics based on both numerical and experimental data during the compression phase. A multi-zone convection model is developed to optimize wall temperatures, while intake pressure is adjusted using physics-informed considerations. In order to refine the quasi-D heat transfer coefficients, the method incorporates a hybrid optimization approach that combines simulated annealing with interior point methods. The methodology is tested on a heavy-duty port fuel injection spark-ignition engine fueled by compressed natural gas. The results demonstrate the effectiveness of this methodology in improving the accuracy of CFD simulations for internal combustion engines, providing a robust framework for boundary condition optimization. This allows the number of 3D simulations required for model tuning and calibration to be reduced compared to trial-and-error or DoE approaches, enhancing model predictive capability.