Spark Ignited (SI) combustions engines in combination with different degrees of hybridization are expected to play a major role in future vehicle propulsion. Due to the combustion principle and the related thermodynamic efficiency, it is especially challenging to meet future CO2 targets. The layout and optimization of the overall system requires novel methods in the development process which feature a seamless transition between real and virtual prototypes. Herein, engine models need to predict the entire engine operating range in steady-state and transient conditions and must respond to all relevant control inputs. In addition, the model must feature true real-time capability.
This work presents a holistic and modular modeling framework, which considers all relevant processes in the complex chain of physical effects in SI combustion. The basis is a crank-resolved cylinder model which describes gas exchange and compression to determine the thermodynamic state and turbulence conditions at spark-advance. Ignition and flame front combustion are modeled by a mechanistic, quasi-dimensional combustion model with a detailed consideration of combustion chamber geometry for flame-wall interaction. Cycle-to-cycle variations are imposed in a semi-empirical manner in order to provide realistic boundary conditions for the thermo-chemical knock model.
The models are validated against engine measurements for a passenger car sized TGDI engine in a wide range of operating conditions covering the entire engine map. Emphasis is put on comparing pressure and heat release traces, not only for the mean cycle, but for the range of stochastic variations of 100 measured cycles. The validation results confirm a good level of agreement between measured and simulated results. To demonstrate capabilities of the proposed modeling concept, a model-based optimization is performed in a computational study, aiming at an optimization of engine efficiency under knocking constraints. The study examines two motoric measures, namely water injection and variable compression ratio. Finally, the optimized model runs in a transient drivecycle simulation. The test is performed on a HiL system to prove the model’s real-time capability.