SI Engine Combustion and Knock Modelling Using Detailed Fuel Surrogate Models and Tabulated Chemistry 2019-01-0205
In the context of today’s and future legislative requirements for NOx and soot particle emissions as well as today’s market trends for further efficiency gains in gasoline engines, CFD models need to further improve their intrinsic predictive capability to fulfill OEM needs towards the future. Improving fuel chemistry modelling, knock predictions and the modelling of the interaction between the chemistry and turbulent flow are three key challenges to improve the predictivity of CFD simulations of Spark-Ignited (SI) engines. The Flamelet Generated Manifold (FGM) combustion modelling approach addresses these challenges. By using chemistry pre-tabulation technologies, today’s most detailed fuel chemistry models can be included in the CFD simulation. This allows a much more refined description of auto-ignition delays for knock as well as radical concentrations which feed into emission models, at comparable or even reduced overall CFD run-time. The FGM model has a high level of intrinsic predictive capability, as already demonstrated for many academic cases as well as for industrial burners, gas turbines and Diesel engines. The application to gasoline engines has not been much investigated however, mainly due to the difficulty of model implementation and chemistry tabulation technology.
In this work, the FGM model in AVL FIRE™ is assessed on multiple load points of a modern SI engine, using various levels of detail for the fuel chemistry, ranging from industry-standard reduced chemical schemes to state-of-the-art gasoline surrogate models. The investigated configuration represents a turbocharged gasoline engine with direct injection and homogeneous charge combustion. The engine simulation involved gas-exchange, fuel spray injection, spark ignition and combustion phases. The computational grid provides a mesh resolution in the range of 0.6–0.3 mm within the cylinder.
CFD simulation has been started at the beginning of the exhaust stroke allowing to correctly predict the residual gas mass fraction, the mixture fraction as well as the temperature field – the parameters which can directly affect the knocking effects. Firstly, the comparison between the simulated pressure curves and available experimental data for a spark timing sweep were discussed. The special attention was dwelled on an ability of the FGM based knock model to characterize the inherent knock combustion features including a knock probability and providing an information on knock cycle indicator. Comparative results and analysis will be provided in the full paper.
Dmitry Goryntsev, Ferry Tap, Mijo Tvrdojevic, Peter Priesching