DEVELOPMENT OF ADVANCED MODELS FOR COMBUSTION SIMULATIONS IN IC ENGINES

CFD simulation of the combustion process is very important for a correct prediction of both engine performance and pollutant emissions such as soot and NOx. However, combustion modeling involves many complex phenomena that need to be handled in detail. Among them, the most important ones are represented by initial flame kernel formation, flame propagation, fuel oxidation and turbulencechemistry interaction. Furthermore, the rapid evolution of SI and Diesel combustion systems requires a continuous improvement of the existing numerical and physical models, to correctly account for the effects of stratified or very diluted mixtures, multiple injections and complex fuel kinetics. In this work, the proposed approaches to model both compression-ignition and spark-ignition combustion are illustrated. In particular, the representative interactive flamelet (RIF) model is used for Diesel combustion. It accounts for both detailed chemistry and turbulence-chemistry interaction. The possibility to use a multiple number of flamelets makes the model also suitable for the simulation of advanced combustion modes including multiple injections. For what concerns SI combustion, the Extended Coherent Flamelet Model (ECFM) is used in combination with a Lagrangian model to describe the first stages of the flame propagation process, where the flame kernel formation and growth processes are influenced also by energy transfer from the spark and heat conduction inside the gas phase. The proposed set of models has been embedded into the Lib-ICE code, which is a set of libraries and solvers for the simulation of IC engines based on the OpenFOAM® technology. Experimental validation was carried out by simulating spray combustion in a constant-volume vessel and premixed flame propagation in an optical engine. Diesel spray combustion modeling Simulation of non-premixed, reacting turbulent flows requires a detailed description of the interaction between turbulence and chemistry, that affects both auto-ignition and mixing-controlled combustion phases. Detailed chemistry is also necessary to account for the complex fuel oxidation process under a wide range of operating conditions (temperature, pressure, equivalence ratio and residual gas fraction). Among the different proposed approaches, flamelet models are widely XXXV Meeting of the Italian Section of the Combustion Institute 2 used. Their main assumption is that both auto-ignition and combustion of a turbulent diffusion flame can be predicted by solving the reaction-diffusion problem in the mixture fraction space for an equivalent igniting laminar diffusion flame, where effects of turbulence and flow field are taken into account by means of the scalar dissipation rate [1]. To better describe the flame structure, the possibility to use a multiple number of flamelets was also introduced [2,3]. In particular, each flamelet represents a certain portion of the injected fuel mass and evolves independently with its own scalar dissipation rate. In this way, effects of flow field are better accounted for, with the possibility to predict local flame extinction close to the nozzle, where gas velocities and mixture fraction gradients are very high. In the proposed approach, for each flamelet, the reaction diffusion problem is solved for both chemical species and sensible enthalpy assuming unit Lewis number. Equations are solved on a 1D mesh representing the mixture fraction space. To reduce the computational time and allow the use of detailed chemistry, the TDAC algorithm was also employed to limit the operation of the ODE stiff solver when chemistry has to be integrated [4]. The chemical composition in each computational cell is computed from flamelet species profiles in the mixture fraction space and flamelet marker functions :