Development of methodology for predictive diesel combustion simulation using 0D stochastic reactor model

Stringent exhaust emission limits and new vehicle test cycles require sophisticated operating strategies for future diesel engines. Therefore, a methodology for predictive combustion simulation, focused on multiple injection operating points is proposed in this paper. The model is designated for engine performance map simulations, to improve prediction of NOx, CO and HC emissions.

The combustion process is calculated using a zero dimensional direct injection stochastic reactor model based on a probability density function approach. Further, the formation of exhaust emissions is described using a detailed reaction mechanism for n-heptane, which involves 56 Species and 206 reactions. The model includes the interaction between turbulence and chemistry effects by using a variable mixing time profile. Thus, one is able to capture the effects of mixture inhomogeneities on NOx, CO and HC emission formation.

The mixing time model is parameterized using transfer functions for engine operating parameters, e.g., injection mass, injection duration, air fuel ratio, start of injection and speed. These functions are calibrated for nine operating points using multi objective simulated annealing optimization combined with fast running metamodels that speed up the optimization process. The calibrated transfer functions are validated for nine additional operating points. The results for the calibration and validation points show a good match of the combustion heat release rate. Especially the main injection heat release rate is well predicted by the model. The NOx and CO emissions reflect the experimental trends and are in close range to the measurements. Finally, the model is tested for triple injection operating points. The results match the measurements, which show the applicability of the stochastic reactor model in conjunction with the mixing time transfer functions for engine performance map simulations.