The paper illustrates and validates a novel predictive combustion model for the estimation of performances and pollutant production in CI engines. The numerical methodology was developed by the authors for near real-time applications, while aiming at an accurate description of the air mixing process by means of a multi-zone approach of the air-fuel mass. Charge stratification is estimated via a 2D representation of the fuel spray distribution that is numerically derived by an axial one-dimensional control-volume description of the direct injection. The radial coordinate of each control volume is reconstructed a posteriori by means of a local distribution function. Fuel mass clustered in each zone is further split in ‘liquid’, ‘unburnt’ and ‘burnt’ sub-zones, given the local properties of the fuel spray control volumes with respect to space-time location of modelled ignition delay, liquid length, and flame lift-off. Multiple injections are described on the same numerical grid to account for jet-to-jet axial interactions, whose effects reflect on improved ignition characteristics. The multi-zones are open systems which are discretized on the equivalence ratio; mass is allowed to travel from one to another, causing a 2D charge stratification. For each zone, local thermodynamic properties and NOx production are determined to estimate cylinder-average performances and emissions. The apparent heat release rate, in-cylinder pressure, BSFC and NOx emissions are validated against experimental data of full map of a light-duty engine. The computational effort of the model is relatively low, which makes the approach suitable for static optimization, to be used in 1-D simulation codes for transient’s optimization and can be run in parallel with a real time 1-D gas model for the simulation of long driving cycles.
