Split injection strategy for diesel sprays: Experiment and modelling

An experimental programme to characterise Diesel fuel sprays was conducted in a Proteus high-pressure rapid compression machine (RCM), at Sir Harry Ricardo Laboratories at University of Brighton. The Proteus experiments aimed to simulate realistic Diesel engine working conditions whilst allowing visualisation of in-cylinder processes by various optical and laser diagnostics techniques. The spray penetration was explored by laser diagnostics methods such as Laser Induced Fluorescence (LIF) and the Mie scattering techniques. High-speed video (HSV) images of the spray were also recorded to show the temporal evolution of the spray. Both liquid and vapour phases of the spray were captured by the LIF technique whilst Mie scattering only recorded the liquid part of the spray. In the current study, a 7-hole injector of the solenoid type was injecting 20mm3 of Diesel fuel each cycle at an injection pressure of 100MPa and in-cylinder pressures 2MPa and 6MPa. The fuel was injected in a split mode with various dwells between the 10mm3 + 10mm3 splits (or individual fuel injections). The instantaneous injection rate was measured by the long-tube rate technique. These data were taken as an input to the numerical model tracking the centre-of-mass (CoM) of the injected fuel. The modelling is based on the conservation of momentum of injected fuel mass in the presence of a realistic drag force acting on the whole spray as a physical body. This approach is particularly suitable for the dense sprays near the nozzle as an asymptotic case for the strong interaction between the spray droplets. Hence the CoM approach is seen as complementary to the traditional Lagrangian modelling for dispersed sprays widely employed by Computational Fluid Dynamics (CFD) codes. Air entrainment was modelled by the exponential decay of liquid fraction in the spray with a characteristic time t. Under the conventional assumption of the conical shape of the spray, the penetration of spray tip was associated with the height of the cone. This allowed the calculation of the frontal area A required in the expression for the drag force. The numerical CoM model was validated against the experimental CoM data in the range of in-cylinder pressures and dwells between two consecutive injections (or splits). The following four cases were calculated and validated against the experimental data: in-cylinder pressures (ICP) 2MPa and 6MPa, split injection strategy with dwells 0.425ms and 0.625ms between injections. In all cases, the injection pressure was 100MPa; under cold intake conditions of ambient air. For validation purposes the image processing software was extended to characterise the position of the centre-of-mass of injected fuel. The ratio of tip penetration to the position of the centre-of-mass, b was assessed from LIF images for ICP = 2MPa and dwells 0.425ms and 0.625ms. An average value of the ratio with a corresponding standard deviation b =1.85 ± 0.3was accepted for the validation of the model calculations versus experiment for all cases under consideration. An uncertainty corridor was constructed for the model validation against the HSV experiment. The corridor was formed by the curves corresponding to b = 2.15 and b =1.55 with the curve for b =1.85 in the middle. A good agreement was observed between the calculated and experimental CoM penetration. The same set of modelling parameters including spray dispersion time t = 0.15ms was taken by the model for all the cases under consideration.