Evaluation of the Predictive Capability of Diesel Nozzle Cavitation Models 2007-01-0245

The predictive capability of Lagrangian and Eulerian multi-dimensional computational fluid dynamics models accounting for the onset and development of cavitation inside Diesel nozzle holes is assessed against experimental data. These include cavitation images available from a real-size six-hole mini-sac nozzle incorporating a transparent window as well as high-speed/CCD images and LDV measurements of the liquid velocity inside an identical large-scale fully transparent nozzle replica. Results are available for different cavitation numbers, which correspond to different cavitation regimes forming inside the injection hole. Discharge coefficient measurements for various real-size nozzles operating under realistic injection pressures are also compared and match well with models' predictions.

The calculations performed have indicated that the two Eulerian models predict a large void zone inside the injection hole and fail to capture the transition from incipient to fully developed cavitation, while the Lagrangian model predicts a more diffused and gradual vapour distribution in agreement with the experimental data. However, all models have predicted similarly the velocity increase inside the injection hole caused by the presence of vapour, and a similar reduction in the nozzle discharge coefficient. Liquid turbulence was significantly underestimated by the Eulerian models in the cavitation zone showing decreasing trends in contradiction with experimental observations while this was better simulated by the Lagrangian model.

Following the comparison with experiment, the effect of cavitation model assumptions, numerical implementation, discretisation scheme, model of turbulence grid resolution and cavitation physical sub-models on the predicted results on the nozzle discharge coefficient and hole exit %blockage are evaluated. The latter includes the bubble break-up and its internal initialisation pressure, the influence of the proximity of solid boundaries on the Rayleigh-Plesset equation and the percentage nuclei volume present in the liquid. Overall, it has been found that cavitation modelling has matured to the level where it can usefully identify many of the effects of cavitation on nozzle performance, and so significantly contribute to nozzle design and optimisation.