Development of correlation model for cavitating spray using Eulerian simulations

Abstract

Modern high-pressure injection systems for heavy-duty applications are prone to cavitation, which may have a detrimental effect on the system integrity. The present study aims to develop reliable correlations for Lagrangian engineering simulations to predict the onset of cavitation. Using the extensive data generated from the high-fidelity Eulerian volume-of-fluid (VOF) simulations of a single-hole injector for a wide range of operating conditions, a proper set of discharge (C_d) and contraction (C_a) coefficients are determined. In particular, four significant factors characterizing nozzle geometry are considered: the K factor, inlet curvature radius, nozzle exit radius, and channel length, as well as ambient and injection pressures. The results showed that a larger K factor and inlet curvature radius yielded higher C_d and C_a values, while the exit radius and channel length had a less significant impact on the cavitation dynamics. The smaller pressure difference between the nozzle inlet and exit mitigated the cavitation formation. These trends were well described using the mathematical regression model. Furthermore, two other correlation models based on the regression learner and artificial neural network methods were also developed. Of these three models, the mathematical regression model yielded the best agreement with an external validation dataset.