To reveal the generation mechanism of oil-film cavitation in the main bearing of diesel engines under dynamic load, this paper adopts large eddy simulation combined with overlapping mesh technology, considers the correction of physical property parameters of lubricating oil under high pressure and high temperature, realizes the transient calculation of the three-phase flow field of bearing bush oil film under dynamic boundary conditions, and verifies the accuracy of the calculation method through experiments. The results show that the cavitation of the main bearing oil film in diesel engines mainly comes from three types of mechanisms: high-speed jet cavitation caused by the contraction of the flow channel when the journal oil hole passes the end of the oil groove, the gradually shrinking and gradually expanding flow channel cavitation caused by journal eccentricity, and the instantaneous negative pressure cavitation resulting from journal radial displacement.

Cavitation flow inside an injector control valve significantly affects the injection performance and operational reliability of diesel engines, and the selection of turbulence and cavitation models is crucial for accurately predicting cavitating flows. To evaluate the predictive performance of different model combinations, a numerical model was established based on the Winklhofer microchannel experiment. Three large eddy simulation (LES) subgrid-scale (SGS) models were coupled with three cavitation models, and their performance was assessed by comparing the predicted outlet mass flow rate, centerline pressure distribution, and cavitation distribution with the experimental results. The results show that the outlet mass flow rate increases with the inlet–outlet pressure difference. When the pressure difference reaches approximately 70 bar, the growth rate decreases significantly, indicating the onset of typical cavitation-induced flow choking. Under low pressure differences, the prediction results of different model combinations are generally similar, whereas significant discrepancies are observed under high-pressure-difference conditions with intense cavitation. In terms of mass flow rate prediction, the Dynamic Smagorinsky + FRP model combination shows the best agreement with the experimental results in the pressure difference range of 70–85 bar, with an error of 3.06%. For pressure distribution prediction, the Dynamic Smagorinsky + FRP and Smagorinsky + FRP model combinations exhibit relatively smaller errors than the other combinations. The cavitation distribution results indicate that, compared with the LES subgrid-scale model, the cavitation model has a more pronounced influence on the extent and evolution of the cavitation region. Overall, the Dynamic Smagorinsky + FRP model demonstrates the best comprehensive predictive performance. Furthermore, this model was applied to the simulation of cavitating flow in a two-dimensional injector control valve, and the predicted results agree well with the visualization experiments. The present study provides a useful reference for the numerical simulation and structural optimization of cavitating flows in injector control valves.