The aim of this article is to assess the impact of turbulence and cavitation models on the prediction of diesel injector nozzle flow. Two nozzles are examined, an enlarged one, operating at incipient cavitation, and an industrial injector tip, operating at developed cavitation. The turbulence model employed includes the re-normalization group k–ε, realizable k–ε and k–ω shear stress transport Reynolds-averaged Navier–Stokes models; linear pressure–strain Reynolds stress model and the wall adapting local eddy viscosity large eddy simulation model. The results indicate that all Reynolds-averaged Navier–Stokes and the Reynolds stress turbulence models have failed to predict cavitation inception due to their limitation to resolve adequately the low pressure existing inside vortex cores, which is responsible for cavitation development in this particular flow configuration. Moreover, Reynolds-averaged Navier–Stokes models failed to predict unsteady cavitation phenomena in the industrial injector. However, the wall adapting local eddy viscosity large eddy simulation model was able to predict incipient and developed cavitation, while also capturing the shear layer instability, vortex shedding and cavitating vortex formation. Furthermore, the performance of two cavitation methodologies is discussed within the large eddy simulation framework. In particular, a barotropic model and a mixture model based on the asymptotic Rayleigh–Plesset equation of bubble dynamics have been tested. The results indicate that although the solved equations and phase change formulation are different in these models, the predicted cavitation and flow field were very similar at incipient cavitation conditions. At developed cavitation conditions, standard cavitation models may predict unrealistically high liquid tension, so modifications may be essential. It is also concluded that accurate turbulence representation is crucial for cavitation in nozzle flows.
A semi-implicit discrepancy model of Reynolds stress in a higher-order tensor basis framework for Reynolds-averaged Navier–Stokes simulations
