• cavitation
• injector
• LES
• turbulence

The aim of the present thesis is making a comparison between different numerical codes for simulations of injector internal flows in cavitating conditions for automotive applications. The numerical codes considered are the commercial software SIEMENS-STARCCM+ and the in-HOUSE code AEROCAV.

Injector flows are characterized by cavitation interacting with turbulence in channels of very small size, making simulations and experiments very challenging. Moreover, in automotive applications, a further source of complexity is due to the pressure differences between the inlet and outlet reservoirs which may reach very high values, up to 100/1000 bars.

The reference test-case is a simplified rectangular cross-section throttle geometry for which both numerical and experimental data are available in the literature. This test-case is characterized by a liquid-into-liquid injection, thus reproducing a 2-phase flow condition and, despite the geometrical simplification, contains all the main difficulties encountered in practical applications, such as cavitation, turbulence, small dimensions and high-pressures.

The work is divided in two parts: in the first one, the simulations (cavitating and non-cavitating) performed with STARCCM+ are presented after having introduced the physical modelling and the numerical discretization; in the second part, a non-cavitating simulation carried out with the in-house code AEROCAV is described.

Concerning physical modelling a popular homogeneous flow model has been adopted for both codes. For STARCCM+, the phases are considered incompressible with a transport equation for the fraction of vapour; the mass-transfer occurring between liquid and vapour phases is represented by means of a source term containing the Schnerr-Sauer cavitation model together with the classical Rayleigh-Plesset equation. As for AEROCAV, the physical modelling adopts a barotropic model based on a suited relation between pressure and mixture density.

We adopted a Large-Eddy Simulation approach, which proved to well represent the dynamics of large-scale turbulence structures even for rather complex flow configurations. Cavitating flow conditions are first simulated by using STARCCM+ and the cavitation model calibrated for URANS simulations in a previous work. The impact of turbulence modeling is analyzed by carrying out LES simulations of the 2-phase throttle flow for difference Sub-Grid Stress (SGS) tensor models. Sensitivity to the numerical scheme for the convective terms and to the discretization grid are also carried out.

Then, large-eddy simulations are performed for the same flow configuration in non-cavitating conditions. Comparison are also made with the results obtained by implicit LES simulations carried out with AEROCAV, a in-house code originally developed for the numerical simulations of cavitating flows in turbopumps for space applications.