Stochastic sensitivity analysis of numerical simulations of injector internal flows to cavitation modeling parameters

A stochastic analysis of the cavitation model parameter sensitivity is carried out for internal flows relevant to injector configurations. Stochastic methodologies, namely generalized polynomial chaos and stochastic collocation, are used to obtain continuous response surfaces of the quantities of interest in the parameter space starting from a limited number of simulations. Cavitation is modeled through a transport equation for the void fraction closed by the Schnerr-Sauer relation, containing four free parameters. As for turbulence, the URANS equations are considered, together with two different closure models. The sensitivity to the cavitation model parameters is investigated, first, for a throttle geometry, for which experimental and LES data are available. First, two out of the four parameters are identified as the most important through a preliminary analysis based on 2D simulations, namely the vaporization and condensation factors. Then, the sensitivity of 3D simulation results to the previously identified most important parameters is investigated. The stochastic range of variability of the results contains the reference data. Thus, a parameter optimization is carried out in order to obtain the values giving the best agreement with the LES data. It is then shown that the cavitation parameter sensitivity is practically independent of the working fluid. Finally, it is shown that the calibrated cavitation model can be successfully applied to a different configuration, characterized by the hydraulic flip phenomenon, namely a 3-phase case in which liquid N-heptane flows from an inlet reservoir through a circular channel in an outlet reservoir where air is present.