Surrogate-based modeling of cryogenic turbulent cavitating flows

The cryogenic cavitation has critical implications on the performance and safety of liquid rocket engines. In this study, a systematic investigation based on the surrogate modeling techniques is conducted to assess and improve the performance of a transport-based cryogenic cavitation model. Based on the surrogate model, global sensitivity analysis is be conducted to assess the role of model parameters regulating the condensation and evaporation rates, and uncertainties in material properties, specifically, the vapor density and latent heat. The surrogate models considered include the response surface, radial basis neural network, Kriging, and a weighted average composite model combining all surrogates. It is revealed that the vapor density and the model parameter controlling the evaporation rate are more critical than latent heat and the model parameter controlling the condensation rate. Based on the recommended model parameter values, better prediction of the cryogenic turbulent cavitation can be attained.