In this paper, a general workflow for the global Sensitivity Analysis (SA) has been proposed based on the coupling of VUQ toolkit DAKOTA and Multiphase Computational Fluid Dynamics (MCFD) solver boilEulerFoam. A surrogate model is first constructed based on sampling simulations from boilEulerFoam. This surrogate is based on Gaussian Processes Model (GPM) and is validated and proved to have good properties. The Morris Screening method is then applied based on the surrogate to those interfacial momentum closure terms for SA, including drag, lift, turbulent dispersion, wall lubrication, and virtual mass. Two different cases are considered, one is on low-pressure adiabatic flow, and the other is on high pressure boiling flow. Each case has its experimental background with data support. The radial void fraction distribution, gas velocity, relative velocity and liquid temperature (only for high pressure boiling case) are chosen as the Quantities of Interest (QoIs) which are of key interests for two-phase flow simulation and boiling crisis prediction. The interfacial force coefficient of each closure term is chosen as the input parameter. For the boiling case, the bubble diameter effect is also analyzed.
Three remarks are drawn from this work on SA. First, it demonstrates the feasibility of surrogate model in the VUQ work for models in MCFD solver. The computational cost can be significantly reduced by employing the surrogate model. Secondly, through the Morris Sensitivity measurements, the importance of interfacial forces on different QoIs and regions can be analyzed and ranked for the two cases. Such analysis is also helpful for further model parameter calibration. Last but not least, the limitation of current work and the desired future work are discussed.