A Validation and Uncertainty Quantification Framework for Eulerian-Eulerian Two-Fluid Model based Multiphase-CFD Solver. Part I: Methodology (1806.03373v2)

Abstract: In this paper, a validation and uncertainty quantification (VUQ) framework for the Eulerian-Eulerian two-fluid-model based multiphase-computational fluid dynamics solver (MCFD) is formulated. The framework aims to answer the question: how to evaluate if a MCFD solver adequately represents the underlying physics of a multiphase system of interest? The proposed framework is based on total data-model integration (TDMI) approach that uses Bayesian method to inversely quantify the uncertainty of the solver predictions with the support of multiple experimental datasets. The framework consists of six steps with state-of-the-art statistical methods, including: 1). Solver evaluation and data collection; 2). Surrogate model construction; 3). Sensitivity Analysis; 4). Parameter selection; 5). Uncertainty quantification with Bayesian inference; and 6). Validation metrics calculation. Those steps are formulated in a modular manner and using non-intrusive methods. Such features ensure the applicability of the flexible framework to different scenarios and modeling of multiphase flow and boiling heat transfer, as well as the extensibility of the framework to support VUQ of different MCFD solvers.