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

TL;DR

This paper introduces a comprehensive validation and uncertainty quantification framework for multiphase CFD solvers based on the Eulerian-Eulerian two-fluid model, utilizing Bayesian methods and data integration to assess physics representation.

Contribution

It presents a modular, non-intrusive VUQ framework that can be applied to various multiphase flow models and enhances the reliability of CFD predictions.

Findings

Framework effectively quantifies solver uncertainty.

Applicable to different multiphase flow scenarios.

Supports validation with multiple experimental datasets.

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.