Modeling gas–particle interaction in an Eulerian–Lagrangian frame involves many approximate models. Hence, understanding the model form error is essential to improve the prediction capability. Often, a model by itself consists of several submodels where each submodel is subject to potential error. The potential errors can superimpose or compensate for each other over time, which makes it difficult to find a possible source of model form error. This study aims to quantify, rank, and isolate the contribution of error in each submodel to the error in the model prediction. The concept of global sensitivity analysis is extended by using it as a tool to isolate the most influential potential error in drag force submodels of a transient multiphase dispersed flow on particle cloud position. The possible errors of the submodels are modeled as uniformly distributed epistemic uncertainty, and their contributions to the quantity of interest are calculated in terms of sensitivity indices. We found that the most influential potential submodel error depends on time and the characteristic of the particle cloud.
Prioritizing Possible Force Models Error in Multiphase Flow Using Global Sensitivity Analysis
AIAA Journal ; 59 , 5 ; 1749-1759
2021-03-10
11 pages
Aufsatz (Zeitschrift)
Elektronische Ressource
Englisch