Error propagation and conditioning analysis of DNS data of turbulent viscoelastic channel flows

In the context of drag reduction phenomenon, direct numerical simulation (DNS) of turbulent viscoelastic flows is a very useful tool able to provide information that is not possible with experiments and that can be used for modeling purposes. A rare and challenging task is to provide measures for st...

Full description

Saved in:
Bibliographic Details
Published in:Journal of non-Newtonian fluid mechanics 2021-10, Vol.296, p.104632, Article 104632
Main Authors: Martins, Ramon Silva, Andrade, João Rodrigo, Brener, Bernardo Pereira, Thompson, Roney Leon, Sampaio, Luiz Eduardo Bittencourt, Mompean, Gilmar
Format: Article
Language:English
Subjects:
Channel flow
Conditioning
Conditioning analysis
Convergence
Direct numerical simulation
Drag reduction
Error analysis
Fluid dynamics
Fluid mechanics
Mathematical models
Mean
Mechanics
Physics
Plane channel flow
Propagation velocity
Reynolds stress
Statistical errors
Tensors
Viscoelastic turbulence
Viscoelasticity
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In the context of drag reduction phenomenon, direct numerical simulation (DNS) of turbulent viscoelastic flows is a very useful tool able to provide information that is not possible with experiments and that can be used for modeling purposes. A rare and challenging task is to provide measures for statistical errors that are present in DNS data. In the present communication, we employ a method to quantify the statistical convergence of the DNS turbulent viscoelastic channel flows from the unbalance on the mean momentum equation. In addition, we conduct a conditioning analysis of the resultant numerical linear system of this problem. We compare results from different research groups. We found that DNS databases of viscoelastic flows can induce high error propagation in the mean velocity. As a consequence, larger sampling times are needed to achieve a statistically converged Reynolds stress tensor. The conditioning analysis has shown that elasticity enhances conditioning. •We discuss the DNS importance for turbulence modeling.•Uncertainty and statistical error are analyzed in the DNS context.•Equations for mean viscoelastic turbulence (VT) are ill-conditioned.•Error propagation in the mean velocity field related to VT is high.•Statistical quantities, like Reynolds stress, in VT need to be more converged.
ISSN:0377-0257
1873-2631
DOI:10.1016/j.jnnfm.2021.104632