Partially-averaged Navier-Stokes (PANS) Method for Turbulence Simulations: Near-wall Modeling and Smooth-surface Separation Computations

TL;DR

This paper evaluates the Partially-averaged Navier-Stokes (PANS) method for turbulence simulation, demonstrating improved accuracy over RANS and near LES quality at lower computational costs, especially for complex separation flows.

Contribution

The study develops a mathematically consistent G2-PANS turbulence model with variable resolution, enabling accurate flow predictions with reduced computational effort.

Findings

G2-PANS accurately predicts smooth surface separation flows.

G1-PANS and G2-PANS outperform RANS in critical benchmark flows.

Reduced computational cost compared to LES while maintaining high accuracy.

Abstract

The goal of this dissertation is to investigate the PANS model capabilities in providing significant improvement over RANS predictions at slightly higher computational expense and producing LES quality results at significantly lower computational cost. The objectives of this study are: (i) investigate the model fidelity at a fixed level of scale resolution (Generation1-PANS/G1-PANS) for smooth surface separation, (ii) Derive the PANS closure model in regions of resolution variation (Generation2-PANS/G2-PANS), and (iii) Validate G2-PANS model for attached and separated flows. The separated flows considered in this study have been designated as critical benchmark flows by NASA CFD study group. The key contributions of this dissertation are summarized as follows. The turbulence closure model of varying resolution, G2-PANS, is developed by deriving mathematically-consistent commutation residues and using energy conservation principles. The log-layer recovery and accurate computation of Reynolds stress anisotropy is accomplished by transitioning from steady RANS to scaled resolved simulations using the G2-PANS model. Finally, several smooth-separation flows on the NASA turbulence website have been computed with high degree of accuracy at a significantly reduced computational effort over LES using the G1-PANS and G2-PANS models.