Hostname: page-component-7b7567c8c-wxvwp Total loading time: 0 Render date: 2026-11-16T21:16:25.612Z Has data issue: false hasContentIssue false

Adaptive detached eddy simulation of transition under the influence of free-stream turbulence and pressure gradient

Published online by Cambridge University Press:  29 March 2021

Zifei Yin*
Affiliation:
School of Aeronautics and Astronautics, Shanghai Jiao Tong University, Shanghai 200240, PR China
Xuan Ge
Affiliation:
Convergent Science, Inc., Madison, WI 53718, USA
Paul Durbin
Affiliation:
Department of Aerospace Engineering, Iowa State University, Ames, IA 50011, USA
*
†Email address for correspondence: yinzifei@sjtu.edu.cn

Abstract

The basic question of whether a hybrid method for turbulence simulation can plausibly capture laminar-to-turbulent transition is addressed. The $\ell ^2$–$\omega$ adaptive detached eddy simulation model (Yin & Durbin, Intl J. Heat Fluid Flow, vol. 62, 2016, pp. 499–509) does so. It dynamically adjusts a model constant, based on local mesh resolution and instantaneous flow features. In a laminar flow, the adaptive procedure returns zero subgrid viscosity, and large-scale low-frequency perturbations are resolved on the grid. However, rather than fully simulating transition, the hybrid model switches on at transition; small-scale turbulence is not resolved. It is found that the correct transitional behaviour is captured because the adaptive formulation responds to the initiation of small-scale components in the field of velocity gradient. The current work addresses flat-plate transition under the influence of free-stream turbulence and pressure gradient, encompassing bypass and separation-induced transition. First, the transition prediction mechanism of the adaptive model is explained. Then, the ability to predict transition statistically is evaluated, along with sensitivity studies of boundary conditions and mesh resolution.

Information

Type
JFM Papers
Copyright
© The Author(s), 2021. Published by Cambridge University Press
Figure 0

Figure 1.Total (modelled $+$ resolved) and resolved turbulent intensity decay.

Figure 1

Figure 2.Contours of instantaneous flow quantities at $Re_y=720$ for mesh 3-d-i.

Figure 2

Figure 3.Velocity perturbation vector (x, y components) coloured by $f_d$, superimposed on Reynolds shear stress ($-R_{xy}/U_\infty ^2$) contour. All quantities are instantaneous.

Figure 3

Figure 4.Instantaneous fields along sample line at $Re_y=720$ on figure 3.

Figure 4

Figure 5.Computational domain and mesh 0-a-i for T3A.

Figure 5

Table 1.Computational grids for the T3A case.

Figure 6

Figure 6.Normalized instantaneous streamwise velocity fluctuation ($u'/U_\infty$) for the T3A case on $Re_y=180$ ($y=0.005$ m) plane. From top to bottom are meshes 1-b-i, 1-d-i and 3-d-i.

Figure 7

Figure 7.Contours of instantaneous flow quantities at $Re_y=720$ for mesh 1-b-i.

Figure 8

Figure 8.Streamwise evolution of resolved and modelled turbulent kinetic energy at $Re_y=720$ for meshes 1-b-i and 3-d-i, averaged in time and spanwise direction.

Figure 9

Figure 9.T3A results. (a) Time-averaged skin friction coefficient along the bottom wall. (b) Instantaneous skin friction coefficient along the bottom wall. (c) Instantaneous spanwise velocity component at $Re_y\approx 10$. (d) Mesh resolution in plus units; red is streamwise and black spanwise.

Figure 10

Table 2.Summary of inlet conditions for the T3 flat-plate cases.

Figure 11

Table 3.Computational grids for the T3B case.

Figure 12

Figure 10.T3B results. (a) Time-averaged and (b) instantaneous skin friction coefficient ($C_f$) along the bottom wall.

Figure 13

Table 4.Computational grids for the T3A- case.

Figure 14

Figure 11.T3A- results. (a) Time-averaged and (b) instantaneous skin friction coefficient ($C_f$) along the bottom wall.

Figure 15

Figure 12.Computational domain and mesh 1 for T3C1; every other grid point is shown.

Figure 16

Table 5.Computational grids for T3C cases.

Figure 17

Figure 13.Time-averaged surface skin friction coefficient for T3C series. Circles represent experimental data.

Figure 18

Table 6.Computational grids for the flat-plate separation case. Meshes 1, 2, 3 and 4 share the same stretching ratio. Meshes 1-a and 2-a have more clustering placed near the leading edge.

Figure 19

Figure 14.Mesh 1 for the separated flat-plate case, showing every other grid line for the coarsest mesh.

Figure 20

Figure 15.Resolved turbulent intensity profiles along the streamwise direction at $y/L = 0.065$ on mesh 4: (a) at various targeted inflow turbulent intensities, compared with LES; and (b) at varying mesh resolution for the $Tu=1\,\%$ case using consistent boundary condition.

Figure 21

Figure 16.$Tu = 0\,\%$ results: (a) skin friction coefficient on the bottom wall; and (b) separation bubble.

Figure 22

Figure 17.$Tu = 1\,\%$ results: (a) skin friction coefficient on the bottom wall; and (b) separation bubble.

Figure 23

Figure 18.$Tu = 2\,\%$ results: (a) skin friction coefficient on the bottom wall; and (b) separation bubble.

Figure 24

Figure 19.Isosurfaces of $u'=\pm 0.2U_\infty$ on meshes 1-a (lower) and 4 (upper) for (a) $Tu = 1\,\%$ and (b) $Tu = 2\,\%$.

Figure 25

Figure 20.Sensitivity with inflow condition of skin friction: (a) total (resolved $+$ modelled) TKE; and (b) eddy frequency.

Figure 26

Figure 21.Skin friction predicted on mesh 3-d-i of T3A, with dynamic procedure or just forced $C_{lim}$.

Figure 27

Figure 22.Normalized modelled TKE ($k/U_\infty ^2$) at the leading edge on meshes 0-a-i and 1-b-i for T3A.