Hostname: page-component-7b7567c8c-jswmv Total loading time: 0 Render date: 2026-11-15T21:53:24.658Z Has data issue: false hasContentIssue false

Formation mechanism of a secondary vortex street in a cylinder wake

Published online by Cambridge University Press:  31 March 2021

Hongyi Jiang*
Affiliation:
School of Engineering, The University of Western Australia, 35 Stirling Highway, Perth, WA 6009, Australia
*
†Email address for correspondence: hongyi.jiang@uwa.edu.au

Abstract

This study identifies and explains the formation mechanisms (FMs) of the secondary vortex street in the far wake of a circular cylinder. Direct numerical simulation and transient growth analysis are performed for a Reynolds number (Re) in the range of 100–200. Unlike most of the earlier studies, which have attributed the formation of the secondary vortices to either the hydrodynamic instability of the mean flow (FM1) or a merging of the two-layered vortices (FM2), the present study demonstrates that both FM1 and FM2 are at play. Specifically, FM1 is observed for Re = 150–160, whereas FM2 is observed for Re = 200. Interestingly, both FM1 and FM2 exist (in an alternate manner) for Re = 170 and 180. The underlying physical mechanism controlling the manifestation of either FM1 or FM2 is explored. In short, secondary vortices that emerge after and before the annihilation of the two-layered vortices are formed from FM1 and FM2, respectively. The variation in the streamwise locations for the emergence of the secondary vortices with Re is quantified and physically explained. Fundamentally, the emergence of the secondary vortices through either FM1 or FM2 is induced by the convective instability of the shear layers in the intermediate wake, which is further induced by an obvious increase in the shear rate of the shear layers as the wake gradually transitions from a primary vortex street to a two-layered vortex street. The general conclusions drawn from a circular cylinder are expected to be applicable to other bluff bodies.

Information

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

Figure 1.Macro-element mesh near the cylinder.

Figure 1

Table 1.Mesh dependence check of key results for Re = 200.

Figure 2

Figure 2.Time-averaged streamwise velocity profile sampled along the wake centreline for Re = 150.

Figure 3

Figure 3.Streamwise variation of the amplitude of uy at fK for Re = 150. In the curve fitting, R2 is the coefficient of determination.

Figure 4

Figure 4.A short sequence of instantaneous vorticity fields at Re = 200. The vorticity contours are shown from −0.4 to 0.4. The vorticity ranges of −0.3 to −0.2 as well as 0.2 to 0.3 are shown with a small interval of 0.005 to highlight the vortex merging process. The same vortices that appear in different snapshots are labelled with the same numbers.

Figure 5

Table 2.Numbers of vortices observed in the primary and secondary vortex streets for Re = 200 over 70 primary vortex shedding periods.

Figure 6

Figure 5.Time evolution of the secondary vortices at Re = 200. The same vortices that appear in different snapshots are labelled with the same numbers. The labels above and below the vortices correspond to the negative and positive vortices, respectively. The symbols + and * denote merging of the two-layered and secondary vortices, respectively.

Figure 7

Figure 6.Instantaneous vorticity field for Re = 160 at a typical time instant of t* = 3000. The vorticity field is shown for (a) x/D = 120–140 with |ω| = 0.11–0.14, (b) x/D = 175–205 with |ω| = 0.07–0.10 and (c) x/D = 350–390 with |ω| = 0.03–0.045.

Figure 8

Figure 7.Vortex patterns for Re = 160 at the time instant t* = 3000: (a) spatial distribution of the vortex centres and (b) evolution of the spatial scale of the vortices with distance downstream.

Figure 9

Figure 8.Time evolution of the secondary vortices at Re = 170. The same vortices that appear in different snapshots are labelled with the same numbers. The labels above and below the vortices correspond to the negative and positive vortices, respectively.

Figure 10

Figure 9.Ranges of x/D where the secondary vortices are formed from either FM1 or FM2. The entire range of x/D for the emergence of the secondary vortices is shaded. The numerical results of Re = 250 and 300 in the extended laminar regime are reproduced from Jiang & Cheng (2019).

Figure 11

Figure 10.Characteristics of the ux profiles for various Re values: (a) the ux profiles sampled along the wake centreline, (b) the ux profiles sampled at x/D = 40 and (c) the maximum shear rate of the ux profiles sampled at various x/D locations.

Figure 12

Figure 11.Optimal energy growth G(τ) as a function of the time interval τ for various Re values.

Figure 13

Figure 12.Optimal energy growth curve log10G(τ) − τ for Re = 180, together with the linear energy growths E(t*)/E(0) evolved from the optimal initial perturbations for various τ values.

Figure 14

Figure 13.Time evolution of the linear perturbation vorticity field developed from the optimal initial perturbation for Re = 180 and τ = 100. The linear perturbation vorticity fields are symmetric about the wake centreline. Positive and negative vorticity values are in red and blue, respectively. The range of vorticity in panel (a) for t* = 0–50 is 100 times smaller than the range in panel (b) for t* = 50–400. In panel (a), the vertical line at x/D = 10 represents the left end of the domain.

Figure 15

Figure 14.Linear energy growth E(t*)/E(0) evolved from the optimal initial perturbation for τ = 200.

Figure 16

Figure 15.Movement of the centroidal location of the perturbation energy over time for τ = 200 and various Re: (a) the relationship between xc/D and t* and (b) the advection velocity of xc/D as a function of xc/D.

Figure 17

Figure 16.Linear energy growth E(t*)/E(0) evolved from the optimal initial perturbation for τ = 200, with the horizontal axis being transformed from t* to xc/D.

Figure 18

Table 3.Optimal energy growths computed with different base flows.

Figure 19

Figure 17.Optimal energy growths G(τ) for Re = 200 predicted based on the full domain and half domain.