Hostname: page-component-7b7567c8c-tnkh4 Total loading time: 0 Render date: 2026-11-16T03:36:20.027Z Has data issue: false hasContentIssue false

Flow states and transitions in flows past arrays of tandem cylinders

Published online by Cambridge University Press:  15 January 2021

Negar Hosseini*
Affiliation:
Department of Mechanical and Product Design Engineering, Swinburne University of Technology, John Street, Hawthorn, VIC 3122, Australia
Martin D. Griffith
Affiliation:
Department of Mechanical and Product Design Engineering, Swinburne University of Technology, John Street, Hawthorn, VIC 3122, Australia
Justin S. Leontini
Affiliation:
Department of Mechanical and Product Design Engineering, Swinburne University of Technology, John Street, Hawthorn, VIC 3122, Australia
*
†Email address for correspondence: negar.mhoseini@gmail.com

Abstract

Direct numerical simulations at $Re=200$ have been conducted of the flow past rows of tandem cylinders. It is shown that when the pitch between the two upstream cylinders is large, the wake downstream is characterised by a two-row vortex structure. Placing a third body on the wake centreline in the majority of this two-row structure has basically no impact both upstream and downstream – the third body is cloaked. However, a region is identified where the placement of a body suppresses vortex shedding from the first cylinder and the two-row structure is destroyed, globally broadcasting the presence of the third body. The effect is shown to occur for different third-body shapes. To understand the existence of this broadcasting region, local instability analysis is conducted which shows the majority of the two-row structure to be convectively unstable, with only a small region adjacent to the rear of the second cylinder that is absolutely unstable. This suggests only bodies placed close to the second body will trigger the global change, and this is supported by a global sensitivity analysis and observation from the simulations. However, neither the local analysis nor the global sensitivity analysis explains the presence of a lower limit for the third-body position that will trigger a global change. However the simulation results clearly show that a third body placed very close to the second body does not trigger this change.

Information

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

Figure 1.Validation of the calculation of the sensitivity field of the leading global mode on the steady flow past a cylinder at $Re = 50$. Contours of (a) direct mode magnitude, (b) adjoint mode magnitude and (c) the sensitivity field $\lambda$. Dark/light colours represent low/high values. The comparison with the result of the same calculation in Giannetti & Luchini (2007) is excellent.

Figure 1

Table 1.Eigenvalue data for the validation calculation of the sensitivity field for the global mode in the steady flow past a cylinder at $Re = 50$. The multiplier $\mu _{T=1}$ is the ratio of the mode from one sample to the next where samples are separated by a period $T = 1$, generated by evolving the governing equations for the direct/adjoint mode forwards/backwards in time by an increment $T$. The frequency $f$ is that associated with the oscillatory mode and $f_{GL}$ is the frequency measured by Giannetti & Luchini (2007).

Figure 2

Figure 2.Schematic description of the three parts of the present study.

Figure 3

Figure 3.(a) Instantaneous flow and (b) mean flow of the two-cylinder system for $p=1.5, 3.0, 4.0$ and $5.0$. Red/blue contours mark positive/negative vorticity in the range $\varOmega D/U = \pm 1$. Solid lines on the mean flow for $p = 5$ mark separating streamlines.

Figure 4

Figure 4.Calculated real ($\bullet$) and imaginary ($\blacktriangle$) frequencies from local instability analysis for the two-cylinder system with $p=5.0$ as a function of the distance from the centre of the second cylinder ($d$). The value ${L_{{A}}}$ marks the distance below which the flow is absolutely unstable.

Figure 5

Figure 5.Instantaneous flow visualisations for three-body systems. The first two bodies are cylinders separated by $p=5.0$ while a third body (cylinder or plate) is placed different distances ${p_2}$ from the second cylinder. All images are at maximum lift on the second cylinder. Dashed lines mark the end of the absolutely unstable region ${L_{{A}}}$ in the two-cylinder set-up.

Figure 6

Figure 6.Instantaneous flow visualisations for three-body systems in the broadcasting, transition and cloaking states for $p=5.0$ and ${p_2} =2.0$, $2.1$ and $2.2$, respectively. All images are shown at an instant of maximum lift on the second cylinder.

Figure 7

Figure 7.A plot of $\overline {C_D}$ as a function of ${p_2}$ for a three-cylinder system with consistent $p=5.0$ and different ${p_2}$. Vertical solid lines mark locations of ${L_{{A}}}$ calculated from the local stability analysis and of ${L_{{R}}}$ calculated from the mean-flow streamlines.

Figure 8

Figure 8.Mean recirculation length ${L_{{R}}}$ ($\bullet$) and the length of the locally absolutely unstable region ${L_{{A}}}$ ($\triangle$) for two-cylinder system as a function of $p$ using the local stability analysis. The ${L_{{A}}}$ data evaluated by simulation are presented with $\times$ and $+$ for upper and lower boundaries, respectively. The dashed lines connecting these data are cubics of best fit in the form of $L_{A_{max}} = -0.694p^3 + 11.071p^2 - 58.829p+ 105.826$ and $L_{A_{min}} = 2.083p^3 - 36.25p^2 + 210.666p - 407.45$.

Figure 9

Figure 9.Instantaneous flow visualisation of the flow before, across and after the broadcasting region for $p=5.2$ and varying ${p_2}=1.2$, 1.3, 1.7, 2.1, $2.2$ and $2.3$. All the visualisations are presented at the instant of the maximum lift on the second body.

Figure 10

Figure 10.The time histories of (a,c,e,g) $C_D$ and (b,d,f,h) $C_L$ on the first cylinder for three-cylinder system with $p_1=5.2$ and (a,b) $p_2=1.2$, (c,d) $p_2=1.3$, (e,f) $p_2=1.4$ and (g,h) $p_2=1.5$. Note the drop in the mean drag and lift fluctuation between $p_2 = 1.2$ and $p_2 = 1.3$ as the lower bound is crossed – the range on the plots has been adjusted to show any variation beyond the lower bound.

Figure 11

Figure 11.Instantaneous flow visualisation of the flow before, across and after the broadcasting region for $p=5.8$ and varying ${p_2}=1.9$, $2.0$ and $2.1$. All the visualisations are presented at the instant of the maximum lift on the second body.

Figure 12

Figure 12.The contours of mean drag coefficient $\overline {C_D}$ as a function of $p$ and ${p_2}$ for the first cylinder (left), second cylinder (centre) and third cylinder (right). All the contours are produced using a common colour map. The dashed lines are the upper and lower boundaries obtained from figure 8.

Figure 13

Figure 13.The contours of maximum lift coefficient $C_{L_{{max}}}$ as a function of $p$ and ${p_2}$ for the first cylinder (left), second cylinder (centre) and third cylinder (right). All the contours are produced using a common colour map. The dashed lines are the upper and lower boundaries obtained from figure 8.

Figure 14

Figure 14.The contours of primary frequency $f$ as a function of $p$ and ${p_2}$ for the first cylinder (left), second cylinder (centre) and third cylinder (right). All the contours are produced using a common colour map. The dashed lines are the upper and lower boundaries obtained from figure 8.

Figure 15

Figure 15.Left: vorticity contours of the mean flow where red/blue contours mark positive/negative vorticity overlaid with separating streamlines. Right: contours of the sensitivity field for the leading global mode on this mean, where dark/light colours mark low/high values. Four cases for the two-cylinder system are shown with (a) $p=4.8$, (b) $p=5.2$, (c) $p=5.6$ and (d) $p=6.0$. The dashed circles represent the location where the most upstream point of the circle coincides with the location of the lower boundary and the solid circles have their most upstream point coinciding with the location of the upper boundary as shown in figure 8.

Figure 16

Figure 16.The mean flow (top) and sensitivity field (bottom) for three-cylinder system with $p_1=5.2$ and (a) $p_2=1.2$ and (b) $p_2=1.3$.

Figure 17

Figure 17.The vorticity of the real component of the global mode, frequency measured from the simulations and the frequency of the leading forward (global) mode for (a) the two-cylinder system with $p=5.2$, (b) the three-cylinder system with $p=5.2$ and $p_2=1.2$ and (c) the three-cylinder system with $p=5.2$ and $p_2=1.3$.