Hostname: page-component-7b7567c8c-tnkh4 Total loading time: 0 Render date: 2026-11-16T05:52:57.356Z Has data issue: false hasContentIssue false

Wakes of wall-bounded turbulent flows past patches of circular cylinders

Published online by Cambridge University Press:  14 April 2020

C. Nicolai*
Affiliation:
Aerodynamics and Flight Mechanics Group, University of Southampton, University Road, SouthamptonSO17 1BJ, UK
S. Taddei
Affiliation:
Aerodynamics and Flight Mechanics Group, University of Southampton, University Road, SouthamptonSO17 1BJ, UK
C. Manes
Affiliation:
Department of Environment, Land and Infrastructure Engineering, Politecnico di Torino, Corso Duca degli Abruzzi, 24, 10129, Torino, Italy
B. Ganapathisubramani
Affiliation:
Aerodynamics and Flight Mechanics Group, University of Southampton, University Road, SouthamptonSO17 1BJ, UK
*
†Email address for correspondence: C.Nicolai@soton.ac.uk

Abstract

The properties of the wake generated by a porous body fully immersed in a turbulent boundary layer are experimentally assessed. The body consists of an array of cylinders, with diameter $d$, covering a circular patch of diameter $D$. For fixed $d$ and $D$, by increasing the number of cylinders, $N_{c}$, within the patch, the wake properties are systematically tested under different levels of density ($\unicode[STIX]{x1D719}$ $=$ covered planar area per total surface) and compared to the flow past a solid body of equivalent diameter and height ($H$). Some insights on the complex flow developing in the wake are captured: $\unicode[STIX]{x1D719}$ varying in the range 2 %–24 % results in the flow meandering among the cylinders and bleeding from the top, the sides and the trailing edge of the patch. The interplay between trailing edge and top bleeding prevents wake entrainment, locking the wake longitudinal extent to 5–7 patch diameters, regardless of the density level. Due to the finite body vertical extent, a third shear layer develops from the top of the patch. The interaction between the top shear layer and the lateral ones leads to a mutual alteration, namely a nonlinear growth not captured by the classical mixing layer theory. Nevertheless, on the horizontal plane at the patch mid-height, the mean flow recovers, exhibiting a self-similar decay. Surprisingly, the recovery is well described by the classical planar wake theory and the characteristic scales, namely the maximum velocity deficit and the wake half-width, evolve linearly as proposed by Wygnanski et al. (J. Fluid Mech., vol. 168, 1986, pp. 31–71).

Information

Type
JFM Papers
Copyright
© The Author(s), 2020. Published by Cambridge University Press
Figure 0

Figure 1.(a) Trailing edge and lateral bleeding formation. (b) Vertical bleeding formation.

Figure 1

Table 1.(a) Main impinging flow parameters evaluated with hot-wire anemometry at $x=0$: turbulent boundary layer thickness at 99 % of the free-stream velocity, friction velocity, equivalent roughness length, displacement thickness, momentum thickness and Reynolds number. (b) Body geometrical parameters: $N_{c}$ is the cylinder number in the porous body, $\unicode[STIX]{x1D719}$ is the corresponding density. In addition, the body height and overall diameter are $H=D=100$ mm and the inner cylinder diameter is $d=0.05D$. Here $aD$ is the non-dimensional frontal area per unit volume.

Figure 2

Figure 2.Schematic of the experiment and coordinate system: the coloured planes represent the field of view of the planar PIV studies.

Figure 3

Figure 3.(a) Mean velocity profile of the incoming boundary layer (symbols): $u_{bl}$ is the time-averaged velocity, $D$ is the patch diameter, $U_{\infty }$ is the free-stream velocity, $y$ is the vertical coordinate, $u_{\unicode[STIX]{x1D70F}}$ is the friction velocity, $y_{0}$ is the equivalent roughness length. The black solid line is the log law of the wall, $u_{bl}/u_{\unicode[STIX]{x1D70F}}=k^{-1}\ln (y/y_{0})$, with von Kármán constant, $k=0.41$. (b) Vertical profile of the turbulence intensity: $u_{bl}^{\prime }$ is the standard deviation of the longitudinal velocity component.

Figure 4

Figure 4.(a) Drag coefficient, $C_{D}=F/(0.5\unicode[STIX]{x1D70C}U_{\infty }^{2}DH)$, against $\unicode[STIX]{x1D719}$ for submergence ratios $\unicode[STIX]{x1D6FF}/H=(3.58,4.75,7.14)$ according to legend. (b) Normalized drag coefficient $C_{D}(\unicode[STIX]{x1D719})/\max (C_{D}(\unicode[STIX]{x1D719}))$.

Figure 5

Figure 5.Normalized velocity deficit, $(u_{bl}-u)/U_{H/2}$, where $U_{H/2}$ represents the free-stream velocity at the patch mid-height. Panels (a–f) represent the deficit behaviour for $\unicode[STIX]{x1D719}=0.0175,0.05,0.0975,0.16,0.2375,1$, respectively.

Figure 6

Figure 6.Sketches of wake height and length in panels (a) and (b), respectively.

Figure 7

Figure 7.(a) Vertical bleeding at increasing density. (b) Horizontal bleeding at increasing density. (c) Vertical wake extent, $H_{W}$, as a function of $\unicode[STIX]{x1D719}$ (the crosses bounding the symbols refer to the sensitivity of the wake extent to the threshold level (=20 %) defining the wake edge). (d) Wake length, $L$, as detected on the horizontal plane at mid-height.

Figure 8

Figure 8.(a–e) Lines refer to the evolution of the velocity deficit, $\unicode[STIX]{x0394}u(y)$, at specific wall-normal locations, $y/H=[0.1;0.9]$, as identified by the colour code reported at the side of the bottom panel. Symbols highlight the maximum deficit and hence $L_{1}$.

Figure 9

Figure 9.Curves described by the maximum velocity deficit location $L_{1}$ in the wall-normal plane for the assessed densities.

Figure 10

Figure 10.(a) Map of the Reynolds shear stress, reported as $-u^{\prime }v^{\prime }/U_{H/2}^{2}$ for display purposes, for model $C39$, assuming this is representative of the general trend. (b) Symbols represent the velocity deficit evolution downstream the patch in sections $x/D\approx (0.5,0.75,1,1.25,1.5,1.75,2,2.25,2.5,2.75,3)$ (as marked by the grey lines on the top map), superimposed on the Reynolds shear stress profiles (lines) at the same locations.

Figure 11

Figure 11.(a) Schematic of the shear layer forming at the top of the body. (b) Typical velocity difference normalized by the convective velocity, $U_{c}=0.5(U_{2}+U_{1})$. (c) Shear layer magnitude, $\max (-u^{\prime }v^{\prime })/U_{2}^{2}$, as absolute maximum in the wall-normal plane. (d) Mixing layer edge $y_{1}$.

Figure 12

Figure 12.(a) Schematic of the shear layer forming at the sides of the body. (b) Normalized velocity difference at increasing $\unicode[STIX]{x1D719}$. (c) Mixing layer edge $z_{1}$ for the tested solidities.

Figure 13

Figure 13.Comparison between $L_{1}$, location of the maximum deficit, and $L_{1,exp}$, location where the horizontal shear layers merge (symbols); the line marks the $L_{1}$ for 2-D conditions, namely the 1 : 1 line agreement.

Figure 14

Figure 14.Velocity deficit evolution on the horizontal plane at the body mid-height; the increasing density corresponds to cases $C7,C20,C39,C64,C95,CS$.

Figure 15

Figure 15.Evolution of velocity deficit profiles at increasing density: (a) $C20$; (b) $C39$; (c) $C64$; (d) $C95$.

Figure 16

Figure 16.(a) Deficit profiles at $x/D=3$ for cases $C20$–$C95$ (markers) scaled up with the characteristic velocity and length scale, $\unicode[STIX]{x0394}u_{0}$ and $z_{1/2}$; solid lines represent the least squares fitting of data through $f=\text{e}^{-\unicode[STIX]{x1D6FC}(z/z_{1/2})^{2}}$. (b, top) The $\unicode[STIX]{x1D6FC}$ parameter in several downstream locations compared to the typical value for a 2-D wake $\unicode[STIX]{x1D6FC}\approx 0.693$. (b, bottom) Goodness of fit as the root mean square error (RMSE).

Figure 17

Figure 17.(a) Universal scaling law for the wake half-width from (3.5). (b) Scaling for maximum velocity deficit as from (3.6). $x_{0}(\unicode[STIX]{x1D719})=(3.9,3.9,3.1,2.7)D$ is the virtual origin.

Figure 18

Figure 18.Value of $C_{D}=F/0.5\unicode[STIX]{x1D70C}D^{2}U_{\infty }^{2}$, as a function of $\unicode[STIX]{x1D719}$, as measured in Taddei et al. (2016).