You are on page 1of 21

RESEARCH ARTICLE | AUGUST 17 2021

Coupling response of flow-induced oscillating cylinder with


a pair of flow-induced rotating impellers 
Hongjun Zhu (朱红钧)  ; Xin Chu (褚鑫); Zhiyin Yan (颜知音); Yun Gao (高云)

Physics of Fluids 33, 083608 (2021)


https://doi.org/10.1063/5.0063029

CrossMark

 
View Export
Online Citation

02 December 2023 10:08:45


Physics of Fluids ARTICLE scitation.org/journal/phf

Coupling response of flow-induced oscillating


cylinder with a pair of flow-induced rotating
impellers
Cite as: Phys. Fluids 33, 083608 (2021); doi: 10.1063/5.0063029
Submitted: 11 July 2021 . Accepted: 29 July 2021 .
Published Online: 17 August 2021

Hongjun Zhu (朱红钧),1,a) Xin Chu (褚鑫),1 Zhiyin Yan (颜知音),2 and Yun Gao (高云)3

AFFILIATIONS
1
State Key Laboratory of Oil and Gas Reservoir Geology and Exploitation, Southwest Petroleum University, 8 Xindu Road, Chengdu,
Sichuan 610500, China
2
Natural Gas Purification Plant, PetroChina Southwest Oil and Gasfield Company, 542 Longshan Road, Chongqing 400021, China
3
School of Ocean Engineering, Harbin Institute of Technology, 2 Wenhua West Road, Weihai 264209, China

a)
Author to whom correspondence should be addressed: zhuhj@swpu.edu.cn

ABSTRACT

02 December 2023 10:08:45


An innovative device transforming the active control of rotating rods to passive control with a pair of impellers is proposed and numerically
examined in this paper. The coupling response of a vortex-induced vibrating (VIV) circular cylinder symmetrically equipped with two impel-
lers that are free to rotate is analyzed based on the results of computations that carried out for a reduced velocity range of Ur ¼ 2–14 at a low
Reynolds number of 150. In comparison with the bare cylinder, both the in-line and cross-flow responses are significantly augmented in the
VIV initial branch with the introduction of a pair of passively rotating impellers, which is mainly attributed to the unstable rotation response
in both direction and speed and the wake adjustment including the reduction in vortex formation length and broadening of flow wake. In
the VIV lower branch, although the response amplitude is close to that of a bare cylinder, the strong interaction between two directional
responses occurs with the same dominant frequency locking on the natural one. Nevertheless, the coexistence of multiple vibration frequen-
cies leads to irregular oscillation trajectories and irregular vortex shedding. Moreover, the secondary vortex street is observed in the whole Ur
range, but the number of merged vortices for the formation of secondary vortex street varies with Ur, depending on the response amplitude
and the interaction between the shear layers of the main cylinder and impellers. In terms of time-averaged rotation, the symmetrical inward
counter-rotating pattern is achieved despite the intermittent alteration of rotation direction. Furthermore, the vibration–rotation coupling is
demonstrated from the variation of time-averaged rotation speed that closely follows the variation of vibration amplitude against Ur.
Published under an exclusive license by AIP Publishing. https://doi.org/10.1063/5.0063029

I. INTRODUCTION layers or modifying the recirculation region and vortex formation.


Vortex-induced vibration (VIV) is a benchmark issue extensively Nevertheless, these wake controllers are sensitive to the flow orienta-
encountered in engineering applications as a result of the alternating tion, i.e., they successfully eliminate vibration in a narrow operation
vortex shedding generated in the wake of a bluff body and the induced range while they present the negative effect out of the range. In con-
fluctuating fluid forces.1 Particularly, fatigue, noise, and structural col- trast, active control approaches consume power to effectively and
lapse probably occur owing to the presence of resonance when the timely adjust the flow separation and vortex shedding. Common
shedding frequency coincides with the structural natural frequency. A examples of active control include suction and blowing, structural
series of active and passive control manners has been proposed to oscillation, acoustic excitation, as well as rotation of control rods.8
reduce the fluid forces and to attenuate the VIV in past decades.2 The usage of a pair of counter-rotating smaller control cylinders
Passive control is mainly achieved by the installation of an accessory (CCs) is one of the most effective active control methods for altering
device or geometry modification and is preferred in practice due to the the flow pattern around the main cylinder (MC) placed upstream and
nonexpenditure of energy input.3,4 Splitter plates,5 fairings,6 and con- reducing the oscillating forces exerted on its surface.9 This control
trol rods7 are typical passive wake controllers which suppress the method can be traced back to a century ago when Ludwig Prandtl
structural vibration through altering the confluence point of shear illustrated his ship of zero viscous resistance by using rotating rods.10

Phys. Fluids 33, 083608 (2021); doi: 10.1063/5.0063029 33, 083608-1


Published under an exclusive license by AIP Publishing
Physics of Fluids ARTICLE scitation.org/journal/phf

Through a series of studies on the control effectiveness of rotating rods, arranged behind a circular cylinder with the location angle of 120
Modi et al.11–13 reported that the pair of CCs rotating at sufficiently that is measured from the front stagnation point of the main cylinder
high speed can delay the separation of boundary layers and hence nar- (MC), because boundary layers typically separate around 120 in a
row the wake flow. Mittal14,15 pointed out some momentum is injected wide Re range and hence 120 is a recommended location for control
into the already existing boundary layer by the CCs, contributing to the rods.17,18,20 The diameter ratio of the control rod to the main cylinder
delay of the adverse pressure gradient generated by the MC and the vor- is d/D ¼ 0.1, where d is the diameter of the control rod with reference
tex shedding. This active control method is therefore termed the moving to the reported studies.14–16 Each control rod is evenly attached with
surface boundary-layer control (MSBC). It is demonstrated both three curved blades along its circumference, yielding a typical vertical-
numerically and experimentally that the control effect depends on the axis 3-vane impeller that is widely used in energy harvesting. Impeller-
Reynolds number (¼ uinD/t, where uin is the free-stream velocity, D is 1 and impeller-2 refer to the upper and lower impellers, respectively.
the cylinder diameter and t is the fluid kinematic viscosity) as well as Each blade has a semicircle profile with internal radius of R ¼ 0.1D
the combination of structural parameters, including the diameter ratio and thickness of d ¼ 0.01D, referring to the conventional size of a 3-
of CCs to MC, the gap between CCs and MC, the location of CCs and vane impeller.21 Therefore, the sweeping diameter of each impeller is
the rotating speed.16–20 A set of complicated mechanical devices, such as ds ¼ 0.32D and the gap between the main cylinder surface and the
motor driver system, and specific feedback system of actuators and sen- sweeping boundary of an impeller is G ¼ 0.4D. The orientation of
sors is required to achieve an optimal control result in the cost of energy the concave surface with respect to the incoming flow is converse for
input. An innovative idea then arises as to whether the power input for the two impellers, in order to achieve the inward counter-rotating con-
driving the rotation of control rods can be extracted from the ambient figuration.14,15 The two impellers are freely rotating without consider-
flow instead of external energy consumption. ing the friction between the impeller and control rod.25–27 The circular
The representative energy harvesting devices are turbines or impel- cylinder with a pair of impellers is elastically mounted with linear
lers that can convert the kinetic energy of ambient flow to power, includ- spring and damper, treated as a mass-spring-damping system that can
ing the horizontal-axis and vertical-axis turbines.21 The horizontal-axis vibrate in both in-line (x) and cross-flow (y) directions. The two
turbines are widely used in the onshore utility-scale wind power genera- impellers vibrate synchronously with the circular cylinder as a system.
tion, requiring an active orientation control to keep turbines always facing A low mass ratio of 1–10 is common in hydroelastic applications
the flow,22 while the vertical-axis turbines are independent of the flow where the fluid is water.28,29 In this work, the mass ratio m* of the sys-
direction and possess compacter size and lower noise emission in com- tem is thus set as 6.0 (i.e., the cylinder and impellers are made of the

02 December 2023 10:08:45


parison with the former.23 The performance of vertical-axis turbines has same material), while the structural damping is assumed to be zero to
been extensively studied in previous literature, including the typical 3- excite a high-amplitude oscillation. The two-degree-of-freedom VIV is
vane impeller,21 Bach-type turbine,24,25 and some creative ones such as examined in a reduced velocity range of Ur ¼ 2–14, where Ur ¼ uin/
dartlike overlay26 and pentagram impeller.27 It is demonstrated that the fnD and fn is the natural frequency of the structure.
energy efficiency is related to the shape and size of the impeller as well as For the current two-dimensional simulations at a low Re of 150, a
the flow velocity. Inspired by the rotation response of impellers immersed rectangular computational domain of 60 D  20D is employed with the
in flow, each smaller control rod enveloped by an impeller is proposed cylinder placed centrally in the transverse direction, as shown in Fig. 1(b).
with the intent of driving the control rod to rotate without extra energy Consequently, the ratio of the cylinder diameter to the width of the com-
support. Such an approach converts the VIV control manner from active putational domain, termed blocking ratio, is 5%. Additionally, the dis-
to passive but achieves the approximate effect as the active one. Thus, it tance between the inlet boundary and the cylinder center is 20 D. The
can be termed a quasi-active control method. Additionally, the impellers inlet and outlet boundaries are defined as the Dirichlet-type boundary
could be used to harvest energy simultaneously, having a great practical condition (u ¼ uin and v ¼ 0, where u and v are the velocity components
application foreground. Nevertheless, a new problem arises that whether along the x and y directions, respectively) and the Neumann-type bound-
there is an interaction between the vortex-induced vibration (VIV) of the ary condition (@u/@x ¼ 0 and @v/@x ¼ 0), respectively. The two lateral
controlled main cylinder and the flow-induced rotation (FIR) of the boundaries and the structure surface are imposed with the symmetry
impellers. The understanding of the vibration–rotation interaction and boundary condition and non-slip boundary condition, respectively.
the wake interference between the main cylinder and the pair of impellers The two-dimensional incompressible flow around the cylinder
is extremely limited as the associated literature involving both the VIV with a pair of impellers is predicted using the incompressible unsteady
and flow-induced rotation is scanty. Navier–Stokes (NS) equations, including the continuity and momen-
In this work, the coupling response of vortex-induced oscillating tum equations, expressed as
cylinder with a pair of flow-induced rotating impellers is numerically @u @v
investigated at a low Reynolds number of 150. The aim of this study is þ ¼ 0; (1)
@x @y
to evaluate the performance of such a quasi-active control device and !
to establish a comprehensive understanding of the VIV-FIR interac- @u @u @u 1 @p @2u @2u
tion and the associated wake flow interaction features as well as the þu þv ¼ þ þ ; (2)
@t @x @y q @x @x2 @y2
evolution of vortex street. !
@v @v @v 1 @p @2v @2v
II. COMPUTATIONAL MODEL AND METHOD þu þv ¼ þ þ ; (3)
@t @x @y q @y @x2 @y2
A. Problem description and governing equations
Figure 1 schematically depicts the geometrical configuration of where t is the flow time, p is the pressure, q and t are the fluid density
the physical problem. A pair of identical control rods is symmetrically and kinematic viscosity, respectively. The flow governing equations

Phys. Fluids 33, 083608 (2021); doi: 10.1063/5.0063029 33, 083608-2


Published under an exclusive license by AIP Publishing
Physics of Fluids ARTICLE scitation.org/journal/phf

02 December 2023 10:08:45


FIG. 1. Schematic of the controlled cylinder: (a) definition of flow over a cylinder with a pair of impellers; (b) computational domain and boundary conditions.

are solved by a blend of the finite-volume discretization (FVD) The sequential fluid-structure interaction iteration is employed in
method and the pressure-implicit with splitting of operators (PISO) present simulations until enough stable or periodic results are obtained
algorithm. In the iteration, a fourth-order cubic scheme and a for statistical analysis. In each time step, the flow governing equations
second-order linear scheme are adopted to discretize the convective are first solved to get the hydrodynamic forces (drag and lift forces)
term and the diffusion term, respectively, while a blended scheme via conducting an integration involving both the pressure and viscous
composed of the second-order Crank–Nicolson scheme and the stresses. The hydrodynamic forces are substituted into the structural
first-order Euler implicit scheme is employed to discretize the time motion equations to calculate the vibration displacement using the
derivative term.30,31 fourth-order Runge–Kutta method.32 The new position of the cylinder
The vortex-induced motion of the structure meets Newton’s sec- is located according to the displacement, and the computational mesh
ond law, expressed as is updated in light of the new position for the calculation of the next
time step.33 In the iteration, the normalized time step is set as 0.0025,
mX 00 þ 2mfx0 X 0 þ mx20 X ¼ fD ðtÞ; (4)
ensuring the Courant–Friedrichs–Lewy (CFL) number is less than 0.5,
00 0
mY þ 2mfx0 Y þ mx20 Y ¼ fL ðtÞ; (5) and the residual in the control volume below 10−5 is defined as the
convergent criteria for each parameter. The VIV of a bare circular cyl-
where m is the mass of the system, f is the damping ratio (f ¼ 0), x0
inder is also simulated as the baseline case for comparison.
is the natural circular frequency of the structure (x0 ¼ 2pfn), X, X0 ,
and X00 represent the in-line displacement, motion velocity, and accel-
eration of the controlled cylinder, respectively, while Y, Y0 , and Y00 B. Computational mesh
denote the same quantities associated with the cross-flow motion, fD(t) The computational domain is divided into three parts: a circle of
and fL(t) are the drag and lift forces exerted on the structure surface by diameter 0.46D enveloped each impeller adopted as an accompanying
the ambient flow. The initial displacements in both in-line and cross- moving zone, a square of 20D  20D around the circular cylinder
flow directions are set as zero. employed as a dynamic mesh zone, and the remainder of the

Phys. Fluids 33, 083608 (2021); doi: 10.1063/5.0063029 33, 083608-3


Published under an exclusive license by AIP Publishing
Physics of Fluids ARTICLE scitation.org/journal/phf

FIG. 2. Grid distribution (I: dynamic zone; II: static zone; III: accompany moving zone): (a) the whole grid system; (b) the grids near the impeller; and (c) the mesh in the

02 December 2023 10:08:45


accompany moving zone.

FIG. 3. Comparison of the grids around the upper impeller among the five different meshes those are denoted by G1, G2, G3, G4, and G5 with increasing the mesh resolution.

Phys. Fluids 33, 083608 (2021); doi: 10.1063/5.0063029 33, 083608-4


Published under an exclusive license by AIP Publishing
Physics of Fluids ARTICLE scitation.org/journal/phf

TABLE I. Grid resolution test for the vibration response of a circular cylinder with two free-to-rotate impellers at Ur ¼ 3.

Mesh Elements AX/D AY/D CDmean CLrms

G1 32 894 0.0088 0.418 1.152 3.141


G2 43 642 0.0094 (6.82 %) 0.425 (1.67 %) 1.156 (0.35 %) 3.218 (2.45 %)
G3 50 890 0.0098 (4.26 %) 0.429 (0.94 %) 1.154 (0.17 %) 3.271 (1.65 %)
G4 66 512 0.010 (2.04 %) 0.432 (0.70 %) 1.152 (0.17 %) 3.302 (0.95 %)
G5 78 094 0.010 (0.00 %) 0.432 (0.00 %) 1.153 (0.09 %) 3.299 (0.09 %)

computational domain treated as a static mesh zone, as shown in C. Method validation


Fig. 2. The triangular-unstructured mesh is employed to tessellate the The employed computational method is validated by comparing
two accompanying moving zones and the dynamic mesh zone, exclud- the calculated VIV response of a bare circular cylinder at Re ¼ 150
ing the boundary-layer mesh around both cylinder and impellers with those reported in previous literature.34–36 The mass and damping
which is divided by quadrilateral structured elements. Three-layer ratios are m* ¼ 2 and f ¼ 0, respectively, in the validation case.
mesh is used to envelop each impeller with the size of the first layer of Figure 4 compares the normalized cross-flow amplitude (A*Y ¼ AY/D)
0.003D and the expansion ratio of 1.05, while five-layer mesh is
vs the reduced velocity. It is seen that the present results coincide well
adopted to envelop the circular cylinder with first layer height of
with the literature, presenting the three typical vibration branches at
0.0077D and the same expansion ratio as the former, resulting in
low Re: the initial branch (IB), the lower branch (LB), and the
y+ ¼ 0.1. Each accompanying moving zone follows the vibrational and
desynchronization branch (DB).
rotational motion of the associated impeller so that the mesh in the
accompanying moving zones remains undeformed, contributing to III. RESULTS AND DISCUSSION
more accurate and effective calculation of the hydrodynamic forces
A. Hydrodynamic forces and flow-induced vibration
acting on the impellers’ surface. In contrast, the triangular-
unstructured mesh is deformed or/and remeshed in each time step The time histories of hydrodynamic coefficients, including CD

02 December 2023 10:08:45


using the spring-based smoothing or/and the local-remeshing and CL, for the cylinder with a pair of impellers (labeled WI) are com-
approaches. Quadrilateral structured mesh is employed to tessellate pared with the bare cylinder (labeled BC) in Fig. 5. It is seen that both
the static mesh zone where the mesh remains static in the calculations. CD and CL of the bare cylinder fluctuate over time in a stable cycle,
The mesh resolution depends on the number of nodes along the and the fluctuation frequency of CD is approximately double that of
structure surface and the expansion ratio. A grid independency test is CL, coinciding with the previous literature.34–36 In contrast, the fluctu-
thus conducted to pick up a reasonable mesh resolution. Five sets of ation presents multifrequency characteristics when the circular cylin-
mesh (G1–G5) are assessed in the test for the VIV of a circular cylinder der is arranged with a pair of impellers, signifying the effect of rotating
with a pair of free-to-rotate impellers at Ur ¼ 3, as depicted in Fig. 3. impellers on the fluid forces acting on the structure. Furthermore,
The influence of mesh resolution on the key results, including the the fluctuation frequency of CD is consistent with that of CL at Ur  7,
response amplitudes and the hydrodynamic coefficients, is suggesting the strong coupling of fluid forces in the streamwise and
evaluated in Table I, where AX/D and AY/D represent the in-line
transverse directions. Nevertheless, the variation of hydrodynamic
and cross-flow amplitudes, respectively, and CDmean and CLrms are
the time-averaged drag coefficient and the root-mean-squared lift
coefficient, expressed as

1X N
2fD ðtÞ
CDmean ¼ ; (6)
N i¼1 qu2in D
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
#ffi
u N "
u 1 X 2fL ðtÞ 2
CLrms ¼t ; (7)
N i¼1 qu2in D

where N is the number of samples in the time series for statistics. In


Table I, the relative deviations in parentheses illustrate the difference
between the present case and the former one as the number of ele-
ments increases. It is clearly seen that the calculation results converge
at G4 with the differences between G4 and G5 are less than 0.1%,
suggesting the further increase in mesh resolution has a negligible
influence. Therefore, the mesh resolution of G4 is adopted in present
simulations, where the cylinder perimeter is evenly divided by
240 nodes and the expansion ratio is less than 1.10 over the entire FIG. 4. Validation of the numerical model for the VIV of a bare circular cylinder in
domain. comparison with the reported results.

Phys. Fluids 33, 083608 (2021); doi: 10.1063/5.0063029 33, 083608-5


Published under an exclusive license by AIP Publishing
Physics of Fluids ARTICLE scitation.org/journal/phf

coefficients of the equipped cylinder against the reduced velocity is coefficient (CD0 ) of WI is slightly smaller than that of BC, and its occur-
similar to that of a bare cylinder, presenting the peak amplitude rence is delayed to Ur ¼ 6. The same trend is observed for the time-
around Ur ¼ 5 and small amplitude at Ur  7. As shown in Fig. 6, averaged drag coefficient (CDmean). Moreover, CDmean of the equipped
although the amplitude of lift coefficient (CL0 ) of WI experiences the cylinder is generally less than that of the bare cylinder at Ur  7, indi-
same rise and drop as BC when Ur increases from 2 to 7, the value of cating the drag reduction.
the former is nearly triple that of the latter in this Ur range, implying a Figure 7 plots the variation of vibration amplitudes in both in-
greater lift force is exerted on the cylinder due to the existence of rotat- line and cross-flow directions vs the reduced velocity. It is unexpected
ing impellers. On the contrary, the maximum amplitude of drag to find that both the in-line and cross-flow oscillations are enhanced

02 December 2023 10:08:45

FIG. 5. Time histories of the hydrodynamic coefficients of a circular cylinder with a pair of impellers (WI) in comparison with that of a bare cylinder (BC).

Phys. Fluids 33, 083608 (2021); doi: 10.1063/5.0063029 33, 083608-6


Published under an exclusive license by AIP Publishing
Physics of Fluids ARTICLE scitation.org/journal/phf

FIG. 8. Comparison of the dominant frequencies vs the reduced velocity between


a cylinder with a pair of impellers (WI) and a bare cylinder (BC).

when the cylinder is equipped with a pair of passively flow-induced


rotating impellers, especially in the initial branch of VIV (Ur  6). In
comparison with the bare cylinder, the maximum in-line and cross-
flow amplitudes are amplified 5.02 and 1.09 times, respectively.
Additionally, the occurrence of the maximum amplitude shifts from

02 December 2023 10:08:45


Ur ¼ 5 to Ur ¼ 6, corresponding to the shift of CD0 and CDmean. Even
in the lower branch of VIV, the response amplitude of WI is slightly
higher than that of BC. Unlike the vibration suppression of CCs, the
arrangement of impellers is counterproductive, which is associated
with the wake adjustment that will be discussed later. The variations of
oscillation amplitudes do not strictly follow the trend of hydrodynamic
forces that acting on the cylinder surface, taking Figs. 6 and 7 into
account. The possible reason is that the hydrodynamic forces exciting
FIG. 6. Comparison of the hydrodynamic coefficients vs the reduced velocity the vibration of equipped cylinder consist of the fluid forces acting on
between a cylinder with a pair of impellers (WI) and a bare cylinder (BC). both the cylinder surface and the impellers’ wall while only the former
is considered in the calculation of CD and CL.
The dominant frequencies of the equipped cylinder are compared
with the bare cylinder in Fig. 8. It is seen that the cross-flow frequency
of the bare cylinder follows the line of Strouhal number (St ¼ 0.162) in
the initial and desynchronization branches of VIV except the lower
branch where the vibration frequency is locked on the natural fre-
quency (fY*  1). The in-line frequency presents the same trend, but
the value is doubled. Consequently, the vibration exhibits the figure-
eight trajectories, as depicted in Fig. 9. The phase angle between the
cross-flow amplitude and the lift coefficient jumps from near zero to
180 at Ur ¼ 5, illustrating the transition from the initial branch to the
lower branch,37–39 as shown in Fig. 10. In contrast, such phase jump
shifts to occur at Ur ¼ 6 when a pair of impellers is mounted, demon-
strating the delayed appearance of the lower branch. The in-line and
cross-flow frequencies of the controlled cylinder are close to fX* ¼ 2
and fY* ¼ 1 in the initial branch, respectively, explaining the significant
augment of oscillation amplitude (as seen in Fig. 7). Furthermore, the
dominant response frequencies are locked on the natural frequency in
both in-line and cross-flow directions at Ur  7, indicating the absence
FIG. 7. Comparison of the vibration amplitudes vs the reduced velocity between a of desynchronization branch and the coupling between two directions
cylinder with a pair of impellers (WI) and a bare cylinder (BC). (fX*  fY*). Nevertheless, the oscillating orbits of WI become irregular

Phys. Fluids 33, 083608 (2021); doi: 10.1063/5.0063029 33, 083608-7


Published under an exclusive license by AIP Publishing
Physics of Fluids ARTICLE scitation.org/journal/phf

FIG. 9. Comparison of the oscillating tra-


jectories between a cylinder with a pair of
impellers (WI) and a bare cylinder (BC).

the occurrence of wake evolution into two-layered vortices when


Ur ¼ 5 due to the vigorous response. Such wake evolution of self-
excited oscillating bare cylinder is in consistent with the observation in
previous literature.36 It is seen that the positive and negative transverse

02 December 2023 10:08:45


flow velocities (v) alternately occur in the wake, in correspondence
with the alternately shed vortices. The positive peak v appears between
a clockwise vortex and the adjacent followed anticlockwise vortex,
while the negative peak v emerges between the anticlockwise vortex
and the next shed clockwise vortex. The alteration of the contours of
transverse velocity in the wake corresponds to the wake evolution
from one-row 2S street (primary vortex street) to the two-layered vor-
tices at Ur ¼ 5. Furthermore, the second evolution from the two-
layered vortices to the secondary vortex street occurs around X ¼ 32D.
As reported in Jiang et al.,40 the occurrence of the secondary vortex
street locates at X ¼ 60.4D for a stationary circular cylinder at
Re ¼ 300 and this location is pushed downstream as Re decreases. By
contrast, the early appearance of the secondary vortex street at
Re ¼ 150 is mainly attributed to the vigorous oscillation at this Ur.
As displayed in Fig. 11, the wake flow is significantly modified
with the arrangement of a pair of impellers in the whole considered Ur
range. At Ur ¼ 3, the two-layered vortices are formed in the wake
FIG. 10. The variation of the phase difference between the cross-flow amplitude
and the lift coefficient with the reduced velocity.
instead of the one-row 2S street. In comparison with the bare cylinder,
the widened wake and the shortened vortex formation mainly contrib-
ute to the increase in vibration amplitude at this Ur. In addition, the
at Ur  7 due to the coexistence of multiple frequencies, as shown in two-layered vortices evolve into the secondary vortex street in the far
Fig. 9. wake as a result of the merging of several adjacent same-sign vortices
in the same layer into a new vortex.40 The appearance of the enlarged
transverse velocity contours suggests the onset position of the second-
B. Wake flow evolution ary vortex street.41,42 The occurrence of the secondary vortex street
Figure 11 compares the vorticity contours behind the structure moves upstream from X ¼ 13.8D at Ur ¼ 3 to X ¼ 12D at Ur ¼ 5, sig-
when it returns to the equilibrium position from the lower side, in nifying the influence of enhanced oscillation (as seen in Fig. 7) on the
accompany with the instantaneous cross-flow velocity at different wake flow transition. When Ur is increased to 6, the wake width is sig-
reduced velocities. The typical one-row 2S (two vortices are shed per nificantly amplified due to the maximum response amplitude.
cycle, one vortex clockwise and the other counterclockwise) vortex Nevertheless, the transition from two-layered vortices to the secondary
shedding mode is observed in the wake of the bare cylinder, excluding vortex street moves downstream in comparison with that at Ur ¼ 5.

Phys. Fluids 33, 083608 (2021); doi: 10.1063/5.0063029 33, 083608-8


Published under an exclusive license by AIP Publishing
Physics of Fluids ARTICLE scitation.org/journal/phf

02 December 2023 10:08:45


FIG. 11. The contours of vorticity and cross-flow velocity corresponding to the equilibrium position at different reduced velocities: (a) Ur ¼ 3–7; (b) Ur ¼ 9–13.

The widened wake possibly delays the merging of vortices and hence the clockwise vortex labeled I + G+D + B. Consequently, the secondary
the onset of the secondary vortex street. When Ur is larger than 7, the vortex street is formed with a frequency approximately quarter of the
wake flow evolution is similar to Ur ¼ 5 in spite of the relatively nar- vortex shedding frequency. The interaction between the shear layers
row wake. Overall, the installation of impellers gives rise to the occur- separated from the cylinder and the impellers during one vortex shed-
rence of secondary vortex street, the decrease in vortex formation ding cycle is displayed in Fig. 12(b). At moment #1, the equipped cyl-
length, and the increase in wake width. inder approximately moves upward to the equilibrium position,
Figures 12–14 depict the wake flow evolution at three representa- resulting in the downward deflection of shear layers (labeled S1 and S2,
tive reduced velocities. It is seen from Fig. 12(a) that an anticlockwise corresponding to the shear layers separated from the upper and lower
is generated from the lower shear layer of the equipped cylinder at the sides, respectively). The upper impeller rotates clockwise at this
moments labeled in odd numbers, resulting in a clockwise velocity moment with approximately the maximum rotation speed (n), leading
along the cylinder surface and hence the maximum upward lift force. to the negative vorticity contours filling the gaps among the three
In contrast, the maximum downward lift force occurs when a clock- blades. Nevertheless, positive vorticity contours cover the impeller due
wise is formed from the upper shear layer at the moments labeled in possibly to the upward motion of the whole system. Two downward
even numbers. The vortex shedding occurs in an alternate manner tilted shear layers (labeled s1 and s2) like the main cylinder are gener-
between the two sides, generating a two-row 2S vortex street at Ur ¼ 3. ated from the upper impeller, but the upper layer s1 quickly curves
It is interesting to find that four adjacent same-sign vortices in the upward, attributed to the clockwise rotation. In contrast, two distinct
same layer are merged into a new vortex during four vortex shedding downward tilted shear layers (labeled s’1 and s’2) are separated from
cycles, as the new-born anticlockwise vortex labeled F + E+C + A and two blade tips of the lower impeller, which basically does not rotate at

Phys. Fluids 33, 083608 (2021); doi: 10.1063/5.0063029 33, 083608-9


Published under an exclusive license by AIP Publishing
Physics of Fluids ARTICLE scitation.org/journal/phf

02 December 2023 10:08:45

FIG. 12. Wake flow evolution and shear layers interaction at Ur ¼ 3: (a) the wake flow evolution behind the cylinder with a pair of impellers; (b) the interaction between the
shear layers separated from the circular cylinder and the impellers.

this moment. As the structure moves upward, the enclosed negative #3, when its rotation speed drops to near zero. Meanwhile, the lower
vorticity contours inside the upper impeller break through the positive shear layers from the main cylinder and the lower impeller connect,
contours and connect to the upper shear layer s1 at moment #2. After labeled S2s20 . At moment #4, the equipped cylinder starts to move
that, this shear layer rolls up to generate a clockwise vortex at moment downward. Consequently, the two layers of the upper impeller tilt

Phys. Fluids 33, 083608 (2021); doi: 10.1063/5.0063029 33, 083608-10


Published under an exclusive license by AIP Publishing
Physics of Fluids ARTICLE scitation.org/journal/phf

02 December 2023 10:08:45


FIG. 12. (Continued.)

upward because of the zero-speed rotation. On the contrary, due to from the shear layer S1s1 to the vortex A is cut off by the upper impel-
the drawn of the layer s10 by the anticlockwise rotation of the lower ler that is rotating clockwise, resulting in the shed of vortex A. The
impeller, the connected layer S2s20 is cut off into two parts with one fill- shed of the clockwise vortex occurs when the rotation speed of the
ing the blades’ gap of the lower impeller and the other left outside the upper impeller reaches the maximum value, while the shed of the anti-
impeller. From moment #4 to moment #7, the vortex (labeled A) con- clockwise vortex happens when the maximum speed is achieved for
nected to the shear layer s1 grows gradually and this layer is linked the lower impeller. Therefore, the vortex shedding of an equipped cyl-
with the upper layer of the main cylinder, labeled S1s1. Additionally, inder is associated with the rotation response of the pair of impellers,
the enclosed positive vorticity contours inside the lower impeller are which is closely related to the oscillation process.
released at moment #7 as the rotation is ceased. From moment #8 to Figure 13(a) depicts the wake flow evolution at Ur ¼ 6. It is seen
moment #9, the cylinder moves upward from the lowest position to that a pair of anticlockwise vortices (labeled B and C) is shed from the
the equilibrium position. As the result of the roll-up of shear layer S2s20 , lower side at moment #1, and simultaneously a pair of clockwise vorti-
an anticlockwise vortex (labeled B) is formed. It is mainly attributed to ces is formed behind the cylinder, contributing to the negative lift coef-
the zero-speed rotation of the lower impeller. Meanwhile, the supply ficient (CL). At moment #2, the pair of clockwise vortices (labeled D

Phys. Fluids 33, 083608 (2021); doi: 10.1063/5.0063029 33, 083608-11


Published under an exclusive license by AIP Publishing
Physics of Fluids ARTICLE scitation.org/journal/phf

FIG. 13. Wake flow evolution and shear


layers interaction at Ur ¼ 6: (a) the wake
flow evolution behind the cylinder with a pair

02 December 2023 10:08:45


of impellers; (b) the interaction between the
shear layers separated from the circular
cylinder and the impellers.

and E) is shed from the upper side, in accompany with the positive lift same-sign vortices in the same row merge into a new vortex while
coefficient. Therefore, the vortices are shed alternately from the two migrating downstream, such as the clockwise vortices labeled D + E
sides in a 2P mode (two pairs of vortices are shed per cycle). A two- and H + I in the upper row and the anticlockwise vortex labeled F + G + C
row vortex street is formed in the wake. Nevertheless, several adjacent in the lower row. The difference in the number of merging vortices

Phys. Fluids 33, 083608 (2021); doi: 10.1063/5.0063029 33, 083608-12


Published under an exclusive license by AIP Publishing
Physics of Fluids ARTICLE scitation.org/journal/phf

FIG. 13. (Continued.)

02 December 2023 10:08:45


is possibly attributed to the violent response with multiple frequen- fluctuation curve. It is seen from Fig. 13(b) that all the shear layers
cies at this reduced velocity. The secondary vortex street also occurs of the main cylinder and the pair of impellers tilt upward at
in the far wake like that at Ur ¼ 3, but the merging process is differ- moments #1 and #2 as the cylinder moves downward with nearly
ent from the latter due possibly to the multifrequency vibration. The zero-speed rotation for both two impellers. At moment #3, the cylin-
two clockwise vortices labeled D + E and H + I in the upper row der reaches the lowest position. An anticlockwise vortex (labeled b)
merge into a new one labeled H + I+D + E at moment #12, while the is formed due to the roll-up of the lower shear layer (labeled S2) of
anticlockwise vortices labeled F + G+C and K in the lower row the main cylinder, and a pair of anticlockwise vortices (labeled a and
merge into a new vortex labeled K + F+G + C at moment #15. It c) is generated from the lower shear layer (labeled s20 ) of the lower
illustrates that the occurrence of the secondary vortex street is due impeller. After that, a clockwise vortex (labeled e) is created from
to the merging of three pairs of vortices. As a result, the frequency of the upper layer (labeled s1) of the upper impeller as the combination
the secondary vortex street is approximately 1.5 times of the vortex result of the clockwise rotation of the upper impeller and the
shedding frequency, which is also demonstrated from the lift equipped cylinder’s upward motion at moment #4. Furthermore, as

Phys. Fluids 33, 083608 (2021); doi: 10.1063/5.0063029 33, 083608-13


Published under an exclusive license by AIP Publishing
Physics of Fluids ARTICLE scitation.org/journal/phf

02 December 2023 10:08:45

FIG. 14. Wake flow evolution and shear layers interaction at Ur ¼ 14: (a) the wake flow evolution behind the cylinder with a pair of impellers; (b) the interaction between the
shear layers separated from the circular cylinder and the impellers.

the cylinder turns to move upward, all the shear layers present a cylinder moves upward from moment #4 to moment #6, it is noted
downward deviation, and three anticlockwise vortices (labeled a, c, that the three same-sign vortices (labeled b, c, and d) merge into a
and d) are cut off from the layer s20 and the anticlockwise vortex new vortex (labeled b, c, and d), which accompanies with vortex a,
(labeled b) is shed from the lower shear layer (labeled S2) of the forming the pair of anticlockwise vortices shed in the wake (as seen
main cylinder because of the interaction of shear layers. As the in figure a). Similarly, two clockwise vortices (labeled k and j)

Phys. Fluids 33, 083608 (2021); doi: 10.1063/5.0063029 33, 083608-14


Published under an exclusive license by AIP Publishing
Physics of Fluids ARTICLE scitation.org/journal/phf

02 December 2023 10:08:45

FIG. 14. (Continued.)

generated from the upper layer (labeled s1) of the upper impeller consequence of shear-layer interaction between the main cylinder
and one clockwise vortex (labeled i) separate from the upper layer and the impellers.
(labeled S1) of the main cylinder merge together, in accompany with The wake flow evolution at Ur ¼ 14 is depicted in Fig. 14(a). It is
the former shed clockwise vortex (labeled h) from the layer s1, yield- seen that the vortices are shed from the shear layers of two impellers
ing a pair of clockwise vortices shed during the cylinder’s downward while no vortex is formed from the shear layers of the main cylinder,
movement. It indicates that the 2P vortex shedding mode is the which is similar to the wake flow of three circular cylinders arranged

Phys. Fluids 33, 083608 (2021); doi: 10.1063/5.0063029 33, 083608-15


Published under an exclusive license by AIP Publishing
Physics of Fluids ARTICLE scitation.org/journal/phf

in a triangular configuration with a small gap.43 It indicates the vortex shear layers separated from the circular cylinder and the impellers at
shedding of the main cylinder is suppressed by the pair of rotating Ur ¼ 14. It illustrates that the shear layers have no obvious upward or
impellers, contributing to the small response amplitude at this reduced downward deviation as Ur ¼ 3 and Ur ¼ 6, owing to the limited
velocity (as seen in Fig. 7). The three adjacent clockwise vortices response amplitude. It is seen that vortices are shed from each shear
(labeled D, B, and A) shed from the upper shear layer of the upper layer of each impeller despite the suppression of vortex shedding from
impeller merge into a new one (labeled D + B + A) at moment #3, the main cylinder. Nevertheless, the vortices on the inner sides are
while the two adjacent anticlockwise vortices (labeled E and C) shed quickly decayed while migrating downstream, resulting in the observa-
from the lower shear layer of the lower impeller merge into a new anti- tion of vortex shedding from the outer sides in figure a. In addition,
clockwise vortex (labeled E + C) at moment #4. At moment #7, the rotation speed varies asynchronously in an unequal frequency for
another anticlockwise vortex (labeled G) catches up with the vortex the two impellers, contributing to the asymmetrical vortex shedding
E + C and merges into the latter. From moment #7 to #10, the clock- and different number of shed vortices.
wise vortex D + B+A decays gradually and is absorbed by the followed
vortex (labeled F), forming a new vortex (labeled F + D+B + A) at C. Flow-induced rotation
moment #10. Such asymmetrical vortex merging process with respect The flow-induced rotation responses of the upper impeller at two
to the wake centerline is possibly attributed to the multifrequency representative reduced velocities (Ur ¼ 6 and Ur ¼ 7), corresponding
oscillating of the cylinder, as illustrated from the lift coefficient curve to the maximum vibration amplitude of VIV initial branch and the
and the oscillating trajectories in Fig. 10. The clockwise vortex F + D onset of VIV lower branch (as seen in Fig. 7), are depicted in Figs. 15
+ B + A and anticlockwise vortex G + E + C migrate downstream in a and 16. It is seen from Fig. 15 that the upper impeller rotates clockwise
pair, signifying the occurrence of the secondary vortex street. It is for the majority of the time, which is associated with the pressure dis-
noted that the vortex pattern varies from 2S to P + S in the secondary tribution around the impeller. In comparison with the convex surface,
vortex street at moment #13, indicating the unstable vortex merging in the greater pressure force acting on the concave surface of the impeller
the wake transition. Figure 14(b) depicts the interaction between the is the main cause of clockwise rotation. The amplified pressure

02 December 2023 10:08:45


FIG. 15. The flow-induced rotation
response of the upper impeller at Ur ¼ 6.

Phys. Fluids 33, 083608 (2021); doi: 10.1063/5.0063029 33, 083608-16


Published under an exclusive license by AIP Publishing
Physics of Fluids ARTICLE scitation.org/journal/phf

difference between the convex and concave surfaces results in the in the rotation. At Ur ¼ 6, the rotation becomes irregular without
acceleration of rotation at moments #4 and #5. The anticlockwise rota- distinct periodicity due to the intermittent switching of rotation
tion intermittently occurs when the pressure force imposed on the direction, as illustrated in Fig. 15. When the response enters the
convex surface becomes heavier than that exerted on the concave sur- lower branch (Ur  7), the fluctuation frequency of rotation is
face, such as moment #6. The intermittent change of rotation direction much higher than that of vibration. Nevertheless, the direction of
suggests the unstable flow-induced rotation response, which is a possi- rotation is unchanged and the speed is smaller than that in the ini-
ble contribution to the amplitude amplification at this reduced velocity tial branch.
as compared to the bare circular cylinder. In spite of the intermittent switching of rotation direction, the
In contrast, as shown in Fig. 16, the upper impeller keeps upper impeller (impeller-1) rotates clockwise in most time, resulting
rotating clockwise during one vibration cycle at Ur ¼ 7, playing a in the negative time-averaged rotation speed (n ), as plotted in Fig. 18.
positive role in injecting momentum into the boundary layer of the The lower impeller (impeller-2) mostly rotates in the anticlockwise
main cylinder16 and hence the small response amplitude (as seen in direction, presenting the positive time-averaged rotation speed with
Fig. 7). Additionally, the rotation speed at Ur ¼ 6 is approximately the same absolute value as the upper impeller. It indicates that the
25 times of that at Ur ¼ 7. It implies that too fast unstable rotation flow-induced rotation in terms of time statistics is symmetrical for the
of impellers is the main reason for the unexpected vibration pair of impellers symmetrically arranged behind the cylinder although
enhancement. the rotation speed varies over time. In addition, the variation of the
Figure 17 depicts the time histories of the rotation speed as time-averaged rotation speed with Ur follows the trend of the response
well as the vibration amplitude at representative Ur cases. It is seen amplitude, i.e., increasing with Ur in the VIV initial branch and keep-
that the fluctuation of rotation speed basically processes the same ing a low speed generally unchanged in the lower branch. It implies
dominant frequency as the vibration in the initial branch (Ur ¼ 3 that the VIV suppression may be achieved by a pair of impellers rotat-
and 5), although there are some secondary frequencies taking part ing at a reasonable low speed.18

02 December 2023 10:08:45


FIG. 16. The flow-induced rotation
response of the upper impeller at Ur ¼ 7.

Phys. Fluids 33, 083608 (2021); doi: 10.1063/5.0063029 33, 083608-17


Published under an exclusive license by AIP Publishing
Physics of Fluids ARTICLE scitation.org/journal/phf

(1) The VIV attenuation is not achieved as expectation in the whole


Ur range with the introduction of a pair of three-blade impellers
mounted on control rods. In the VIV initial branch, the
response amplitudes in both in-line and cross-flow directions
are significantly enlarged with the presence of passively rotating
impellers, and the occurrence of peak amplitude shifts to a
higher Ur as opposed to the bare cylinder. Even in the VIV
lower branch, the response amplitude of equipped cylinder is
also slightly higher than the bare cylinder. Furthermore, both
the in-line and cross-flow frequencies are locked on the natural
frequency in the lower branch, suggesting the strong interaction
between the two directional responses. Nevertheless, the coexis-
tence of multiple frequencies results in irregular oscillation tra-
jectories, unlike the figure-eight orbits in the initial branch.
(2) The typical one-row 2S vortex mode is observed in the wake of
the bare cylinder, except the Ur ¼ 5 case which presents two
wake evolutions: the first evolution from the primary street into
the two-layered vortices and the second evolution into the sec-
ondary vortex street. In comparison with the flow around a sta-
tionary cylinder, the vigorous vibration mainly contributes to
the early occurrence of the secondary vortex street. In contrast,
the secondary vortex street occurs in the whole Ur range
with the absence of the one-row 2S street when the cylinder is
equipped with a pair of impellers. The onset position of the sec-
ondary vortex street could be identified from the enlarged
transverse velocity contours. In comparison with the bare cylin-

02 December 2023 10:08:45


der, the shortened vortex formation and the widened wake
mainly contribute to the vibration enhancement.
(3) Unlike the bare cylinder, the formation of secondary vortex
street is not limited to the result of two adjacent same-sign vor-
tices merging in each layer of the equipped cylinder due to the
violent multifrequency vibration. Moreover, the merging pro-
cess of vortices in the upstream wake varies with Ur, which is
FIG. 17. Comparison between the rotation speed of the upper impeller and the
associated with the response amplitude and the interaction
vibration amplitude at representative Ur cases.
between the shear layers separated from the main cylinder and
the two impellers. The shear layers are easily cut off by rotating
impellers, affecting the vortex formation and shedding.
Nevertheless, the rotation speed fluctuates over time in a vari-
able cycle, depending on the fluid force acting on the blades
and the vibration process. The intermittent switching in the
rotation direction and the unstable fast rotation of impellers are
the possible causes of vibration amplification. In terms of time
statistics, the symmetrical inward counter-rotating pattern is
achieved with the arrangement of two impellers. Additionally,
the time-averaged rotation speed closely follows the response
amplitude against the variation of Ur, suggesting the coupling
FIG. 18. Rotation speeds of the two impellers vs the reduced velocity. of vibration and rotation.
Further extensive studies are required to investigate the coupling
vibration–rotation response of the equipped cylinder, taking into
IV. CONCLUSIONS
account the mixed effects of dimensions and location of impellers and
This study pioneers the numerical investigation of the coupling rotational damping due to friction, and to quantitatively evaluate the
response of vortex-induced vibration (VIV) of a circular cylinder sym- energy deficit over a wider Re range.
metrically equipped with a pair of flow-induced rotating (FIR) impel-
lers at a low Reynolds number of 150. The objective of this paper is to ACKNOWLEDGMENTS
examine an innovative idea that transforms the active control of rotat-
ing rods to passive control with aid of impellers to drive the rotation of This research was supported by the National Natural Science
rods. The main conclusions are as follows: Foundation of China (Grant No. 51979238). The work was carried

Phys. Fluids 33, 083608 (2021); doi: 10.1063/5.0063029 33, 083608-18


Published under an exclusive license by AIP Publishing
Physics of Fluids ARTICLE scitation.org/journal/phf

out in the computer cluster of the laboratory of offshore oil and gas REFERENCES
engineering at Southwest Petroleum University. 1
H. Zhu, C. Zhang, and W. Liu, “Wake-induced vibration of a circular cylinder
at a low Reynolds number of 100,” Phys. Fluids 31, 073606 (2019).
2
H. Choi, W. P. Jeon, and J. S. Kim, “Control of flow over a bluff body,” Annu.
NOMENCLATURE Rev. Fluid Mech. 40, 113–139 (2008).
3
H. Zhu, G. Li, and J. Wang, “Flow-induced vibration of a circular cylinder with
AX/D normalized response amplitude in the in-line direction splitter plates placed upstream and downstream individually and simulta-
AY/D normalized response amplitude in the cross-flow neously,” Appl. Ocean Res. 97, 102084 (2020).
direction
4
L. Ding, X. X. Mao, L. Yang, B. W. Yan, J. L. Wang, and L. Zhang, “Effects of
installation position of fin-shaped rods on wind-induced vibration and energy
CDmean time-averaged drag coefficient
harvesting of aeroelastic energy converter,” Smart Mater. Struct. 30, 025026
CLrms root-mean-squared lift coefficient (2021).
CD0 fluctuating amplitude of drag coefficient 5
H. Zhu, Z. Liao, Y. Gao, and Y. Zhao, “Numerical evaluation of the suppres-
CL0 fluctuating amplitude of lift coefficient sion effect of a free-to-rotate triangular fairing on the vortex-induced vibration
D diameter of control rod of a circular cylinder,” Appl. Math. Model. 52, 709–730 (2017).
D cylinder diameter
6
H. Zhu and J. Yao, “Numerical evaluation of passive control of VIV by small
ds sweeping diameter of impeller control rods,” Appl. Ocean Res. 51, 93–116 (2015).
7
G. R. S. Assi, P. W. Bearman, and M. A. Tognarelli, “On the stability of a free-
fD(t) drag forces exerted on the structure surface by ambient
to-rotate short-tail fairing and a splitter plate as suppressors of vortex-induced
flow vibration,” Ocean Eng. 92, 234–244 (2014).
fL(t) lift forces exerted on the structure surface by ambient flow 8
H. Zhu, T. Tang, H. Zhao, and Y. Gao, “Control of vortex-induced vibration of
fn natural frequency a circular cylinder using a pair of air jets at low Reynolds number,” Phys.
fX* normalized vibration frequency in the in-line direction Fluids 31, 043603 (2019).
9
f Y* normalized vibration frequency in the cross-flow direction D. Fan, L. Yang, Z. Wang, M. S. Triantafyllou, and G. E. Karniadakis,
“Reinforcement learning for bluff body active flow control in experiments and
G gap between main cylinder surface and sweeping bound-
simulations,” Proc. Natl. Acad. U. S. A. 117, 26091–26098 (2020).
ary of impeller 10
S. R. Munshi, V. J. Modi, and T. Yokomizo, “Fluid dynamics of flat plates and
m mass of the system rectangular prisms in the presence of moving surface boundary-layer control,”
m* mass ratio J. Wind Eng. 79, 37–60 (1999).
11
V. J. Modi, “Moving surface boundary-layer control: A review,” J. Fluids Struct.

02 December 2023 10:08:45


n rotation speed of impeller
N number of sample in the time series for statistics 11, 627–663 (1997).

n time-averaged rotation speed of impeller
12
V. J. Modi, M. S. U. K. Fernando, and T. Yokomizo, “Moving surface boundary
layer control as applied to two-dimensional and three-dimensional bluff bod-
p pressure ies,” J. Wind Eng. Ind. Aerodyn. 38, 83–92 (1991).
R internal radius of blade 13
V. J. Modi, M. S. U. K. Fernando, and T. Yokomizo, “Moving surface boundary
Re Reynolds number layer control: Studies with bluff bodies and application,” AIAA J. 29,
St Strouhal number 1400–1406 (1991).
t flow time
14
S. Mittal, “Control of flow past bluff bodies using rotating control cylinders,”
u velocity components along the x direction J. Fluids Struct. 15, 291–326 (2001).
15
S. Mittal, “Flow control using rotating cylinders: Effect of gap,” ASME J. Appl.
uin free-stream velocity
Mech. 70, 762–770 (2003).
Ur the reduced velocity 16
H. Zhu, J. Yao, Y. Ma, H. Zhao, and Y. Tang, “Simultaneous CFD evaluation of
v velocity components along the y direction VIV suppression using smaller control cylinders,” J. Fluids Struct. 57, 66–80 (2015).
x in-line direction 17
S. Muddada and B. S. V. Patnaik, “An active flow control strategy for the sup-
X in-line displacement pression of vortex structures behind a circular cylinder,” Eur. J. Mech. B Fluids
X0 in-line motion velocity of the structure 29, 93–104 (2010).
X00 in-line acceleration of the structure
18
H. Zhu and Y. Gao, “Effect of gap on the vortex-induced vibration suppression
of a circular cylinder using two rotating rods,” Ships Offshore Struct. 13,
y cross-flow direction 119–131 (2018).
Y cross-flow displacement 19
I. Korkischko and J. R. Meneghini, “Suppression of vortex-induced vibration
Y0 cross-flow motion velocity of the structure using moving surface boundary-layer control,” J. Fluids Struct. 34, 259–270
Y00 cross-flow acceleration of the structure (2012).
d thickness of blade
20
H. Zhu and Y. Gao, “Vortex-induced vibration suppression of a main circular
f damping ratio of the system cylinder with two rotating control rods in its near wake: Effect of the rotation
direction,” J. Fluids Struct. 74, 469–491 (2017).
q fluid density 21
H. Zhu and Y. Gao, “Vortex induced vibration response and energy harvesting
t kinematic viscosity of fluid of a marine riser attached by a free-to-rotate impeller,” Energy 134, 532–544
u phase angle between the lift coefficient and the cross-flow (2017).
amplitude 22
L. Y. Pao and K. E. Johnson, “Control of wind turbines,” IEEE Control Syst.
x0 natural circular frequency of the structure 31, 44–62 (2011).
23
L. Wang and R. W. Yeung, “On the performance of a micro-scale Bach-type
turbine as predicted by discrete-vortex simulations,” Appl. Energy 183,
DATA AVAILABILITY 823–836 (2016).
24
S. Roy and A. Ducoin, “Unsteady analysis on the instantaneous forces and
The data that support the findings of this study are available moment arms acting on a novel Savonius-type wind turbine,” Energy Convers.
from the corresponding author upon reasonable request. Manage. 121, 281–296 (2016).

Phys. Fluids 33, 083608 (2021); doi: 10.1063/5.0063029 33, 083608-19


Published under an exclusive license by AIP Publishing
Physics of Fluids ARTICLE scitation.org/journal/phf

25
H. Zhu, Y. Zhao, and J. Hu, “Performance of a novel energy harvester for 34
I. Borazjani and F. Sotiropoulos, “Vortex-induced vibrations of two cylinders
energy self-sufficiency as well as a vortex-induced vibration suppressor,” in tandem arrangement in the proximity-wake interference region,” J. Fluid
J. Fluids Struct. 91, 102736 (2019). Mech. 621, 321–364 (2009).
26
H. Zhu, H. Zhao, J. Yao, and Y. Tang, “Numerical study on vortex-induced 35
J. Wu, C. Shu, and N. Zhao, “Numerical investigation of vortex-induced vibration
vibration responses of a circular cylinder attached by a free-to-rotate dartlike of a circular cylinder with a hinged flat plate,” Phys. Fluids 26, 063601 (2014).
overlay,” Ocean Eng. 112, 195–210 (2016). 36
Y. Bao, D. Zhou, and J. Tu, “Flow interference between a stationary cylinder
27
H. Zhu, Y. Zhao, and T. Zhou, “CFD analysis of energy harvesting from flow and an elastically mounted cylinder arranged in proximity,” J. Fluid Struct. 27,
induced vibration of a circular cylinder with an attached free-to-rotate penta- 1425–1446 (2011).
gram impeller,” Appl. Energy 212, 304–321 (2018). 37
N. Jauvtis and C. H. K. Williamson, “The effect of two degree of freedom on
28
A. Khalak and C. H. K. Williamson, “Dynamics of a hydroelastic cylinder with vortex-induced vibration at low mass and damping,” J. Fluid Mech. 509, 23–62
very low mass and damping,” J. Fluids Struct. 10, 455–472 (1996). (2004).
29
A. Khalak and C. H. K. Williamson, “Investigation of the relative effects of 38
C. H. K. Williamson and R. Govardhan, “Vortex-induced vibrations,” Annu.
mass and damping in vortex-induced vibration of a circular cylinder,” J. Wind Rev. Fluid Mech. 36, 413–455 (2004).
Eng. Ind. Aerodyn. 6971, 341–350 (1997). 39
H. Zhu, P. Lin, and Y. Gao, “Vortex-induced vibration and mode transition of
30
H. Jiang, L. Cheng, S. Draper, H. An, and F. Tong, “Three-dimensional direct a curved flexible free-hanging cylinder in exponential shear flows,” J. Fluid
numerical simulation of wake transitions of a circular cylinder,” J. Fluid Mech. Struct. 84, 56–76 (2019).
801, 353 (2016). 40
H. Jiang and L. Cheng, “Transition to the secondary vortex street in the wake
31
H. Zhu, T. Tang, T. Zhou, H. Liu, and J. Zhong, “Flow structures around trape- of a circular cylinder,” J. Fluid Mech. 867, 691–722 (2019).
zoidal cylinders and their hydrodynamic characteristics: Effects of the base 41
H. Zhu, J. Zhong, and T. Zhou, “Wake structure characteristics of three tandem
length ratio and attack angle,” Phys. Fluids 32, 103606 (2020). circular cylinders at a low Reynolds number of 160,” Phys. Fluids 33, 044113
32
H. Zhu and K. Wang, “Wake adjustment and vortex-induced vibration of a cir- (2021).
cular cylinder with a C-shaped plate at a low Reynolds number of 100,” Phys. 42
H. Zhu and W. Liu, “Wake structure and evolution of flow over a finned circu-
Fluids 31, 103602 (2019). lar cylinder,” Phys. Fluids 33, 073613 (2021).
33
H. Zhu, T. Tang, Y. Gao, T. Zhou, and J. Wang, “Flow-induced vibration of a 43
W. Chen, C. Ji, M. M. Alam, J. Williams, and D. Xu, “Numerical simulations of
trapezoidal cylinder placed at typical flow orientations,” J. Fluids Struct. 103, flow past three circular cylinders in equilateral-triangular arrangements,”
103291 (2021). J. Fluid Mech. 891, A14 (2020).

02 December 2023 10:08:45

Phys. Fluids 33, 083608 (2021); doi: 10.1063/5.0063029 33, 083608-20


Published under an exclusive license by AIP Publishing

You might also like