Professional Documents
Culture Documents
1 s2.0 S1290072916318245 Main
1 s2.0 S1290072916318245 Main
A R T I C L E I N F O A B S T R A C T
Keywords: In this paper, the effect of vortex induced vibration (VIV) on convective heat transfer from an elastically
Vortex induced vibration mounted rigid circular cylinder in cross-flow is investigated numerically. The motion of cylinder is modeled by
Convective heat transfer using a mass-spring-damping system. The effect of reduced velocity and damping ratio on the vortex formation,
Circular cylinder vortex shedding process, cylinder displacement amplitude, Nusselt number and the position of maximum local
Reduced velocity
Nusselt number is studied. In this study the Reynolds Number and the mass ratio are 150 and 2, respectively.
Damping ratio
Furthermore, different values of reduced velocity and damping ratio are investigated (Ur = 3, 4, 5, 6, 7, 8 and ζ =
0. 0.01, 0.05, 0.1). The numerical results demonstrate that the reduced velocity and damping ratio can affect the
heat transfer considerably. The beating phenomenon is occurs at Ur = 4 and ζ = 0.05 which leads to changes in the
displacement amplitude and Nusselt number widely with respect to the time. In the beating phenomenon the
total Nusselt number decreases in comparison with a stationary cylinder. The maximum heat transfer en-
hancement is obtained at Ur = 4 , ζ = 0 .
∗
Corresponding author. Department of Mechanical Engineering, Faculty of Engineering, Persian Gulf University, Bushehr, Iran.
E-mail address: e.izadpanah@pgu.ac.ir (E. Izadpanah).
https://doi.org/10.1016/j.ijthermalsci.2017.12.011
Received 24 December 2016; Received in revised form 10 July 2017; Accepted 11 December 2017
Available online 21 December 2017
1290-0729/ © 2017 Elsevier Masson SAS. All rights reserved.
E. Izadpanah et al. International Journal of Thermal Sciences 125 (2018) 405–418
overlay on the VIV of a circular cylinder. They found that the rotation of
dartlike overlay is counterclockwise and leads to disturb the boundary
layer near the body and suppress the vibration response to critical re-
duced velocity.
Recently, the effect of the free surface on VIV of the cylinder [21],
the Wake-induced vibration of a small cylinder located in the wake of a
larger cylinder [22] and flow-induced vibrations of circular cylinders
arranged in a tandem configuration [23] were investigated, numeri-
cally.
As aforementioned, the influence of VIV on the aerodynamic and
hydrodynamic aspects of cylinders was studied in the literature, widely.
Fig. 1. Schematic diagram of the computational domain and its boundary condition. However, there are many situations where the heat transfer char-
acteristics of the cylinder are essential. For example, when the tubes of
amplitude and frequency at Re = 150 for two inclination angles are the heat exchangers vibrate due to the vortex shedding, their heat transfer
same, both inside and outside of the lock-in region. But for Re = 1000 coefficient are changed. The change in the heat transfer coefficient is
these parameters are slightly different. The effect of using a pair of critical for heat exchanger and should be examined. The heat transfer
synthetic jets (SJs) on control the VIV of cylinder at the Reynolds characteristics of stationary cylinders was studied by many authors
number of 100, the mass ratio mr =2 and zero-damping were studied by [24–28]. But based on the authors' knowledge, the influence of VIV of
Wang et al. [12]. They observed that the VIV of cylinder can be effec- cylinders on the heat transfer coefficient are not studied, yet. In the
tively suppressed by using SJ control with three parameter combina- present work, the effect of cylinder oscillation on the quantity of heat
tions. transfer are investigated. In this study, the fluid is taken as water, and
Cui et al. [13] investigated the flow induced vibrations (FIV) of a different values of damping coefficient and the reduced velocity are
square and rectangular cylinder with different incidence angle. They considered. The Reynolds number of the present work is taken to be
found that for square and rectangular cylinder the galloping, a self- 150. Since this value is under the critical Reynolds number, the 2D
excited instability with strong vibration at a lower frequency, occurs at numerical solution is performed [10].
the incidence angle of zero. The galloping was not observed at the in-
cidence angle of 45 for rectangular cylinder and incidence angles of 2. Problem description and formulation
22.5 and 45 for a square cylinder. Joly et al. [14] studied about the
galloping phenomenon, too. Zhao et al. [15,16], Zhou et al. [17], The external flow over the circular cylinder investigated numeri-
Franzini et al. [18] and Jain and Modarres-Sadeghi [19] reported about cally. The cylinder is allowed to vibrate only in cross-flow direction.
the effect of the flow incidence angle on the VIV of the cylinder in their The computational domain is schematically shown in Fig. 1. At the
researches. Zhu et al. [20] studied the effect of a free-to-rotate dartlike inlet, the x-component of the velocity and the temperature are con-
sidered uniform U∞ and isothermal T∞, respectively. The cylinder is
406
E. Izadpanah et al. International Journal of Thermal Sciences 125 (2018) 405–418
Table 1 ∂u ∂u ∂p ∂ 2u
ρ ⎡ i + (uj − uj ) i ⎤ = − + μ ⎛⎜ 2i ⎞⎟
Details of numerical simulations results at different grid size for Re = 150 , m* = 2 , ζ = 0
and Ur = 4 .
⎢ ∂t
⎣ ∂x j
⎥
⎦ ˆ∂x i ⎝ ∂x i ⎠ (2)
Table 3 On top and bottom boundaries: Slip boundary conditions are used,
Comparison of present results with literature values at case 2.
∂U ∂T
= 0 , V = 0 and =0
A ∂y ∂y (7)
D
407
E. Izadpanah et al. International Journal of Thermal Sciences 125 (2018) 405–418
3. Numerical methodology method for the pressure linked equations) scheme is applied for pres-
sure-velocity coupling. The second order upwind scheme is employed to
The presented results are obtained numerically using the commer- discretize the convection terms in momentum and energy equations. A
cial software ANSYS FLUENT (version 17.0). The SIMPLE (semi-implicit User-Defined Function (UDF) is utilized to calculate the motion of the
408
E. Izadpanah et al. International Journal of Thermal Sciences 125 (2018) 405–418
cylinder. In each time step, after obtaining the lift forces by solving the of an optimal time step is significant to achieve reliable and accurate
UΔt
fluid dynamic equations (Eqs. (1) and (2)), the displacement and the results. The time step study is conducted for ur = 4 and D < 0.002
velocity of the cylinder are determined by solving the equation of the [11] and the results are independent on the time step size.
cylinder motion (Eq. (4)). The smoothing method of the dynamic mesh The multi-block structured quadrilateral cells with non-uniform
which adjusts the mesh by moving the boundary is selected for cylinder mesh spacing are used. Enough number of cells is considered around
motion. The convergence criteria of 10−6 for the continuity, x- and y- the cylinder to cover the high velocity and temperature gradient near
components of the velocity and energy equations is considered. The that. In addition, to increase the quality of the grid its size grows
first order implicit scheme is used for the time discretization. Selection smoothly. The generated computational mesh and its enlarged view
409
E. Izadpanah et al. International Journal of Thermal Sciences 125 (2018) 405–418
around the cylinder are displayed in Fig. 2. is smaller than 4.2%. In the second validation test case, the vibration of
For the grid dependence study, four grids with a different number of cylinders in the transverse direction at Re = 150, m∗ = 2 and , ζ = 0.01
cells are considered. The non-dimensional vibration amplitude is studied. The cross-flow vibration amplitude A/D are compared with
(A/ D = (ymax − ymin )/2D) , the total time averaged Nusselt number (Nu ) literature for Ur =3–8 in Table 3. An excellent agreement can be ob-
and the amplitude of the lift coefficient (CL) for Re = 150 , m* = 2, ζ = 0 served between the present values and the results of Wang et al. [12]
and Ur = 4 , at different grids are presented in Table 1. The symbols G1, (mean discrepancy of 1.95%). The small difference between the results
G2, G3 and G4 are illustrated the total number of cells in the domain. of present work and corresponding values of references Jiang et al. [34]
The maximum deviation between results of grids G3 and G4 is less than (mean discrepancy of 5.38%) and Bao et al. [35] (mean discrepancy of
0.6%. Therefore, the grid G3 sufficiently refined to resolve the flow and 3.82%) is related to the use of different numerical methods and grid
temperature fields with acceptable levels of accuracy. generation.
In the third validation test case, the verification is performed for the
4. Verification of the results vibration of cylinders in the cross-flow direction at Re = 200, m∗ = 10
, ζ = 0.01 and Ur = 5. According to Table 4, the maximum deviation is
The present results are validated with the published data for three found to be less than 2.4%. By comparison of present results with lit-
different cases. In the first validation case, the flow around stationary erature in three cases, it can be declared the results are reliable and
heated circular cylinder at Re = 100, 150, 200 and Pr = 0.7 is in- have acceptable levels of accuracy.
vestigated. The values of Nu , St , Cd and Cl are compared with the
published results [29–31] in Table 2. In addition, the Nusselt number is
5. Result and discussion
compared with the correlations which are presented by Knudsen and
Katz [32] eq. (13) and Churchill and Bernstein [33] eq. (14).
In this section, the effect of VIV on the heat transfer from circular
Nu = 0.638 Re 0.466Pr 1/3 (13) cylinder is investigated for Ur = 3, 4, 5, 6, 7, 8 and ζ = 0. 0.01, 0.05,
0.1 at Re=150 and mr =2. Vorticity and temperature contours at ζ = 0
0.62 Re1/2Pr 1/3 ⎡1 + ⎛ Re ⎞ ⎤
5/8 4/5
and Ur = 4 for five different cylinder position, namely a-e, in one period
Nu = 0.3 + of the cylinder displacement are depicted in Fig. 3. The location of these
[1 + (0.4/ Pr )2/3]1/4 ⎢ ⎝ 282000 ⎠ ⎥ (14)
⎣ ⎦
points (a-e) are shown in the time history curves of the non-dimensional
According to Table 2, the maximum deviation from literature values displacement and the surface-average Nusselt number (top of Fig. 3).
410
E. Izadpanah et al. International Journal of Thermal Sciences 125 (2018) 405–418
Dash-line in the contours indicate the initial position of the cylinder at vortices in a row is 3.24D and the distance between two rows of the
y∗ = 0 . Vorticity contours show that the vortex-shedding pattern is 2S vortex street is 2.86D. The temperature and vorticity contours have the
mode (two single vortices shed per the cycle of the displacement, like same features. The effect of VIV on thermal boundary layer and con-
the classic Karman street) with two horizontal rows of vortex street. The sequently on the thermal energy transfer by the flow can be seen in
vortices at the top and bottom rows have clockwise and counter- temperature contours and the time history curve of the surface-average
clockwise rotation, respectively. The space between two consecutive Nusselt number. The asymmetry of the vortexes behind the cylinder in
411
E. Izadpanah et al. International Journal of Thermal Sciences 125 (2018) 405–418
412
E. Izadpanah et al. International Journal of Thermal Sciences 125 (2018) 405–418
413
E. Izadpanah et al. International Journal of Thermal Sciences 125 (2018) 405–418
Table 5
Numax , Numin , ΔNu and fNu for Ur = 3–8 and ζ =0–0.1.
414
E. Izadpanah et al. International Journal of Thermal Sciences 125 (2018) 405–418
Fig. 13. a) Time histories of surface-average Nusselt number and b) maximum value of local Nusselt number versus angle for ζ = 0 and Ur = 3 − 5 .
damping ratio and reduced velocity can change the vibrations of the histories of cylinder displacement and Nusselt number fluctuate peri-
cylinder. Therefore, they can changes the amplitude and the frequency odically over constant values but at Ur = 4 the variation of A/ D and Nu
of variations of Nusselt number, too. As can be seen, oscillating cylinder are different, see Fig. 9. At this reduced velocity, the beating phe-
with Ur =4, 5, 6 has more heat transfer value with respect to the sta- nomenon occurs. This interesting phenomenon occurs when the forcing
tionery one; Conversely, for the oscillating cylinder with Ur =3, 8 the frequency is close to but not precisely identical to the natural frequency
heat transfer is lower than stationary one and for Ur = 7 both oscillating of the system. The beating phenomenon impose an unfavorable effect
and stationary cylinders have similar values of the heat transfer. It is on the heat transfer and caused to the vibration amplitude and average
seen that Ur =3 and 4 have minimum and maximum vibration amplitude Nusselt number periodically changes in a wide range with time (0.126D
(and heat transfer) respectively. to 0.461D for amplitude vibration and 13.26 to 14.97 for surface-
As shown in Fig. 7, the non-dimensional displacement of cylinder average Nusselt number). At Ur =4, each period of the displacement
and the surface-average Nusselt number diagrams for ζ = 0.01 are si- includes about 18 local peaks where in the other cases each period of
milar to the corresponding diagrams for ζ = 0 (Fig. 6), but the Nusselt displacement contains only one local peak.
number at Ur = 7 for ζ = 0.01 is less than the stationary cylinder one. As seen in Fig. 10 time histories of the non-dimensional displace-
The influence of reduced velocity on A/ D and Nu for ζ =0.05 is ment of cylinder and the surface-average Nusselt number at ζ = 0.1 for
shown in Fig. 8. In all reduced velocity (except Ur = 4) the time Ur =3, 5, 6, 7, 8 is similar to those at ζ = 0.05. Even though for Ur = 4
415
E. Izadpanah et al. International Journal of Thermal Sciences 125 (2018) 405–418
Fig. 14. a) Time histories of surface-average Nusselt number and b) maximum value of local Nusselt number versus angle for ζ = 0 and Ur = 6 − 8.
some differences are observed. According to Fig. 11 the vibration am- ΔNu is affected by cylinder displacement and vortex formation. The
plitude at ζ = 0.1 varies between 0.312D to 0.328D which is smaller maximum value of variation is observed For Ur = 4 and ζ =0.05 because
than the corresponding one at ζ = 0.05. Also the number of peaks at of the beating phenomenon. The frequency of variation of the surface-
each displacement period for ζ = 0.1 is lower than corresponding one average Nusselt number is twice of the frequency of lift forces and for
for ζ = 0.05. The surface-average Nusselt number for Ur = 4 and ζ = 0.1 stationary cylinder its value is fNu = 0.012. The values of fNu for
varies from 14.01 to 14.104 which decreases in comparison with those Ur = 3–5 are greater than those of stationary cylinder that indicate the
for ζ = 0.05. Also the maximum value of the surface-average Nusselt higher rate of variation of the Nusselt number in comparison with
number is changed from 14.97 for ζ = 0.05 to 14.104 for ζ = 0.1. The stationary cylinder. For all the damping ratio the maximum value of the
surface-average Nusselt number at Ur = 4 and ζ = 0.1 is lower than the fNu is belong to Ur = 4.
other cases, even than the stationary cylinder. Fig. 12 shows the effect of reduced velocity and damping ratio on
The maximum (Numax ), minimum (Numin ) and the range of variation the non-dimensional vibration amplitude ( A D ) and the total average
(ΔNu = Numax − Numin ) of the surface-average Nusselt number and the Nusselt number (Nu) . For all damping ratios by increasing the reduced
frequency ( fNu ) of its variation for Ur = 3–8 and ζ = 0–0.1 are presented velocity from Ur = 3, the vibration amplitude increases and reaches to a
in Table 5. The maximum and minimum values of ΔNu is belong to maximum value at a specified reduced velocity (Ur =4 for ζ = 0 and 0.01;
Ur = 4 and 8 (for all the damping ratio), respectively. The variation of Ur = 5 for ζ =0.05 and 0.1). Then, the vibration amplitude decreases by
416
E. Izadpanah et al. International Journal of Thermal Sciences 125 (2018) 405–418
417
E. Izadpanah et al. International Journal of Thermal Sciences 125 (2018) 405–418
and liquids to a circular cylinder in crossflow, J Heat Tran 99 (2) (1977) 300–306. [36] J.S. Leontini, M.C. Thompson, K. Hourigan, The beginning of branching behaviour
[34] R. Jiang, J. Lin, Z. Chen, Vibrations of cylindrical objects obstructing a Poiseuille- of vortex-induced vibration during two-dimensional flow, J Fluid Struct 22 (6–7)
type flow, Phys Rev 88 (2) (2013) 023009. (2006) 857–864.
[35] Y. Bao, C. Huang, D. Zhou, Two-degree-of-freedom flow-induced vibrations on [37] M.H. Bahmani, M.H. Akbari, Effects of mass and damping ratios on VIV of a circular
isolated and tandem cylinders with varying natural frequency ratios, J Fluid Struct cylinder, Ocean Eng 37 (5–6) (2010) 511–519.
35 (2012) 50–75.
418