Thermophysics and Aeromechanics, 2005, Vol. 12, No.

VISUALIZATION OF FLOW AROUND

SPHERE FOR REYNOLDS NUMBERS
BETWEEN 22 000 AND 400 000
1 2
V. BAKIĆ and M. PERIĆ
1
”VINČA” Institute of Nuclear, Belgrade, Serbia and Montenegro
2
Computational Dynamics, Nűrnberg, Germany
(Received )

The experiments on visualization of complex flows in a wind tunnel and water channel have been con-
ducted. The structures of wake, separation, and laminar-turbulent transition were determined and investigated
in a separated boundary layer downstream of a sphere for Reynolds number from 22 000 to 400 000

INTRODUCTION

The problem of flow visualization in fluid mechanics is of considerable interest, es-

pecially the visualization of complex flows, which is used to verify existing theories.
Most objects of practical interest belong to the class of bluff bodies. The sphere is a typi-
cal example of bluff body without sharp edges which usually fix the flow separation,
thus backing to flows similar to those found around many vehicle types. In the present
work, the investigations of visualization of a flow around the sphere are carried out for
Reynolds number from 22 000 to 400 000
The main characteristic of flow around sphere is the existence of turbulent wake
with recirculation, which has a dominant effect on the drag and lift of the sphere. The
extent of this region is dependent on the shape, orientation, and size of the body, the ve-
locity and viscosity of the fluid, and may be influenced by a wide variety of small flow
disturbances, which may be generated in various ways.
Flow past spheres has been the subject of numerous experiments, especially at low
Reynolds number. Depending on the Reynolds number there is a variation of the wake
structure behind a sphere. Taneda [1] used dye visualization methods to study the wake
of a sphere for 5<Re<300. He determined that when the Reynolds number is lower than
about 20, the flow is laminar everywhere and separation does not occur. According to
him, separation from the rear of a sphere occurs at Re ≈ 24 and results in the generation
of an axisymmetric vortex ring. This stationary vortex ring is visible at Reynolds num-
bers between 20 and 400 [1, 2]. In experiments of [3] and [4] the same vortex ring struc-
ture were obtained, however, the rings remained stable and axisymmetric only up to Re
= 210. At this Reynolds number the wake makes a transition from a single-thread wake
to the double-thread one. In the range 210 < Re < 270 the wake is steady and non-
axisymmetric (double-thread wakes). At Re = 270 the double-thread wake was observed
to become unstable, and vortex loops began to shed from the sphere as so-called hairpin
vortices, forming a laminar wake. When the Reynolds number is further increased to
around 800, the vortex loops diffuse very rapidly, and the wake flow becomes turbulent [5].

© V. Bakić and M. Perić, 2005

307
 The first known visualization of the flow past a sphere at high Reynolds numbers
was done by Wieselberger [6]. He extended Prandtl's [7] experiments and showed that
transition to turbulence in the boundary layer substantially decreased the wake width.
A more detailed picture of flow past a sphere was obtained in [8]. For Reynolds numbers
higher than 1500, it was found that vortices were shed in rings whose axes were parallel
to the direction of flow in free stream. For Reynolds number higher than 2000 this flow
pattern became well established. Effects of support position and turbulence intensity in
the free stream on the flow near the surface of a sphere were analyzed in [9]. The oil-
flow method was used for flow visualization. It was found that, for low turbulence inten-
sity, the separation occurs at about αs = 81° from the stagnation point for a subcritical
flow, and at about αs = 102o for a supercritical flow. The same visualization techniques
were used in [10]. It was that the boundary layer separates at αs = 80° for Re = 23 000
and position of separation remains almost unaltered over the range 104 <Re < 3.5·105.
When the Reynolds number is raised to about 350 000, the position of laminar separation
changes from 92° to 110°, the re-attachment from 107° to 127°, and the turbulent separa-
tion from 123° to 147°. The existence of a critical Reynolds number beyond which the
drag drops dramatically (often called “drag crisis”), the separation is postponed, and the
size of the wake is substantially reduced, although the critical value of Reynolds number
varies between 250 000 and 350 000 from experiment to experiment. It is also well un-
derstood that an early transition of boundary layer from laminar to turbulent (e.g.,
through surface roughness or tripping devices) leads to later separation and drag reduc-
tion even at much lower Reynolds number ⎯ as low as 15 000.
Many papers have been published concerning the vortex shedding from a sphere,
[2, 5, 8, 10, 11]. Although the results are fairly different from paper to paper, in all these
studies the existence of two Strouhal numbers, the low-mode and the high-mode one, has
been documented. The authors of [11] have tried to explain these two Strouhal modes
with small-scale instability of the separation shear layer and with large-scale instability
of the wake.
The subject of this study is to try to determine characteristics of the transition proc-
ess of the separated boundary layer and wake structure of a sphere at Reynolds numbers
between 22 000 and 400 000 using flow visualization methods.

1. EXPERIMENTAL APPARATUS AND METHODS

The flow visualization has been carried out in a wind tunnel and in a water channel.
The wind tunnel used can be operated in two modes, with a closed circuit flow or with
an open outlet behind the impeller and an inlet in front of the settling-chamber. In our
experiments the wind tunnel was operated with open circuit flow. The nozzle opening
was 1.75 × 1.05 m. The working section is preceded by a 10:1 contraction with three
turbulence-suppressing screens upstream of the contraction and was 5.5 m long. The
tunnel was operated in the range from U0 = 2 to 38 m/s. The turbulence level in the free
stream was 0.89 %. The sphere with D = 0.165 m in diameter was supported from the
rear by a sting, which had a length 5D and a diameter d = 0.12D.
The dimension of the water-channel cross section used in another set of visualiza-
tion studies was 2.1 × 1.1 m. The sphere, with the same diameter as in the wind tunnel,
was attached to a towing carriage and was supported from the rear by the same sting as
in the wind tunnel. The blockage ratio was 1.15 % for the wind tunnel and 0.8 % for the
water channel. The carriage was carefully designed to minimize mechanical vibrations
when towed. The motion was initially accelerated and the sphere traveled a couple of
meters before the final steady towing speed was reached. The range of speed of the car-
riage was U0 = 0.03 to 2.3 m/s. The Reynolds number based on sphere diameter and ve-
locity of air in the wind tunnel, or speed of towing carriage in water channel, could be

308
varied from 2.2⋅104 to 4⋅105 in the wind tunnel and from 5⋅103 to 4⋅105 in the water
channel.
Flow visualization in the wind tunnel has been conducted by injecting smoke in the
separation zone through holes in the supporting sting. The smoke particles were illumi-
nated by a thin sheet of light created with a coherent laser beam and a cylindrical lens.
The light source was a 10 W argon-ion laser operated in the multi-line mode, under
which the smoke particles had a bright green color.
In the water channel dye was introduced into the boundary layer through holes with
a diameter dh = 0.8 mm at two locations positioned at angles 45° and 135°, measured
from the front stagnation point. Images of the smoke traces and dye were recorded
with a MotionBlitz digital high-speed camera (500 frames/s) with a resolution of
256 × 228 pixels. Up to 2964 frames could be continuously recorded (6 seconds of real
time). Digital images (bitmap-files) can be processed to produce animation in various
formats (avi, mpg etc.). Digital color photographs with a resolution of 1024 × 768 pixels
have also been taken, as well as color VHS videos at 50 frames/s.

2. FLOW VISUALIZATION IN A WIND TUNNEL

The smoke particles were illuminated in a thin laser light sheet which could be posi-
tioned in the (x, y), (x, z) or (y, z) plane, x being in the flow direction.
Figure 1 shows visualization of the flow around sphere in the (x, y) plane for Rey-
nolds numbers 22 000 and 50 000. The observed near-wake recirculating region is large
and the wake performs a progressive wave motion. At the lower Reynolds number large
coherent structures are clearly visible. The videos and the photographs show the roll-up
of the separated shear layer and pairing of vortices. The roll-up and the pairing processes
can be more clearly observed in the water-channel visualization (see next section) for
Reynolds numbers up to 175 000. At higher Reynolds numbers, the turbulence structure
in the mixing layer is so fine that roll-up and pairing processes were not visible with the
present camera allowing 500 frames/s.
In order to determine the three-dimensional structure of the wavy wake, the laser
sheet was positioned in the plane normal to the flow direction, (y, z) plane, 5.5D down-
stream of the sphere. Figure 2 shows the changes of vortex patterns during one charac-
teristic time interval D/U0 = 82.5 ms, where D = 0.165 m is the sphere diameter and
U0 = 2 m/s is the free-stream velocity, and for Re = 22 000. It was found in [2], by ana-
lyzing hot wire signals, that the vortex separation points rotate around the sphere for
Reynolds numbers above 6 000. This suggests a single helical wake configuration.
The existence of a helical vortex system does not respect Thomson's circulation
theorem. This theorem states that if the incident flow is free from vorticity, the net flux
of the vorticity across planes perpendicular to the wake axis must be zero. This led Pao
and Kao [12] to propose a complex double helix which unwinds in the opposite sense.
However, from pictures in Fig. 2 and from corresponding films, no regularity of rotation
of vortices could be observed, and also for Re = 50 000 [13]. The wake is not completely
filled by smoke since it was introduced through holes at the upper and lower side of the
sting. The visible smoke in the wake cross section shown in Fig. 2 changes its shape and

Fig.1. Visualization of flow around sphere using smoke in a wind tunnel

for Re = 22 000 (left) and Re = 50 000 (right).

309
Fig.2. Structure of the sphere wake in a plane normal to the wake axis 5.5D downstream from
sphere, and for Re = 22 000.

310
 Fig. 3. Long exposure visualization of the flow around sphere for Re = 50 000 and Re = 400 000.

moves left-right and up-down, but no regularity could be identified in this motion over
the continuous recording of 6 s of real time. The conclusion about no systematic
arrangement and rotation of vortices in the plane normal to wake axis agree with results
[5, 10 and 11].
Whit the Reynolds numbers increasing from sub-critical 22 000 to critical 350 000, the
point of boundary layer separation moves from about 82° to 132°. At the same time the
position of transition moves towards the sphere and the rate of entrainment tends to in-
crease. Once the transition point comes close enough to the sphere wall as the Reynolds
number is increased, itcauses the boundary layer to become turbulent further down-
stream; thus, laminar separation is often followed by turbulent re-attachment after a short
distance, with turbulent boundary layer remaining attached before it separates again fur-
ther downstream. Hence, the near-wake recirculating region shrinks suddenly and the
wavy motion of the wake vanishes. The vanishing of progressive wavy motion of the
wake behind the sphere is obvious at Reynolds number 400 000. The transition to turbu-
lence in the boundary layer substantially decreases the wake size, as can be seen
in Fig. 3.

3. FLOW VISUALIZATION IN A WATER CHANNEL

The characteristics of boundary layer separation and laminar turbulent transition

have also been investigated in a water channel at Reynolds numbers from 5 000 to
380 000. The boundary layer was colored by dye. Due to the difference in kinematics
viscosity, the same Reynolds number is achieved at lower velocity than in air, allowing a
better time resolution with the camera at 500 frames/s.
The separation, laminar-turbulent transition and growth of a separated shear layer de-
pend on many factors and were subject of many experimental and theoretical investiga-
tions. Figure 4 shows these processes for different Reynolds numbers. The shear layer
formed after the separation of boundary layer is subject to Kelvin ⎯ Helmholtz instabil-
ity, which causes the roll-up of vortex rings at nearly regular time intervals. It appears
that for all cases with laminar separation, the first two vortex rings seen are laminar; the
third one is already turbulent and further downstream no regular structure can be ob-
served any more. From these pictures and corresponding film, it is obvious that, with
increasing Reynolds number, the turbulence structures in the mixing layer become finer
and transition of the separated shear layer occurs closer to the separation point. For

311
Fig. 4. Separation and transition of a boundary layer over sphere for different Reynolds numbers.

Re = 250 000, the movie shows that at the separation point the boundary layer is not
completely laminar. There are intervals of time when the boundary layer is turbulent and
separation occurs later. At the critical Reynolds number around Re ≈ 300 000, the transi-
tion from laminar to turbulent boundary layer occurs before separation and the turbulent
boundary layer separates much later, at an angle close to 135°. It appears that, after be-
coming turbulent, the rolled-up vortices start interacting with the wall when the Rey-
nolds number is high enough so that the transition happens close to the separation point.
This can be observed from Re = 200 000 onwards; this causes the separated boundary
layer to re-attach and become separated further downstream. Such re-attachment and
subsequent separation can be seen in Fig. 4 at Re = 250 000.
Figure 5 shows that besides the roll-up of vortices, pairing of vortices is also pre-
sent in the free shear layer of a sphere. The roll-up includes the folding, rolling ,and
break-up of a continuous vortex sheet into discrete vortices, while pairing means a se-
quential amalgamation of these vortices into larger vortex structures. If experiment is
carefully controlled and prepared, it is possible to more than two vortices combine to
form still larger vortices [14]. The roll-up process of a shear layer occurs at a location at

Fig. 5. The process of vortex pairing at Re = 50 000..

312
Fig. 6. Distribution of low-mode and high-mode Strouhal numbers as a function of Reynolds
number.

which the fundamental frequency (natural instability frequency) attains its maximum
amplitude and is followed by the generation of a sub-harmonic frequency, as described
in [15]. The roll-up and pairing of vortices observed here is similar to the processes
found in mixing layers.
Successive amalgamation of pairs of neighboring vortices marked by dye is evident
in Fig. 5. During the pairing, vorticity is continuously being redistributed into larger and
larger eddies, whose wavelength and size double at each interaction. Winant and Brow-
and [16] have shown that merging of vortices was indeed the primary process governing
the streamwise growth of the mixing layer thickness. However, sub-harmonics alone can
not explain the complicated nature of shear layer development. According to [17] the
location at which pairing occurs coincides with the location at which the amplitude of
the sub-harmonic mode frequency attains a maximum. The value of the natural instabil-
ity frequency of the shear layer for different Reynolds numbers is obtained from the pre-
sent experiments by counting roll-up of vortices with time in the high speed digital
video. These results, together with results from other researchers are shown in Fig. 6.
This figure shows the distribution of low-mode and high-mode Strouhal number versus
Reynolds number Str = fD / U 0 , where f is the frequency, D diameter of sphere and U0
the free-stream velocity.
Experimental investigations of frequencies in a sphere wake [11], obtained from
spectra of velocity fluctuations detected by a hot-wire in unexcited flow, showed the
presence of several dominant frequencies: low frequency, a higher frequency, and often
the sub-harmonic of the higher frequency. The higher frequency (high-mode of Strouhal
number), was detectable only in the region of the wake immediately downstream of the
sphere, but low frequency (low-mode of Strouhal number) could be observed in a much
larger region. The sub-harmonic of the higher frequency obtained in this experiment was
produced by pairing of vortices, which is obvious from our flow visualization.
Kim's and Durbin's [11] data does not show a unique power-law dependence of fre-
quency of the Kelvin ⎯ Helmholz shear layer instability fKH on Reynolds number: for
103 < Re < 104 frequency scales approximately with Re0.75 while at higher Reynolds
numbers the exponent is closer to 0.66. Our experiments, for Reynolds number
Re > 22 000, give a power law:

313
 S h = 0.0039 Re0.695 . (1)
This result is in good agreement with Kelvin ⎯ Helmholtz frequency of shear layer
instability for the flow around a circular cylinder obtained in [18]. This frequency is usu-
ally normalized by the shedding or Karman frequency fK of the large vortices in the
wake. They found that the frequency scales for Re<105 with power law:
f KH
= 0.0235 Re0.67 , (2)
fK
where the exponent and the constant are found by analyzing experimental data. In [17],
data analysis indicates a higher exponent for Re > 105:
f KH
≈ Re0.7 . (3)
fK
According to our flow visualization and an experimental investigation by Kim and Dur-
bin [11] we can conclude that the measured higher frequency, high mode of Strouhal
number, corresponds to the instability frequency of the shear layer. However, flow visu-
alization showed no evidence of discrete vortex shedding behind the sphere; for that rea-
son we can agree with [11] that the measured low frequency, the low mode of Strouhal
numbers, corresponds to the large scale (sinus) instability of the sphere wake.

4. CONCLUSIONS

Visualization experiments were carried out in a wind tunnel and in a water channel
to determine the wake structure, separation, and laminar-turbulent transition of a sepa-
rated boundary layer behind a sphere. The range of Reynolds numbers covered varied
from 22 000 to 400 000 in a wind tunnel and from 50 000 to 300 000 in a water channel.
For flow visualization the smoke and dye methods were used.
The results of flow visualization indicated that the wake structure of a sphere, and
the vortex configuration, is more complicated than the simple helical or double helical
vortex structure that were reported in some earlier publications. Films and photographs
showed the roll-up of the separated shear layer and the pairing of vortices, indicating that
besides the natural instability frequency which defines the high-mode Strouhal number,
there also exists a sub-harmonic of this frequency. This agrees with experimental inves-
tigations of frequencies in a sphere wake [11], based on spectra of velocity fluctuations
obtained from hot wire signals. Results of flow visualization also showed that there is no
vortex shedding like in the case of a cylinder. But this doesn't mean that there is no vor-
tex shedding. Particle image velocimetry will answer this question. With increasing
Reynolds number the turbulent structures in the mixing layer become finer, and transi-
tion of the separated shear layer occurs closer to the separation point. The near wake
recirculating region shrinks suddenly at critical Reynolds number and the wavy motion
of the wake behind sphere vanishes in the supercritical range of Reynolds numbers.

Acknowledgments. The authors gratefully acknowledge to Prof. Peter Bradshaw

for valuable advice during this work and Mr. Dita Brikman for his help during experi-
ments. Experiments were done at the Technical University Hamburg Harburg, Germany.

REFERENCES
1. S. Taneda, Experimental investigation of the wake behind sphere at low Reynolds number, J. of Phys. So.
Japan, 1956, Vol. 11, P. 1104–1108.
2. E. Achenbach, Vortex shedding from spheres, J. Fluid Mech., 1974, Vol. 13, P. 209-221.
3. R.H. Magarvey and L.R. Bishop, Transition ranges for three-dimensional wakes, Canadian J. Phys., 1961,
Vol. 39, P. 1418–1422.

314
4. I. Nakamura, Steady wake behind a sphere, Phys. Fluids, 1968, Vol. 19, P. 5–8
5. H. Sakamoto and H. Haniu, A study on vortex shedding from a spheres in a uniform flow, J. Fluid Eng.,
1990, Vol. 112, P. 386–392.
6. C. Wieselberger, Der Luftwiderstand von Kugeln, Zeitschrift fűr Mechanics, 1914, Vol. 5, P. 140–144.
7. L. Prandtl, Der Luftwiderstand von Kugeln,Gőtttingen Nachrichten für Mathematik und Phys., 1914,
S. 177–199.
8. W. Mőller, Experimmentelle Untersuchung zur Hydrodynamik der Kugel, Physik Zeitschrift, 1938, Vol. 39,
P. 57–80.
9. G.D. Raithby, E.R.G. Eckert, The effect of support position and turbulence intensity on the flow near the
surface of a sphere, Wärme und Stoff, 1968, Vol. 1, P. 87–94.
10. S. Taneda, Visual observations of the flow past a sphere at Reynolds numbers between 104 and 106.
J. Fluid Mech., 1978, Vol. 85, P. 187–192.
11. J.H. Kim and P.A. Durbin, Observations of the frequencies in a sphere wake and of drag increase by
acoustic excitation, Phys. Fluids, 1988, Vol. 31, P. 3260–3264.
12. H.P. Pao and T.W. Kao, Vortex structure in the wake of a sphere, Phys. Fluids, 1977, Vol. 20, No. 2,
P. 187–191.
13. V. Bakic, Experimental investigation of turbulent flows around a sphere, Ph. D. Thesis, Technical Univ.,
Hamburg-Harburg, Germany, 2003.
14. J.H. Ferziger, Direct and large eddy simulation of turbulence, Lecture note: Understanding, Modelling and
Simulation of Turbulence, Hamburg, 1999.
15. C.M. Ho and P. Huerre, Perturbed free shear layers, Annual Rev. Fluid Mech., 1984, Vol. 16,
P. 365–424.
16. C.D. Winant and F.K. Browand, Vortex pairing, the mechanism of turbulent mixing-layer growth at
moderate Reynolds numbers, J. Fluid Mech., 1974, Vol. 63, P. 237–255.
17. C.M. Ho and L.S. Huang, Subharmonics and vortex merging in mixing layers, J. Fluid Mech., 1982,
Vol. 119, P. 443–473.
18. A. Prasad, H.K. Williamson, The instability of the shear layer separating from a bluff body, J. Fluid
Mech., 1997, Vol. 333, P. 375–402.

315