ME360 Lab

Flow Visualization: Stationary Cylinder

Experiment

Visualizing Flow Field of Cylinder Experiment

For Reynolds Number between 140 and 2000
 Table of Contents
1.0 Abstract ………………………………………………….. 2
2.0 Nomenclature ……………………………………….……. 3
3.0 Introduction and Background..…………………………… 3
4.0 Model Installation………….…………………………….. 15
4.1 Stationary Cylinder………………………………. 15
5.0 Experimental Procedure ………………………………….. 17
5.1 Stationary Cylinder……………………………….. 17
6.0 References and Additional Reading ……………………… 24

1.0 Abstract

The purpose of this experiment is to study and visualize the dominant aspects of the flowfield
about a stationary cylinder. The aerodynamics of a circular cylinder have been an area of
significant interest and intensive research since the days of Isaac Newton. An understanding of
the flow about a cylinder can explain the trajectory of a curveball, the collapse of the Tacoma
Narrows bridge, and why power lines hum in the wind. Although the shape of a cylinder contains
no sharp edges or flat surfaces, it is not considered to be an aerodynamically streamlined shape
due to its highly rounded, blunt trailing-edge. The flowfield is dominated by the effect of a very
thin layer of fluid close to the surface called the boundary-layer. The state and condition of the
boundary-layer determines the amount of flow separation about the cylinder and the resultant
forces on the cylinder. This experiment will study the dynamics and behavior of the stationary
cylinder flowfield through visualization.
2.0 Nomenclature
Symbol Description

D Diameter of the cylinder

l Characteristic length
n Kármán vortex street shedding frequency, (Hz)
r Radial distance from the origin
R Radius of the cylinder
Re Reynolds number, ratio of inertial to viscous forces, (Vd/)
S Strouhal number, (S=nV/D)
V Velocity (ft/s)
z Vertical location of dye wand from the tunnel floor

Greek Symbols

 Polar coordinate angle

 Free-stream density, (slugs/ft3)
 Free-stream viscosity, (lbf s/ft2)
 Circulation, (ft2/s)

Subscripts and Superscripts

    Per unit span
 Free-stream condition
r radial component
    Tangential component

3.0 Introduction and Background

Interest in the aerodynamic characteristics of a cylinder dates back to the 1600’s. Isaac Newton
published a paper on the forces effecting a spinning ball in 1671. The flowfield about a stationary
and rotating cylinder provides an introduction into some of the most fundamental aspects of
aerodynamics including flow separation, lift, drag, and the effect of viscosity. Although the shape
of a cylinder contains no sharp edges and is generally smooth, it is not considered to be an
aerodynamically streamlined shape due to its highly rounded, blunt trailing-edge and as a result is
termed a blunt body. The theoretical development of the flowfield about the cylinder will first
consider the idealized, inviscid, potential approach. The effect of viscosity and boundary-layer
separation will then be introduced, followed by rotation and the generation of lift.

The idealized study of the potential flow about a stationary cylinder is very instructive as to the
effects of viscosity and provides a basic understanding of the circular cylinder flowfield. A
potential flowfield is one where the fluid is assumed to be incompressible, irrotational, and
inviscid. Although initially these restrictions might seem very limiting, they allow a closed form
analytical solution of the flowfield to be generated. First, a flow is said to be incompressible if the
density is constant everywhere in the flowfield. A flow is said to be irrotational if fluid elements
have no angular velocity relative to each other and their motion through space is purely
translational. Although fluid elements may follow a curved path, the paths of separate fluid
elements are parallel to each other. For a fluid to be irrotational it must be inviscid. The effect of
viscosity is to impart a shearing force between fluid elements. This shearing force produces
vorticity, or rotation between fluid elements. The assumptions of incompressibility and
irrotationality allow certain flowfields to be described by Laplace’s equation. Laplace’s equation
is a linear, second order partial differential equation. Laplace’s equation is one of the most
fundamental and extensively studied equations in mathematics for which a wealth of solutions
exist which are well understood.1 A complete discussion and derivation of Laplace’s equation and
its relationship and applicability to potential flow solutions is better left to more comprehensive
treatments found in a fluid dynamic text books, and the reader is referred to Anderson1 or Bertin
and Smith2.

A closed form, analytical solution for the potential flowfield about a stationary cylinder can be
derived using Laplace’s equation. Since the equation is linear, the solution for several types of
flows can be combined together to form an overall solution for a given flowfield. For the case of a
cylinder the solution for a uniform parallel flow is added to that of a doublet. A doublet is the
special case of a source and sink pair that leads to a singularity. Again the reader is referred to
either Anderson1 or Bertin and Smith2 for a more complete treatment. Due to the circular
geometry of the cylinder, it is customary to use polar coordinates. A schematic of a cylinder and
the polar coordinate system used is shown in Fig. 1.
V
Vr

Point P(r,)

r
R

V 

Cylinder

Figure 1: Schematic showing cylinder and polar coordinate system.

From Fig. 1, some point P in the two-dimensional flowfield can be located using its radius, r from
the origin and angle . The velocity at point P can be described by its radial component, Vr, and
corresponding tangential component V. Using the solution for a uniform flow and doublet from
Laplace’s equation, the radial and tangential velocity for any point in the potential flowfield about
a stationary cylinder is given by Eqn. 1.

 R2 
Vr 1  r 2 V cosθ 

Eqn. (1)
 R2 
Vθ 1  r 2 V sin θ 
 
The resulting solution for the potential velocity field about a stationary cylinder is exquisitely
simple. Solving Eqn. 1 for a series of r and  combinations one can generate the velocity field
about a circular cylinder of radius R as shown in Fig. 2.

Figure 2: Flowfield of a nonlifting stationary cylinder showing contours of velocity

and streamlines.

From Fig. 2, the velocity field is shown using contours of the local velocity divided by the free-
stream velocity. Streamlines are also shown. A streamline is defined as a curve which is tangent
to the local velocity vector at any point. As the flow approaches the cylinder it senses an
increasing pressure due to the presence of the cylinder and decelerates. The dividing streamline is
shown to go through the stagnation points. Any point in a flow where the velocity goes to zero,
V=0, is called a stagnation point. As fluid particles move along the dividing streamline towards
the cylinder and decelerate, they come to a point of zero velocity on the surface. This point is
termed a stagnation point. The cylinder shows a leading and trailing-edge stagnation point. The
dividing streamline which goes through the stagnation points is called the stagnation streamline.
Flow above the stagnation streamline travels over the upper surface of cylinder where flow below
the stagnation streamline travels over the lower surface. From Fig. 2, the potential flow about the
cylinder is symmetrical about both the horizontal and vertical axes. Above and below the
stagnation streamline the flow accelerates on the upstream half of the cylinder and decelerates on
the downstream half. As the flow accelerates on the upstream half of the cylinder, the pressure
decreases. Conversely, as the flow decelerates over the aft half of the cylinder the pressure
increases. Recall from Bernoulli’s equation that pressure is inversely proportional to velocity.
Since the flow is symmetrical, the pressures on the upper half of the cylinder are equal to the
pressures on the lower half and the resulting net lift force is zero. In a similar fashion, the
pressures on the front of the cylinder are exactly equal to the pressures on the rear of the cylinder
resulting a net drag force of zero. This is most interesting and can be confirmed using Eqn. 1 to
integrate the pressures in the axial direction to calculate drag. Any body immersed in a real fluid
is known to experience a drag force. This paradox between the theoretical zero drag and the
acknowledgement that in real life a finite drag exists was first discussed in 1744 by Jean Le Rond
d’Alembert and is known as d’Alembert’s paradox. This paradox confounded fluid dynamic
researchers through the 1800’s. It was not until the very early 1900’s with understanding and
inclusion of viscous effects that d’Alembert’s paradox was explained.
Viscosity plays an incredibly important and complex roll in fluid dynamics. All fluids have a
finite viscosity. Viscosity is a dissipative factor and creates friction between fluid particles and
retards their motion. Due to the frictional and dissipative effects of viscosity, the velocity of any
fluid directly on the surface of an object is zero. The fact that the fluid velocity is zero directly on
the surface is called the “no slip” condition. The influence of viscosity, however, is confined to a
very thin layer near the surface. As we move off the surface and away from the body the influence
of viscosity lessens. At a point sufficiently far away from the body the effect of viscosity becomes
negligible and the velocity is most influenced by pressure gradients induced by the geometry of
the body. This region near the surface where the viscosity has a significant influence is called the
boundary-layer and was first proposed by Ludwig Prandtl in 1904. The thickness of this layer is
generally very small and governed by a dimensionless parameter called the Reynolds number.
The Reynolds number is a ratio of the importance of inertial forces to viscous forces and takes the
form Re=Vl/, where l is some characteristic length. For a cylinder the characteristic length is
the diameter. For very high Reynolds numbers the inertial effects are dominant and viscous
effects small. The boundary layer for high Reynolds number flows is very thin. As the Reynolds
number decreases the effects of viscosity on the flow field become more important. The
boundary-layer thickness is not constant. The boundary-layer begins with zero thickness at the
leading-edge stagnation point and grows in height with increasing downstream distance along the
surface.

The flow within the boundary-layer can take one of three forms: laminar, turbulent, or transitional.
Boundary-layers usually begin at the stagnation point as “laminar” and transition to “turbulent” at
some point downstream. When the boundary-layer is in its laminar form, it is relatively thin with
limited mass and momentum transfer. The momentum transfer in a laminar boundary-layer is
accomplished through molecular diffusion (i.e. molecule-to-molecule, low energy). There is a
relatively low velocity gradient near the wall which produces relatively low skin friction. A
turbulent boundary-layer, however, is much thicker with significant mass transfer. Momentum
transfer in a turbulent boundary-layer is a bulk transfer process, and the flow is considerably more
“energetic” than the laminar case. The turbulent momentum transfer in is orders of magnitude
greater than that present in a laminar boundary-layer. There are higher velocities near the wall and
thus higher skin friction. The flow transitions from laminar to turbulent (through a transitional
state) due to instabilities inherent in the laminar flow.

For the flow about a circular cylinder, the effect of viscosity and the presence of a boundary layer
have a significant and marked effect on the actual real world flowfield. As the flow experiences a
decreasing pressure over the front of the cylinder it accelerates and the boundary-layer remains
“attached”. As the flow reaches the top, or bottom of the cylinder at its maximum velocity, it sees
an increasing pressure on the back side of the cylinder and begins to decelerate. Due to the
bluntness of the downstream side of the cylinder, the pressure increases rapidly and the flow is
forced to decelerate rapidly. The flow cannot negotiate the rapid pressure increase and
deceleration and “separates” from the surface. The separating boundary-layer forms a free-shear
layer. The shear layer is drawn back to the rear of the cylinder due to the low pressure created by
the separation, which causes the shear layer to roll up and form a vortex. The vortex separates and
moves down stream. The boundary-layer separation point is greatly dependent upon the state of
the boundary-layer and Reynolds number. If the boundary-layer is laminar, the flow has little
energy and cannot negotiate the adverse pressure gradient beginning on the upper, or lower
portion of the cylinder very well and separates earlier. The more energetic turbulent boundary-
layer, however, can remain attached longer which moves the separation point aft. Increasing the
Reynolds number generally forces the flow to transition from laminar to turbulent earlier on the
front of the cylinder and allows the flow to remain attached longer. For low Reynolds numbers
(less than approximately ReD=30,000), the flow is laminar up to the separation point and
separation occurs at 100(80 from the leading-edge stagnation point on the front of the
cylinder).2 At higher Reynolds numbers, ReD>1.5x106, where the front of the cylinder is turbulent,
the separation occurs further aft at 60<<70 on the rear of the cylinder.2 A schematic of the
circular cylinder flowfield depicting the boundary-layer, separation, and vortex roll up is shown in
Fig. 3.
Separation Point

Stagnation Point

V

Streamlines

Boundary-Layer Vortex Roll Up

Figure 3: Schematic of the boundary-layer, separation, and vortex roll up around

a circular cylinder.

From Fig. 3, the vortices which separate from the upper surface rotate clockwise with the lower
surface vortices rotating counter clockwise. These vortices shed in an alternating pattern. At a
distance away from the body these shed vortices form a regular pattern. The pattern is called a
Kármán vortex street after a paper describing the behavior published by Theodore von Kármán in
1911. A regular Kármán street is observed only in the range of Reynolds numbers from about
ReD=60 to 5000. Below this range the flow in the wake is laminar and above this range there is
complete turbulent mixing in the wake.3 Two photographs showing this alternating separation
pattern at different Reynolds numbers using slightly different visualization techniques are shown
in Fig. 4.

a) ReD=140 b) ReD=10,000
Figure 4: Kármán vortex street shedding from a circular cylinder at two different Reynolds
numbers.4

From Fig. 4, the flowfield is quite different from the theoretical potential flowfield shown in Fig.
2. The alternating vortex street is extremely regular. The difference in the amount of turbulence
present in the wake is clearly evident between the two Reynolds numbers shown in Fig. 4a and b.
Although the mean, time averaged lift is zero for the stationary cylinder, the alternate upper and
lower surface shedding creates an oscillatory force on the body. In reference to the potential
flowfield shown in Fig. 2, the non-separating symmetrical flowfield does actually exist for some
special cases. These include extremely low Reynolds number cases or creeping flow cases, but
these are special cases and will not be discussed here as most real world flows exhibit the Kármán
vortex street shown in Fig. 4.

The shedding of a Kármán vortex street is not special to a cylindrical geometry. Any blunt body
in the correct Reynolds number range will shed a vortex street. The frequency of the shedding is
extremely regular and dependent upon the velocity and a characteristic length of the body. The
shedding frequency is usually reported in a dimensionless form:

nL
S Eqn. (2)
V

This dimensionless frequency, S, is called the Strouhal number and is dependent only upon the
Reynolds number. The variable n in Eqn. 2 is the number of sheds per second and includes both a
lower and upper surface shed vortex pair. For a cylinder, the characteristic length is the diameter.
A plot showing the Strouhal number as a function of the Reynolds number is given in Fig. 5.

Strouhal Number For a Kármán Vortex street

0.30
Strouhal Number, S=(nD/V)

0.28
0.26
0.24
0.22
0.20
0.18
0.16
0.14
0.12
0.10
10 100 1000 104 105 106
Reynolds Number

Figure 5: The Strouhal number, S, as a function of Reynolds number for a

Kármán vortex street.

An understanding of the fluid dynamics of blunt body shedding and the Kármán vortex street can
explain several observed natural behaviors. The dance of a deftly thrown “knuckle” ball. The
“singing” of high tension power lines in a strong wind is a result of the vortex street shed from the
wires. At the right wind velocity and power line diameter the shedding frequency is in the audible
range. Several cloud formations observed from space exhibit a vortex street as the cloud systems
move past mountainous terrain or an islands. Two NASA photos obtained from the space shuttle
are shown in Fig. 6 and show the vortex street formed on the leeward side of islands.

Figure 6: NASA space shuttle photos of clouds forming a Kármán vortex street
on the leeward side of islands.

The most famous example of a Kármán vortex street involves the Tacoma Narrows bridge in
Washington state. Known for its tendency to sway in windstorms the Tacoma Narrows bridge
became known as Galloping Gertie. Oscillatory forces due vortex shedding of the bridge deck in
the wind would cause the deck to sway. On November 7, 1940 a large windstorm brought the
Kármán vortex street shedding frequency into the same range as the natural structural frequency of
the bridge which resulted in resonance and a catastrophic failure of the structure. Photographs of
the swaying bridge deck and collapse are shown in Fig. 7.

Figure 7: Photographs of the oscillatory swaying and failure of the Tacoma Narrows bridge on
November 7, 1940 as a result of unsteady forces induced by the bridge deck shedding a Kármán
vortex street in high winds.

The purpose of the cylinder experiment is to visualize and study the flowfield dynamics of the
stationary cylinder. This involves observing the formation and shedding of a Kármán vortex
street.
4.0 Model Installation
4.1 Stationary Cylinder Installation

The stationary cylinder model should be installed in the wall mount model support using the
angle-of-attack adjustable push-rod. The wall mount is composed of two major pieces, one for
each side of the tunnel. The most complicated piece, the one containing both the hand crank and
the push-rod, is mounted on the side of the tunnel opposite the speed control. Two screws attach
each side to the tunnel, which are accessed through holes in the top of the rectangular, black,
mounting frame. There are two possible mounting positions for the wall mount. The stationary
cylinder should be mounted in the set of holes towards the front of the test section. The forward
mounts allow the wake behavior of the cylinder to be viewed. The stationary cylinder is inserted
into the wall mount by sliding the stainless steel pivot pins into the ball bearings at the bottom of
the wall plates. The experiment must be mounted in the wall mount before the mount is attached
to the tunnel. The push rod should be connected by sliding it over the rear control pin on the
model, also prior to installation in the tunnel. A photograph showing this installation is shown in
Fig. 8.

Figure 8: Photograph showing stationary cylinder in the forward

wall mount location.

The angle-of-attack push-rod should be adjusted so that the upper and lower dye ports are
equidistant from the horizontal axis of the cylinder. Only two colors of dye are used for the
stationary cylinder. The three upper dye ports are interconnected through a single manifold, as are
the three lower dye ports. The upper and lower ports should be connected to dye tubes of a
separate color. The separate colors help to visualize the Kármán street through color distinctions
in the upper and lower surface shed vortices. It is useful to tape the dye tubes to the upper
crossbar of the tunnel after they exit the water as shown in the photograph. The stationary
cylinder set-up can be installed when the tunnel is either empty or full.

Care should be taken when filling the tunnel with water to remove all air bubbles in the
honeycomb flow conditioners before testing begins. The presence of air bubbles in the
honeycomb will adversely affect the flow quality in the test section. Air bubbles are best removed
during the filling process by directing flow from the hose being used to fill the tunnel through the
honeycomb from the inlet baffle side. Any spraying mechanism, however, can be used to remove
air bubbles. It should be noted that as the tunnel runs, more oxygen tends to come out the solution
so air bubbles may reform periodically.

After filling the tunnel, the dye tubes should be cleared of air. This is accomplished by first
pressurizing the accumulator bottle with 5-6 pumps of the handle. The tunnel should also be set to
a nominal speed value of around 4in/s. This will help in determining which ports have air
bubbles. The stationary cylinder has a set of three ports on the upper surface and three ports on the
lower surface. Each of these sets are controlled by a single needle valve. Open the first needle
valve and watch dye flow from the valve to the cylinder model. Dye will usually exit only one of
the 3 ports initially. Place a finger over the port with dye emanating from it. This will force the
air bubbles out of another port and allow the dye to flow. This process should be repeated for the
remaining port until dye is flowing from all 3 ports for a given valve. After closing the first
needle valve, the second needle valve should be opened with the same clearing procedure used for
first needle valve.

5.0 Experimental Procedure

This experimental procedure will discuss two visualization experiments using the stationary and
rotating cylinder models. The purpose of the stationary cylinder experiment is to visualize and
study the dynamics and of the Kármán vortex street. Although other experiments can be
performed with the stationary cylinder model, only the effect of Reynolds number and tunnel
speed on the shedding frequency of the stationary cylinder will be discussed in detail here.

5.1 Stationary Cylinder

This experiment will visualize and calculate the Kármán vortex street shedding frequency of the
stationary cylinder as a function of the tunnel speed. The shedding frequency will then be
compared to the frequency predicted by the Strouhal number. The diameter of the stationary
cylinder is 1.0”. A plot of the cylinder Reynolds number versus tunnel speed is given in Fig. 9.
Stationary Cylinder Reynolds Number vs. Tunnel Speed
5000

4000
D
Reynolds Number

3000

2000

1000 For Water at 68° F

Re Based On Cylinder Diameter=1.0"
0
0 1 2 3 4 5 6
Tunnel Speed (in/s)
Figure 9: Stationary cylinder Reynolds number versus tunnel speed.

From Fig. 9, the Reynolds number range between 1.0in/s and 5.3in/s is within the range of
Reynolds numbers (ReD=60 to 5000) to observe a regular and well defined Kármán vortex street.
The experiment is best performed starting at the maximum speed then decreasing (5.3in/s then
4.0in/s … 3.0in/s …etc.).
For viewing of the vortex street it is best to have both upper and lower dye ports running. A high
dye injection rate will produce the best visualization of the street. The high injection rate,
however, will cloud the water quickly. The shedding frequency is based on an upper and lower
surface pair of vortices being shed; hence, one cylce of shedding includes both an upper surface
and lower surface vortex. A photograph showing the Kármán vortex street shed from the cylinder
at a tunnel speed of 5.3in/s is shown in Fig. 10.

Figure 10: Photograph of the Kármán vortex street shed from the
cylinder at a tunnel speed of 5.3in/s

When calculating the shedding frequency it is helpful to turn either the upper or lower surface
ports off. This will increase the amount of time before the water becomes clouded and also eases
determination of the shedding frequency. A stop-watch or timing device is required to calculate
the shedding frequency. To calculate the shedding frequency at a set tunnel speed, a given number
of either upper or lower surface sheds should be timed. For example, use the stop-watch to
determine the amount of time required to shed 10 lower surface vortices. The shedding frequency
is then equal to the number of sheds divided by amount of time required. This process should be
repeated several times for each tunnel speed and the results averaged. Typical results versus
tunnel speed are given in Table 1.

Time For 10 Cycles:

10 Sheds
Shedding Frequency, n 
Average Time

Tunnel Speed ReD Run #1 Run #2 Run #3 Run #4 Run #5 Avg. Time Frequency
(in/s) (s) (s) (s) (s) (s) (s) (Hz)
2.0 1274 23.14 22.63 23.12 23.34 22.86 23.02 0.43
3.0 1911 15.78 15.20 15.44 15.44 15.06 15.38 0.65
4.0 2548 11.87 11.88 12.07 11.60 11.32 11.75 0.85
5.3 3376 9.03 8.60 8.72 8.93 8.74 8.80 1.13

Table 1: Typical shedding results versus tunnel speed.

From Table 1, the shedding frequency is seen to increase with increasing tunnel speed as expected.
Over the Reynolds number range shown in Table 1, the Strouhal number S, from Fig. 5 is constant
at 0.21. From Eqn. (2), the Strouhal predicted shedding frequency is equal to n=SV/D. Using
S=0.21, the predicted Strouhal shedding frequency versus the experimentally measured values are
plotted in Fig. 11.
Kármán Vortex Street Shedding Frequency
Comparison of Strouhal Predicted and Measured Shedding Frequency
2.0
Shedding Frequency, n (Hz)

Strouhal Shedding Frequency

1.5 Experimental Shedding Frequency

1.0

0.5
Circular Cylinder: 1.0" Diameter
S =0.21, n=S V /D

0.0
0 1 2 3 4 5 6
Tunnel Speed (in/s)
Figure 11: Comparison of Strouhal predicted and measured Kármán vortex
street shedding frequency.
From Fig. 11, the experimentally measured values of the shedding frequency compare very well
with the Strouhal predicted values. Since the Strouhal number is constant over this Reynolds
number range the shedding frequency is linear with increasing tunnel speed.

6.0 References and Additional Reading

1
Anderson, J. D., Fundamentals of Aerodynamics, McGraw-Hill, New York, 1984.
2
Bertin, J. J. and Smith, M. L., Aerodynamics For Engineers 2nd ed., Prentice Hall, New Jersey,
1989.
3
Schlichting, H., Boundary-Layer Theory 7th ed., McGraw-Hill, New York, 1987.
4
Van Dyke, M., An Album of Fluid Motion, The Parabolic Press, Stanford, California, 1982.

Thwaites, B., Incompressible Aerodynamics: An Account of the Theory and Observation of the
Steady Flow of Incompressible Fluid past Aerofoils, Wings, and Other Bodies, Dover
Publications, Inc., New York, 1960.

White, F. M., Viscous Fluid Flow, 2nd ed., McGraw-Hill, Inc., New York, 1991.