Professional Documents
Culture Documents
ABSTRACT
It is well known that at certain intensities of wind flow velocity acting on a structure, the
response of the latter in the transverse direction of flow is induced by alternating vortices. The
most important parameters that contribute to the cross wind response are: the intensity of
turbulence, the duration of the gusts and the magnitude of the wind speed in the floe direction.
Nevertheless, there is not exist an unified criteria to evaluate the cross-wind response and
several methodologies has been proposed. In this paper, two of those methodologies for the
calculation of total displacements in the transverse direction of the wind flow, are analyzed and
described. Also, some procedures proposed in different international design codes to evaluate
the cross-wind response of cylindrical structures, are applied to the case of a chimney for
comparison. Finally, it is concluded that the method that best estimates the cross response due to
vortices, with respect to the results reported in the literature and obtained experimentally in full
scale prototypes, is the one proposed in the Danish code.
INTRODUCTION
A high speed flow passing around a body with arbitrary shape (Figure 1) produces a wake vortex
on the back with alternating movement from one side to another, a phenomenon known as wake
vortices Bnard - von Karman who are credited with this observation.
In many structures not only the dynamic response in the along-wind direction is
important, but also the response due to vortex shedding in the leeward side that produce crosswind displacements and which must be considered in the total response. The longitudinal
vibrations of the structure are caused by the natural turbulence of the wind, but the cross-wind
ones are caused, besides the natural turbulence of the wind, by vortex shedding. This
phenomenon can be presented in structures like lattice towers with a high solidity ratio
( 0 .5) , in rectangular prismatic structures like buildings (Figure 2) and slim bridges, but
mainly it appears in structures with cylindrical cross-sectional section like cylindrical towers,
poles, masts and chimneys.
A great number of failures in structures, mainly with circular cross-section [1], have been
reported in literature. In steel chimneys, the cross-wind effects produce important displacements
perpendicular to the wind direction and these at the time, increase the cross-wind base
overturning moment in the foundations, and therefore in the joint stresses, as they are the
anchors, nuts and the base plate. Because the vortex shedding produced fluctuating forces, this is
translated in a series of cycles of load that can generate a fatigue failure in the material. Some
examples are shown in Figure 3.
vortex shedding method. Before describing these two methodologies, it is necessary to remark
some concepts described in the following.
The vortex shedding frequency depends on the body shape, flow velocity, surface
roughness and the flow turbulence. The frequency of vortex shedding is given by
St V
(1)
b
Where St is the Strouhal number (dimensionless), V (m/s) is the mean wind velocity
crit
and b (m) is the characteristics width of the cross-section; for circular cylinders the
characteristics width is the mean external diameter. The vortex shedding effect on a circular
cylinder depends on the Reynolds number, which is given by:
Vb
(2)
Re
Where V and b are defined as in Equation 1 and is the kinematic viscosity of the air,
which is approximately 15 x 10-6 m2/s corresponding to a temperature of 20 centigrade. The
path of vortex wake in the leeward is important, mainly those that occur regularly and move
alternately from one side to another side (Figure 4).
Figure 4: Experimental evidence of the vortex shedding on a leeward side of circular section
When Reynolds number increases the flow shifts from laminar to transitory turbulence.
Achenbach [2] identified four intervals depending on the flow behavior of the boundary layer,
these are: subcritical, critical, supercritical and transcritical. The Reynolds number that defines
each scheme has not been determined accurately because it depends on various factors like the
roughness of the cylinder, the intensity of streamlines, and aspect ratio of the cylinder (Figure 5).
In Table 1, this intervals and their corresponding approximated Reynolds number are related.
For circular cross-section, Strouhal number varies with flow velocity and therefore with
Reynolds number. In Table 2, StrouhalReynolds number empirical relationships are shown [3],
where y ln Re 1.6 x 10 3 .
In general, Strouhal number ranges from 0.18 to 0.20. In most practical cases that arise in
structures, it may be considered a constant Strouhal number equal to 0.2.
In the following paragraphs the spectral and resonance vortex shedding methodologies
are described.
SPECTRAL METHOD
The spectral method gives adequate results for relatively rigid structures such as concrete
silos, concrete chimneys with large diameter and low height [4]. In this case, the procedure of the
vortex shedding response is based on the spectral modal analysis.
The generalized force on the structure, due to the vortex shedding, is:
Regime
Reynolds Number
Subcritical
Critical
Supercritical
Transcritical
Strouhal Number
S t 0.2139 4.0 Re
Reynolds Number
S t 0 .1848 8 .6 x 10 -4 R e 1 .5 x 10 5
Q(t ) Fv ( z , t ) ( z ) dz
0
(3)
Where the inertia force per unit length in cross-wind direction, Fv ( z , t ) , is given by:
Fv ( z , t ) q ( z ) b( z ) C L ( z , t )
(z )
h
(4)
In the Equation 4, q (z ) is the velocity wind pressure, b (z ) is the external diameter and
C L ( z, t ) is a dimensionless factor. If it is considered that e (Hz) is the natural modal frequency,
then the structure deflexion can be written as
y ( z , t ) ( z ) a (t )
(5)
Where a (t ) is the modal displacement, which is considered as stochastic process with a
power spectral density given by
S a ( ) H ( )
h 2 q b C L ref
2
e Bref
J ( )
(6)
1 ( z ) 2
s
exp
(8)
2
(
B( z ) s ( z )
B
z )
If the correlation length is small, the joint aerodynamic admittance can be approximated
q ( z ) b( z ) C L ( z ) ( z )
g ( z , )
q b C L ref
by:
J ( ) 2
bref
Bref
0 g ( z, )
h
dz
(9)
h2
Where b ref is the correlation length.
The standard deviation of the displacement for a white noise excitation can be
approximated by:
y ( z) ( z)
S a ( ) d
(10)
Or
y ( z)
( z ) C L ,ref
2 mref ref
Bref
1 h
g ( z, e )2 dz
0
h
(2 e ) 2 v
(11)
1 h m( z ) 2 ( z )
dz
2
h 0 mref ref
(12)
a 2 K a
2
bref
me
(13)
Where is the air density, b ref is the reference width, K a is an aerodynamic parameter,
which is positive if the aerodynamic damping is negative and me is the equivalent mass per unit
length given by:
me
m( z ) 2 ( z ) dz
(14)
2 ( z ) dz
The inertia force per unit length, Fv (z ) , acting perpendicular to the wind direction, can
be obtained by:
Fv ( z ) m( z ) (2 e ) 2 k p y ( z )
(15)
Where k p is the peak factor and y (z ) is the standard deviation of the displacement
given in Equation 11.
RESONANT VORTEX SHEDDING METHOD
The modal force for a dynamic system is the same that is obtained with the Spectral
Method (Equation 3). Nevertheless, the Resonant Vortex Shedding Method [5] establishes that
inertia force per unit length, in cross-wind direction, is:
Fv ( z , t ) q ( z ) d ( z ) c F ( z ) sin 2 s t ( z )
(16)
Ymax .
Fe
2
( 2 e ) me s
(17)
Fe max .
q( z ) b( z ) c F ( z ) ( z ) dz
2 ( z ) dz
(18)
Where max . is the maximum amplitude of the modal shape. The Equation 17 can be
written as:
Ymax .
max .
bref.
q ( z ) b( z )
c F ( z ) ( z ) dz
q ref. bref.
1 1
S c S t2
4 2 ( z ) dz
(19)
Where S c and St are the Scruton and Strouhal numbers, respectively. Scruton number is
given by:
Sc
2 s me
2
bref.
(20)
St
e bref.
(21)
Vref.
The vibration amplitude and the limited correlation of load, described by the correlation
length, shows that maximum load does no occur simultaneously along the structure.
Ruscheweyh [5] takes in account this last effect by integrating the maximum amplitude of load
along the entire length, L , so that L 2 is the correlation length that is equal to the integral of
correlation function from zero to infinite. The maximum amplitude of the load is calculated in
the nodal points near the maximum deflection; this is for considering the aeroelastic effects and
the wind action that produces the maximum response. If variations, throughout structure height,
of wind pressure and width of structure are negligible and, still more, assuming that the modal
shape has the same sign, the integral of the numerator of Equation 18 can be approximated by:
h
c
0
( z ) ( z ) dz clat. k p ( z ) dz
(22)
Where clat. is the standard deviation of the load. It has seen that the maximum load can
be equal to the standard deviation multiplied by the peak factor. Ruscheweyh [5] considers the
peak factor by means integration of the modal shape over the effective correlation length, L e ,
defined by:
( z ) dz k ( z ) dz
p
(23)
Le
Thus, the effective correlation length incorporates the influences of the correlation of
load and peak factor. Substituting Equations 22 and 23 in Equation 19, is obtained
Ymax .
1 1
K K w clat .
bref.
S c S t2
(24)
Where K and K w are constants. For example, in the Euro Code [6], these constants are
defined, respectively, as:
h
K max .
( z) dz
4 ( z ) dz
0
(25)
Kw
( z) dz
Le
( z ) dz
(26)
For modes that do not have constant sign, it is assumed that load acts in the same
direction as the modal deflection, so the definition of K and K w should be modifying as
proposed in Reference [4].
Vcrit.
e b
(27)
St
Ymax
C 3 ( b 2 / me )
b
s C 2 ( b 2 / me )
b
h
(28)
Vcrit . 5 e b
(29)
Ymax .
K bt
Sc
(30)
Where K is the factor for maximum tip deflection, taken as 0.5 for circular
cross-sections, and bt is the average breadth of the top third of the structure.
The equivalent static wind force per unit length for chimneys is given as follows:
Fv ( z ) m ( z ) ( 2 e ) 2 ( z ) Ymax .
(31)
(32)
The standard deviation of the displacement related to the width b at the point with the
largest deflection can be calculated by:
y
c1 c12 c 2
(33)
Where the constants c1 and c2 are calculated using Equations 33 and 34, respectively.
c1
c2
a L2
2
Sc
1
4 K a
(34)
b 2 a L2 C c2 b
(35)
me K a S t4 h
1 1
K mod K lat K w
Ymax . b 2
St Sc
(36)
Where K mod is the mode shape factor, K lat is the lateral force coefficient and K w is the
effective correlation length factor.
The mode shape factor is calculated with:
m
K mod
j 1 l j
( z ) dz
(37)
4 dz
j 1 l j
2
j
Where m is the number of antinodes of the vibrating structure in the considered mode
shape, j (z ) , and l j is the length of the structure between two nodes. If one considers only the
first vibration mode for a cantilever structure, j 1 , m 1 and l j h . If it assumed
that 1 ( z ) z h , then the previous equation gives K mod 5 12 0.13 .
2
The lateral force coefficient is shown in Table 4. K lat 3 2 .4 V crit . V1, L1 K lat , 0
Constant
R e 10 5
R e 5 x 10 5
R e 10 6
Cc
0.02
0.005
0.01
K a , max .
0.5
aL
0.4
0.4
0.4
The constants C c and K a , max . are assumed to vary linearly with the logarithm of the
5
5
5
6
Reynolds number for 10 R e 5 x 10 and for 5 x 10 R e 10 .
K lat
K lat K lat , 0
K lat 0
In Table 4, V1, L1 is the mean wind velocity in the centre of the effective correlation
length, L1 , which is obtained from Table 5 as a function of vibration amplitude for first vibration
mode, Y (s1 ) . The basic value K lat , 0 of the lateral force coefficient is given Figure 6 for circular
cylinders.
The effective correlation length factor, for first vibration mode of a cantilever structure, is
given by:
L /b L /b 1 L /b
K w 3 1 1 1 1
1
1
3 1
Where 1 h b
Y (s1 ) b
L1 b
0.1
0.1 to 0.6
0.6
12
(38)
K lat ,0
Re (Vcrit . ) for
circular cylinders
S t 0.20 for h / b 15 . For h / b in the range between 6 and 15, St is assumed to vary as the
logarithm of h / b . For structures with varying cross-wind dimensions, values in Equation 27
corresponding to the point with maximum movement are used.
The effect of resonant vortex shedding depends on the turbulence intensity of the wind.
For 10 min. mean wind velocities larger than approximately 15 m/s the turbulence intensity of
the wind is calculated by:
1
1
I v ( z)
if z z min
(39a)
ct ( z ) z
ln
z0
I v ( z ) I v ( z min )
if z z min
(39b)
Where c t (z ) is the topography factor, which is taken as unity in this paper work. The
roughness length z0 and the minimum height z min are shown in Table 6.
The inertia force per unit length and the maximum deflection in cross-wind direction are
given by Equations 31 and 32, respectively. The standard deviation can be calculates by means
of Equations 33-35, except that:
Terrain category
Roughness length
z0 (m)
Minimum height
z min (m)
0.01
0.05
0.3
1.0
16
K a K a , max h v ( I v )
(40)
Where:
hv ( I v ) 1 3 I v ( z )
if 0 I v ( z ) 0.25
(41a)
hv ( I v ) 0.25
if I v ( z ) 0.25
(41b)
The turbulence intensity I v (z ) is determined at the height where the movement of the
structure is at a maximum.
When the standard deviation of the deflection is less than approximately 2% of the crosswind dimension, the peak factor can be calculated with:
0.577
k p 2 ln(600 e )
(42)
2 ln(600 e )
For standard deviation exceeding approximately 20% of the diameter, the peak factor can
be taken as 2 .
(3 m) (0.73 Hz)
(3 m) (4.57 Hz)
12.17 m/s ; Vcrit , 2
76.17 m/s Equation 27
0.18
0.18
Vcrit ,1 12.17 m/s 1.25 V 1.25 (36 m/s) 45 m/s The effect vortex shedding need be
investigated
Vcrit , 2 76 .17 m/s 45 m/s The effect vortex shedding need not be investigated
Vcrit ,1
Vcrit ,1
12.17 m/s
0.34 0.83
K lat K lat , 0
Table 4
V
36 m/s
(3 m )(12.17 m/s )
Table 3
Re (Vcrit ,1 )
2.434 x 10 6 K lat K lat , 0 0.2
15 x 10 -6 m 2 /s
K mod 0.13
Equation 37
Sc
Equation 20
1
1
Ymax (3 m)
(0.13) (0.2) K w 0.542 K w m
2
(0.18) 4.44
1 h b 60 m 3 m 20 K w
L1 L1
L12
1
20 60 10,800
Equation 36
Equation 38
As L1 depends onf Ymax b , one must proceed in an iterative way with equations that
includes this values. Thus, the maximum cross-wind deflection is Ymax 0.366 m .
Spectral Method
Because Re 2.434 x 10 6 10 6
4.44
1
0.0517
4 (1)
c1
(0.4) 2
2
c2
Equation 34
Equation 35
y
b
Equation 33
0.3217 y 0.965 m
Sc
k p 2 1 1.2 arctan 0.75
4 K a
4.44
2 1 1.2 arctan 0.75
1.854
4 (1)
Equation 32
I v ( z)
1
1
0.141
1 60 m
ln
0.05 m
hv ( I v ) 1 3 (0,141) 0.577
60 m 4 m
Equation 39a
0 0.141 0.25
Equation 41a
(0.4) 2
2
Equation 40
4.44
1
0.031
4
(
0
.
577
)
Equation 34
Equation 35
y
b
Equation 33
0.2494 y 0.748 m
k p 2 1.414
Equation 42
Equation 32
Ymax
Equation 29
0.5 (3 m)
0.338 m
4.44
Equation 30
0.0039
The Canadian Code and ISO 4354:1997(E) [10] point out that in this case, large
amplitudes up to value of b may results. Therefore, Ymax could outcome 3 m.
A summary of the total displacements due to cross-wind response for a chimney with
typical characteristics, applying the wind design codes previously described, is shown in Table 7.
In this table, it can be seen that the results obtained applying the Canadian and the
ISO 4356:1997(E) Codes, can be very conservative. Likewise, the spectral method proposed by
Euro Code ignored the turbulence intensity influence; therefore the results obtained are
conservative as well.
On the other hand, the Danish code certainly considers the turbulence intensity, reason
why the displacements acquired with this code are more reliable. About this, spite of it doesnt
appear in this paperwork, the results that have been obtained in other research works [5, 11-12]
demonstrate that the Danish code provides more accurate values.
Table 7: Results obtained for the case of study
Code
Australian/New Zealand
Canadian and ISO 4356:1997(E)
Euro
Resonant vortex shedding Method
Spectral Method
Danish
Terrain category 2
Ymax (m)
0.338
3
0.366
1.789
1.058
CONCLUSIONS
In this work the methodology proposed by different wind design codes has been analyzed to
calculate the maximum response that can take place in a cylindrical structure in cross wind
direction, due to the vortex shedding. Considering the turbulence intensity and, for hence, the
roughness place in where the structure will be built; the results are more congruent between the
data obtained with experimental tests and the data resultant from the environment vibration tests
made on steel chimneys. In conclusion, the methodology proposed by the Danish code for vortex
shedding effects is the more accurate.
REFERENCES
[1] Tranvik P. and Goran A., Dynamic behaviour under wind loading of a 90 m steel chimney, Alstom
Power Sweden AB, Vaxjo, Report S-0141, 2002.
[2] Achenbach E., Influence of surface roughness on the cross-flow around a circular cylinder, J. Fluid
Mechanics, 1971, 46, p. 321-335.
[3] Norberg C., Fluctuating lift on a circular cylinder: Review and new measurements, J. Fluids Struct.,
2003, 17, 57.
[4] Dyrbye G. and Hansen S., Wind load on structures, John Wiley and Sons, ISBN 0-471-9565-1, 1997.
[5] Ruscheweyh H. and Sedlacek G., Crosswind vibrations of steel stacks-critical comparison between
some recently proposed codes, Journal of Wind Engineering and Industrial Aerodynamics, 1988, 30,
p. 173-183
[6] BS EN 1991-1-4-4:2005, Euro Code 1: Actions on structures, Part 1-4: General actions-Wind
actions, British Standard, 2005.
[7] NRCC 48192, National Research Council Canada, Users Gudie-NBC 2005 Structural Commentaries
(Part 4 of Division B), ISBN 0-660-19506-2, 1993.
[8] AS/NZS 1170.2:2002, Australian/Neo Zealand Standard, Structural design actions, Part 2: Wind
Actions, 2005.
[9] DS410 E:2004, Code of Practice for Loads for the Design of Structures, Danish Standard
Association, 2004.
[10] ISO 4354:1997(E), International Standard, Wind Actions on Structures, 1997.
[11] Ciesielski R., Gaczek M. and Kawecki J., Observation results of cross-wind response of towers and
steel chimneys, Journal of Wind Engineering and Industrial Aerodynamic, 1992, 41-44, p. 2205-2211.
[12] Vickery B. J. and Basu R. I., Simplified approaches to the evaluation of the across-wind response of
chimneys, Journal of Wind Engineering and Industrial Aerodynamics, 1983, Vol. 14, p. 153-166.