Wind Loading On Curved Circular Cylinder Structures
Wind Loading On Curved Circular Cylinder Structures
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/267567804
READS
138
2 authors:
Piotr Mieczysaw Szczepaniak
Agnieszka Padewska
18 PUBLICATIONS 9 CITATIONS
9 PUBLICATIONS 0 CITATIONS
SEE PROFILE
SEE PROFILE
th
Abstract
The paper presents the way of estimating the wind force acting on straight or curved elements with circular
cross-section. These elements can be positioned at any angle to the wind direction. They may also be bent into
the form of a torus or a helix, laid horizontally or sloped. The first part of the work shows the analytical
approach to the problem, solved by decomposing the wind velocity vector to the normal, binormal and tangent
components, estimating the pressure distribution around the elements cross-section and finally integration of the
pressure over the whole surface. The second part of the article briefly presents the results of the air flow
computer simulations. Because there were observed significant differences between the data obtained from the
analytical and numerical method, some empirical correction functions had to be attached to the analytical
equations. The last part consists of the engineering applicable advices, presented on diagrams and tables of
coefficients.
Key Words
wind load; drag force; curved structure; numerical air flow computations
INTRODUCTION
Wind load is one of the most important load cases, acting on building structures. Procedures for calculating
the values of wind forces are precisely described in Eurocode 1 Part 1-4 [1]. However there is a specific, but
quite popular type of structure which is not covered by these standards. It means the building objects made
of various straight or curved circular cylinder elements, such as waterslides or other amusement ride devices
(Fig.1). For this type of structure the wind load is often the leading variable action, especially if the supports
have a static scheme of a vertical, fixed column with horizontal beams, where the bending moments
at the foundations level of the columns are the most important internal forces.
The problems with calculations of the wind load are caused by the fact, that the longitudinal axis of these
structures is rarely perpendicular to the wind blows direction, as it is assumed in section 7.9 of [1]. So in
the current paper there is presented the way of estimating the value and direction of the wind force acting on
a straight cylinder, positioned at certain angle to the wind direction, and on a torus shaped structure, laid
horizontally or sloped.
PhD Eng. Piotr Szczepaniak, Silesian University of Technology, Faculty of Civil Engineering, Department of
the Theory of Building Structures, ul. Akademicka 5, 44-100 Gliwce, Poland, tel.: +48 608 524 333, e-mail:
[email protected].
2
MSc Eng. Agnieszka Padewska, Silesian University of Technology, Faculty of Civil Engineering, e-mail:
[email protected].
In Eurocode 1 [1] the standard procedure for calculating the wind load acting on straight cylindrical elements
is described with the following basic formulas:
Fw cscd
qp ( ze ) Aref
(1)
elements
(2)
Re
b v ( ze )
Re 5 105 cf ,0 1.2
2 qp ( ze )
0.18 log(10 k / b)
1 0.4 log( Re / 106 )
(3)
(4)
with its value slightly oscillating, which is caused by the von Krmn vortex shedding. In other cases, mainly
if the axis of the cylinder is positioned at different (not straight) angle to the wind blows, the drag force
is reduced, but at the same time appears a significant lift force, which is not taken under consideration in [1].
To cover this situation a more sophisticated procedure has to be developed.
At the first stage it is necessary to create a mathematical representation of the axis and surface of the element.
In most engineering structures made of cylindrical members its axes can be defined as a simple flat curve on
the Oxy plane, which is afterwards rotated and translated to the final position, as it is shown below.
r( ) rx ( ); ry ( ); rz ( )
(5)
r(a) MRot rb ( ) r0
(6)
where: r() parametric vector equation of the axis in a Cartesian coordinate system,
rb() base function,
MRot rotation matrix,
r0 translation vector
M Rot
0
0
cos( z ) sin( z ) 0 cos( y ) 0 sin( y ) 1
sin( z ) cos( z ) 0
0
1
0 0 cos( x ) sin( x )
0
1 sin( y ) 0 cos( y ) 0 sin( x ) cos( x )
0
(7)
(8)
(9)
T
(10)
( )
(11)
r' ( )
T1 T1 ( )
0 B1 B1 ( )
r ' ( )
r ' ( )
(12)
(13)
0 r' ( ) v 0 B1 B1 ( )
N1 N1 ( ) B1 ( ) T1 ( )
r ' ( ) v
r ' ( ) v
(14)
(15)
b
S( , ) r N1 cos B1 sin
2
(16)
n1 n1 , N1 cos B1 sin
(17)
b b
b b
(18)
where:
b
T
n1
dA
(19)
(20)
(21)
( v) 2
vze 2
v1 v1
vze 2
(22)
(23)
Much more complicated is the pressure distribution resulting from the perpendicular air flow. It strongly depends
on the value of Reynolds number and the surface roughness, as it is presented on figure 7.27 in [1]. Similar
graphs, obtained from experiments in wind tunnel or numerical simulations, can be found in [3,4,5].
Unfortunately there are no equations of the pressure distribution, and the only one applicable formula (25) has
been obtained from [2].
pqBN qBN cp0 qp (v12B v12N ) cp0
(24)
(25)
0 arctanv1N , v1B ,
(26)
2 b c
p0,PN
cf, PN
cos d
cp,cor
cf,0
cf,0
(27)
(28)
dF
dFw0(BN)
dFw
dF
cs cd w0 cs cd w0(T)
cs cd f w0
d
d
d
f w0
dFw0(T) dFw0(BN)
(30)
dFw0(T)
pqT n1 dA
b2
(29)
qp v12T N1 r' ( ) d
(31)
dFw0(BN)
pq n dA
BN
c n dA
p0
(32)
b
2.2112 1.9752 cos2 0 cos 0 N1
qp (v12B v12N ) cf,0 b
4
1.9752
sin 2 0 sin 0 B1 r ' ( ) d
4
The resultant wind force can be obtained by integration of the distributed wind load fw0 over parameter .
2
Fw0 f w0 d
(33)
Fw cs cd Fw0
(34)
Along with the analytical calculations, a large number of numerical air flow simulations were done. They were
made using the ANSYS software, especially the Fluent module. Computations were performed with the aid of
the PL-Grid Infrastructure.
5.1 Main parameters
First, there were modelled series of air flows past a straight cylinders with a diameter b = 1.0m, surface
roughness k = 0.15mm, positioned at the angle against the wind direction (z) in range from 30 to 90,
at 15 steps. The computations were performed at the free flow speed (v) equal to 11.0, 15.0, 22.0 and 33.5 m/s.
It gives the Reynolds number in range from around 7.5105 to 2.5106, which indicates the supercritical type of
air flow, with a strong influence of the turbulent shear flow. To avoid the necessity of a very dense FVM mesh
near the cylinder surface, the k-/SST model has been chosen. The remaining boundary conditions are presented
on Fig. 4.
At the second stage there has been modelled the air flow past the half of a torus, with the same diameter, surface
roughness and wind speed as mentioned above. It has been positioned at slope angle (y) in range from 0 to 90,
at 22.5 steps. The radius of curvature equals R = 3.0m Fig. 5.
5.2 Results of the air flow simulations
Results of the numerical air flow simulations are presented in Tab. 1 and 2. There are also shown adequate
values obtained from the analytical calculations. It can be easily noticed, that as far as a straight cylinder is
concerned, the analytical underestimation of the wind force never exceeds 3 N/m, which is a completely
insignificant value in civil engineering calculations. In most cases the wind force is slightly overestimated,
however the difference rarely exceeds 10%.
Much worse convergence of the results shows up in Tab. 2. However in the case of the not sloped torus (y = 0)
the analytical results are acceptable, in other cases appears a significant underestimation of the y and z wind
force components. Thats why an empirical correction function has to be attached. Because it shouldnt be an
explicit function of the main parameters, such as and y angles, the following formulas are proposed:
b2
qp cf ,0 a3 r '
2
a3 23.69 a34 36.43 a32 9.09 a1 0.87 2.5 B1
f w0
2
2
2
2
2
3
(35)
7.1 a22 N1
where: a1 arccos(v1N ) ,
a2 arcsin( v1B ) ,
a3 v1B ,
f w0,cor
dFw0(T) dFw0(BN)
d
f w0
(36)
These functions were obtained by finding a constant multipliers to B1 and N1 (fB, fN) in successive ranges
of integration (i, i+1), which give a minimal difference of the wind force vector (37). Next these multipliers
were interpolated, using the arccos(v1N) as the independent variable. The improved analytical values are
presented in Tab. 3.
i1
i1
num,i
Fw0 | ii1 f w0 d f B(i ) B1 f N(i ) N1 d Fw0
i
i
Top view
y
z
1m
x
vx=v
(37)
vy=vz=0
periodic b.c.
num
fw0
p=0
periodic b.c.
b
Side view
z
vx=vy=vz=0
vx=v
vz=0
p=0
vy=vz=0
vz=0
Fig. 4. Boundary conditions for the air flow simulations straight cylinder
Top view
i+1
vx=v
vy=vz=0
vy=0
num,i
Fw0
p=0
vy=0
x
2R
Side view
vz=0
y
vx=v
vy=vz=0
num,i
Fw0
vx=vy=vz=0
p=0
2R
vz=0
Fig. 5. Boundary conditions for the air flow simulations half of a torus
Rotation
angle
z []
30
45
60
75
90
30
45
60
75
90
30
45
60
75
90
30
45
60
75
90
f wnum
0 , x [N/m]
f wnum
0 , y [N/m]
f wan0, x [N/m]
f wan0, y [N/m]
-10.85
-17.72
-18.80
-12.10
0
-21.06
-34.39
-36.48
-23.49
0
-47.38
-77.37
-82.06
-52.84
0
-114.50
-186.97
-198.31
-127.70
0
Tab. 1. Wind force components vs straight element rotation angle and wind speed
Slope
angle
y []
Limits of
integration
67.5
,i
Fwnum
0 , z [N]
Fwan0,,xi [N]
Fwan0,,yi [N]
Fwan0,,zi [N]
-19.54
-33.56
-3.49
34.86
45.76
54.39
51.07
21.59
,i
Fwnum
0 , x [N]
,i
Fwnum
0 , y [N]
0
22.5
45.0
67.5
90.0
112.5
135.0
157.5
22.5
45.0
67.5
90.0
112.5
135.0
157.5
180.0
87.1
52.0
5.5
-2.5
12.9
38.4
68.9
53.5
-16.5
-33.3
-1.9
37.4
58.8
49.9
42.9
11.9
102.07
53.49
4.51
-6.17
9.46
37.37
77.98
110.37
315.8
149.2
389.08
151.09
91.1
59.5
20.0
11.0
19.1
39.4
83.9
72.9
-13.9
-27.4
2.5
48.6
64.4
42.4
22.8
13.5
24.4
37.2
51.9
18.3
10.8
26.4
38.6
31.5
103.85
62.31
17.93
3.33
15.78
44.18
82.12
110.99
-17.10
-29.77
-4.25
31.12
49.10
57.36
51.19
21.12
17.49
26.66
30.91
21.96
11.20
4.10
-5.40
-13.61
396.9
152.9
239.1
440.48
158.78
93.32
22.5
45.0
67.5
90.0
112.5
135.0
157.5
180.0
98.6
79.3
55.7
43.2
43.8
56.6
66.8
50.0
-6.0
-8.8
6.9
34.3
66.7
63.7
30.3
8.8
70.9
77.2
71.9
50.0
32.9
3.6
-13.2
-22.5
108.11
83.87
54.89
40.69
46.53
67.38
93.85
112.60
-10.44
-17.97
-0.41
28.57
50.70
58.88
48.80
19.22
31.03
43.35
52.63
47.49
31.31
10.69
-10.70
-25.45
(i )
w0
494
195.9
270.8
607.92
177.35
180.35
110.0
103.3
94.4
84.2
77.0
72.7
70.0
66.4
12.3
35.5
52.0
54.3
53.2
49.3
35.1
12.6
78.1
68.0
53.4
40.6
22.3
-6.7
-31.9
-42.3
112.39
105.51
96.44
90.95
92.11
99.09
108.10
114.29
-1.37
0.85
13.18
31.55
46.76
50.65
39.39
14.90
38.26
44.29
48.79
43.87
27.55
3.90
-19.54
-34.25
678
304.3
181.5
818.88
195.91
152.87
22.5
45.0
67.5
90.0
112.5
135.0
157.5
180.0
107.0
105.0
103.2
102.4
102.4
103.4
105.2
107.2
11.1
36.9
61.0
76.7
76.5
60.7
36.0
11.0
76.7
61.0
36.9
11.1
-11.0
-36.0
-60.7
-76.5
114.59
114.59
114.59
114.59
114.59
114.59
114.59
114.59
7.75
22.07
33.03
38.96
38.96
33.03
22.07
7.75
38.96
33.03
22.07
7.75
-7.75
-22.07
-33.03
-38.96
(i )
w0
835.8
369.9
1.5
916.75
203.60
0.00
0
22.5
45.0
67.5
90.0
112.5
135.0
157.5
22.5
45.0
67.5
90.0
112.5
135.0
157.5
180.0
0
22.5
45.0
67.5
90.0
112.5
135.0
157.5
0
22.5
45.0
67.5
90.0
112.5
135.0
157.5
22.5
45.0
67.5
90.0
112.5
135.0
157.5
180.0
(i )
w0
90
i+1 []
(i )
w0
45
i []
(i )
w0
22.5
0
22.5
45.0
67.5
90.0
112.5
135.0
157.5
Slope
angle
y []
Limits of
integration
Fwan0,,xi [N]
Fwan0,,yi [N]
Fwan0,,zi [N]
-16.5
-33.3
-1.9
37.4
58.8
49.9
42.9
11.9
102.07
53.49
4.51
-6.17
9.46
37.37
77.98
110.37
-19.54
-33.56
-3.49
34.86
45.76
54.39
51.07
21.59
315.8
149.2
389.08
151.09
91.1
59.5
20.0
11.0
19.1
39.4
83.9
72.9
-13.9
-27.4
2.5
48.6
64.4
42.4
22.8
13.5
24.4
37.2
51.9
18.3
10.8
26.4
38.6
31.5
89.47
58.48
16.40
5.04
24.59
47.00
66.10
83.81
-14.16
-25.85
4.34
50.19
63.34
50.12
33.15
12.97
25.57
33.59
41.21
35.52
26.50
26.49
29.12
28.76
396.9
152.9
239.1
390.89
174.10
246.77
22.5
45.0
67.5
90.0
112.5
135.0
157.5
180.0
98.6
79.3
55.7
43.2
43.8
56.6
66.8
50.0
-6.0
-8.8
6.9
34.3
66.7
63.7
30.3
8.8
70.9
77.2
71.9
50.0
32.9
3.6
-13.2
-22.5
94.45
81.32
60.77
44.35
46.98
59.31
66.07
69.54
-1.78
-2.01
8.44
37.14
63.32
57.99
34.56
11.46
81.04
76.38
67.59
53.25
28.34
4.83
-6.51
-11.82
(i )
w0
494
195.9
270.8
522.78
209.12
293.10
110.0
103.3
94.4
84.2
77.0
72.7
70.0
66.4
12.3
35.5
52.0
54.3
53.2
49.3
35.1
12.6
78.1
68.0
53.4
40.6
22.3
-6.7
-31.9
-42.3
111.10
102.63
91.08
81.68
75.06
72.10
73.08
74.72
7.03
24.25
44.60
57.12
57.97
53.48
40.15
15.01
83.47
81.12
69.62
46.09
18.53
-9.16
-35.22
-51.14
678
304.3
181.5
681.44
299.62
203.30
22.5
45.0
67.5
90.0
112.5
135.0
157.5
180.0
107.0
105.0
103.2
102.4
102.4
103.4
105.2
107.2
11.1
36.9
61.0
76.7
76.5
60.7
36.0
11.0
76.7
61.0
36.9
11.1
-11.0
-36.0
-60.7
-76.5
113.49
113.49
113.49
113.49
113.49
113.49
113.49
113.49
14.30
40.74
60.97
71.91
71.91
60.97
40.74
14.31
71.91
60.97
40.74
14.31
-14.31
-40.74
-60.97
-71.91
(i )
w0
835.8
369.9
1.5
907.90
375.84
0.00
i+1 []
,i
Fwnum
0 , x [N]
,i
Fwnum
0 , y [N]
0
22.5
45.0
67.5
90.0
112.5
135.0
157.5
22.5
45.0
67.5
90.0
112.5
135.0
157.5
180.0
87.1
52.0
5.5
-2.5
12.9
38.4
68.9
53.5
0
22.5
45.0
67.5
90.0
112.5
135.0
157.5
22.5
45.0
67.5
90.0
112.5
135.0
157.5
180.0
(i )
w0
45
67.5
0
22.5
45.0
67.5
90.0
112.5
135.0
157.5
0
22.5
45.0
67.5
90.0
112.5
135.0
157.5
22.5
45.0
67.5
90.0
112.5
135.0
157.5
180.0
(i )
w0
90
0
22.5
45.0
67.5
90.0
112.5
135.0
157.5
,i
Fwnum
0 , z [N]
i []
(i )
w0
22.5
CONCLUSIONS
For simplifying the manual calculations, the total effects of the axis curvature and rotation can be compressed
into a single vector coefficient , named position coefficient.
2
~ c c q c A
~;
~;
~ T
Fw cs cd Fw0 cs cd f w0,cor d cs cd qp cf ,0 Aref
s d
p
f ,0
ref
x
y
z
(38)
Aref b l b
r' ( ) d
(39)
2
T
~ ~
x ; ~
y ; ~
z
w0,cor
(40)
qp cf ,0 b r ' ( ) d
1
x ; y ; z
T
f w0,cor
qp cf ,0 b r' ( )
(41)
2
2
2 3/ 2
2
2
2
2
cos y cos z sin z sin y cos y sin z
(42)
Slope
angle
y []
22.5
45
67.5
90
Limits of integration
i []
i+1 []
~x [-]
0
22.5
45.0
67.5
90.0
112.5
135.0
157.5
0
22.5
45.0
67.5
90.0
112.5
135.0
157.5
0
22.5
45.0
67.5
90.0
112.5
135.0
157.5
0
22.5
45.0
67.5
90.0
112.5
135.0
157.5
0
22.5
45.0
67.5
90.0
112.5
135.0
157.5
22.5
45.0
67.5
90.0
112.5
135.0
157.5
180.0
22.5
45.0
67.5
90.0
112.5
135.0
157.5
180.0
22.5
45.0
67.5
90.0
112.5
135.0
157.5
180.0
22.5
45.0
67.5
90.0
112.5
135.0
157.5
180.0
22.5
45.0
67.5
90.0
112.5
135.0
157.5
180.0
0.891
0.467
0.039
-0.054
0.083
0.326
0.680
0.963
0.781
0.510
0.143
0.044
0.215
0.410
0.577
0.731
0.824
0.710
0.530
0.387
0.410
0.518
0.577
0.607
0.970
0.896
0.795
0.713
0.655
0.629
0.638
0.652
0.990
0.990
0.990
0.990
0.990
0.990
0.990
0.990
~y
-0.170
-0.293
-0.030
0.304
0.399
0.475
0.446
0.188
-0.124
-0.226
0.038
0.438
0.553
0.437
0.289
0.113
-0.016
-0.018
0.074
0.324
0.553
0.506
0.302
0.100
0.061
0.212
0.389
0.498
0.506
0.467
0.350
0.131
0.125
0.355
0.532
0.628
0.628
0.532
0.355
0.125
[-]
~z
[-]
0.223
0.293
0.360
0.310
0.231
0.231
0.254
0.251
0.707
0.667
0.590
0.465
0.247
0.042
-0.057
-0.103
0.728
0.708
0.607
0.402
0.162
-0.080
-0.307
-0.446
0.628
0.532
0.355
0.125
-0.125
-0.355
-0.532
-0.628
ACKNOWLEDGEMENT
This research has been partially supported by PL-Grid Infrastructure.
REFERENCES
[1]
EN 1991-1-4:2005: Eurocode 1: Actions on structures - Part 1-4: General actions - Wind actions. CEN,
Brussels, 2005.
[2]
[3]
Mallick M. Kumar A.: Study on drag coefficient for the flow past a cylinder, International Journal of
Civil Engineering Research, Vol. 5, No. 4 (2014), pp. 301-306.
[4]
Merrick R. Bitsuamlak G.: Control of flow around a circular cylinder by the use of surface roughness:
A computational and experimental approach, Internet publication at http://www.ihrc.fiu.edu/wpcontent/uploads/2014/03/MerrickandBitsuamlak_FlowAroundCircularCylinders.pdf
[5]
Lakehal D.: Computation of turbulent shear flows over rough-walled circular cylinders. Journal of Wind
Engineering and Industrial Aerodynamics, Vol. 80, Issues 1-2 (March 1999), pp. 47-68.