Journal of Applied Science and Engineering, Vol. 19, No. 3, pp. 285-292 (2016)
DOI: 10.6180/jase.2016.19.3.06
Characteristics of Wind Pressures on a Cooling
Tower Exposed to Stationary and Translating
Tornadoes with Swirl Ratio 0.54
Shuyang Cao1, Jin Wang2,3 and Jinxin Cao4*
1
State Key Lab of Disaster Reduction in Civil Engineering,
Tongji University, Shanghai, P.R. China
2
Department of Bridge Engineering,
Tongji University, Shanghai, P.R. China
3
Glenn Department of Civil Engineering, Clemson University, SC, USA
4
State Key Lab of Disaster Reduction in Civil Engineering,
Tongji University, Shanghai, P.R. China
Abstract
Current wind-resistant design of wind-sensitive structures including large-scale cooling towers
is generally carried out with respect to synoptic boundary-layer-type strong winds. A swirling tornado
can produce significantly different wind pressures than conventional boundary-layer wind. This paper
presents both stationary and translating tornado effects on a cooling tower in a tornado vortex
simulator developed at Tongji University, China. Wind pressures acting on the external surface of
cooling tower model were measured at a fixed swirl ratio (S = 0.54) in the present study. Different
radial distances between a cooling tower and stationary tornado vortex center were considered.
Translating tornadoes with three different translation speeds (u = 0.04 m/s, 0.12 m/s and 0.2 m/s) were
simulated. The results show that a tornado vortex can produce high negative wind pressures on a
cooling tower surface due to the negative pressure drop accompanying a tornado. A cooling tower
exposed to a tornado experiences combined effects of pressure drop accompanying a tornado and
aerodynamic flow-structure interaction.
Key Words: Tornado Vortex, Cooling Tower, Wind Pressure, Aerodynamics
1. Introduction
Although tornadoes in China are not as strong as
those in the USA, they do occur in eastern China. Golden
and Snow [1] reported that China experiences about 10
to 100 tornadoes per year. Featured by a funnel three-dimensional vortex structure, the wind characteristics of
a tornado are quite different than those of conventional
boundary-layer wind, which indicates the necessity to investigate the tornado-induced wind loads on structures.
Many attempts have been made to physically model
*Corresponding author. E-mail: [email protected]
tornado vortices [2-4]. Some studies have been conducted on a stationary tornado to investigate tornadostructure interaction [5-8]. Recently, some research work
has been performed considering translating tornado effects [9-11]. However, most previous researches focused
on the interactions between tornadoes and low-rise structures. With the fast development of the Chinese economy,
public consciousness and expectations of a structure’s
safety are rising, especially where increasing energy demand requires large cooling towers to be built in tornado-prone areas. This has increased the demand to study
tornado-cooling tower interaction. Cao et al. [12] investigated stationary tornado effects on a cooling tower ex-
286
Shuyang Cao et al.
posed to a stationary tornado vortex and found that the
tornado-induced wind pressure on a cooling tower is significantly different than that in conventional straight-line
winds.
In the present study, both a stationary tornado vortex and a translating tornado vortex with three different
translation speeds were modeled in a tornado vortex generator at one swirl ratio (S = 0.54). Tornado-cooling
tower aerodynamic interactions in a stationary tornado
vortex at different radial distances relative to the cooling
tower structure are presented. Particular attention is devoted to the difference between tornado-induced wind
pressure and that of conventional boundary-layer wind
flow. In addition, the effects of a translating tornado vortex on wind pressures are studied and discussed.
2. Experimental Setup
Figure 1 shows a schematic diagram of the tornado
vortex simulator at Tongji University, whose mechanism
for generating a tornado-like flow is similar to that of the
tornado vortex simulator at Iowa State University, USA
[4]. A circular duct 1.5 m in diameter and 1.009 m in
height is suspended overhead with a 0.5 m-diameter updraft hole (ro = 250 mm) holding a controlling fan to
generate a strong updraft. The simulator floor could be
adjusted up and down, enabling a range of heights for the
inflow layer (H = 150 mm~550 mm). Both the fan and
the guide vanes are placed on the top of the simulator,
which allows more spaces in which to conduct model
Figure 1. Schematic diagram of tornado vortex simulator at
Tongji University.
tests to determine the tornado effects. In addition, this
tornado vortex simulator can translate along the ground
plane at a given speed (maximum translation speed is 0.4
m/s).
Figure 2 shows the cooling tower model, whose
height is 143.3 mm. The radius of the throat part of the
model is about 33.3 mm, and the largest radius is 52.7
mm. The cooling tower model is fitted with a total of 36
pressure taps distributed evenly over the external surface
at three levels as shown in Figure 2. At each level, twelve
pressure taps are distributed uniformly around the circumference.
In the present study, the inflow height was fixed at
H = 400 mm below the exit of the outer duct. Swirl ratio
is the most important factor that controls the flow structure, and is defined as S = tanq/(2a), where q and a are
guide vane angle and aspect ratio a = H/ro, respectively
[13,14]. Church et al. [2] and Matsui and Tamura [14]
pointed out that tornado-like vortices are laminar when
the swirl ratio is smaller than 0.3 and become turbulent
when the swirl ratio is 0.3-0.6. Therefore, the present
measurements were performed at swirl ratio 0.54, which
responded to a turbulent tornado-like vortex that was
thought closer to a natural tornado. In addition, swirl ratio 0.54 is the highest value available in the simulator
with the given inflow height. The fan speed was fixed at
1500 rpm and the pressure data were measured at a rate
of 300 Hz in both a stationary and translating tornado.
The maximum tangential velocities at different elevations above the floor are given in Table 1. In the present
study, the maximum tangential velocity and corresponding core radius at model height (Z/ro = 0.57) were selected as the reference velocity to calculate pressure coefficients and to determine the relative locations be-
Figure 2. Cooling tower model.
Wind Pressures on a Cooling Tower Exposed to a Tornadoe-like Vortex with Swirl Ratio 0.54
Table 1. Simulator settings and major parameters of
generated tornado-like vortex
Vane angle
q (°)
60
H
(m)
0.4
ro
(m)
0.25
S
Z/ro
Vtmax
(m/s)
rc/ro
0.54
0.10
0.30
0.40
0.57
11.96
11.19
10.98
09.85
0.28
0.36
0.36
0.44
tween the cooling tower model and the tornado vortex.
The scaling parameter for modeling the translational
motion is a key issue when investigating the effects of
translational motion. Two proposals were used for scaling the translational velocity in previous studies. Refan
et al. [15] proposed the ratio between maximum tangential velocity and the translational velocity of a tornado
while Haan et al. [4] focused on the wind loads on structures and proposed the duration of tornado wind load
on a structure as the scaling parameter to be maintained
between the prototype and simulated translating tornado. In the present study, the latter scaling was adopted
because the tornado effects on structures were of interest. The tornado-like vortex radius available in the simulator varies from 70 mm to 110 mm at S = 0.54 while the
core radii of real tornados in China are approximately
20-50 meters, so a scale ratio lL of core radii between a
model tornado and a design tornado in 1/200-1/700 is
necessary. In order to make the translational time ratio
lT equal to unity, the translational velocity ratio lVT
must be equal to the length scale lL. The three selected
translational velocities, i.e. u = 0.04 m/s, 0.12 m/s and
0.2 m/s, correspond to translational velocities in the range
of 8 m/s to 140 m/s.
287
pressure coefficients of a cooling tower are defined as in
Eq. 1:
(1)
where Pe is the local pressure acting on the external
cooling tower surfaces, P¥ is the reference static pressure used to calculate the pressure coefficient, r is the
density of air, and Vtmax is the maximum tangential velocity at the tower height as shown in Table 1. It should
be noted that the atmospheric pressure far from the tornado is considered as the reference pressure in the current calculation.
Figure 3 illustrates the relative locations between
cooling tower model and tornado vortex. The coordinate
origin is set at the center of the model indicating that
negative r means the tornado is located on the left side of
the cooling tower model and positive r means the tornado is located on the right side. Figure 4 shows the external wind pressures acting on the cooling tower model
exposed to a stationary tornado at fixed swirl ratio S =
0.54 for different radial locations (r/rc = 0, 0.64, 1.18 and
1.91). Note that rc is the core radius at model height for
swirl ratio 0.54. The radial locations r/rc = 0.64, 1.18
and 1.91 indicate that the cooling tower is located inside
the tornado core (|r/rc| < 1), near the tornado core radius
(|r/rc| » 1) and outside the tornado (|r/rc| > 1), as shown
in Figure 3. When the tornado vortex is located in different radial locations relative to the cooling tower center, wind pressures exhibit different characteristics. The
pressure decreases from pressure tap No. 6 (stagnation
point) to pressure tap No. 10 and then recovers a little
until pressure tap No. 12 (base point) at r/rc = 0. At other
3. Wind Pressures Acting on a Cooling Tower
Model
3.1 Stationary Tornado Vortex-induced Wind
Pressures
A tornado features two regions: an inner core and an
outer region. In the inner core, the tangential velocity increases with distance from the vortex center. After reaching a peak value at the vortex core radius, the tangential
velocity component reduces [12]. Wind pressure coefficients at three levels are presented. The external wind
Figure 3. Illustration of relative location between model and
tornado vortex.
288
Shuyang Cao et al.
Figure 4. Wind pressure coefficients on cooling tower at different radial locations (a) r/rc = 0; (b) r/rc = 0.64; (c) r/rc = 1.18; (d)
r/rc = 1.91.
radial locations (i.e. r/rc = 0.64, 1.18 and 1.91), the pressure coefficient decreases from pressure tap No. 7 (stagnation point) to pressure tap No. 10 or No. 11 and then
recovers a little until pressure tap No. 1 (base point). The
characteristics of wind pressure distribution show that
the pressures acting on a cooling tower may be regarded
as the sum of two individual parts: the negative pressure
drop accompanying the tornado and the aerodynamic
force acting on the cooling tower model. The pressure
when the tornado vortex was located on the left side of
the model (negative r) was also measured. Similar findings were observed.
Figure 5 shows the variations of maximum and minimum external pressure coefficients (Cpemax and Cpemin)
around the cooling tower at the middle level with radial
distance r. The minimum and maximum pressure coefficients were obtained from the twelve pressure taps at the
middle level. With increasing distance from the tornado,
the maximum and minimum pressure coefficients recover gradually. However, the maximum pressure coefficient remains negative over a wide range of r/rc. In addition, it is interesting to note that the minimum pressure
Figure 5. Minimum and maximum pressure coefficients at
middle level.
Wind Pressures on a Cooling Tower Exposed to a Tornadoe-like Vortex with Swirl Ratio 0.54
coefficients exhibit a W shape distribution at S = 0.54,
with the minimum pressure coefficient existing at about
|r/rc| = 0.8. The phenomenon of vortex breakdown [16],
or in other words, the transition of single-celled vortex to
double-celled vortex [11] which makes the vortex center
to be not necessarily at the geometric center of the simulator is thought to be the reason for this result.
Standard deviations of pressure coefficients are
analyzed with respect to radial locations as shown in
Figure 6. Due to the aerodynamic feature shown in Figure 4, pressure taps 1, 4, 7 and 10 are selected as representative points to describe the standard deviation of
pressure coefficients. As can be seen from Figure 6(a),
the pressure coefficient fluctuation exhibits M shape
versus radial distance r and the peak value exists around
the core boundary |r/rc| » 1. Figure 6(b) compares the
characteristics of wind pressure acting on the cooling
tower in tornado-like vortices with those obtained in
straight-line boundary-layer winds, in which the X coordinate indicates the angle from the stagnation point
where the pressure coefficient is maximum. The experiment in boundary-layer-type winds was conducted on a
pressure model of the same shape as that used for the experiment in tornado-like winds but with a different scale
ratio, by assuming the mean velocity profile of oncoming straight-line wind has an exponent of 0.15. It can be
found that the fluctuation of pressure coefficient in a tornado-like vortex is more significant at r/rc = 0, 0.64 and
1.18 than in a boundary-lay flow. When the model was
located at r/rc = 1.91 outside of the tornado vortex core,
the trend of Cperms distribution is close to the result from
289
boundary-layer wind flow.
3.2 Translating Tornado Vortex-induced Wind
Pressures
Figure 7 presents wind pressures acting on a cooling
tower exposed to a translating tornado vortex with different translation speeds (u = 0.04 m/s, 0.12 m/s and 0.2
m/s) and compares the results with those in a stationary
tornado vortex. Wind pressures at the same four pressure
taps as in Figure 6 are given because they are representative points to describe the aerodynamic characteristics. The reference velocity used to calculate the pressure coefficient and the core radius for normalizing the
relative distance are the maximum tangential velocity and
corresponding core radius at the tower height in a stationary tornado. It should be noted that negative and
positive radial distance r means the tornado vortex is located on the left side and right side of the cooling tower
center, respectively. During the experiment, the moving
velocity of the tornado simulator’s downdraft duct was
confirmed to be constant within -10 < r/rc < 10 after initial acceleration from rest. It is interesting to note that
with the tornado vortex moving fast, greater negative
radial shift is generated that is different than that in Haan
et al. [11], in which the peak pressure was reported to
appear after the tornado simulator’s downdraft duct
reached the model. Because the mechanism of generating a swirling flow of our tornado simulator is fundamentally the same as that of Iowa State University, the
difference in peak pressure shown above was a serious
concern. One possible reason is that our model size rela-
Figure 6. Standard deviation of pressure coefficients at middle level.
290
Shuyang Cao et al.
Figure 7. Cpe time history at four pressure points at middle level for different translation speed.
tive to tornado core diameter was significantly larger
than that of Haan et al. [11], and accordingly the blockage ratio of the present study is much larger than that in
Haan et al. [11]. These larger values may make the present experiment a particular case test, whose results cannot be compared with those in Haan et al. [11]. In addition, another possible reason is that the ground flow in
front of the simulator’s downdraft duct, where the flow
starts to swirl from nearly rest, swirls more distinctly and
forms a swirling front behind which the flow is more turbulent due to flow mixing. The result in which the peak
pressure occurs before the simulator’s downdraft duct
reaches the model was also noticed when the moving direction of the simulator was opposite, implying there is
no asymmetry in the facility itself that may lead to this
result.
Figure 8 shows the variations of minimum external
pressure coefficients Cpemin among 12 pressure taps at
the middle level with respect to radial distance between
the cooling tower model and the tornado vortex. The dif-
ferences between the stationary and translating tornadoes are qualitatively similar to those in Figure 7 for a
single particular pressure tap. The change in peak pressure due to translational motion is not significant. In addition, the distance shift (or time shift) of peak pressure
caused by the translating tornado vortex is obvious at
Figure 8. Minimum pressure coefficients for different translation speeds.
Wind Pressures on a Cooling Tower Exposed to a Tornadoe-like Vortex with Swirl Ratio 0.54
faster translation speed, which is similar to the trend of a
single pressure point.
4. Conclusions
The present research focused on the characteristics
of wind pressures on a cooling tower exposed to a stationary and translating tornado vortex at swirl ratio 0.54.
The conclusions can be summarized as:
(1) High negative pressure is produced on the external
surface due to the pressure drop accompanying a tornado. Higher pressure coefficient fluctuation is generated inside the tornado vortex core compared
with that in boundary-layer wind flow.
(2) The pressure distribution shows that a cooling tower
exposed to a tornado experiences combined effects
of pressure drop accompanying a tornado and aerodynamic flow-structure interaction.
(3) Translational motion does not significantly influence
the minimum external pressures of a cooling tower.
Acknowledgement
The authors wish to acknowledge the financial support from the Natural Science Foundation of China
(NSFC) Grant No. 51478358 and Research Foundation
of State Key Laboratory of Disaster Reduction in Civil
Engineering Grant No. SLDRCE14-A-01.
References
[1] Golden, J. H. and Snow, J. T., “Mitigation Against Extreme Windstorms,” Reviews of Geophysics, Vol. 29,
No. 4, pp. 477-504 (1991). doi: 10.1029/91RG01814
[2] Church, C., Snow, J. T., Baker, G. L. and Agee, E. M.,
“Characteristics of Tornado-like Vortices as a Function
of Swirl Ratio: a Laboratory Investigation,” Journal of
the Atmospheric Sciences, Vol. 36, pp. 1755-1776
(1979). doi: 10.1175/1520-0469(1979)036<1755: CO
TLVA>2.0.CO;2
[3] Ward, N. B., “The Exploration of Certain Features of
Tornado Dynamics Using a Laboratory Model,” Journal of the Atmospheric Sciences, Vol. 29, pp. 11941204 (1972). doi: 10.1175/1520-0469(1972)029<1194:
TEOCFO>2.0.CO;2
291
[4] Haan, F. L., Sarkar, P. P. and Gallus, W. A., “Design,
Construction and Performance of a Large Tornado
Simulator for Wind Engineering Applications,” Engineering Structures, Vol. 30, No. 4, pp. 1146-1159
(2008). doi: 10.1016/j.engstruct.2007.07.010
[5] Jischke, M. C. and Light, B. D., “Laboratory Simulation of Tornadic Wind Loads on a Rectangular Model
Structure,” Journal of Wind Engineering and Industrial Aerodynamics, Vol. 13, pp. 371-382 (1983). doi:
10.1016/0167-6105(83)90157-5
[6] Bienkiewicz, B. and Dudhia, P., “Physical Modeling
of Tornado-like Flow and Tornado Effects on Building
Loading,” Proceedings of the Seventh US National
Conference on Wind Engineering, pp. 95-106 (1993).
[7] Mishra, A. R., James, D. J. and Letchford, C. W.,
“Physical Simulation of a Single-celled Tornado-like
Vortex, Part B. Wind Loading on a Cubical Model,”
Journal of Wind Engineering and Industrial Aerodynamics, Vol. 96, pp. 1258-1273 (2008). doi: 10.1016/
j.jweia.2008.02.027
[8] Rajasekharan, S. G., Matsui, M. and Tamura, Y., “Characteristics of Internal Pressures and Net Local Roof
Wind Forces on a Building Exposed to a Tornado-like
Vortex,” Journal of Wind Engineering and Industrial
Aerodynamics, Vol. 112, pp. 52-57 (2013). doi: 10.
1016/j.jweia.2012.11.005
[9] Sarkar, P. P., Haan, F. L., Balaramudu, V. and Sengupta,
A., “Laboratory Simulation of Tornado and Microburst to Assess Wind Loads on Buildings,” Structures
Congress, St. Louis, MO (2006). doi: 10.1061/40889
(201)11
[10] Sengupta, A., Haan, F. L., Sarkar, P. P. and
Balaramudu, V., “Transient Loads on Buildings in Microburst and Tornado Winds,” Journal of Wind Engineering and Industrial Aerodynamics, Vol. 96, pp.
2173-2187 (2008). doi: 10.1016/j.jweia.2008.02.050
[11] Haan Jr, F. L., Balaramudu, V. K. and Sarkar, P. P.,
“Tornado-induced Wind Loads on a Low-rise Building,” Journal of Structural Engineering, Vol. 136, No.
1, pp. 106-116 (2010). doi: 10.1061/(ASCE)ST.1943541X.0000093
[12] Cao, S., Wang, J., Cao, J., Zhao, L. and Chen, X., “Experimental Study of Wind Pressures Acting on a Cooling Tower Exposed to Stationary Tornado-like Vortices,”
Journal of Wind Engineering and Industrial Aerody-
292
Shuyang Cao et al.
namics, Vol. 145, pp. 75-86 (2015). doi: 10.1016/j.
jweia.2015.06.004
[13] Mitsuta, Y. and Monji, N., “Development of a Laboratory Simulator for Small Scale Atmospheric Vortices,”
Natural Disaster Science, Vol. 6, pp. 43-53 (1984).
[14] Matsui, M. and Tamura, Y., “Influence of Incident
Flow Conditions on Generation of Tornado-like Flow,”
Proceedings of the 11th American Conference on Wind
Engineering, Puerto Rico, USA (2009).
[15] Refan, M., Hangan, H. and Wurman J., “Reproducing
Tornadoes in Laboratory Using Proper Scaling,” Jour-
nal of Wind Engineering and Industrial Aerodynamics, Vol. 135, pp. 136-148 (2014). doi: 10.1016/j.
jweia.2014.10.008
[16] Wang, J., Cao, S. Y., Pang, W. Q., Cao, J. X. and Zhao,
L., “The Effects of Ground Roughness on the Characteristics of Tornado-like Vortices,” Proceedings of 14th
International Conference on Wind Engineering, Brazil
(2015).
Manuscript Received: Feb. 22, 2016
Accepted: Jul. 4, 2016