J. Wind Eng. Ind. Aerodyn.

126 (2014) 7586

Contents lists available at ScienceDirect

Journal of Wind Engineering

and Industrial Aerodynamics
journal homepage: www.elsevier.com/locate/jweia

Wind loads and vortex shedding analysis on the effects

of the porosity on a high slender tower
M. Belloli, L. Rosa n, A. Zasso
Politecnico di Milano, Department of Mechanical Engineering, via G. La Masa 1, 20156 Milan, Italy

art ic l e i nf o a b s t r a c t

Article history: This paper presents a detailed experimental study nalized to evaluate the effect of the porosity on the
Received 9 March 2012 dynamic behavior of a civil structure. The structure considered is an original architectonic complement
Received in revised form erected on top of a new tall building. Its cladding, made of perforated steel plates, forms an intricate
7 January 2014
three-dimensional spiral characterized by ve different diameters which decrease along the height of the
Accepted 18 January 2014
Available online 10 February 2014
structure.
The main objective of the tests is to evaluate the effect of the porosity of the external perforated
Keywords: panels, comparing the dynamic response obtained from the in-service conguration (permeable panels)
Porous surface and the results obtained with non-porous panels. This comparison showed that the porosity of the
Vortex shedding
panels is of great importance and can completely change the dynamic behavior of the whole spire. This
Wind tunnel tests
study has a realistic relevance as the structure could be subjected to atmospheric icing which could block
Wind loads
Aeroelastic model up the panels and change their porosity.
High slender tower Wind tunnel tests were performed on 1:50 scaled rigid and aeroelastic models in smooth and
Atmospheric icing turbulent ow conditions. The effect of the porosity was estimated comparing wind actions in terms of
global wind loads at the base of the spire, local wind load on a panel and the structure's proneness to
vortex induced vibrations. The analysis carried out highlights on the great inuence of porosity in the
dynamic response of the structure, mainly in relation to vortex shedding induced vibrations, which were
almost absent with porous panels, but very strong in the case of non-porous panels.
& 2014 Elsevier Ltd. All rights reserved.

1. Introduction
The tower, or spire, investigated in this study is included in one
of the largest projects underway in Milan for the redevelopment of
an area of abandoned railyards adjacent to the Garibaldi train
station. The spire, which is erected on top of a 139 m tall building,
consists of a supporting lattice framework structure 80.44 m high,
clad with perforated steel and glass plates, Fig. 1. Considering the
height of the building, the spire's maximum elevation from the
ground is 220 m. The external cladding is not regular, but forms an
intricate three-dimensional spiral characterized by ve different
diameters which decrease along the height of the structure. Only a
small area of the panels is made of glass (a portion at the bottom
of the spire), while most of them are perforated steel plates.
The structural response of the tower is governed by the internal
lattice framework. A review of the literature indicates that much
research has been conducted into the aerodynamic behavior of
lattice tower structures, showing that their dynamic response is
n
Corresponding author.
E-mail addresses: marco.belloli@polimi.it (M. Belloli),
lorenzo.rosa@polimi.it (L. Rosa), alberto.zasso@polimi.it (A. Zasso).
highly three dimensional and not easy to predict (Zou et al., 2008).
In particular oscillation amplitudes of displacements and accel-
erations in the cross-wind direction could be as great as or even
larger than those in the along-wind direction (Glanville and Kwok,
1995; Ballio et al., 1992). In addition, the tower considered in this
research has a three dimensional cladding which, even if not
regular and porous, could lead to wind induced vibrations and
fatigue damage (Repetto and Solari, 2010; Irwin et al., 2008).
The present study is focused on the inuence of porosity. In fact
the ow through the panels modies the pressure distribution
and, as a result, the wind loads and the dynamic response of the
whole structure. The realistic relevance of this analysis is that,
owing to its position, the spire can be subjected to atmospheric
icing which could make parts of the external panels non-porous
(Fikke et al., 2006), causing possible severe damage to the
structure (Sundin and Makkonen, 1998; Mulherin, 1996).
Not many experimental data are available on wind interaction
with permeable civil structures. Most of them regard wind loads on
structures such as greenhouses, panels or roofs (Briassoulis et al.,
2010; Robertson et al., 2002; Letchford, 2001; Letchford et al., 2000;
Richards and Robinson, 1999; Cheung and Melbourne, 1988); these
results show that the wind loads on perforated structures are
generally lower than those on the same non-porous surfaces. More

0167-6105/$ - see front matter & 2014 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.jweia.2014.01.004
76 M. Belloli et al. / J. Wind Eng. Ind. Aerodyn. 126 (2014) 7586
Fig. 1. The 139 m tall spire object of this study. (a) The bold nal stage of its erection. (b) The nal structure.
systematic studies regard ow around porous cylinders (Fransson
et al., 2004; Mathelin et al., 2001): in particular they show how
vortex shedding can be reduced or suppressed by manipulating the
ow around the cylinder through the application of suction or
blowing.
The aim of the tests was to evaluate the effects of the porosity
of the panels of this very particular spiral-shape slender structure
on the dynamic response. The analysis was carried out comparing
wind actions in terms of global wind loads at the base of the spire,
local wind load on a panel and the structure's proneness to vortex
induced vibrations.
2. Wind tunnel tests
Wind tunnel tests were performed at the 1.5 MW closed-circuit
wind tunnel of the Politecnico di Milano, Italy. The large dimensions of
the boundary layer test section (4 m high, 14 m wide and 36 m long)
permitted a very large geometric scale L 1=50 model=real,
while maintaining lower blockage effects. Large scale testing is
advantageous in the general context of wind engineering tools for
the following reasons: adverse scaling effects (such as those due to
violation of Reynolds number similitude) can be reduced, and some
building external geometric details are too small to be reproduced in
small-scale tests. This means that as far as concerns vortex shedding
the hypothesis is that the wind tunnel tests performed in the
subcritical region are more conservative than the real full scale
structure condition. On the other hand, the full scale structure will
be in the post-critical region.
In order to consider possible effects due to the singular position
of the spire, the model was placed at a height of 0.7 m from the
oor level, over a base simulating the upper part of the building.
The rst natural frequencies of the building (0.238 Hz 0.254 Hz
and 0.321 Hz, respectively the rst two exural and the rst
torsional modes) are low compared to the lower frequencies of
the spire so that the modes between the building and the spire are
uncoupled. For this reason the building was not dynamically
considered in the present study and only its external shape was
reproduced, Fig. 2(a).
To verify the sensitivity to the Reynolds number, some tests
were repeated changing the wind speed. The analysis was carried
out comparing the overall wind load at the base of the spire in
terms of drag and lift coefcients and on a pressure tap in terms of
pressure coefcient. These coefcients were stable in the range of
wind speeds tested.
2.1. Flow conguration
Wind tunnel tests were performed in smooth and turbulent
ow, respectively called SF and TF below. In the rst condition (SF)
the wind speed is constant along the height of the test section and
the turbulence is lower than 2%, while in turbulent ow the
vertical velocity and turbulence prole were simulated by means
of spire and roughness elements positioned at the entrance of the
test section. The turbulent ow proles are similar to those typical
of an area covered with buildings, but they do not follow a precise
target since the main goal of the work was the comparison
between the two conditions of porosity on dynamic behavior, in
the same ow conditions. Fig. 3 shows the characteristic of the
ow in the test section in terms of mean wind speed (Fig. 3(a)) and
turbulence intensity prole (Fig. 3(b)). Pressure measurements
were carried out in turbulent ow, while smooth ow is the worst
condition considering vortex shedding excitation and the best way
to test in order to discover possible aerodynamic instability.
2.2. Rigid model
The rigid model of the spire, Fig. 2(a), allowed the measure-
ment of the global wind loads, in terms of forces and moments at
its base, and the pressure on some external panels. To ensure high
quality measurements the model was made as stiff as possible, to
ensure high structural frequencies and a quasi-static behavior
under wind loads. It is a static model which reproduces the
geometry of the full-scale structure in terms of aerodynamic
surfaces and details. For this reason great care was taken in scaling
the external surface geometry and porosity, while the internal
lattice framework was not reproduced as it was the real structure.
2.3. Aeroelastic model
The vortex shedding investigation was carried out using the
aeroelastic model, Fig. 2(b). The design and making of the aero-
elastic model were on the basis of the modal parameters obtained
from a nite element model of the real structure. Froude simili-
tude criteria were adopted for scale reduction, leading to a factor
F 7 for frequency scaling model=real and to a factor
 M. Belloli et al. / J. Wind Eng. Ind. Aerodyn. 126 (2014) 7586 77

Fig. 2. (a) The rigid model in wind tunnel test section, smooth ow condition. (b) The aeroelastic model; on the back are visible the four cobra probes used to analyze the
wakes from the spire.

U(z)/Urif Iu 2 FEM, 1st mode

2.6 2.6 Iv Exp. data, 1st
1.8 FEM, 2nd mode
2.4 2.4 Exp. data, 2nd
heigth z from ground, model scale (m)
heigth z from ground, model scale (m)

1.6
2.2 2.2 h spire model scale (m)
1.4
2 2

1.2
1.8 1.8

1
1.6 1.6

1.4 1.4 0.8

1.2 1.2 0.6

1 1 0.4

0.8 0.8 0.2

0.6 0.6 0
0 0.5 1 1.5 2 0 10 20 30 1 0.5 0 0.5 1 1.5
U/Urif () Iu, Iv (%) 1st and 2nd mode ydirecton

Fig. 3. (a) TF: vertical velocity prole. (b) TF: vertical prole of turbulence. (c) First and second mode shape of the spire.

V 1=7 for scaling the velocities. Froude similitude criteria, match the correctly scaled mass. A space of 1 mm was left
which consider the inuence of gravitational forces in the aero- between adjacent modules, in this way the cover-modules do
elastic phenomena and are usually used for testing long span not contribute to the exural stiffness of the tower. Fig. 4(a) shows
bridges, were chosen for these tests as an optimum compromise a modulus of the aeroelastic model, while Fig. 4(b) displays a
among the geometric scale adopted, the range of wind speeds close-up view of the nal part of the aluminium spire during the
available (the maximum velocity in the test section is 14 m/s) and construction phase.
advantages in making and tuning of the aeroelastic model (mass Natural frequencies and damping of the aeroelastic model were
distributions, stiffness and frequencies must be reproducible determined by analyzing decay tests excited by impulsive forces
scaled). In addition this led to a 1:1 scale factor for the accelera- and sweep-tests generated by an electromagnetic shaker. Table 1
tions measured on the model compared to those on the full shows the results of the modal identication.
scale tower. The correspondence between the target and experimental
The aeroelastic model was designed as a spine model: the values is good and the low values of damping obtained are in
elastic properties of the real structure were reproduced by means good agreement with the expected full-scale ones. On the other
of an aluminum bar consisting of rectangular section of four hand, low levels of damping are needed in order to make
different dimensions. An external cover, constituted of 16 modules, aeroelastic phenomena more visible, in terms of both buffeting
was adopted to reproduce the wind interaction shape and to response and vortex induced vibrations. Fig. 3(c) shows the modal
78 M. Belloli et al. / J. Wind Eng. Ind. Aerodyn. 126 (2014) 7586
Fig. 4. The aeroelastic model. (a) A modulus of the external cover. (b) The central aluminium spine.
Table 1
Target frequencies (from FEM), experimental frequencies and non-dimensional
damping obtained on the model.
Mode Freq. target (Hz) Exp. freq. (Hz) Exp. damping (%)
1st Bending x 5.09 5.09 0.45
1st Bending y 5.23 5.09 0.45
2nd Bending x 11.41 12.3 0.45
2nd Bending y 11.80 12.3 0.45
deected shapes for the rst and second exural modes calculated
using the nite element method and identied in the experimen-
tal tests. The gure proves that the model was correctly designed
and the deected shapes are in good agreement with calculations.
2.4. Porosity
Concerning the porosity of the panels, in order to fulll
kinematic similitude, the ow conditions for the model and
prototype should be the same: in other words, the velocities and
accelerations of the ow at corresponding points should have the
same directions and scaled magnitudes (Fay, 1994). In order to
guarantee this condition the perforated panels must be scaled not
geometrically but the same loss coefcient k must be maintained
(Letchford et al., 2000). The loss coefcient k is dened as
pu  pd
k 1
2
U =2
where pu and pd are respectively the upstream and the down-
stream static pressure on either side of the grid, is the density of
the air and U is the mean wind speed. This coefcient is an
indicator of the resistance to ow through a porous surface,
including the effects of the open area (dened as the ratio
between the area open to through ow and the total area of the
panel) as well as the shape of the perforations. The pressure loss
coefcients were evaluated experimentally in a smaller wind
tunnel, in which its entire round cross section (diameter
400 mm) was covered by the grid being tested.
In the real structure the panels have round perforations (radius
7.5 mm) with a triangular pitch of 21 mm, Fig. 5(a). Their open
area is 46.3% and the loss coefcient k is about 4.5 (experimen-
tally evaluated). On the models, round hole perforated plates with
radius 0.4 mm and triangular pitch of 1 mm were used. These
plates were formed by electrical discharge machining, they have
an open area 58% and a loss coefcient k, as shown in Fig. 5(b),
very close to the value of the full scale panels.
3. Model instrumentation
Global forces' measurement was measured using a dynamo-
metric six-component force-balance (RUAG 192-6I) linked to the
base of the model. The exposure angles and the reference system
are shown in Fig. 6(a).
In order to evaluate the local wind loads on the permeable
panels, some areas representative of the whole cladding structure
were instrumented on the rigid model to measure internal and
external pressures, Fig. 7(a). The measurements were performed
with tubes connecting the pressure taps to an electronic pressure
scanner. Pressure data were collected through high-frequency
simultaneous measurements. One can see that these tubes are
quite large, but their presence does not condition our study for the
following reasons: the pressure tubes placed inside the spire (i.e.
the taps measuring the external pressure) do not disturb the ow
since inside the spire the test will show that there is no an air ow,
but a quite constant pressure (constant at the different heights of
the spire). The pressure tubes placed outside the spire (i.e. the taps
measuring the internal pressure) can have some impact on the
measurements of the external pressure but we were interested in
the comparison between the global dynamic response with porous
and non-porous panels. So if the external tubes have some impact
on the ow, this is given in both the porosity conditions and the
comparison is not affected. Lastly the pressure measurements gave
very good results so we consider that their inuence is not
signicant for the results of this work.
The aeroelastic model was instrumented with 6 accelerometers
xed on the structural bar in order to monitor the accelerations
according to its bending modes (x-dir and y-dir). Vortex shedding
excitation was investigated using four Cobra Probes, a multi-hole
pressure probe developed by Turbulent Flow Instrumentation
(TFI). The Cobra Probe can simultaneously measure three unsteady
orthogonal velocity components in subsonic ows elds within a
range of 7451 at frequencies of more than 200 Hz (Mallipudi and
Selig, 2004). They were xed 1.2 m downwind of the spire at the
heights shown in Fig. 6(b); the same gure shows the positions of
the accelerometers.
Measured forces (shears F D;L and moments M D;L ) on the balance
are presented in terms of non-dimensional aerodynamic coef-
cients reduced at the spire base and dened as follows:
F D;L t M D;L t
C D;L t ; C MD;L t ; 2
q h Arif q h Arif b
where Arif is the reference area (Arif 390.6 m2 full scale), b is the
reference dimension (b31.4 m full scale) and q h Pa is the
dynamic pressure at the base of the spire. The peak values
are evaluated using the method of the peak factor for Gaussian
 M. Belloli et al. / J. Wind Eng. Ind. Aerodyn. 126 (2014) 7586 79

10
Model scale microperforated panel
Full scale perforated panel
8

Loss coefficient k ()
6

0
0 5 10 15 20
V model (m/s)

Fig. 5. (a) The perforated panels on the real structure. (b) The loss coefcient k function of the wind speed. Comparison between the full scale and the micro-
perforated panel.

h4

Acc3
h3

h2
Acc2

Acc1
h1

Fig. 6. (a) Exposure angles and reference system. (b) Heights of the cobra probes (h) and the accelerometers (Acc). (c) Position of the pressure tap 007.

process (Hansen and Dyrbye, 1996); in the gures they are plotted
including the mean value and following the same sign convention.
Pressures are presented as non-dimensional pressure coef-
cient C p;k dened, at the kth pressure tap, as follows:
pk t  ps
C p;k t
qh
peak values are calculated by statistical approaches developed by
J. Cook and J.R. Mayne, with methods that obtain the values
from extreme distributions functions. These methods are largely
described and assessed for instance in the references (Holmes,
2004; Cook, 1985; Cook and Mayne, 1979, 1978).
3

where pk(t) is the static pressure time history at the kth pressure
tap and ps is the reference static pressure. The total load on the
panel is given by the net-pressure coefcient, dened as the
difference between the external (ext) and internal (int) pressure
coefcients at each kth pressure tap:
C pnet;k t C pext;k t  C pint;k t
where a positive value of C p  net means a net-positive load on the
panel (a pushing from the outside), while a negative value means a
suction load.
The wind load at a point uctuates in time owing to turbulence
and other causes; it is the sum of a mean value and a uctuating
part. The design values of loading actions required for static
structures are the maximum (largest positive) and minimum
(largest negative) expected in the lifetime of the structure. These
3.1. Evaluation of the effects of the porosity
In order to evaluate the effects of porosity on wind loads as
well as vortex shedding excitation, some tests were duplicated
using adhesive tape to make some panels on the rigid model,
Fig. 7(b), and the entire external surface of the aeroelastic model,
4
Fig. 7(c), impermeable. Great care was taken in order to keep the
surface roughness unchanged adding the tape on the external
surface. The added mass of the tape on the aeroelastic model
caused a slight decrease of the frequencies, while the damping
ratio was reduced. As the structural properties of the aeroelastic
model are the same in the two congurations (porous and non-
porous), the reduction of damping is caused by the air that, during
the decay tests, passes through the micro-holes. This passing of air
generates energy dissipation greater than the air moving around
80 M. Belloli et al. / J. Wind Eng. Ind. Aerodyn. 126 (2014) 7586
Fig. 7. Analysis of the effects of the porosity. (a) Rigid model: some perforated panels instrumented for the evaluation of the net-pressure coefcient. The internal taps are
squared while the external circled. (b) Rigid model, some panels made non-permeable. (c) Aeroelastic model whose panels were fully non-porous.
Table 2
Experimental frequencies and non-dimensional damping obtained on the aero-
elastic model fully covered with aluminium tape.
Mode Exp. freq. (Hz) Exp. damping (%)
1st Bending x 4.5 0.2
1st Bending y 4.5 0.2
2nd Bending x 11.2 0.2
2nd Bending y 11.2 0.2
the circular surface closed with the tape. Table 2 shows the
dynamic properties related to this modied model.
This analysis has an important realistic relevance as the
structure could be subjected to atmospheric icing which could
modify the porosity of the panels and consequently the aerody-
namic behavior of the spire.
4. Results
4.1. Overall wind load
Fig. 8 shows the overall wind load at the base of the aeroelastic
model of the spire obtained in smooth ow. The gure shows the
drag and lift coefcients (mean and peak values) comparing the
porous and non-porous panels' congurations. Owing to the three
dimensional external shape of the cladding both coefcients show
variability as the exposure angle changes.
Considering the drag coefcient CD, Fig. 8(a), the non-porous
conguration shows a moderate increment of the mean value
and a higher increment of peak values. In particular at
301   601, the mean value is practically the same, but
the peak value is nearly doubled. The effect of porosity on the lift
coefcient, Fig. 8(b), is more relevant. The non-porous spire does
not show a modication in mean value, which stays close to zero,
while peak values show considerable increment: the maximum
peak occurs at about  301, where its value is about ten-times
larger than the equivalent with porous panels.
The analysis of the wind load coefcients shows relevant
enhancement of the dynamic drag and dynamic lift (shown in terms
of peak values) associated to the non-porous conguration in smooth
ow. In the following paragraphs a relevant enhancement of vortex
shedding in the non-porous conguration compared to the porous
one will be shown. The measured values conrm the literature data
quoting relevant increase of dynamic drag and lift in smooth ow as
an effect of the vortex shedding presence, e.g. Zdravkovich (1997a),
Chen (1987), and Bishop and Hassan (1964).
4.2. Pressure measurements on the cladding plates
Tap 007, whose position is shown in Fig. 6(c), is used to
illustrate the pressure analysis on the cladding plates and the
comparison between the porous and non-porous panels. The
pressure measurements were carried out using the rigid model
in turbulent ow conditions.
Fig. 9 shows the mean (line ), peak negative (line ) and peak
positive (line ) net-pressure coefcients related to the pressure
tap 007 as a function of the wind exposure angle . The gure
shows the results from the in-service perforated panel and from
the non-porous panel.
The net-pressure coefcient on the porous panel, Fig. 9(a), has
similar suction and pushing values, both for mean and for negative
and positive peaks. Furthermore when the ow is tangential to the
panel (  301, Fig. 6(a)) the mean net-pressure coefcient is close
to zero. This value is quite unusual considering this panel as a
cylindrical shape in tangential ow condition: in fact an equivalent
cylindrical non-porous surface in the same ow conguration is
expected to have a greater suction. This consideration is conrmed
by the net-pressure coefcient obtained on the non-porous panels,
Fig. 9(b), where the negative mean and the peak pressure, if the
panel is tangential to the ow  301, are more negative
and closer to suction values typical of a not-porous cylinder
(Zdravkovich, 1997a, 1997b).
In order to dene the contribution of porosity, it is interesting
to analyze separately the contribution of the external and the
internal pressure. Fig. 10(a) shows the mean and peak external
pressure coefcients measured on the perforated panel (solid line)
and on the non-porous panel (dash-line, full marker). Fig. 10
(b) shows the same comparison made on the internal pressure
coefcients. Fig. 10(a) shows that the increase in suction loads is
due to negative pressures on the external surface which, when the
panel is made non-porous, assumes lower values. On the contrary
Fig. 10(b) shows that the internal pressures measured on the
 M. Belloli et al. / J. Wind Eng. Ind. Aerodyn. 126 (2014) 7586 81

2.5 12
MeanNonporous MeanNonporous
PeakNonporous 10 PeakNonporous
2 MeanPorous MeanPorous
Drag coefficient Cd () PeakPorous PeakPorous
8

Lift coefficient Cl ()
1.5
6

4
1

2
0.5
0

0 2
180 120 60 0 60 120 180 180 120 60 0 60 120 180
Exposure angle (deg) Exposure angle (deg)

Fig. 8. Overall wind load at the base of the spire, smooth ow. Comparison between porous and non-porous panels. (a) Drag coefcient, CD. (b) Lift coefcient, CL (note the
difference in the y-axis scale).

4 4
3 3 nonporous Mean
2 2 nonporous PkSup
1 1 nonporous PkInf

Cpnet 007 ()
Cpnet 007 ()

0 0
1 1
2 2
3 3
4 4
porous Mean
5 porous PkSup 5
6 porous PkInf 6

180 120 60 0 60 120 180 180 120 60 0 60 120 180

Exposure angle (deg) Exposure angle (deg)

Fig. 9. Mean and peak net-pressure coefcients as a function of the wind exposure angle , pressure tap 007. (a) Porous panels. (b) Non-porous panels.

4 4
3 Mean 3
PkSup
2 2
PkInf
1 1
Cpext 007

Cpint 007

0 0
1 1
2 2
3 3
Mean
4 4
PkSup
5 5 PkInf
6 6
180 120 60 0 60 120 180 180 120 60 0 60 120 180
Exposure angle (deg) Exposure angle (deg)

Fig. 10. Mean and peak external and internal pressure coefcients as a function of the exposure angle, pressure tap 007. The solid line is related to the perforated panel, the
dash-line to the non-porous panel. (a) External pressure tap. (b) Internal pressure tap. The ow is tangential to the tap at  301.

porous panel are similar to the pressure measured on the non-
porous panels.
The analysis demonstrates that permeability reduces the wind
load on the panel mainly when the surface is tangential to the
ow. In this condition the ow does not create a strong suction on
the surface because the air ows through the holes of the perme-
able panel, tending to equalize the mean pressure and attenuate
the peak across the panel.
On the contrary when the panel is perpendicular to the ow
601 o o1501 the difference is given by the internal pressure.
The external pressure measured on porous and non-porous panels
is similar, but the internal pressure from the non-porous panel
shows stronger suction. Hence the net-pressure difference across
the panel is increased when the panel is non-porous. However in
this condition (i.e. positive external pressure) the attenuation
given by the permeability is less important. The results are in
agreement with previous studies carried out on porous roofs
(Cheung and Melbourne, 1988) and attest that the wind loads on
the perforated plates are generally lower than those on the same
non-porous plate.
4.3. Vortex shedding investigation
The vortex shedding investigation was carried out on the
aeroelastic model in the smooth ow condition. In the discussion
of the results the cross-wind velocity vi refers to the probe at
height hi placed downwind of a section of the spire characterized
by a diameter Di. In smooth ow conditions the wind speed is
82 M. Belloli et al. / J. Wind Eng. Ind. Aerodyn. 126 (2014) 7586
constant along the height of the test section, so along wind
velocities measured by the four cobras are equal and equal to
the wind speed in the wind tunnel.
4.3.1. Estimation of the Strouhal number
The Strouhal number St depends on the shape of the cross-
section, the surface roughness and the turbulence of the wind and
it is dened as
f VS D
St
U vibration mode of the structure (f2 11.2 Hz) generated by section
where fVS is the frequency of vortex shedding, U is the incoming
wind velocity and D is the specic cylinder section generating
vortex shedding. Knowing the Strouhal number it is possible to
estimate the critical wind velocity U c;i which generates a vortex
shedding in resonant condition with the ith mode of the structure:
f iD
U c;i
St
Experimental evidence of the monitored vortex shedding phe-
nomenon permitted a summary in Table 3 of the critical wind
velocities related to sections D2 and D3 calculated for the porous
structure, while Table 4 shows the critical velocities estimated for
the non-porous structure. The Strouhal numbers were estimated
by analyzing the vortex shedding frequencies obtained from tests
at different wind speeds far away from the resonant frequencies of
the model. Bearing in mind the velocity scale V 1=7, the tables
also show the full scale critical wind velocities. One can see that
these quantities are in the range of possible, very frequent or less
frequent but with very high wind speed, which makes it worth
examining the vortex shedding phenomenon.
4.3.2. Non-porous panels
Fig. 11 shows the spectra of the cross-wind velocities vi in the
wake of the fully non-porous spire, simulating a complete atmo-
spheric icing. The test speed was selected in order to have vortex
shedding generated by the section D3 resonant with the rst
vibration mode of the structure f 1 4:5 Hz : U U D3
c;1 1:28 m=s
(equal to a full scale wind speed UR 9.05 m/s). The clear narrow
bandwidth peak at the frequency f1 in the spectra of the compo-
nent v3, Fig. 11(b), conrms this condition. A narrow bandwidth
peak at the same frequency is also present in the spectrum of the
component v4, Fig. 11(a), while downwind of the sections at the
lower levels, D1 and D2, a peak is also present but it is located at
the Strouhal frequency of the section and not at the rst natural
frequency of the model, Fig. 11(c, d). This analysis reveals that
vortex shedding induced vibration is present, but because of the
very low wind speed, the energy introduced is not enough to
induce signicant vibration of the entire spire.
Table 3
Porous panels. Critical wind velocities (model and full scale) related to the sections D2 and D3, 1st and 2nd bending mode.
Section St
h3 D3 0.049 m 0.185
h2 D2 0.093 m 0.211
Table 4
Non-permeable panels. Critical wind velocities (model and full scale) related to the sections D2 and D3, 1st and 2nd bending mode.
Section St
h3 D3 0.049 m 0.192
h2 D2 0.093 m 0.151
Fig. 12(a) shows the top-spire cross-wind time history displa-
cement (15 s length part) calculated from integration of the
accelerometer signals and knowing the modal shape of the spire.
Constant vibration of the spire due vortex shedding excitation is
evident. The vibration is small, 2 mm model equal to 100 mm
full scale, nevertheless because of the low but very frequent
wind speed, it could be enough to lead to fatigue damage to the
structure.
Fig. 13 shows the same analysis with a test speed selected in
order to have vortex shedding synchronous with the second
5
D2 : U U D2c;2 7:11 m=s (equal to a full scale wind speed
UR 50.3 m/s). Observing the gure one can see an unequivocal
narrow bandwidth peak located at the second natural frequency of
the model at all the levels of the structure except the rst (this is
because the stiffness of the lower part of the of the spire is much
greater than that of the upper part, so no vibrations occur). The
6 presence of this peak in the wake is a clear indication of very
strong vortex shedding induced vibrations, conrmed by the top-
spire cross-wind time history displacement shown in Fig. 12(b).
The level of vibration reached is very high, 30 mm model equal to
1.5 m full scale, but the wind speed associated is much less
frequent.
Lastly, Fig. 14 shows two frames from the video of the motion of
the spire recorded in this resonant condition; the frames are related to
two oscillation peaks. The vibration of the model in its second
vibration mode is evident. The video was recorded with a high-
speed video camera (Olympus i-SPEED) at 100 Hz sample rate.
The non-porous spire also shows a lock-in phenomenon,
typical of circular cylinders (Zasso et al., 2008; Zdravkovich,
1997a, 1997b). Fig. 15 shows the spectra obtained using a test
speed of U6.28 m/s. This speed value is about 12% lower than the
critical velocity and should originate from a vortex shedding
associated with section D2 at the frequency fD2 10.2 Hz and
associated to the section D3 at the frequency fD3 24.5 Hz. Actually
the structural motion interacts with the wind eld in such a way
that the vortex shedding frequency fD2 synchronized with the
second natural frequency of the structure f 11.2 Hz and vortex
induced vibrations occur.
4.3.3. In-service porous panels
Fig. 16 shows the spectra of the cross-wind velocities vi in the
wake of the spire in its in-service conguration (i.e. porous
panels). The test speed was selected in order to generate from
section D2 a vortex shedding resonant with the rst vibration
mode of the structure f 1 5:09 Hz : U U D2
c;1 2:18 m=s (equal to
a full scale wind speed UR 15.5 m/s). The peak at the frequency f1
in the spectra of component v2, Fig. 16(c), conrms this condition.
U c;1 (f1 5.0 Hz) U c;2 (f2 12.5 Hz)
1.3 m/s (9.1 m/s full sc.) 3.2 m/s (22.4 m/s full sc.)
2.1 m/s (14.1 m/s full sc.) 5.3 m/s (37.1 m/s full sc.)
U c;1 (f1 4.5 Hz) U c;2 (f2 11.2 Hz)
1.2 m/s (8.4 m/s full sc.) 2.3 m/s (16.1 m/s full sc.)
2.9 m/s (20.3 m/s full sc.) 7.0 m/s (49.0 m/s full sc.)
 M. Belloli et al. / J. Wind Eng. Ind. Aerodyn. 126 (2014) 7586 83

0.04 0.04

0.03 0.03

D4 v4

D3 v3
f = 4.50Hz f = 4.50Hz
0.02 0.02

0.01 0.01

0 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Freq. (Hz) Freq. (Hz)

0.1 0.1

f = 2.0Hz
D2 v2

D1 v1
0.05 0.05

0 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Freq. (Hz) Freq. (Hz)

Fig. 11. Aeroelastic model, panels fully non-permeable, U U D3

c;1 1:28 m=s, SF. Spectra of the cross-wind velocities vh downwind the spire (model scale data).

6 60
U=1.28 m/s Dispy Top Spire U=7.11 m/s dispy Top Spire
4 40
Displacement y (mm)

Displacement y (mm)

2 20

0 0

2 20

4 40

6 60
20 25 30 35 105 110 115 120
Time (s) Time (s)

Fig. 12. Aeroelastic model, non-porous panels, SF. Top-spire cross-wind displacement time histories calculated from the accelerometers signals (mm) model scale.
(a) U U D3 D2
c;1 1:28 m=s. (b) U U c;2 7:11 m=s.

0.2 0.2
D4 v4

D3 v3

f = 11.2Hz
f = 11.2Hz
0.1 0.1

0 0
0 10 20 30 40 50 0 10 20 30 40 50
Freq. (Hz) Freq. (Hz)

1 1

f = 11.2Hz
D2 v2

D1 v1

0.5 0.5

0 0
0 10 20 30 40 50 0 10 20 30 40 50
Freq. (Hz) Freq. (Hz)

Fig. 13. Aeroelastic model, panels fully non-permeable, U U D2

c;2 7:11 m=s, SF. Spectra of the cross-wind velocities vh downwind the spire (model scale data).

Section D2 has a full length scale of about 20 m, a quarter of the with a 4:1 aspect ratio. For this reason, even if changing the
total height of the structure. Considering its mean diameter of diameter along the height of the structure is known to mitigate
about 5 m, this section can be considered an equivalent cylinder the vortex shedding (Irwin, 2009; Kim et al., 2008; Shimada and
84 M. Belloli et al. / J. Wind Eng. Ind. Aerodyn. 126 (2014) 7586

Fig. 14. Aeroelastic model fully non-permeable. Two frames from the video of the motion of the spire during a strong vortex shedding excitation on the section D2 in
resonant condition with the second vibration mode of the structure, U U D2
c;2 7:11 m=s.

0.2 0.2 f = 11.2Hz f = 24.0Hz

D4 v4

D3 v3

f = 11.2Hz
0.1 0.1

0 0
0 10 20 30 40 50 0 10 20 30 40 50
Freq. (Hz) Freq. (Hz)

1 1

f = 11.2Hz
D2 v2

D1 v1

0.5 0.5

0 0
0 10 20 30 40 50 0 10 20 30 40 50
Freq. (Hz) Freq. (Hz)

Fig. 15. Aeroelastic model, panels fully non-permeable, U U D3

c;2 6:28 m=s, SF. Spectra of the cross-wind velocities vh downwind the spire (model scale data).

Hibi, 1995), this section, as demonstrated in the previous para-
graph, could have a length sufcient for introducing enough
energy to the whole structure and inducing vibrations.
The previous assumption, generally valid for non-porous circular
surfaces, is disproved in the present case inspecting the spectra of
the cross-wind velocities at the other levels. The excitation of the
whole structure due to vortex shedding phenomenon is excluded
by the following considerations (Zdravkovich, 1997a, 1997b):
the spectra of the component v3, Fig. 16(b), show a peak at the
Strouhal frequency of section D3 f 0:185  2:18=D3 8:3 Hz and
not at the rst natural frequency of the model. The peaks in the two
spectra (Fig. 16(b, c)) have a broad bandwidth, indicating vortex
shedding that is not fully synchronized. Lastly, Fig. 17 shows the
cross-wind displacement time histories (mm model scale) calcu-
lated from the integration of the accelerometer signals. The gure
shows that the displacements are much reduced at all the levels
 M. Belloli et al. / J. Wind Eng. Ind. Aerodyn. 126 (2014) 7586 85

0.04 0.04
f = 8.3Hz

D4 v4

D3 v3
0.02 0.02

0 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Freq. (Hz) Freq. (Hz)

0.2 0.2

0.15 f = 5Hz 0.15

D2 v2

D1 v1
0.1 0.1

0.05 0.05

0 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Freq. (Hz) Freq. (Hz)

Fig. 16. Aeroelastic model, porous panels, U U D2

c;1 2:18 m=s, SF. Spectra of the cross-wind velocities vh downwind the spire (model scale data).

Dispy Level 2 Dispy Level 3

Displacement y (mm)

Displacement y (mm)
0.05 0.05 Dispy Level 1

0 0

0.05 0.05

0 10 20 30 40 50 60 0 10 20 30 40 50 60
Time (s) Time (s)

Fig. 17. Aeroelastic model, porous panels, U U D2

c;1 2:18 m=s, SF. Cross-wind displacement time histories calculated from the accelerometers signals (mm) model scale.
(a) Displacement y-direction level Acc2. (b) Displacements y-direction level Acc1 and Acc3.

0.1 0.1

f = 21.1Hz
D4 v4

D3 v3

0.05 0.05

0 0
0 10 20 30 40 50 0 10 20 30 40 50
Freq. (Hz) Freq. (Hz)

0.4 0.4

0.3 0.3
D2 v2

D1 v1

0.2 f = 12.1Hz 0.2

0.1 0.1

0 0
0 10 20 30 40 50 0 10 20 30 40 50
Freq. (Hz) Freq. (Hz)

Fig. 18. Aeroelastic model, porous panels, U U D2

c;2 5:41 m=s, SF. Spectra of the cross-wind velocities vh downwind the spire (model scale data).

considered (the maximum full scale peak to peak displacement is introduced by the wind in section D2 is not enough to induce
5 mm) and no regular vibrations are visible. vibrations in the entire spire.
The absence of strong and constant vortex induced vibration is Fig. 18 shows the same analysis using a test speed selected in
due to the porosity of the panels which modies wake formation and order to have vortex shedding generated by section D2 isofrequent
reduces the intensity of vortex shedding and the overall cross-wind with the second vibration mode of the structure f 2 12 Hz :
load (Zheng and Zhang, 2010; Baek and Karniadakis, 2009; Fransson U U D2
c;2 5:41 m=s (equal to a full scale wind speed UR 38.2
et al., 2004; Mathelin et al., 2001). As a consequence the energy m/s). The spectra of cross-wind velocities at levels D1, D3 and D4
86 M. Belloli et al. / J. Wind Eng. Ind. Aerodyn. 126 (2014) 7586
conrm also in this case the absence of a vortex shedding
synchronization on the whole structure.
5. Conclusions
This paper deals with wind tunnel tests on a tall slender tower
whose cladding forms an intricate three-dimensional spiral made
up of perforated steel plates. The effect of the porosity on the
dynamic response of the tower was the main object of the study.
The contributions of this paper can be summarized as follows:
 The overall wind load at the base of the spire indicates that the
impermeability of the panels slightly increases the mean along-
wind force but does not modify the mean cross-wind compo-
nent which remains close to zero. Otherwise the dynamic
component is considerably increased: the peak drag coefcient
in the impermeable conguration is almost the double while
the cross-wind coefcient can reach at certain exposure angles
a value almost ten times bigger.
 Local wind loads on the porous panel are found to be lower
than those on the same non-porous panel. This happens
because air ows through the porous surface, tending to
equalize the mean pressures and attenuating the peak pres-
sures across the panel, both of which reduce the total wind
loading on the panel. The analysis of the pressures on the
external panels proves that permeability reduces the wind load
(suction) mainly when the surface is tangential to the ow.
 The spire in its in-service conguration (i.e. porous panels) is in
practice not affected by vortex shedding induced vibrations.
Vortices associated with section D2 and D3 were found in the
wake of the structure, but the energy introduced by the wind is
not enough to the induce a synchronous vibration of the entire
structure.
 On the other hand the fully non-porous spire suffers consider-
able vortex shedding induced vibration which, owing to the
presence of four different diameters along the height, occurs in
a wide range of wind speeds and is generated by both low and
very frequent winds and less frequent but very high winds. The
lock-in phenomenon has been also observed.
The analysis carried out has highlighted the signicant inu-
ence of porosity, mainly in relation to vortex shedding induced
vibrations. Owing to its particular position at the top of a tall
building, the structure could be subjected to atmospheric icing. For
this reason, the inuence of porosity is worth considering in the
design stage, which should include an adequate de-icing device.
Acknowledgments
The authors gratefully acknowledge the support given by
Colombo Costruzioni S.p.A, SCT Informatica S.r.L. and S.E.VAL S.r.L.
for this research activity.
References
Baek, H., Karniadakis, G.E., 2009. Suppressing vortex-induced vibrations via passive
means. J. Fluids Struct. 25 (5), 848866.
Ballio, G., Maberini, F., Solari, G., 1992. A 60 year old, 100 m high steel tower: limit
states under wind actions. J. Wind Eng. Ind. Aerodyn. 43 (13), 20892100.
Bishop, R.E.D., Hassan, A.Y., 1964. The lift and drag forces on a circular cylinder
oscillating in a owing uid. Proc. R. Soc. Lond. Ser. A Math. Phys. Sci. 277
(1368), 5175.
Briassoulis, D., Mistriotis, A., Giannoulis, A., 2010. Wind forces on porous elevated
panels. J. Wind Eng. Ind. Aerodyn. 98 (12), 919928.
Chen, S.S., 1987. Flow-induced Vibration of Circular Cylindrical Structures. Hemi-
sphere Pub. Corp., Washington.
Cheung, J.C.K., Melbourne, W.H., 1988. Wind loading on a porous roof. J. Wind Eng.
Ind. Aerodyn. 29 (13), 1928.
Cook, N.J., 1985. The designer's guide to wind loading of building structures.
Building Research Establishment Report. Building Research Establishment,
Department of the Environment; Butterworths, Garston, Watford; London;
Boston.
Cook, N.J., Mayne, J.R., 1978. Acquisition, analysis and application of wind loading
data. Building Research Establishment, Garston, Watford, UK, pp. 13391355.
Cook, N.J., Mayne, J.R., 1979. Novel working approach to the assessment of wind
loads for equivalent static design. J. Ind. Aerodyn. 4 (2), 149164.
Fay, G., 1994. Introduction to Fluid Mechanics. MIT Press, Cambridge, MA.
Fikke, S., Ronsten, G., Heimo, A., Kunz, S., 2006. Cost 727: atmospheric icing on
structures. Measurements and data collection on icing: state of the art.
Publication of MeteoSwiss 75, 110 pp.
Fransson, J.H.M., Konieczny, P., Alfredsson, P.H., 2004. Flow around a porous
cylinder subject to continuous suction or blowing. J. Fluids Struct. 19 (8),
10311048.
Glanville, M.J., Kwok, K.C.S., 1995. Dynamic characteristics and wind induced
response of a steel frame tower. J. Wind Eng. Ind. Aerodyn. 5455, 133149.
Hansen, S.O., Dyrbye, C., 1996. Wind Loads on Structures. John Wiley & Sons,
Chichester.
Holmes, J.D., 2004. Wind Loading of Structures. Spon Press; Taylor & Francis,
London, New York.
Irwin, P., Kilpatrick, J., Robinson, J., Frisque, A., 2008. Wind and tall buildings:
negatives and positives. Struct. Des. Tall Spec. Build. 17 (5), 915928.
Irwin, P.A., 2009. Wind engineering challenges of the new generation of super-tall
buildings. J. Wind Eng. Ind. Aerodyn. 97 (78), 328334.
Kim, Y.-M., You, K.-P., Ko, N.-H., 2008. Across-wind responses of an aeroelastic
tapered tall building. J. Wind Eng. Ind. Aerodyn. 96 (89), 13071319.
Letchford, C.W., 2001. Wind loads on rectangular signboards and hoardings. J. Wind
Eng. Ind. Aerodyn. 89 (2), 135151.
Letchford, C.W., Row, A., Vitale, A., Wolbers, J., 2000. Mean wind loads on porous
canopy roofs. J. Wind Eng. Ind. Aerodyn. 84 (2), 197213.
Mallipudi, S., Selig, M., 2004. Use of a four-hole cobra pressure probe to determine
the unsteady wake characteristics of rotating objects. In: Proceedings of 24th
AIAA Aerodynamic Measurement Technology and Ground Testing Conference,
Portland, Oregon.
Mathelin, L., Bataille, F., Lallemand, A., 2001. Near wake of a circular cylinder
submitted to blowing ii: impact on the dynamics. Int. J. Heat Mass Transf. 44
(19), 37093719.
Mulherin, N., 1996. Atmospheric icing and tower collapse in the united states. In:
7th International Workshop on Atmospheric Icing of Structures (IWAIS'96),
Chicoutimi, Quebec, Canada.
Repetto, M.P., Solari, G., 2010. Wind-induced fatigue collapse of real slender
structures. Eng. Struct. 32 (12), 38883898.
Richards, P.J., Robinson, M., 1999. Wind loads on porous structures. J. Wind Eng. Ind.
Aerodyn. 83 (13), 455465.
Robertson, A.P., Roux, P., Gratraud, J., Scarascia, G., Castellano, S., Dufresne de Virel, M.,
Palier, P., 2002. Wind pressures on permeably and impermeably-clad structures. J.
Wind Eng. Ind. Aerodyn. 90 (45), 461474.
Shimada, K., Hibi, K., 1995. Estimation of wind loads for a super-tall building (ssh).
Struct. Des. Tall Build. 4 (1), 4760.
Sundin, E., Makkonen, L., 1998. Ice loads on a lattice tower estimated by weather
station data. J. Appl. Meteorol. 37 (5), 523529.
Zasso, A., Belloli, M., Giappino, S., Muggiasca, S., 2008. Pressure eld analysis on
oscillating circular cylinder. J. Fluids Struct. 24 (5), 628650 0889-9746.
Zdravkovich, M.M., 1997a. Flow Around Circular Cylinders. 1: Fundamentals. Oxford
University Press, Oxford.
Zdravkovich, M.M., 1997b. Flow Around Circular Cylinders. 2: Applications. Oxford
University Press, Oxford.
Zheng, C., Zhang, Y., 2010. Numerical investigation of wind-load reduction for a
high-rise building by blowing control. J. Build. Struct. 31 (Suppl. 2), S176S181.
Zou, L., Liang, S., Li, Q.S., Zhao, L., Ge, Y., 2008. Investigation of 3-d dynamic wind
loads on lattice towers. Wind Struct. Int. J. 11 (4), 323340.