Fifty Years of Wind Engineering

Prestige Lectures from the

Sixth European and African
Conference on Wind Engineering

Editors: C J Baker, D M Hargreaves, J S Owen and M Sterling

Fifty Years of Wind Engineering: Prestige Lectures from the Sixth
European and African Conference on Wind Engineering

Chairs: J S Owen and M Sterling

Editors: C J Baker, D M Hargreaves, J S Owen and M Sterling

ISBN-13: 9780704428348

Typeset using the LATEX typesetting system by D M Hargreaves

ii
Foreword
Prof. Chris Baker, The University of Birmingham

Much has changed since 1963. In international terms political power was
in the hands of the generation who had lived through the second-world war,
and the scars of that war were still very noticeable across Europe. The stand-
off between east and west known as the cold war was at its height, and the
Cuban missile crisis still fresh in people’s minds. In social terms there was
a broad social democratic consensus across Europe, with the state playing a
major role in industry, education, transport and many other sectors.
The subject of wind engineering, in so far as it had reached any level of
self definition, mirrored the society in which its practitioners lived. Research
work was concentrated in the state owned research laboratories the National
Physical Laboratory and the Building Research Station in the UK for example
- and the concerns were those of the society around it. A glance at the contents
of the 1963 meeting on Wind Effects on Buildings and Structures held at the
National Physical Laboratory in Teddington will reveal these interests. In
that contents list we see papers on the design of long span bridges, necessary
for the development and reinvigoration of long distance transport links; the
need to eliminate conductor cable galloping to make the electricity supply
network more resilient; and the need to develop codes of practice so that
the housing stock could be improved and able to withstand gales such as the
fiercely destructive Sheffield gale of 1962.
Whilst wind engineering has a long history, arising from incidents such as
the Tay Bridge and Tacoma Narrows collapse, with solid foundational work
carried out by the likes of Prandtl and von Karman in Germany, Eiffel in
France, Irminger and Rathbun in the USA, Jensen in Denmark and Scruton
in the UK, in 1963 new ideas and concepts were being formulated, and young
researchers were arriving on the scene who were to have a major influence on
the future of the discipline. Most notable amongst these was Alan Davenport
who made a major contribution to the 1963 NPL meeting, and who, through
the conceiving and the development of the “wind loading chain” was to make
such a major contribution to the general development of the discipline. The
meeting also saw discussion of many things we now take for granted - not least
the development of wind tunnel boundary layer simulations in large environ-
mental wind tunnels, which at that time was still a developing technology, if
no longer a novelty.
The 1963 meeting thus marks a pivotal moment in the development of wind

iii
engineering, and it is appropriate to mark the 50th anniversary of that meeting
as part of the 2013 European and African Conference on Wind Engineering.
The chapters of this volume contain the keynote lectures from this conference
that review the development of six key themes in wind engineering over the
50 years since the 1963 conference, and describe the current state of the
art. Our understanding of the atmospheric boundary layer is fundamental to
Wind Engineering and Prof. Barlow helpfully reminds us that in 1963 there
was significant debate on how best to represent it to determine wind loads
for design. She also highlights how the rapid advance in instrumentation
is providing new data that is transforming our understanding of this key
area today. Such understanding will feed into the next generation of wind
loading codes, which were in their infancy in 1963. Now, in Europe at least,
national codes have been overtaken by the Eurocode and Dr Hansen outlines
in this volume the plans for further harmonisation. The last fifty years have
seen great changes and developments in both buildings and bridges. Prof.
Borri charts our growing understanding of wind effects on bridges and Prof.
Letchford reminds us that we face new challenges and many issues remain
unresolved. In 1963, the tool of choice for wind engineering research was
the wind tunnel. Prof. Diana illustrates how technological developments
have continued to improve the quality and quantity of experimental data and
facilities have now been developed to investigate other extreme wind events
such as downbursts. However, the intervening years have also seen the rise
of Computational Wind Engineering, almost unheard of in 1963 but now, as
Prof. Blocken shows, a major contributor to the development of the discipline.
It will be clear from reading these chapters that wind engineering in 2013
owes a very great deal to those who gathered at Teddington in 1963. Yet, those
six themes remain the focus of much research today and it is also clear that
the most recent developments offer the potential to transform the practice of
wind engineering. In Cambridge in 2013 we gather for the sixth in the series
of regional conferences organised under the auspices of the IAWE. Delegates
from 20 nations around the world have contributed over 120 papers that
continue in the vibrant and successful tradition of the previous conferences
in Guernsey, Genoa, Eindhoven, Prague and Florence. It is our hope that
in 2063 wind engineers will look back on EACWE 2013 as another pivotal
moment in the development of the discipline, but in doing so we remember
the words of one of Cambridge’s greatest scholars, “If I have seen further it is
by standing on the shoulders of giants.”

iv
Distinguished Lecturers

Professor Janet Barlow

Janet Barlow is Professor of Environmental Physics at the Department of Me-

teorology at the University of Reading. She has interests in urban meteorol-
ogy, turbulence, remote sensing of the boundary layer and novel experimental
techniques in both wind tunnel and the field.

Professor Bert Blocken

Bert Blocken is full professor and holds the Chair of Building Physics at
Eindhoven University of Technology in the Netherlands. His main areas of
expertise are Urban Physics and Environmental Wind Engineering.

Professor Claudio Borri

Claudio Borri is full professor of Computational Mechanics of Structures, Uni-

versity of Florence (Italy) and Vice-President of the Interuniversity Research
Centre on Building & Environmental Aerodynamics.

Professor Giorgio Diana

Giorgio Diana is Professor Emeritus of Applied Mechanics at Politecnico

di Milano. He is Director of the Research Centre for Wind Engineering
(CIRIVE), whose wind tunnel is the largest boundary Layer Wind Tunnel
in Europe.

Dr Svend Ole Hansen

Svend Ole Hansen is director and founder of Svend Ole Hansen ApS and has
been a lecturer in wind engineering at the Technical University of Denmark
and Syddansk University. His text books on wind engineering are widely read
and well regarded.

Professor Chris Letchford

Chris Letchford is Professor and Head of Department of Civil & Environ-

mental Engineering at Rensselaer Polytechnic Institute. His research inter-
ests cover a wide range of wind effects on structures including thunderstorm
downbursts and tornadoes.

v
Scientific Committee Members
Dr J Macdonald (Chairman, UK) Dr A Quinn (Secretary, UK)
Prof. G Augusti (It) Dr J Barlow (UK)
Prof. J Bogunovic-Jakobsen (Nor) Prof. C Borri (It)
Prof. A Flaga (P) Dr O Flamand (Fr)
Dr C Georgakis (Dk) Dr C Geurts (NL)
Dr A Goliger (RSA) Dr G Grillaud (Fr)
Prof. R Hoeffer (Ger) Dr J Holmes (Aus)
Prof. N Jones (USA) Prof. A Kareem (USA)
Prof. K Kwok (Aus) Prof. C Letchford (USA)
Prof. J Naprstek (Cz) Prof. H-J Niemann (Ger)
Dr A Palmeri (UK) Prof. U Peil (Ger)
Dr S Pospisil (Cz) Prof. P Richards (NZ)
Prof. G Solari (It) Prof. T Stathopoulos (Can)
Prof. Y Tamura (Jap) Dr I Taylor (UK)
Dr N Waterson (UK) Dr Z-T Xie (UK)
Prof. A Zasso (It)

vi
 Contents

1 The Wind That Shakes the Buildings: Wind Engineering

from a Boundary Layer Meteorology Perspective
J.F. Barlow 1
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Wind Engineering in the 1960s: the 1963 Conference . . . . . . 2
1.3 State of the Art Today . . . . . . . . . . . . . . . . . . . . . . . 9
1.4 Research Challenges for the Future . . . . . . . . . . . . . . . . 26
1.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

2 Wind Loading Design Codes

S.O. Hansen 35
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.2 Eurocodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
2.3 Wind Climate . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2.4 Terrain Effects . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
2.5 Aerodynamic Response . . . . . . . . . . . . . . . . . . . . . . . 44
2.6 Mechanical Response . . . . . . . . . . . . . . . . . . . . . . . . 54
2.7 Design Criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
2.8 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

vii
3 Urbanization and Wind Effects on Buildings
C.W. Letchford and D. Menicovich 69
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
3.2 Urbanization, Cities and Buildings . . . . . . . . . . . . . . . . 70
3.3 Understanding Air Flow and Form . . . . . . . . . . . . . . . . 76
3.4 Wind Loading of Buildings . . . . . . . . . . . . . . . . . . . . 82
3.5 Across-wind Response in Tall Buildings . . . . . . . . . . . . . 86
3.6 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101

4 The Long Way from Teddington (1963) to Cambridge (2013)

Through 50 Years of Bridge Aerodynamics
C. Borri, G. Bartoli and C. Mannini 103
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
4.2 The Phenomena of Major Concern in Bridge Aerodynamics . . 104
4.3 Evolution of Long-span Bridges . . . . . . . . . . . . . . . . . . 122
4.4 Wind Tunnel Facilities for Bridge Aerodynamics . . . . . . . . 128
4.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149

5 Wind Tunnel Testing Developments in the Last 50 Years

G.Diana, M. Belloli, S. Giappino, A. Manenti, L. Mazzola, S. Mug-
giasca and A.Zuin 151
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
5.2 Wind Tunnel Testing in the 1960s . . . . . . . . . . . . . . . . 152
5.3 Progress in Wind Tunnel Facilities . . . . . . . . . . . . . . . . 154
5.4 Subspan Oscillations . . . . . . . . . . . . . . . . . . . . . . . . 160
5.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178

6 50 Years of Computational Wind Engineering: Past, Present

and Future
B. Blocken 181
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
6.2 Some Early CWE Achievements . . . . . . . . . . . . . . . . . 185
6.3 CWE Best Practice Guidelines . . . . . . . . . . . . . . . . . . 186
6.4 CWE Symposia: Historical Background and Scope . . . . . . . 189
6.5 CWE Review Papers . . . . . . . . . . . . . . . . . . . . . . . . 192
6.6 Some Quotes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
6.7 Evaluation of Pedestrian–Level Wind Conditions . . . . . . . . 205

viii
6.8 Summary and Future Perspectives . . . . . . . . . . . . . . . . 214
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227

ix
x
 Wind Loading Design Codes
2
S.O. Hansen

2.1 Introduction

In 1963, at the time of the first international conference on wind effects on

structures held in UK, the wind loading design codes were typically a few
pages long, and they covered only a few simple structures. As an example
the first Danish wind code was published in 1945, and it specified static wind
load acting on selected structural geometries. In the revised Danish wind code
published in 1966, specifications of gust factors for dynamic sensitive struc-
tures and vortex shedding were added. The codified gust factor approach
followed the model originally established by Alan G. Davenport in the be-
ginning of the sixties, see Davenport (1962) and Davenport (1967), and in
the following years, this model was widely acknowledged and applied in many
national wind loading design codes.
In 1963 most pressure and force coefficients were still obtained from wind
tunnel testing carried out in smooth flow. Although Martin Jensen’s model
law “The flow in the wind tunnel should be turbulent in the same way as the
flow in the natural wind” was put forward in the 1950s (Jensen, 1958), the con-
struction of boundary layer wind tunnels applying it was in its initial phase.
One of the first boundary layer wind tunnels able to generate scaled boundary
layers in accordance with Martin Jensen’s model law was constructed by Alan
G. Davenport at University of Western Ontario in Canada in the early sixties,
see Davenport and Isuymov (1968).

35
Wind Loading Design Codes

In the following years boundary layer wind tunnels were constructed world-
wide. Wind tunnel tests were carried out with simple as well as with more
complex structures, and the data obtained in these wind tunnel tests were
used to extend the scope of the wind loading design codes and to make the
specifications in much better compliance with full–scale conditions. However,
there has been a lack of a common worldwide basis with consistent definitions
for wind loading calculations, e.g. the wind climate has not been described
using a common definition of terrain roughness and averaging time of the
wind velocity, and the pressure and force coefficients have not been based on
the same wind loading model. This lack of consistency has led to many wind
loading design codes that provide somewhat different results. However, the
deviations are in most situations relatively minor, and now the time seems
to be mature for a harmonization of the many different wind loading design
codes.
In Europe this process of harmonization was initiated around 1980-90 with
the decision to prepare Eurocodes – see Section 2.2.
Many papers have dealt with comparisons between wind loads calculated
using different wind loading design codes, such as Eurocodes, ISO codes and
the ASCE codes, see e.g. Holmes et al. (2009) and Bashor and Kareem (2009).
In the present paper, the main focus is put on the Eurocodes, and the main
trends expected in future revisions.

2.2 Eurocodes
The present Eurocodes comprise more than sixty parts, one of them being the
Eurocode on wind actions.
The first Eurocode on wind actions, ENV 1991-2-4:1995, was completed in
1995, and it was followed by a revised Eurocode version, EN 1991-1-4:2005 in
2005. The process of drafting the EN Eurocode on wind action was discussed
by Zimmerli (2001) and selected background material of the drafting were
outlined by Geurts et al. (2001) and by Hansen et al. (2001).
In the Eurocode EN 1991-1-4:2005, the specification of the characteris-
tic wind load is structured in accordance with the wind load chain originally
introduced by Alan G. Davenport, see Dyrbye and Hansen (1996) and Fig-
ure 2.1. This was decided in order to make the presentation as user friendly
as possible.
The elements of the wind load chain are as follows:

1. The wind climate is specified by the basic wind velocity defined as the
10-minute mean wind velocity at 10 m height above reference terrain

36
 S.O. Hansen

Figure 2.1: Wind load chain introduced by Alan G. Davenport.

with a roughness length of 0.05 m and having a return period of 50 years

corresponding to an annual probability of exceedance of 0.02.
2. The terrain effects are specified by 5 standard flat terrains spanning
from category 0 with a roughness length of 0.003 m to category IV with a
roughness length of 1 m. Simplified rules for transition between terrains
having different roughnesses and the effect of orography are included in
the Eurocode. The extreme winds and terrain effects specified are used
to calculate the peak velocity pressure, qp , which is the basic parameter
describing the incoming undisturbed airflow approaching the structure.
3. The aerodynamic response is determined by multiplying the peak veloc-
ity pressure by pressure coefficients and force coefficients specified for
the different structural geometries.
4. The mechanical response defines the response of the structure, e.g. in
form of deflections, accelerations and stresses.
5. The design criteria define the dimensioning criteria used to evaluate the
mechanical response calculated.
The following sections focus on each of the different elements of the wind
load chain.
The present Eurocodes do not cover phenomena such as transient wind
conditions and thermal effects on the characteristic wind, e.g. strong arctic
thermal surface inversion, hurricanes or tornadoes. Furthermore, torsional

37
Wind Loading Design Codes

vibrations, e.g. of tall buildings with a central core, vibrations where more
than the fundamental mode needs to be considered, cable supported bridges
and bridge deck vibrations from transverse wind turbulence are not covered
in the Eurocodes. Some of these phenomena are planned to be included in
the next Eurocode revision.
In each European member state the Eurocodes are applied together with
its respective National Annex indicating that the present lack of harmoniza-
tion is found by comparing the different National Annexes across Europe.
National Annexes may provide meteorological information and select the Eu-
rocode procedure that is actually to be applied in the design calculations.
Part of the harmonization is to reduce the number of National Determined
Parameters in future Eurocode revisions.
The European Committee for Standardization, CEN, plans to carry out
a coordinated revision of the Eurocodes during the coming 3-5 years. Work-
ing Groups with representatives from different European countries are being
established and the scope of the revisions is being discussed thoroughly. One
of the key issues is to make the Eurocodes more user friendly by introducing
simplifications wherever possible. In the planned Eurocode revision of EN
1991-1-4:2005 a part of the scope is to identify, and as far as possible to re-
move inconsistencies in determining wind actions and wind action effects on
structures covered in different Eurocode Parts and in ISO standards. Aspects
of the revision planned for EN 1991-1-4:2005 are discussed in the following
sections.
In the Eurocodes partial safety factors are used to convert characteristic
values to design values. For structures exposed to wind actions these partial
safety factors should take the uncertainties of all elements of the wind load
chain properly into account. Detailed calibration studies, see e.g. Vrouwen-
velder and Scholten (2010), have shown that the partial safety factor on wind
actions should be of the order of 1.7 and that the Eurocode partial safety
factor for wind action specified to be 1.5 may actually underestimate the in-
herent uncertainties. However, this underestimation of approximately 10-15%
is more than balanced out by safe estimates of characteristic values in some
of the elements of the wind load chain, e.g. in the definition of design criteria,
see Section 2.7 for further explanation.

2.3 Wind Climate

The wind map in Figure 2.2 shows basic wind velocities included in EN 1991-
1-4:2005. The lack of harmonization across the borders of Europe in 1995 is

38
 S.O. Hansen

Figure 2.2: North European wind map of basic wind velocities (m s−1 ) in EN 1991-
1-4:2005.

obvious. Some of the geographical areas lacking harmonization are:

1. The basic wind iso–velocity curves from different parts of UK are con-
flicting.

2. Denmark and Sweden on each side of the narrow Øresund strait specify
a basic wind velocity of 27 m s−1 and 23 m s−1 , respectively.

3. Denmark and Germany on each side of Fehmarn Belt specify a basic

wind velocity of 27 m s−1 and 32 m s−1 , respectively.

The obvious lack of harmonization shown in Figure 2.2 and the European
aim of establishing harmonized structural codes have initiated many local
studies of basic wind velocities using historical data. These studies have
reduced or removed some of the sudden jumps across borders, see Figure 2.3.

39
Wind Loading Design Codes

Figure 2.3: North European wind map of basic wind velocities (m s−1 ) according to
National Annexes of EN 1991-1-4:2005.

There is now seen to be a reasonable good agreement between basic wind

velocities specified in Norway, Denmark and Sweden. The specifications in
UK are now internally consistent, and they are in reasonable agreement with
extrapolations based on Scandinavian values. Not all parts of Europe show
the same consistency across borders, and this may be shown in an updated
wind map based on basic wind velocities specified in existing National Annexes
and by emphasizing the wind map values, where differences across borders still
exist. A map like that may urge local standards organizations to continue the
harmonization process, e.g. for Germany and Denmark where the change in
basic wind velocity crossing Fehmarn Belt is from 24 m s−1 in Denmark to
30 m s−1 in Germany.
Some member states have had to change their basic wind velocity definition
to the Eurocode basis, i.e. a 10-minute mean wind velocity at 10 m height
above a terrain with a roughness length of 0.05 m. Previously the UK, for

40
 S.O. Hansen

instance, used a 1 hour mean wind velocity at 10 m height above a terrain

with roughness length of 0.03 m. However, the UK did not have to prepare a
new wind map since the effect of changing the averaging time from 1 hour to
10 minutes balances out the effect of changing roughness length from 0.03 m to
0.05 m. The basic wind velocity defined in ISO 4354:2009 refers to a 10-minute
mean wind velocity at 10 m height above a terrain with roughness length of
0.03 m, and unfortunately this definition deviates from the Eurocode basis.
The wind map data illustrated in both Figure 2.2 and 2.3 have mainly
been based on historical data, and the effects of climate change have not been
included on a consistent basis. Recently, extensive research programs have
been carried out in order to establish a basis for determining the effect of
climate change on the wind conditions. As an example the Danish Meteoro-
logical Institute have found that the basic wind velocity may increase by up
to approximately 5% over the next 100 years if the most severe climate change
scenarios turn out to be realistic. The modest influence of climate change on
the basic wind velocity is confirmed by the trend indicated using the historical
data illustrated in Figure 2.4.
The relatively minor influence of climate change seems to have been in-
cluded already by the basic wind velocities conservatively estimated. How-
ever, climate change may increase the magnitude of frequent winds, and this
may increase the importance of fatigue damage of wind sensitive structures.
Climate change will be focused on in future Eurocode revisions.

2.4 Terrain Effects

The terrain effects include wind models for calculation of wind velocities and
peak velocity pressures based on terrain roughness and orography, and also
on terrain roughness transitions. The standard terrain categories specified
in the Eurocode are illustrated in Figure 2.5, taken from EN 1991-1-4:2005.
The illustrations of terrain categories I to IV were originally prepared by the
Danish Standards Foundation.
The European Member States specify many different models for taking
the transition between roughness categories into account. This topic is one of
the main aspects for harmonization in the Eurocode revision planned.

41
Wind Loading Design Codes

Figure 2.4: Annual extreme mean velocity pressures measured at 10 m height at

Kastrup airfield in Denmark from 1953-2012 by the Danish Meteorological Insti-
tute. The annual extremes, ql , have been normalized by the 50-year mean velocity
pressure, q50 , based on the site measurements.

42
 S.O. Hansen

Category 0
Sea and coastal areas exposed to the
open sea.
Roughness length 0.003 m.

Category I
Lakes or areas with negligible vege-
tation and without obstacles.
Roughness length 0.01 m.

Category II
Areas with low vegetation such as
grass and isolated obstacles (trees,
buildings) with separations of at
least 20 obstacle heights.
Roughness length 0.05 m.

Category III
Areas with regular cover of vegeta-
tion or buildings or with isolated ob-
stacles with separations of a max-
imum of 20 obstacle heights (such
as villages, suburban terrain, perma-
nent forest).
Roughness length 0.3 m.

Category IV
Areas in which at least 15% of the
surface is covered with buildings and
their average height exceeds 15 m.
Roughness length 1.0 m.

Figure 2.5: Terrain categories in EN 1991-1-4:2005.

43
Wind Loading Design Codes

Figure 2.6: Recommended Eurocode procedure for determining the external pressure
coefficient cpe for buildings with a loaded area between 1 m2 and 10 m2 .

2.5 Aerodynamic Response

The sections below focus on wind-induced pressures on a one skin façade and
two skin façade, respectively, and the global wind load on a structure is also
analysed.
In the planned revision of the Eurocode a systematic review of internal
and external pressure coefficients, including influence areas and the zoning,
will be carried out based on the current state of the art. Furthermore, force
coefficients for global wind loads, e.g. for the design of foundations, will be
included.

2.5.1 Wind–induced Pressures on a One Layer Façade

Cook (1985) has carried out a large number of wind tunnel measurements and
simulations in order to answer the following question: What is the value of the
loading coefficient that results in a design load of the desired design risk, given
a wind speed of the same risk? Thus, for the external pressures the question
is: Which pressure coefficient cpe provides a characteristic wind pressure cal-
culated by we = cpe qp , in which qp is the characteristic peak velocity pressure?
Cook (1985) found that the pressure coefficient should be obtained as the 78%
fractile in the Gumbel distribution of pressure extremes.
As illustrated in Figure 2.6, the Eurocode specifies external pressure co-
efficients as a function of loaded area, and the tabulated values give cpe,10
representing a loaded area of 10 m2 to be used for the wind load on the main
structural elements, and cpe,1 representing a loaded area of 1 m2 to be used
in the design of fixings, smaller panels, etc.
The majority of the Eurocode pressure coefficients are based on Cook
(1985) and Stathopoulos (1979) for 10 m2 and 1 m2 loaded areas, respectively.

44
 S.O. Hansen

Figure 2.7: Photo of a pressure tap cluster on model. Each pressure tap has a
full-scale equivalent area of less than 1 m2 .

Cook used time averaging for determining 10 m2 wind loads indicating that
these values have not been based directly on a spatial averaging.
After the publication of the EN 1991-1-4 in 2005, a large number of mea-
surements have been carried out in our wind tunnel aimed at determining
both 1 m2 and 10 m2 loads, as well as loads on intermediate areas, on a con-
sistent basis. This has been accomplished by installing pressure tap clusters
on the wind tunnel models, see Figure 2.7. The pressures in each pressure tap
cluster have been measured simultaneously indicating that spatial averaging
could be used to estimate the wind action in different areas of up to approxi-
mately 10 m2 . Thus, the procedure applied facilitates a determination of both
1 m2 and 10 m2 pressure coefficients, which could be compared directly to the
same type of pressure coefficients specified in the Eurocode.
Typical trends for the loads measured on façades are:

• For 1 m2 loaded areas the measurements show larger suctions than the
Eurocode value of -1.4 for façades. This may partly originate from the
fact that each pressure tap has an area of less than 1 m2 .

• For 10 m2 loaded areas the measurements show lower suctions than the
Eurocode value of -1.2 for façades. Thus, the spatial averaging applied
in the wind tunnel gives larger reductions than the Eurocode.

The measurements illustrated in Figures 2.8 to 2.10 show two case studies,

45
Wind Loading Design Codes

(a) (b)

Figure 2.8: The Concert and Conference Centre “Harpa” in Reykjavik, Iceland:
(a) photo of 1:200 wind tunnel scale model; (b) normalised pressure coefficients
measured on the Eastern façade in separated zones with large suctions as function
of load duration and load area. Reproduced by permission of Ramboll Denmark,
Consulting Engineers of the building.

where the influence of both time averaging and spatial averaging of simulta-
neously measured external pressures are illustrated.
The effect of spatial averaging and time duration for pressures measured
at Søndermarken is shown in Figure 2.10. In this figure, the time duration
has been made non-dimensional by multiplying it with the mean wind velocity
divided by the dimension e from the Eurocode – the minimum of 2 heights and
the cross wind dimension determining the sizes of the different zones on the
structure. This choice of length scale has been found to be in good agreement
with measurements carried out in the separated zones with large suctions on
different models.
Figures 2.8(b) and 2.10 show examples of characteristic pressure coeffi-
cients as function of load durations for loaded areas of 1 m2 , 2 to 3 m2 and
10 m2 , respectively. The effect of load duration is significant for all loaded
areas considered, indicating that the largest suction peaks have very short
durations of the order of 0.1 s. Especially for short durations the averag-
ing effect is pronounced. It may be observed that the pressure coefficient is
approximately 30 to 40% smaller for the 10 m2 loads with duration of approxi-
mately 1 s compared to its value for 1 m2 loads with duration of approximately
0.1 s. A consistent approach with data sets including both the effect of dura-
tion and loaded area are needed. The data presented show the importance of

46
 S.O. Hansen

Figure 2.9: Photo of a 1:200 wind tunnel scale model of a block building in
Søndermarken in Copenhagen, Denmark. The pressure tap model is situated in
the middle of the circle. Reproduced by permission of Moe, Consulting Engineers,
and the building owner, KAB.

both load duration and spatial averaging.

More accurate design calculations may be carried out if the wind load data
could be combined with structural resistances as function of load durations.
However, the resistance data available for most structures are not sufficiently
accurate to support a detailed design of this nature. Thus, the structural
resistances are often underestimated for the short duration loads indicating
that the present approach is conservative. However, it is not possible to
quantify the actual degree of conservatism at present. This is discussed further
in Section 2.7.

47
Wind Loading Design Codes

Figure 2.10: Normalised pressure coefficients from measurements of façade wind

loads on a structure. The nondimensional x-axis is averaging time, t, multiplied by
the mean wind velocity, vm , and divided by the Eurocode-defined dimension of e.
Reproduced by permission of Moe, Consulting Engineers, and the building owner,
KAB.

2.5.2 Wind–induced Pressures on a Two Skin Façade

The data provided in the Eurocode for two skin façades is based on ECCS
(1987). However, the background needs to be extended with further test
results in order to cover the typical full-scale situations sufficiently accurately.
The two skin façade considered here has an impermeable inside skin and
a permeable outside skin. The wind-induced pressures on the two skin façade
depend on many geometrical parameters, such as the distance between the
layers and the permeability of the outside skin. The Eurocode recommends
that there are no openings at the extremities of the layer between the skins,
and in this situation the specifications are given for distances between the skins
of less than 100 mm and when the outside skin has approximately uniformly

48
 S.O. Hansen

Figure 2.11: 1:25 scale model of two skin façade of a block building in Søndermarken
in Copenhagen. The photo shows the model in the large wind tunnel of SOH Wind
Engineering in Vermont, USA. The vertical oriented cavity is shown in the middle of
the photo, where the model is constructed of plexiglass. Reproduced by permission
of Moe, Consulting Engineers, and the building owners, KAB.

distributed openings. According to the Eurocode the net pressure coefficient

for overpressure and under pressure are 2/3 and 1/3 of the external pressure
coefficient, respectively.
An example with an inner impermeable skin and an outer permeable skin
at a distance of 50 mm has been carried out in a large wind tunnel on a scale
of 1:25, see Figure 2.11. The permeability of the outer skin is approximately
10%. The air flow in the cavity is mainly able to move vertically due to
vertical barriers from top to bottom.
Figure 2.12 shows the pressures measured in the wind tunnel tests carried
out simulating a relatively low turbulent flow. The net pressure acting on the
outer skin is determined as the difference between the two curves on each of
the two plots in Figure 2.12. The results presented in the figure show net
pressure coefficients on the outer skin being less than 0.1, i.e. much lower
values than specified in the Eurocode. Thus, the pressure equalization is
much more effective than assumed in the Eurocode for the particular geometry
investigated.
The net pressures have also been determined by CFD using boundary

49
Wind Loading Design Codes

Figure 2.12: Normalised external pressures on outside skin and normalised cavity
pressures as function of time for a two skin façade. The net pressures acting on
the outside skin are determined as the pressure difference illustrated by the black
curves. Left-hand side: Pressures in windward side. Right-hand side: Pressures in
separated zone with large suctions on along-wind side. Reproduced by permission
of Moe, Consulting Engineers, and the building owner KAB.

pressure conditions measured in a wind tunnel in turbulent flow on a model

with an impermeable skin. These CFD calculations confirm that the pressure
equalization is very efficient for the present geometrical configuration, and
that the net pressures become insignificant.
Measured pressure coefficients for the many different geometrical configu-
rations possible are needed in order to establish a sufficiently broad code basis.
The results above are one example where the pressure equalisation across the
external outer skin is very effective in reducing the wind action on this skin.

2.5.3 Wind–induced Global Loads

The characteristic wind load, Fw , acting on the structure is determined by
the following expression specified in the Eurocode EN 1991-1-4:2005,

Fw = qp cs cd cf Aref ,

where qp is the characteristic peak velocity pressure at reference height, cs cd

is the structural factor, cf is the force coefficient, and Aref is the reference
area. The structural factor comprises of a size effect expressed by cs and a
dynamic amplification effect given by cd in the Eurocode.
The background of the two models presented in the Eurocode for detailed
calculations of the structural factor in dynamic response is documented by

50
 S.O. Hansen

Solari (1993a,b) and Hansen and Krenk (1999), respectively. The models are
based on the approach originally established by Davenport – see Davenport
(1962) and Davenport (1967).
Figure 2.13 shows a wind tunnel model where different size effects have
been determined based on wind tunnel measurement. The test results illus-
trated below have been measured with a wind direction being perpendicular
to the largest facade. The turbulent boundary layer air flow simulated has a
roughness length of approximately 0.02 m.
The local pressures and suctions measured are illustrated in Figure 2.14.
Figure 2.15 shows time histories of pressures averaged spatially over half of
the windward and leeward façades, respectively. The Eurocode size factor,
cs , applied in Figure 2.15 has been calculated using Eurocode method 2 for
dynamic response with a structure height of 20.2 m and a cross-wind width of
26.8 m (53.6/2) corresponding to the area of the spatially averaged pressures.
The turbulent length scale calculated was 102 m based on the Eurocode for
the air flow simulated.
Figure 2.14 shows that the Eurocode is in good agreement with the local
pressures and suctions measured on the windward and leeward sides of the
building. However, Figure 2.15 shows smaller global force coefficients for both
the windward side and the leeward side compared to the Eurocode values.
In the right-hand figure the Eurocode has been interpreted by multiplying
the size factor of a side with the Eurocode factor of 0.85 taking the lack
of correlation between windward side and leeward side into account. This
interpretation is seen to be conservative for the total force adding pressures
on the windward side to suctions on the leeward side.
The small variations of the averaged suctions on the leeward side are due
to a small length scale mainly originating from the dimensions of the structure
and not from the incoming air flow. The Eurocode dimension of e is often a
relatively good estimate of a representative length scale on sides located in
the zones governed by separated flows, see Section 2.5.1.
As described above the present Eurocode gives guidance for taking lack
of correlation between pressures acting on the windward side and leeward
side, respectively, into account when calculating wind actions on buildings.
However, for reasons of simplification this effect has not been included in the
Eurocode methods for calculating the structural factor for structural elements,
where the forces are specified by force coefficients. This lack of consistency be-
tween forces based on pressure coefficients and force coefficients, respectively,
may be removed in the Eurocode revision planned.

51
Wind Loading Design Codes

Figure 2.13: Wind tunnel model on a scale of 1:75. The façade height × width ×
depth is 20.2 × 53.6 × 27.2 m. Reproduced by permission of SiteCover.

Figure 2.14: Coefficient for local pressure and suction measured at windward and
leeward sides, respectively. The dotted lines indicate Eurocode pressure coefficients
of 1.0 and -0.43. Reproduced by permission of SiteCover.

52
 S.O. Hansen

Figure 2.15: Left-hand figure and middle figure show coefficients for averaged pres-
sure on windward side and leeward side, respectively. Right-hand figure also includes
averages of the total coefficient adding pressures on windward side to suctions on
leeward side. The dotted lines indicate Eurocode pressure coefficients: cs = 0.91,
cP P
pe,10 = 0.77 and cpe,10 = −0.43. Reproduced by permission of SiteCover.

53
Wind Loading Design Codes

2.6 Mechanical Response

The mechanical response described below focus on some of the topics judged
to be most important to modify in the panned Eurocode revision:

1. Influence lines or mode shapes with changing signs.

2. Vortex-induced vibrations.

3. Galloping-induced vibrations.

4. Special issues with cross section ratio 1:2, where vortex-induced and
galloping-induced vibrations interact.

These topics are covered in the following sections.

2.6.1 Influence Lines or Mode Shapes with Changing Signs

The present Eurocode does not cover structures where influence lines or mode
shapes have changing signs. In order to calculate accurate wind actions on
structures such as torsionally sensitive high-rise buildings, multi-span bridges
and guyed masts, specifications for influence lines or mode shapes chang-
ing signs will be included in the revised Eurocode. Thus, the scope of the
Eurocode will be extended by the new specifications. The challenge of es-
tablishing these new specifications is the conversion of existing background
information given in text books, see for example, Dyrbye and Hansen (1996),
to user friendly codified rules.

2.6.2 Vortex-induced Vibrations

For a code, the most important aspect with respect to aeroelastic phenom-
ena may be to specify a simple rule describing whether the structure to be
designed may be susceptible to large vibrations. This is the main focus be-
hind describing the influence of the general non-dimensional mass-damping
parameter ScG defined by
2δs me
ScG = , (2.1)
ρbd
where ρ is the air density, δs quantifies the structural damping by the log-
arithmic decrement, which is approximately 2πξs , where ξs is the structural
damping ratio, b is the crosswind width, d is the alongwind depth and me is
the mass of the structure per unit length. When d is equal to b, this parameter
is identical to the Scruton number, Sc = 2δs me /ρb2 .

54
 S.O. Hansen

Traditionally the Scruton number has been used to determine effects of

motion-induced vortex shedding wind loads acting on structures, but as it
depends on the cross-wind dimension only, and not on the in-wind dimen-
sion, the Scruton number does not contain any information on the wind
load exposed length. Instead Hansen (2013) and Hansen (2011) propose to
base the evaluations on the general non-dimensional mass-damping param-
eter, which includes the wind load exposed length and describes effects of
motion-induced wind load consistently for different structures. This general
non-dimensional mass-damping parameter may be derived by assuming that
the motion-induced wind load is proportional to the wind load exposed length
being equal to the in-wind dimension. Wind tunnel tests carried out with sec-
tion models having the cross sections illustrated in Figure 2.16 confirm that
it describes effects of motion-induced wind load more consistently than the
Scruton number normally applied. The special issues of cross section 1:2
deliberately excluded from Figure 2.16 are described in Section 2.6.4.
Figure 2.17 shows the vortex-induced vibrations measured in the wind
tunnel at the resonance wind velocity after build-up. The vibrations are
expressed by the standard deviation, σy , of the displacement normalised by
the cross-wind dimension, b, and the two illustrations are based on the same
wind tunnel test results, however, plotted as function of the Scruton number
and the general mass-damping parameter, respectively.
For low Scruton numbers galloping vibrations may occur at wind veloci-
ties below the resonance wind velocity for vortex shedding. However, these
Scruton numbers are well below the Scruton numbers at which vortex-induced
vibrations are illustrated in Figure 2.17.
The largest amplitudes in Figure 2.17 are seen to be highly dependent
on the cross-sectional shape. It is not surprising that the bridge-like cross
sections experience the smallest vibrations since increasing vibration ampli-
tudes changes the air flow pattern for even small relative angles of attack for
these types of cross section. The largest vibrations are observed for the cross
sections of 1:1, 2:1 and 4:1. Large vibrations for square cross sections are con-
firmed in studies by others, see e.g. Hjorth-Hansen and Kyrkjeeide (1978),
Kawai (1992), Scruton (1981) and Scruton (1963).
In typical designs the most interesting aspect of Figure 2.17 is to determine
the mass damping parameters at which the vibrations start to occur. It is
seen from the left-hand figure that vibrations starts to occur when the Scruton
numbers are in the range of approximately 5 to 100, i.e. a very large range
of critical Scruton numbers depending on the actual cross section in question.
The general non-dimensional mass-damping parameter applied in the right-
hand figure narrows down the critical mass-damping parameter range to be

55
Wind Loading Design Codes

Figure 2.16: Cross sections tested (Hansen, 2011, 2013). Turbulence intensity in the
tests was approximately 2%.

approximately 25 to 50 when the 1 : 5 bridge-like cross section is excluded.

It seems that the assumption of a damping force being proportional to the
along-wind depth may be too conservative for rectangular bridge-like cross
sections, when the along-wind dimension is more than 4 to 5 times the cross-
wind dimension. This is not surprising remembering that the vortices are
generated at the upstream edges giving vortex shedding forces mainly acting
at the upstream part of the cross section located in the vortex regions at top
and bottom.
The cross sections of 4:1, 2:1 and 1:1 experience very large vibrations and
standard deviations of approximately 10% of their cross-wind dimension occur
for Scruton numbers in the range of approximately 5 to 30. The equivalent
range of the general non-dimensional mass-damping parameter is approxi-
mately 25 to 30, i.e. a much smaller range.
The narrow ranges of the general non-dimensional mass-damping parame-
ter facilitates that vortex-induced vibrations of the rectangular cross sections
may be predicted by a theoretical model having only small variations in the
parameters used in the calculations. Using the general non-dimensional mass-
damping parameter rather than the Scruton number in analytical calculations
hence makes the specific geometry less important when representative aero-

56
 S.O. Hansen

Figure 2.17: Vortex-induced vibrations measured. The Scruton number is used as

x-axis in the left-hand figure, the general mass-damping parameter in the right-hand
figure. From Hansen (2011) and Hansen (2013).

57
Wind Loading Design Codes

dynamic parameters are determined, see Hansen (2013).

Future revisions of the Eurocode on wind actions may include the gen-
eral nondimensional mass-damping parameter predicting vortex-induced vi-
brations of different structures. This will reduce the uncertainties involved in
the response predictions.
BD 49/01 gives rules for calculating vortex-induced response of different
bridge cross sections, and the non-dimensional mass-damping parameter ap-
plied in BD 49/01 is found by replacing db with b1.5 d0.5 in Equation 2.1.
However, for a certain cross section their model specifies response amplitudes
being inversely proportional to the structural damping, which is very differ-
ent from the abrupt change from large to small vibrations indicated by the
measurements presented in Figure 2.17. Furthermore, the results presented
in Figure 2.17 show that BD 49/01 may underestimate the response, e.g. for
1:1 cross sections, where BD 49/01 specifies amplitudes 10 times smaller than
the measurements for ScG = 25. For more bridge-like cross sections of say
1:4 the agreement between BD 49/01 and the results of Figure 2.17 is better.
The special aeroelastic effects for 1:2 cross sections outlined in Section 2.6.4
are not considered by BD 49/01.

2.6.3 Galloping-induced Vibrations

The general non-dimensional mass-damping parameter proposed in Equa-
tion 2.1 may also be used in galloping response predictions. Assuming that the
driving force is proportional to the along-wind depth, d, gives the following
onset wind velocity of galloping, vCG , (Hansen, 2011)

vCG = 2ScG ne b/aGG,

where ne is the natural frequency of structure and aGG is the factor of gal-
loping instability corresponding to the use of the general mass-damping pa-
rameter, ScG ,  
dCL
aGG = − + CD , (2.2)
dθ
which is based on an aerodynamic damping estimate of
ρdvm
δaG = − aGG , (2.3)
4me ne
where CD and CL are the force coefficients for drag and lift respectively, using
the in-wind depth as a reference length, and θ is the angle of incidence of the
wind. The onset wind velocity of galloping corresponds to a total damping

58
 S.O. Hansen

of zero occurring when the negative aerodynamic damping becomes equal to

the structural damping.
The Eurocode specifies a factor of galloping instability aG from 0.4 to 2.0
for rectangular cross sections with the ratio of d/b in the range of 1/3 to
2, and these values are based on the traditional Scruton number definition.
The similar galloping instability factor of aGG based on the general non-
dimensional mass-damping parameter will be in a much more narrow range of
1.0 to 1.5. Thus, the tendency described above for vortex-induced vibrations is
also found for galloping-induced vibrations. For the 1 : 2 cross section with an
in-wind depth being twice the cross-wind dimension, the Eurocode factor aG
of galloping instabilities referring to the Scruton number is 2.0 corresponding
to aGG values in Eq. 2.2 of 1.0.

2.6.4 Aeroelastic Effects for 1:2 Cross Section

The results presented below are from a wind tunnel study focusing on aeroe-
lastic effects of the 1:2 cross section with an in-wind depth being twice of the
cross-wind dimension. The cross section 1 : 2 with an along-wind depth twice
the cross-wind dimension was tested in low turbulent flow with an intensity
of approx. 2% – see the results presented in Figures 2.18 to 2.20. From Fig-
ures 2.18 the galloping instability factor, aGG has been found to be 4.3 at zero
angle of attack corresponding to wind perpendicular to the small side of the
cross section.
For the 1:2 cross section both vortex shedding and galloping gives con-
tributions to the aerodynamic damping. In the simplest form the two con-
tributions of aerodynamic damping are assumed independent, and the total
aerodynamic damping, δa , is expressed by

δa = δKaG + δaG , (2.4)

where δKaG and δaG are the aerodynamic damping originating from vortex
shedding and galloping, respectively. Figure 2.19 shows the aerodynamic
damping determined by subtracting the structural damping from the total
damping measured, and the vortex-induced aerodynamic damping has been
determined by Eq. 2.4 assuming that the aerodynamic damping from gallop-
ing follows Eq. 2.3. For an aerodynamic damping due to vortex shedding
of 2 to 3% shown in Figure 2.19, the non-dimensional damping parameter
(Hansen, 2013)
δKaG me
KaG =
2π ρ bd

59
Wind Loading Design Codes

becomes equal to between 1.8 and 2.7. It is promising to observe that this
value of the aerodynamic damping parameter is in good agreement with the
values of 2.4 and 2.7 found for cross sections of 2:1 and 1:1, respectively
(Hansen, 2013).
In Figure 2.20 measurements of a rectangular cylinder with a cross section
are shown together with estimated values for the onset galloping velocity from
the Eurocode and ISO. The vibrations start around the resonant velocity for
vortex shedding and continue to grow almost linearly with increasing wind
velocity. The anticipated development of the response due to vortex shedding
cannot be seen within the band of mass-damping parameters tested. This
agrees with results obtained by Itoh and Tamura (2002), Kawai (1992) and
Washizu et al. (1978). It may be seen that relative large differences in ScG
do not yield significant differences in response and that the onset galloping
velocity is independent of ScG and depends only on the Strouhal number.
For larger ScG , ScG > 50, separation between vortex-induced vibrations
and galloping have been seen, and for really small ScG , ScG < 3, vortex-
induced vibrations have been seen at v/nb equal to approximately 6 – see
Itoh and Tamura (2002) and Washizu et al. (1978).
Figure 2.20 shows that the Eurocode prediction of the onset wind velocity
for galloping is unsafe for general non-dimensional mass-damping parameters
larger than approximately 9. For mass-damping parameters in the range of
9.5 to 39.5, the response starts to increase for reduced wind velocities of
approximately 15. The most extreme deviation between the Eurocode and
the test results occurs for a general mass-damping parameter of 39.5. The
Eurocode onset wind velocity of galloping corresponds to a reduced wind
velocity of 2 × 39.5 = 79, where the test results show initiation of galloping
instabilities for a reduced wind velocity of approximately 15, i.e. at a wind
velocity being approximately 5 times smaller than the Eurocode predictions.
With the higher aGG of 4.3 based on Figure 2.18, the predictions fits well
with the measured data except if the mass-damping-parameter is really small.
The reason for this can be found in Figure 2.19. Here it can be seen that be-
fore the critical velocity there is a damping effect of the vortices. Thus the
model will stay still until resonance speed is reached and start vibrating from
there with increasing wind speed. Figure 2.19 also shows that galloping be-
comes the only driving force at v/nb & 20, where the aerodynamic damping
due to vortex shedding becomes zero. At the highest mass-damping param-
eter number the damping is so large that the vortex induced vibrations are
suppressed, and therefore the model starts to vibrate at the onset galloping
velocity. ISO 4354:2009 is seen to agree well with the measurements for all
mass-damping parameters.

60
 S.O. Hansen

Figure 2.18: The lift and drag coefficient as a function of angle of attack. In-wind
depth is used as a reference length for both the drag and lift coefficients.

The Eurocode specifies that interaction effects between vortex shedding

and galloping are likely to occur when the calculated onset wind velocity of
galloping is in the range of 0.7vr to 1.5vr , where vr is the resonance wind
velocity for vortex shedding. The measuring results shown in Figure 2.20
indicate that this range should be much larger for the 1:2 cross section.

2.7 Design Criteria

For the dimensioning criteria reference is made to the code parts on different
materials, such as steel, concrete and timber. In the present planned Eu-
rocode revision, a focus area is wind loads relevant for structures, where their
resistance increases for shorter load durations, e.g. glass panels.
The present codes represent the wind action on a structure by an equiva-
lent static wind load calculated by the procedures outlined above. This static
wind load is used to calculate structural stresses to be compared with the
resistance specified in the code parts on different materials, such as steel, con-
crete, timber and glass. Most materials experience an increased resistance

61
Wind Loading Design Codes

Figure 2.19: Negative aerodynamic damping – positive y-values – expressed by a

logarithmic decrement. δKaG is the negative vortex-induced aerodynamic damping
estimated by subtracting the galloping-induced aerodynamic damping from the total
aerodynamic damping, δa , originating from both vortex shedding and galloping.

62
 S.O. Hansen

Figure 2.20: Wind-induced vibrations at different mass-damping parameters to-

gether with the predictions from the Eurocode EN 1991-1-4:2005 and ISO 4354:2009.
The Eurocode and ISO instabilities described by the onset wind velocity of galloping
are indicated by the markings on the x-axis.

when the load fluctuates fast. The time constants may be different indicating
that the importance depends on the load in question. Future codes are ex-
pected to focus much more on the structural resistance relevant for fluctuating
wind effects. Some examples relevant for ultimate limit states are described
below.
Most steel structures are very ductile, and wind effects may initiate yield-
ing, but the short duration of the extreme wind effects are not able make
the structure collapse. Typically, for lattice structures the governing load
situation is the wind-induced compression forces in the structural members.
Besides permanent actions these compression forces originate from mean wind
load and the influence of background turbulence and resonance turbulence.
The mean wind load acts over 10 minutes and the effects of background tur-
bulence often have frequencies of the order of 0.1 Hz corresponding to typical
time periods of 10 seconds, and these durations seem to be relatively long com-
pared to the response time of the lattice structure. However, the resonance

63
Wind Loading Design Codes

turbulence giving vibrations at the natural frequency, which for the lattice
structure typically is of the order of 1 Hz, may have duration too short for an
actual collapse to develop.
For glass panels experiments have been carried out with an experimental
arrangement as shown in Figure 2.21. The glass plate, which is subject to
the test, is placed on an iron frame. Underneath the plate is placed a strain
gauge force transducer, on a piston pulling in the glass plate with a given
force. The pressure is delivered by a 10 bar compressor, which is not shown
in the picture. The static tests of two types of glass panels have been carried
out by increasing the load gradually in cycles, and then unloading the glass
plate. The dynamic tests of the glass have been carried out by repeating the
load in series of 100 repetitions. This is done repeatedly until the glass panels
fail.
The tests carried out showed that the failure when the fluctuating wind
loads were applied were approximately twice the failure load for static con-
ditions. Further testing will provide more statistical information on glass
resistance, and it will enable the background for establishing appropriate load
duration specifications in the European wind code.
If more accurate design criteria are established following the ideas de-
scribed above, appropriate partial safety factors taking all relevant uncertain-
ties of the wind load chain properly into account should be determined by
detailed calibration studies, see chapter 2. These studies may give slightly
increased partial safety factors expected to remove only a minor part of the
increased structural resistance obtained by the more advanced design criteria
described above.

2.8 Conclusion
The present wind loading design codes have reached a stage where they pro-
vide specifications that typically only deviate slightly. However, further har-
monization is still needed, and a number of focus areas should be addressed
in future revisions. Some of these focus areas are a consistent approach for
determining the influence of both load duration and spatial averaging on pres-
sure coefficients, wind loads on two skin façades and influence lines or mode
shapes with changing sign.
Preferably, the codes should be transparent and this is not always the situ-
ation. A considerable improvement in the quality of building codes, including
the Eurocode, is possible if a quality check is done before data are selected to
be included. The quality check should consider the following items:

64
 S.O. Hansen

Figure 2.21: Experimental arrangement for testing glass resistance as function of

load duration.

65
Wind Loading Design Codes

• Building codes should be probability based, so data used in the codes

should also be probability based.

• The analysis techniques should be consistent with the choices made for
the level of safety in the building codes.

• Values of the pressure coefficients in codes should be based on analysis

of extreme values.

• The relation between averaging time and spatial averaging time and the
choice of extreme value analysis should be consistent with the proba-
bilistic approach applied.

• Flow conditions and measurements techniques should be known and

within the range of applicability of the codes.

The general non-dimensional mass-damping parameter described in the

paper will be a helpful design parameter used to predict whether structures
may be sensitive to vortex-induced and galloping-induced vibrations. The
use of the general non-dimensional mass-damping parameter instead of the
Scruton number normally applied may reduce the uncertainties involved in
code predictions of vortex-induced vibrations and galloping-induced vibra-
tions considerably.
The combined effects of vortex shedding and galloping will for some cross
sections give larger vibrations than predicted by the Eurocode. Improved
theoretical models may be the basis for subsequent Eurocode updates taking
this phenomenon more accurately into account.

Acknowledgements
The allowance of using the model scale results presented and the help of
my colleagues at Svend Ole Hansen ApS and SOH Wind Engineering LLC
preparing the wind tunnel tests carried out are greatly acknowledged.

References
R. Bashor and A. Kareem. Comparative study of major international standards. In
Proceedings of the Seventh Asia-Pacific Conference on Wind Engineering, Novem-
ber 8-12, Taipei, Taiwan., 2009.
N. J. Cook. The designers guide to wind loading of building structures. Part 2:
Static structures. Building Research Establishment, 1985.

66
 S.O. Hansen

A. G. Davenport. The response of slender, line-like structures to a gusty wind.

Proceedings of the Institution of Civil Engineers, 23:389–408, 1962.
A. G. Davenport. Gust loading factors. Journal of the Structural Division, ASCE,
93(3):11–34, 1967.
A. G. Davenport and N. Isuymov. The application of the boundary layer wind tunnel
to the prediction of wind loading. In Wind Effects on Buildings and Structures,
Toronto, pages 201–230, 1968.
C. Dyrbye and S. O. Hansen. Wind loads on structures. John Wiley & Sons, 1996.
ECCS. Recommendations for calculating the effect of wind on constructions, 2nd
Edition. European Convention for Constructional Steelwork, 1987.
C. Geurts, B. Zimmerli, S. Hansen, P. Van Staalduinen, G. Sedlacek, M. Hortmanns,
P. Spehl, and P. Blackmore. Transparency of pressure and force coefficients. Pro-
ceedings of 3rd European and African Conference on Wind Engineering, Eind-
hoven, The Netherlands, July 2–6, pages 165–172, 2001.
S. Hansen, P. Blackmore, C. Geurts, M. Hortmanns, G. Sedlacek, P. Spehl, P. Staal-
duinen, and B. Zimmerli. Eurocode procedures for predicting vortex excitation.
In Proceedings of 3rd European and African Conference on Wind Engineering,
Eindhoven, The Netherlands, July 2–6, 2001.
S. O. Hansen. Vortex-induced and galloping-induced vibrations of simple structures.
In Proceedings of the International Conference on Wind Engineering, Amsterdam,
The Netherlands, 2011.
S. O. Hansen. Vortex-induced vibrations the Scruton number revisited. Structures
and Buildings Journal – Institution of Civil Engineers, 2013.
S. O. Hansen and S. Krenk. Dynamic along-wind response of simple structures. J.
Wind Eng. Ind. Aerod., 82(1):147–171, 1999.
E. Hjorth-Hansen and A. Kyrkjeeide. Aeroelastic response of a square tower model
having solid or perforated walls. In The University of Trondheim, Norway, 1978.
J. Holmes, Y. Tamura, and P. Krishna. Comparison of wind loads calculated by
fifteen different codes and standards, for low, medium and high-rise buildings. In
Proceedings of the 11th Americas Conference on Wind Engineering, San Juan,
Puerto Rico, June 22-26, 2009.
Y. Itoh and T. Tamura. The role of separated shear layers in unstable oscillations
of a rectangular cylinder around a resonant velocity. J. Wind Eng. Ind. Aerod.,
90(4):377–394, 2002.
M. Jensen. The model-law for phenomena in natural wind. 1958.
H. Kawai. Vortex induced vibration of tall buildings. J. Wind Eng. Ind. Aerod., 41
(1):117–128, 1992.
C. Scruton. On the wind-excited oscillations of stacks, towers and masts, 1963.
C. Scruton. An introduction to wind effects on structures. Oxford University Press,
Engineering Design Guides 40, 1981.
G. Solari. Gust buffeting. I: Peak wind velocity and equivalent pressure. Journal of
Structural Engineering, 119(2):365–382, 1993a.
G. Solari. Gust buffeting. II: Dynamic alongwind response. Journal of structural

67
Wind Loading Design Codes

engineering, 119(2):383–398, 1993b.

T. Stathopoulos. Turbulent wind action on low-rise buildings. PhD thesis, The
University of Western Ontario, London, Ontario, Canada, 1979.
T. Vrouwenvelder and N. Scholten. Assessment criteria for existing structures. Struc-
tural Engineering International, 20(1):62–65, 2010.
K. Washizu, A. Ohya, Y. Otsuki, and K. Fujii. Aeroelastic instability of rectangular
cylinders in a heaving mode. J. Sound and Vibration, 59(2):195–210, 1978.
B. Zimmerli. Process of drafting the Eurocode on wind action. In Proceedings of
the Third European & African Conference on Wind Engineering. Eindhoven, The
Netherlands, July 2-6, pages 157–163, 2001.

68