J. Wind Eng. Ind. Aerodyn.

104–106 (2012) 12–24

Contents lists available at SciVerse ScienceDirect

Journal of Wind Engineering

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

Alan G. Davenport’s mark on wind engineering

Nicholas Isyumov n
The Boundary Layer Wind Tunnel Laboratory, The University of Western Ontario, London, Ontario, Canada N6A 5B9

a r t i c l e i n f o a b s t r a c t

Available online 12 May 2012 This paper overviews the contributions to wind engineering made by the late Professor Alan G.
Keywords: Davenport. These include an engineering model of strong wind in the atmospheric boundary layer and
Davenport’s contributions its simulation in wind tunnels. His mean and turbulent wind profiles and his spectrum of longitudinal
Wind buffeting turbulence are known world-wide. He developed the theory of wind buffeting of line-like structures
Loads and responses and introduced the gust factor method, which determines the magnitude of the peak wind-induced
Wind tunnel testing response, including the effects of wind-induced resonance. Also he has made key contributions to the
Development of the wind loading chain development of statistical methods which are needed in the prediction of wind loads and wind-induced
effects. These include the method of up-crossings, which predicts the probability of exceedance of
particular extreme values. Foremost, he will be remembered for the wind tunnel model studies of many
of the world’s milestone buildings and structures.
Professor Davenport’s approach to wind loading was to combine its key components in a chain of
thought, analogous to a physical chain, whose strength is determined by its weakest link. In recognition
of this and his many other achievements, the International Association of Wind Engineering has
officially named his approach to wind loading as the ‘‘Alan G. Davenport Wind Loading Chain’’.
& 2012 Elsevier Ltd. All rights reserved.

1. Introduction pioneered at the Boundary Layer Wind Tunnel Laboratory (BLWTL)

The wind engineering community lost one of its most distin-
guished members, when Alan Garnett Davenport passed away on
July 19, 2009. Accounts of Professor Alan Davenport’s life and his
contributions to wind engineering have since appeared in print.
His obituary was published in the International Journal of Wind
Engineering (Isyumov, 2009). A special edition of the Journal of
Wind Engineering of Japan was published in Alan Davenport’s
memory. This included a paper which described his close colla-
boration with Japanese researchers (Isyumov et al., 2010). A
special edition of the Canadian Journal of Civil Engineering (of
which Professor Davenport was its Founding Editor) has also been
published in 2011, in which several of his former students and
colleagues have written papers on their areas of research which
have been most influenced by Professor Davenport (Canadian
Journal of Civil Engineering, 2011).
Alan Davenport first presented his ideas on the action of wind
on buildings and structures in his doctoral thesis 50 years ago and
has since left an indelible mark on wind engineering. His name is
one of the most recognized in wind engineering, both for his many
theoretical contributions, as well as for the many wind engineering
studies of buildings and structures which he and his colleagues
n
Tel.: þ1 519 661 3338; fax: þ 1 519 661 3339.
E-mail address: ni@blwtl.uwo.ca
at the University of Western Ontario. Alan founded the BLWTL in
1965 and was its director throughout its ‘‘heyday’’ years. This paper
discusses Alan G. Davenport’s many achievements and contribu-
tions, as I remember them. It is expected that the interested reader
will turn to the technical literature for further details. Fortunately,
Professor Alan Davenport has published widely and left a broad and
long paper trail of his ideas and theories in papers, books and
conference proceedings worldwide.
Whilst history is most likely to remember Alan Davenport for
his pioneering work in wind engineering, he also made important
contributions to the fields of meteorology, structural dynamics
and earthquake engineering. Alan was committed to improve the
safety of the built environment and found time and energy to
become involved with the United Nations’ declared International
Decade for Natural Disaster Reduction. Among his many lifelong
interests were the reliability and safety of structures. Towards
this end, he contributed to building codes and standards of many
countries and served on numerous national and international
committees, charged with their improvement. Alan Davenport
was instrumental in the establishment of the Centre for Studies in
Construction and the Institute for Catastrophic Loss Reduction at
the University of Western Ontario. These sister organizations
added critical mass and complemented the activities of the
BLWTL. He was a gifted and dedicated researcher and a much
respected teacher and mentor to a countless number of students,
both at the undergraduate and graduate levels. In recognition of

0167-6105/$ - see front matter & 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jweia.2012.02.007
 N. Isyumov / J. Wind Eng. Ind. Aerodyn. 104–106 (2012) 12–24
his pre-eminence in both research and academia, Professor
Davenport was awarded honourary degrees by 10 different
universities and was appointed a Member of the Order of Canada
for his lifetime achievements.
This paper expands the keynote address given at the ICWE-13
in Amsterdam in Professor Davenport’s memory and honor. This
is done with additional material which describes his approach for
evaluating the action of wind on buildings and structures. In this
process the action of wind is determined by a rational procedure
which includes estimates of the local wind climate; the shaping of
the site specific wind by the influence of local terrain and
topography; project specific aerodynamic data; the recognition
of dynamic effects, including wind-induced resonance and the
establishment of criteria for evaluating acceptability. Alan’s
approach to wind loading has been named as the ‘‘Alan G.
Davenport Wind Loading Chain’’. This wind engineering termi-
nology has been formally announced by the International
Association of Wind Engineering at ICWE-13, held at Amsterdam
in July 2011. This was done in order to recognize Alan G.
Davenport’s many contributions to wind engineering and to
honor his memory.
2. Analytical methods
2.1. Wind loading chain concept
Alan Davenport’s doctoral research (Davenport, 1961a) laid the
foundation of today’s wind engineering. His approach was based
on the chain of thought which recognized that the wind loading
experienced by a particular building or structure is determined by
the combined effects of the local wind climate, which must be
described in statistical terms; the local wind exposure, which is
influenced by terrain roughness and topography; the aerody-
namic characteristics of the building shape and the potential for
load increases due to possible wind-induced resonant vibrations.
He also recognized that clear criteria must be in place for judging
the importance of the consequences of the predicted wind action.
This included the effects of wind on the integrity of the structure
and the exterior envelope and various serviceability considera-
tions, such as the control of the wind-induced drift, the effects of
building motions on occupants and the usability of outdoor areas
at and near particular buildings and structures. In his papers Alan
Davenport referred to this process of evaluating the effects of
wind action as the ‘‘wind loading chain’’. This was in recognition
that the evaluation of the wind loading and its effects relies on
several interconnected considerations, each of which requires
scrutiny and systematic assessment. With analogy to a physical
chain, the weakest link or component in this process determines
the final outcome. Little is gained by embellishing strong links but
much is lost by not paying attention to the weak ones.
This symbolic chain of thought is shown in Fig. 1. Alan used
this ‘‘chain’’ analogy throughout his practice, lectures and pub-
lications. Not only does the wind loading chain provide a firm
basis for the evaluation of wind loads it also forms the basis for
the chain of thought needed to define and solve other wind
engineering problems. The format of Alan’s wind loading chain
has been adopted in the specification of wind loads in most
building codes and standards worldwide.
The chain approach permits systematic estimates of the
statistical variability of the predicted wind action. The link with
Fig. 1. Alan G. Davenport’s wind loading chain.
13
the largest uncertainty dominates the reliability of the entire
process. Alan’s papers detailed this approach and demonstrated
the improvement in reliability that can be made by strengthening
weak links in the wind loading process. He clearly showed the
benefits which result from improvements in the description of the
local wind climate and how the use of aerodynamic data from a
wind tunnel model simulation instead of information from
building codes improves reliability. As part of his wind loading
chain, Alan also suggested criteria for evaluating the effects of
wind induced accelerations on the comfort of occupants in tall
buildings (Davenport, 1972) and the effects of wind speeds on the
comfort and safety of pedestrians in outdoor areas (Davenport,
1975).
2.2. New direction
A major change in the technology of buildings and structures
occurred in the post World War II years. This was most evident
from the building boom experienced in North America. These new
or ‘‘modern’’ buildings and structures were becoming taller and/
or of greater span, more daring in shape and concept and more
slender in proportions. In addition, more powerful structural
computation methods, the emergence of new materials and
fastening methods with welding replacing the traditional use of
rivets in steel construction, collectively contributed to the emer-
gence of lighter weight and more flexible elastic structures with
less inherent energy dissipation or damping. However, while
more economic than traditional construction, buildings and
structures of this new generation were more susceptible to the
dynamic wind action. Building codes of those days described
wind loads as static and did not address possible wind-induced
resonant vibrations which can increase the effective loads. Alan
Davenport’s innovative approach recognized the stochastic char-
acteristics of natural boundary layer wind in both the time and
space domains and the importance of allowing for both its static
and dynamic actions. This required an improved understanding of
the nature of the wind itself, a better description of the aero-
dynamics of buildings and structures and the use of structural
vibration theory to determine their wind-induced responses.
Alan’s ideas first presented in his Ph.D. Thesis and shored up by
a burst of key papers in internationally available journals and in
conference proceedings, were very much needed in order to
support these rapid changes in the technology of the built
environment.
Alan’s approach to wind loading was able to correctly evaluate
the dynamic action of wind on buildings and structures and to
determine their oscillatory behavior in the presence of wind-
induced resonant vibrations. This set a new direction in wind
engineering. Also, it made Alan Davenport an internationally
recognized authority on the action of wind on buildings and
structures and a much sought-after consultant on the character-
istics of boundary layer winds and their effects on wind sensitive
tall structures, such as skyscrapers, towers and long span bridges.
His ideas and methods were put into practice in the design of
such landmark projects as the World Trade Center in New York,
the Sears Tower in Chicago (now Willis Tower) and the CN Tower
in Toronto. Many others projects followed and the skylines of
many of the world’s major cities are graced with buildings and/or
structures, whose designs were influenced by Alan’s ideas and the
efforts of the Laboratory which he founded.
2.3. Engineering model of natural wind
The cornerstone of Alan Davenport’s approach was a rational
and computationally manageable framework for describing the
spatial and temporal characteristics of wind. This is essential for
14 N. Isyumov / J. Wind Eng. Ind. Aerodyn. 104–106 (2012) 12–24
estimates of the wind-induced dynamic loads and responses and
the prediction of their extremes for design. Alan’s model recog-
nized that wind speeds are driven by weather fluctuations which
vary on a macro-meteorological scale with periods in the order of
hours and days. Details of the wind structure during particular
storms are largely shaped by the surface drag of the earth’s
surface. During strong winds, the flow within the surface bound-
ary layer becomes less dependent on thermal gradients and is
similar to the flow over an aerodynamically rough plate.
The roughness of the earth’s surface determines the vertical
profile of the time averaged or sustained wind and the intensity
and structure of turbulence or ‘‘gusts’’ within this atmospheric
boundary layer (ABL). These gusts occur at a micro-meteorologi-
cal scale with periods of seconds and minutes.
The distinctly separated time scales of the macro- and micro-
meteorological phenomena allowed Alan to describe wind within
the ABL as a slowly varying mean flow with superimposed
turbulent fluctuations or gusts. Measurements of the spectrum
of the horizontal wind speed at Brookhaven, New York were
reported by Van der Hoven (1957) in the late 1950s. Fig. 2, which
Alan presented at ICWE-1, held in Teddington in 1963
(Davenport, 1965), clearly shows a spectral peak at a period of
about 4 days, which corresponds to the passage of synoptic storm
systems and a spectral peak at about 1 min, which corresponds to
the local turbulence generated within the atmospheric boundary
layer. The flow fluctuations due to these phenomena are sepa-
rated by a distinct spectral gap at periods of about 10 min to 1 h.
Alan’s hypothesis was that, estimates of loads and responses of
Fig. 2. Spectrum of wind speed variations within the atmospheric boundary layer.
Fig. 3. Davenport’s power law profiles of the mean wind speed over different terrains.
buildings and structures can be best made by describing wind as a
locally stationary random process with a mean speed determined
from observations made with an averaging time of 10 min to 1 h
and a surface drag generated superimposed turbulence. Alan used
Taylor’s (1935) statistical theory of turbulence to describe the
wind gusts.
This overall concept remains central in current day descrip-
tions of natural wind for most wind engineering applications. It
permits the separation of natural wind into variations caused by
the weather and those due to the turbulence, which is generated
by flow within the ABL. At time scales of most buildings and
structures the slow changes in the mean wind speed and direc-
tion can be disregarded. Details of this mean wind climate can be
statistically predicted from available historic wind data. Alan used
statistical models, such as the Rayleigh and Weibull distributions,
with azimuth dependent parameters to describe the parent
statistics of local mean wind climate. These models correctly
recognized the directional preferences of the wind in a particular
region or the local directionality. Extreme value theory was used
to predict the occurrence rate of extreme winds, as needed in
design (Davenport, 1964).
During strong winds, both spatial and temporal variations of
atmospheric turbulence are dominated by the action of the
surface drag and its structure is similar to the mechanically
generated turbulence over an aerodynamically rough plate. Alan
estimated surface drag coefficients for different types of terrain
and classified their roughness based on their surface roughness
length z0. Alan proposed the use of power law profiles to describe
the variation of the mean wind speed with height within the ABL.
His early estimates of the power law exponent at the gradient
height for typical types of terrain (Davenport, 1975) are shown in
Fig. 3. The power law profile is more user friendly in comparison
with the logarithmic profile and provided a convenient method
for transferring wind climate information, determined from
weather station observations; typically located in open country
terrain; to wind speeds at other heights and to locations in other
types of terrains. Alan’s approach to the classification of terrain
roughness and its effect on the mean velocity profile remains in
use till today (Davenport et al., 2000; Wieringa et al., 2001). His
profiles and his concept of a gradient wind, reached near the top
of the ABL whose height is determined by the local terrain
roughness, remains sufficient for the design of the vast majority
of the built environment. Nevertheless, it is important to
recognize that the mean wind speed increases above the ABL.
 N. Isyumov / J. Wind Eng. Ind. Aerodyn. 104–106 (2012) 12–24 15

Fig. 4. Davenport’s spectrum of the longitudinal component of turbulence.

This becomes important in the design of super-tall buildings and

towers. A model of the vertical distribution of the mean wind
speed within the entire planetary boundary layer (PBL) has been
proposed for neutrally stable thermal stratifications by Harris and
Deaves (1986). Their model has been adopted by ESDU (1990) for
extrapolating surface wind speed measurements upwards. Albeit,
these extrapolations are based on mathematical models of the
ABL and PBL and confirmations with actual upper level wind
speed measurements are very much needed.
Alan’s mathematical model for the spectrum of longitudinal
gustiness was first presented in his doctoral dissertation
and subsequently published by the Royal Meteorological Society
(Davenport, 1961b) and presented at ICWE-1 (Davenport, 1965a),
see Fig. 4. His spectrum of the gustiness of strong winds and his
terrain roughness categorization to determine his power law
profiles continue to be used by the National Building Code of
Canada (2005).
When first presenting his model of the structure of wind in the
ABL, Alan had to make assumptions about some of the details of Fig. 5. Davenport’s coherence function.
turbulence. This was necessary due to the lack of data. In
particular, limited information was available on the vertical and
horizontal scales of turbulence and spatial correlations of wind and as cyclones in the Indian Ocean and in Australia. The
gusts in built-up areas (Davenport, 1965b). These are required in occurrences of such severe tropical storms at particular locations
the evaluation of the dynamic action of atmospheric turbulence are rare events, which are unlikely to be captured in the historic
on wind sensitive structures, such as tall masts and buildings and records of individual weather stations. Alan’s approach was to
long span bridges. The square root of the coherence of the determine the regional statistics of tropical storm properties by
longitudinal component of turbulence at 3 heights above ground pooling together the track data from historically recorded tropical
in open grassland terrain is shown in Fig. 5. These data were storms which occurred with the surrounding area. These tropical
computed by Alan and reported in his Ph.D. Thesis (Davenport, storm statistics would then be used in a Monte Carlo simulation
1961a). His proposed exponential form of the variation of the to predict future storm occurrences. These would be tracked to
coherence of the simultaneous wind speeds at locations separated determine the wind speed and direction at particular locations of
in horizontal or vertical directions has stood the test of time. interest. Simulations would be continued over many thousands of
Understandably, the initially proposed constants in these func- years in order to provide statistically reliable predictions of
tions improved, as more information became available. extreme winds and their action on buildings and structures.
Alan’s description of wind was intended for neutral stratifica-
tions of the ABL, when the structures of both the mean and
turbulent components of the flow are largely determined by the 2.4. Dynamic wind action
surface drag. This provided a good model for strong synoptic
winds experienced in extra-tropical low pressure systems. Alan With mathematical models of the wind climate and structure
realized that the wind structure would deviate from his model in in place, Alan turned to structural dynamics and random vibration
other types of wind storms. He was particularly concerned about theory to determine the wind-induced loads and responses of
the effects of severe tropical storms, referred to as hurricanes in buildings and structures. The approach for determining the
North America; typhoons in Japan and other parts of the Far East; resonant dynamic response of structures in particular modes of
16 N. Isyumov / J. Wind Eng. Ind. Aerodyn. 104–106 (2012) 12–24

Fig. 6. Davenport’s approach for determining the wind-induced resonant response of buildings and structures.

vibration was described in his paper on gust effect loading

(Davenport, 1967) and is summarized in Fig. 6.
Central to the evaluation of the wind-induced resonant
dynamic response is the linearization of the relationship between
the time-varying component of the wind speed and the dynamic
wind load, namely it is assumed that
2
FðtÞ ¼ 1=2rC F U 2 ðtÞ  1=2rC F U ð1þ 2uðtÞ=UÞ ð1Þ

where the instantaneous wind speed is the sum of the time

average wind speed and fluctuating component, due to turbu-
lence; namely UðtÞ ¼ U þuðtÞ and where r, CF, U(t), t, U and u(t)
are, respectively, the air density; the aerodynamic shape coeffi-
cient, the instantaneous wind speed, time, the time-average of the
wind speed and the fluctuating component of the wind speed in
the direction of U, at time t. The term involving u2(t) is assumed to
be small and therefore negligible.
This crucial linearization, which allows enormous simplifica- Fig. 7. Statistical parent and extreme value distributions of a random variable. (a)
tions for calculations made in the frequency domain, has stood Description of site specific wind climates, (b) influence of surrounding terrain and
the test of time for the purpose of engineering estimates. Also, it is topography, (c) sources of aerodynamic data, (d) wind-induced dynamic effects
and (e) criteria for evaluating acceptability.
assumed that the wind loading is quasi-static, namely that
aerodynamic data used to obtain the time average wind loads
also apply for the dynamic loads. Alan’s buffeting approach was
provided in the NBCC for the calculation of buffeting loads due
intended for estimates of wind-induced along-wind or drag
to wind-induced drag on vertical structures. It assumes that the
forces, where the time average or mean value of the wind load
wind-induced resonant vibrations are in the fundamental mode
is significant. His premise was that the maximum value of the
of vibration, assumed to have a mode shape which varies linearly
wind load or the resulting effects W, can be related to its time
above ground. This captures the essence of the dynamic wind
average mean value, namely
loading process for most buildings and provides a framework for
^ ¼ Cg W
W ð2Þ extending the procedure to higher modes of vibration which may
also be excited by buffeting induced drag loads.
where W ^ and W are, respectively, the maximum and time-
The procedure, outlined in Fig. 6, is readily used for structures
average values of W and Cg is the gust effect factor. The gust where the wind-induced drag dominates. This includes most
effect factor is then defined as buildings with reduced frequencies of (fD/VH) greater than about
C g ¼ 1 þg p ðsW =WÞ ð3Þ 1/7 to 1/5. Here f, D and VH are, respectively, the frequency of
vibration of the building, the frontal width of the building and the
2 2 1/2
where sW ¼(s s Bþ Ress)and where sB, sRess and gp, are respec-
, design mean wind speed at the top of the building with height H.
tively, the standard deviation of the background or non-resonant The vast majority of buildings and structures are in this category
component of W, the standard deviation of the resonant compo- and their maximum wind induced loads are drag dominated.
nent of W and the peak factor which is determined by extreme Nevertheless, there are structures where other wind-induced
value theory (Davenport, 1964). The peak factor depends on the forces such as those acting in the across-wind direction or in
cycling rate or natural frequency n of the process and the interval torsion become important. These can be caused by vortex shed-
of time during which the peak value is observed, see Fig. 7. ding induced forces, the buffeting by wakes of other structures or
Alan’s approach for the evaluation of the drag-induced buffet- in situations of aerodynamic instability. Estimates of the drag-
ing dynamic wind loads and their effects, has been included in the induced loads alone are insufficient in such situations and other
NBCC (2005) and other wind codes. Detailed information is methods for evaluating wind action become necessary.
 N. Isyumov / J. Wind Eng. Ind. Aerodyn. 104–106 (2012) 12–24
2.5. Details of the wind loading chain
Professor Davenport introduced the wind loading chain con-
cept early in his professional life. He continued to improve and
sharpen the details of this chain throughout his career. In fact
details of the chain, shown in Fig. 1, continue to evolve as wind
engineering matures and as more information becomes available.
Some further developments and refinements were introduced by
Alan himself, some were made by others. The various components
or links of the wind loading chain and some of their refinements
warrant further mention. This is done ‘‘link: by link’’.
In the ‘‘wind climate’’ link, see Fig. 1, Alan realized that
historical data at particular weather stations did not provide
sufficient information on the risk of experiencing hurricane or
typhoon winds at a particular project site. He conceived the use of
Monte Carlo computer simulations of tropical storms using
tropical storm statistics for the region or storm basin in which
the project is to be located. Initial statistical methods relied on
Rayleigh and Single Weibull models to describe the parent
distribution of wind speed and direction at particular locations.
The Double Weibull distribution was introduced quite recently in
order to improve the fitting of historical wind data in regions of
mixed wind climates (Xu et al., 2008). Computer weather recon-
struction modeling is a promising emerging technique for
improving the description of natural wind (Irwin, 2007; Gatey
et al., 2011).
Power law and logarithmic profiles continue to be used in
many building code versions of the Davenport wind loading
chain. For most buildings and structures, Alan’s gradient wind
concept with typical gradient heights of Zg E270,360 and 450 m
in homogeneous open country, suburban and urban terrains is
sufficient to describe the increase of the wind speed above
ground. However, it becomes insufficient for very tall buildings
which protrude above these heights. The Harris and Deaves wind
profile, as adopted by ESDU, has become the wind profile of
choice as it permits winds over all terrains to reach the same
speed at the geostrophic height. Procedures to estimate wind
speeds in non-homogeneous terrains and the effects of topogra-
phy are further improvements which have been made in order to
describe the wind structure at particular locations. Wind tunnel
model testing has expanded the aerodynamic data base for
buildings and structures. New types of tests which have emerged
include high frequency force balance model tests for tall build-
ings, taut-strip model tests for long span bridges, spatially
averaged loads to determine instantaneous pattern loads for
low-rise buildings, etc. Also, tests in large-scale facilities are on
the horizon.
Alan’s linearization of the relationship between the fluctuating
wind force and turbulent component of the wind speed, see
Eq. (1), made the analysis of the wind loading process possible
in the frequency domain. This permitted progress to be made
with modest computational power. With increased computing
power, it is now possible to do the computations in the time
domain without the need to linearize. Also, it is now possible to
examine the response of non-linear structures and to evaluate the
accumulation of stress cycles, as required in the evaluation of
fatigue damage.
It is important to have criteria in place for evaluating the
acceptability of predicted wind loads and responses, as well as
other wind effects so that design adjustments can be made. Alan
proposed acceleration criteria for judging the comfort of occu-
pants of tall buildings and criteria for assuring the safety and
comfort of pedestrians Finally, it is important to stress that Alan’s
wind loading chain lends itself to different levels of sophistication
of analysis. In its simplest form, the wind loads and or wind
effects can be evaluated for a code prescribed wind speed,
17
without taking wind directionality into account. The chain
equally lends itself to probabilistic predictions, using statistical
descriptions of speed and direction and recognizing the direc-
tional sensitivity of structures.
3. Wind tunnel methods
3.1. Need for wind tunnel tests
Wind tunnel model tests are essential for unusual buildings
and structures and in particular those structures where the
maximum wind loads are not likely to be due to the buffeting
action of turbulence alone. Essentially two approaches are fol-
lowed in such situations. One approach is to carry out wind
tunnel model tests which provide aerodynamic data, such as
aerodynamic force and/or pressure coefficients, which are subse-
quently used in analytical evaluations of wind loads and
responses. Unfortunately, all inclusive analytical models of wind
loading are complex and difficult to develop. The second approach
is to carry out wind tunnel tests which provide direct estimates of
the wind loads and/or wind-induced responses. The latter
approach does not require particular analytical models of the
wind loading process but relies on the quality of the physical
model simulation achieved in the wind tunnel.
The aerodynamic data required in the prediction of wind loads
must come from wind tunnel model tests which are carried out in
simulated turbulent boundary layer flow. Such tests cannot be
properly made in conventional aerodynamic wind tunnels but
require the use of special boundary layer type wind tunnels in
which the effects of the surface drag on the wind flow can be
modeled. The importance of properly simulating the boundary
layer characteristics of wind was demonstrated by Dr. Martin
Jensen of the Technical University of Denmark. His wind tunnel
experiments clearly showed that it is essential to carry out tests
at a geometric scale which assures the correct modeling of the
various characteristic lengths of the atmospheric boundary layer.
These include the roughness length of the surface, the depth of
the boundary layer and the scales of turbulence. Jensen subse-
quently published his model law for phenomena in natural wind
(Jensen, 1958).
Alan realized the need of wind tunnel model testing for
buildings and structures which are unusual due to their shape,
height or horizontal span, their siting and/or their structural
properties. Consequently, he built his own boundary layer wind
tunnel at the University of Western Ontario. His first wind tunnel
was an open return boundary layer wind tunnel where the
characteristics of natural wind were shaped both with terrain
roughness and special spires and turbulence generators at the
entrance to the working section. This unique facility for its time
opened on November 1965 at the University of Western Ontario.
A larger, closed circuit boundary layer wind tunnel was subse-
quently constructed at the BLWTL and opened in 1984. Sche-
matics of both of these facilities are shown in Fig. 8.
Much effort was spent in the early days of boundary layer wind
tunnel testing on the development of new technology; including
the modeling of natural wind at a reduced geometric scale, model
construction, new testing procedures and the necessary instru-
mentation. The various measured quantities were probabilistic
rather than deterministic and had to be described in statistical
terms. It took a number of years for the wind engineering
profession to reach consensus on the similarity requirements
which have to be satisfied and the procedures which have to be
followed in order to provide reliable wind loading information.
Alan and his group at the BLWTL were very much part of that
effort and contributed to the writing of ASCE Manual no. 67 on
18 N. Isyumov / J. Wind Eng. Ind. Aerodyn. 104–106 (2012) 12–24
Fig. 8. Wind tunnel facilities at the Boundary Layer Wind Tunnel Laboratory (BLWTL), University of Western Ontario (Bottom: open circuit tunnel opened November 1965;
Top: closed circuit tunnel opened 1984).
‘‘Wind Tunnel Studies of Buildings and Structures’’ (1999). Its first
edition came out in 1987 and a second expanded edition appeared
in 1999.
The BLWTL contributed to the development of new testing
methods and instrumentation required for experiments to deter-
mine various aspects of wind action. This included the use of
computer based data acquisition and control systems for the
reliability and efficiency of experiments and new computer
controlled machines for wind tunnel model construction. Some
of these initiatives were driven and supported by specific com-
mercial projects. Others were achieved by graduate students as
part of their research.
One of these was the development of the high frequency force
balance (HFFB) model method, which has become a standard
state-of-the-art tool for evaluating the structural wind loads for
tall buildings. This research required the construction of a very
sensitive balance or transducer for measuring the wind excitation
and the development of theoretical and computational methods
for using this information to predict the effective wind loads and
structural responses (Tschanz, 1982). Dr. Tony Tschanz, a former
principal of the structural engineering firm Magnusson Klemencic
Associates, was Alan’s graduate student for this important
initiative.
3.2. Early days at the boundary layer wind tunnel laboratory
(BLWTL)
The early days at the Laboratory were demanding on time and
energy yet exhilarating by the achievements made. Experiments
on the behavior of buildings and structures in turbulent boundary
layer flow were cutting edge research and required new ideas,
new experimental methods and appropriate instrumentation. An
equally important requirement is to develop a methodology for
dealing with the massive amounts of data which are generated in
turbulent boundary layer tests and to relate these model data to
prototype applications. The BLWTL operated much like a clinical
hospital. Ideas were born, theories were formulated and incu-
bated and technology was developed to assist in new designs and/
or to provide solutions to existing problems. The exposure to the
design process of cutting edge buildings and structures broadened
the education process and brought in funds needed for novel
instrumentation and facilities. This in turn fueled research activ-
ities. The synergism resulting from the interaction of education,
research and engineering applications has been the key to the
success of the BLWTL. Alan was able to convince the University
administration that commercially sponsored research added rele-
vance to engineering education and therefore had an important role
on the University campus. The BLWTL became a show piece and a
visit with Professor Davenport in order to glimpse work-in-progress
on unique new buildings and structures became a top stop on the
University’s tour for dignitaries and important visitors.
The Laboratory attracted many wind engineering researchers
and professionals. Some stayed and expanded the critical mass of
the Laboratory. Others made contributions and then moved on.
The BLWTL became a Mecca for wind engineering researchers
from all parts of the world. There was a continuous stream of
academics, researchers, post-doctoral fellows, students, practising
engineers and the curious who were attracted by Alan’s ideas and
then carried the wind engineering torch back to their own lives. In
fact, few budding, as well as, established wind engineers did not
visit the BLWTL during its zenith years.
4. Wind engineering conferences
Interest in industrial aerodynamics and its application to the
evaluation of wind effects on buildings and structures was fueled
worldwide by conferences organized by international and
national wind engineering organizations. The first and probably
the most influential of these was held at the National Physical
Laboratory (NPL) in Teddington, England in June 1963. The NPL
was the hub of industrial aerodynamics at the time. The proceed-
ings were subsequently published by Her Majesty’s Stationary
Office in 1965. Alan presented two important papers at that
conference. One of these was ‘‘The relationship of wind structure
to wind loading’’ (Davenport, 1965a); the other was ‘‘The buffet-
ing of structures by gusts’’ (Davenport, 1965b). Participants at
that key conference included Christopher Scruton and Roger
Whitbread from the NPL; George Vincent of the US Bureau of
Public Works; Professor Arne Selberg of the Norges Tekniske
Hogskole in Norway; R.I. Harris of the Electrical Research Associ-
ates at Cranfield, England; John Skilling of the firm Worthington,
Skilling, Helle, Jackson and subsequently also Robertson, who
were the structural engineers for New York World Trade Center;
Professor A. Hirai of the University of Tokyo; Professor I. Konishi
of Kyoto University; Professor Ackeret of the Institute of Aero-
dynamics at Zurich; Dr. Martin Jensen of Denmark; Professor
Franz Leonhardt of Germany and many other pioneers of this
emerging new discipline. It is noteworthy that Ian Harris later
became the 1st editor of the International Journal of Wind
Engineering and Industrial Aerodynamics. Also, Kit Scruton and
Martin Jensen were posthumously honored by naming the non-
dimensional parameters Sc ¼4 mpz/rD2 and Je ¼z0/h as the
Scruton and Jensen numbers, respectively.
At the 5th of these international conferences, chaired by
Professor Jack Cermak and held in Fort Collins, Colorado, this
new discipline was formally labeled as wind engineering.
The International Conferences on Wind Engineering (ICWE)
 N. Isyumov / J. Wind Eng. Ind. Aerodyn. 104–106 (2012) 12–24
became catalysts in the formation of an international network of
wind engineering research activities and contacts. This influenced
further development and promoted the acceptance of the emer-
ging new concepts and ideas by practising engineers. Alan was
honored at the 9th ICWE in New Delhi, India in January 1995 by
the publication of a separate Davenport 60th Birth Anniversary
Volume, entitled ‘‘A State of the Art in Wind Engineering’’. This
volume includes invited papers by internationally recognized
researchers and a citation to Alan written by Krishna (1995).
Alan attended each of the first twelve of these conferences and
became an active and influential member of the growing wind
engineering community. He was instrumental in holding ICWE-2
in Canada in 1967 and organized and chaired ICWE-8 held at his
own base at the University of Western Ontario in London, Ontario,
Canada. At the 12th of these conferences, held at Cairns, Australia
in 2007, Alan was honored by the announcement of the Alan G.
Davenport Medal, which is to be awarded at ICWE to a prominent
wind researcher and a junior award to a promising young wind
engineering achiever. A tribute from the International Association
for Wind Engineering to Alan Garnett Davenport, which relates to
this medal, was provided in a paper by Solari et al. (2008). The
ICWE-13 is the first of these conferences held without Alan.
5. Wind tunnel based studies
5.1. Overview
In the space available it is not possible to mention all aspects
of wind engineering which have been influenced by Alan Daven-
port’s direct or indirect involvement. Attention is therefore given
to achievements made in the mainstream research (basic and
applied) areas of the Laboratory. One of these is the action of wind
on long span bridges. These structures first sparked Alan’s interest
in wind loading and continued to be one of his life long interests.
The buffeting theory which he developed is particularly suited for
these line-like structures. Another area is the action of wind on
vertical structures, such as tall buildings and structures. Alan’s
buffeting theory is difficult to apply to tall buildings. Typically,
their aspect ratios tend to be H/D r10 and the flow around the
building is more 3- rather than 2-dimensional. This 3-dimen-
sioanality is compounded by the downwash which occurs on the
windward face. As a result Alan turned to wind tunnel model
studies to provide wind loading information for specific tall
buildings and structures. Some of the iconic tall structures which
have been studied at the BLWTL are presented below. Some other
noteworthy specific structures are mentioned and brief com-
ments are made about generic studies made in support of
codification. It is in the area of wind action on tall buildings
where Alan’s work may be most remembered.
Fig. 9. Views of the aeroelastic wind tunnel model and the prototype of the A. Murray McKay Bridge, Halifax, N.S.
19
5.2. Long span bridges
Studies of long span bridges at the BLWTL started with the
wind tunnel model testing of the A. Murray MacKay suspension
bridge in Halifax, Nova Scotia, Canada. These tests were the first
ever made with an aeroelastic model of the full bridge in scaled
turbulent boundary layer flow. Fig. 9 shows this aeroelastic model
in the wind tunnel. The results indicated that the dynamic
behavior of the full model was distinctly different from that
predicted from conventional section model tests in aerodynamic
wind tunnels.
The A. Murray MacKay suspension bridge has a main span of
14000 (427 m) and side spans of 5130 (156 m). The bridge deck is
relatively light as its deck is orthotropic, designed to work in
combination with two truss girders located at either side of the
deck. It was studied at the BLWTL in order to confirm its design
and to determine its performance at various stages of erection.
Conventional section model tests, made with 1:40 and 1:320 scale
models indicated a coupled vertical-torsional oscillatory instabil-
ity, typical of truss-stiffened bridge sections. The critical wind
speeds determined at these two different geometric scales were
quite similar. This assured that the effects of geometric scale and
corresponding differences in Reynolds number were minimal. The
full bridge model, constructed at a geometric scale of 1:320 was
tested in uniform smooth flow and was found to be dynamically
stable. However, it exhibited a divergent instability at a speed
well in excess of the critical speed found in the section model
tests for the oscillatory instability. The 1:320 scale full bridge
model was also tested in turbulent boundary layer flows, modeled
to simulate natural wind over fetches of open and built-up
terrains. These turbulent boundary layer tests showed random
oscillatory motions in both vertical and along-wind directions
with no instability or significant torsional motions. These differ-
ences between the response of the full bridge and the conven-
tional section model tests were disconcerting. This raised doubts
about the validity of section model tests, which were routinely
relied upon in the design of long span bridges.
The notable differences in behavior were attributed to the
influence of atmospheric turbulence and the correct modeling of
the modal deflections or ‘‘mode shapes’’ achieved with the full
aeroelastic model. This resulted in a correct simulation of the
wind-induced generalized forces for various modes of vibration.
In contrast, conventional section model tests are carried out in
uniform smooth flow using mass scaled rigid models which are
mounted on springs, chosen to scale the frequency of motion, and
dashpots to provide damping. Corrections for mode shape effects
are made analytically, when converting the model data to full
scale. The combined effect of this difference in mode shape and
the absence of turbulent flow fluctuations were believed to have
caused the observed differences.
20 N. Isyumov / J. Wind Eng. Ind. Aerodyn. 104–106 (2012) 12–24
To overcome these difficulties, Alan devised the concept of a
flexible ‘‘taut-strip’’ model, which would simulate the model
displacements of the bridge deck and would be constructed at a
geometric model scale, which would permit testing in turbulent
boundary layer flow. The response of such taut-strip models
includes both the effects of turbulence and mode shape. Like full
aeroelastic models, taut strip models permit testing for wind
directions other than that normal to the bridge deck. Furthermore,
it does not have the disadvantage of testing full aeroelastic models
of suspension bridges at very low wind speeds, in order to satisfy
Froude number scaling. Comparisons of taut-strip and section
model tests were made for the A. Murray MacKay Bridge, as well
as the Golden Gate Bridge in San Francisco and the 1st Tacoma
Bridge in Seattle. Details of this taut-strip model approach and the
findings of these tests were presented at ICWE-3, held in Tokyo in
1971 (Davenport et al., 1971). The theory of taut-strip model
testing was subsequently revisited with further tests made during
the studies of the Bronx-Whitestone Bridge in New York. These
findings were reported at ICWE-5, held at Fort Collins, Colorado
(Davenport et al., 1979). Subsequently, the taut-strip model
became an important addition to the arsenal of wind tunnel tests
of long span bridges at the BLWTL.
The BLWTL provided wind loading information for many of the
world’s long span suspension, cable-stayed and other types of
bridges. Most frequently such tests relied on full aeroelastic
models, which simulated the entire structure, including cables,
towers and where necessary for the participation of the tower
foundations. Some of these, like the early study for the A. Murray
McKay Bridge, also examined bridge performance during stages of
construction. Partially constructed bridges tend to be most
vulnerable to wind action. Usually provided information are
maximum forces and moments at key locations for wind speeds
well above the design speed, the distribution of effective wind
loads over the entire structure for selected return period(s), deck
motions and deck accelerations required in the considerations of
the drivability of the bridge during strong winds. Other studies,
which contributed to the overall development, are those for the
Bronx-Whitestone Bridge in New York; the Sunshine Skyway
Bridge across Tampa Bay; the Tsing Ma Bridge in Hong Kong
and the multi-span concrete girder Confederation Bridge between
Prince Edward Island and the Canadian mainland. Alan Davenport
was a consultant and an advisor in the preliminary design of the
Messina Strait Bridge. Wind tunnel model tests were carried out
under his direction at the BLWTL in order to examine the
influence of the shoreline topography on the wind flow.
5.3. Iconic tall structures
5.3.1. World Trade Center, New York
Alan Davenport was hired as a wind consultant by Worthing-
ton, Skilling, Helle, Jackson and Robertson, structural engineers
for the New York World Trade Center (WTC) Towers. This was in
the early 1960s and at the time when he had completed his
doctoral presentation and had published some of the key papers,
which describe his approach to wind loading. His ideas on wind
loading influenced the design of these iconic towers, which were
to become the world’s tallest buildings at that time, albeit only for
a short time. Alan did not have his own wind tunnel at that time
and wind tunnel model studies were carried out in the Colorado
State University (CSU) Meteorological Wind Tunnel at Fort
Collins, Colorado. This was one of the few wind tunnels of that
time which could generate a representative model of turbulent
boundary layer wind. A two degree-of-freedom or ‘‘stick’’ aero-
elastic model, which simultaneously replicated the dynamic
properties in two fundamental sway modes of vibration, was
used to evaluate the response of each of the two Towers for a
Fig. 10. Aeroelastic models of the New York World Trade Center Tower in the CSU
Wind Tunnel with Alan Davenport (on left) and with Jack Cermak.
range of wind speeds and for different compass directions. These
tests were made under the direction of CSU Professor Jack
Cermak. Fig. 10 shows Alan Davenport (on the left) and Jack
Cermak with the aeroelastic stick models of the WTC Towers.
The development of wind loads for the WTC Towers relied on
cutting edge technology and all details required close scrutiny.
Comparison of wind tunnel model tests were made at the
National Physical Laboratory, Teddington, England. Information
on wind speeds and directions for the project was assembled from
available historical observations at nearby airports and compared
to the requirements of the New York City. Alan was careful in the
recommendation of the design wind speed and wind profile and
made comparisons with data available from an anemometer on
top of the Whitehall Building, in lower Manhattan. The wind
tunnel tests showed that the action of wind on the towers was
driven by across-wind forces. Close attention was given to the
dynamic action of wind on the Towers, including the aerodynamic
interference for wind directions, where one of the Towers was
upwind of the other. In such situations the downstream Tower
experienced higher dynamic wind loads. Visco-elastic dampers
were installed in the two Towers in order to control the wind-
induced motions and to achieve acceptable performance. Special
subjective moving room tests were made in order to indicate
levels of horizontal motion which would be acceptable for human
comfort The motion perception levels, determined in that study,
continue to be relied upon today (Chen and Robertson, 1972).
5.3.2. Sears Tower, Chicago (now Willis Tower)
There were a number of ground-breaking studies in the late
sixties and early seventies which used multi-degree of freedom
models to evaluate the wind-induced dynamic response of build-
ings and structures. In the case of tall buildings, such models were
necessary in order to provide information on the torsional wind
loads and responses. These could not be measured with base
pivoted ‘‘stick’’ aeroelastic models, such as used in the New York
WTC Tower studies. There were several pivotal projects during
which the methodology of aeroelastic model design and testing
was developed and honed. The most significant of these was
the Sears Tower in Chicago. This building was designed by
Skidmore Owings and Merrill and became the world’s tallest
building. Wind engineering studies started in 1970 and were
carried out in BLWTL’s first wind tunnel. The BLWTL worked with
 N. Isyumov / J. Wind Eng. Ind. Aerodyn. 104–106 (2012) 12–24
Dr. Fazlur Khan and his colleagues Hal Iyengar and John Zils to
provide information on structural loads and responses and on
local loads on the exterior envelope. The effect of the building on
ground level winds was also examined. The most noteworthy
aspect of that study was the design and testing of a 21 degree-of-
freedom aeroelastic model, which provided predictions of the
overall peak structural wind loads and the wind-induced deflec-
tions and accelerations of the Tower. Views of the structural
skeleton which simulated the structural system of the building
and the final model clad with a non-structural skin are shown in
Fig. 11.
The Sears Tower was to be located in the highly built-up
downtown ‘‘loop’’ portion of Chicago, with Lake Michigan in
relatively close proximity to the East. The roughness of the
surrounding terrain changed abruptly for some wind directions
and the mean and turbulent wind profiles were difficult to
determine. A small scale topographic model of the area was tested
in order to determine appropriate upstream exposures for the
main wind tunnel tests. Velocity measurements, made with this
topographic model, also assisted in relating wind speeds at nearby
weather stations and development of a design wind climate,
which predicted the joint occurrence of the mean wind speed
and direction for different probability levels. An important com-
ponent of this process was a study of historical upper level balloon
data at Peoria, Illinois, some 200 km southwest of Chicago.
The resonant dynamic sway response of the Tower was
governed by wind-induced vibrations in the two fundamental
sway modes. Both along-wind and across-wind dynamic forces
were found to be important. The shape of the building changes at
heights, where 7 of the 8 outside tubes of the bundled tube
structural system are discontinued. The plan form of the building
is square with 9 square tubes at the base. These shape changes and
the turbulence generated by nearby tall buildings were beneficial
in reducing the effectiveness of the across-wind excitation. Con-
tributions to the dynamic loading from higher sway modes were
found to be minimal. This confirmed previous assumptions of the
dominance of the fundamental modes—wind finds it more diffi-
cult to excite higher sway modes. On the other hand, the wind-
Fig. 11. Structural skeleton and fully clad model of the 21 D.O.F. aeroelastic wind tunnel model of the Sears Tower, Chicago, Illinois.
21
induced excitation of the fundamental torsional mode of vibration
was found to be significant.
An important consequence of this study was the confirmation
of the joint action of wind loads in orthogonal principal sway
directions of the structure and in torsion. Influence lines were
used to relate wind forces in selected columns to the measured
base moments and torque. This permitted the evaluation of the
instantaneously acting wind forces in different parts of the
structure and the development of joint action factors which allow
for the simultaneous action of wind loads determined in each of
the two principal sway axes and in torsion. This important concept
of the joint action of along-wind, across-wind and torsional wind
loads has since been adopted by most building codes.
5.3.3. CN Tower, Toronto
This iconic tall structure was only recently overtaken by the
Burj Khalifa in Dubai as the world’s tallest free-standing structure.
Replica aeroelastic models were first studied for an initial design
with three constant diameter concrete tubes tied together with
shear braces. This proved to be overly costly. Detailed wind
loading information was then developed for the finally con-
structed version of the tower. Structural wind load and wind-
induced response data were developed with reliance on replica
aeroelastic model tests. Views of the prototype structure and its
aeroelastic wind tunnel model are shown in Fig. 12. Other types of
wind tunnel models were used to provide information on local
pressures, winds at base level, performance of the air-supported
micro-wave radomes, etc.
Here too, the development of a design wind climate was one of
the main challenges. Surface data were available from the Toronto
City Airport located just to the south of the site of the Tower.
However, data from an anemometer located at 10 m above
ground alone was judged insufficient for the design of this very
tall structure. Small scale topographic wind tunnel model study
was carried out for the project area and other data from a surface
anemometer at the Lester B. Pearson International Airport and a
tower based anemometer located to the north of the downtown
22 N. Isyumov / J. Wind Eng. Ind. Aerodyn. 104–106 (2012) 12–24
Fig. 12. View of the aeroelastic wind tunnel model and the prototype of the CN-Tower, Toronto, Ontario.
area were included. Also evaluated were upper balloon data from
Buffalo, New York.
The aeroelastic model of the 553 m high tower was constructed
at a geometric scale of 1:450 in the early 1970s. By that time the
BLWTL had acquired much expertise and experience with the
design and construction of intricate aeroelastic models. The model
of the concrete shaft, which extended to a height of 450 m, was
constructed as a replica model from 0.25 mm thick sheets of a
commercially available metalized epoxy material known by the
trade name DEVCON A. The sheets were cut and glued to form the
tapered box girder shaft. This material has a low modulus of
elasticity and a density and inherent damping, which are similar
to those of reinforced concrete. The structural properties of the
antenna superstructure were simulated with a central spine
machined from aluminum tubing. This spine was enclosed with
non-structural radomes machined to give correct exterior dimen-
sions. Details of the wind tunnel model testing and the subsequent
full scale measurements made to evaluate prototype performance
were reported at the 12th IABSE Congress (Isyumov et al., 1984).
Aeroelastic wind tunnel models are usually constructed to
simulate the elastic properties of a structure, typically obtained
below working stress levels. Estimates of performance at higher
stress levels are then extrapolated. This involves assumption of the
‘‘softening’’ or loss of stiffness of the structure and the increase in
energy dissipation or inherent damping at higher stress levels. These
effects of structural softening and increased structural damping
were found to offset each other. Consequently, tests with an elastic
model with a constant damping provided good estimates of struc-
tural performance at high stress levels. A paper written after the
25 year anniversary of the completion of the CN Tower summarizes
the performance of the prototype Tower and also provides informa-
tion on this important structure (Mardukhi et al., 2001).
5.4. Other studies
5.4.1. Buildings
Some other major building projects at the BLWTL deserve
mention. A pivotal study in the seventies was for the
John Hancock Building in Boston. The original wind tunnel model
studies for this building were made elsewhere, as the BLWTL was
judged to be too expensive. However, when the problems with
the building surfaced during its construction, Bruno Thurliman, a
Swiss engineer was brought in by the architects to do a complete
review of the structure. Bruno insisted on involving Alan and his
team. Other notable studies include; the Hong Kong Bank (the
first commercial application of the force balance concept) and the
Bank of China Tower in Hong Kong; the principal Towers of the
Toronto Bank District; the entire Canary Wharf Complex in
London, England; the Jin Mao and the Shanghai World Financial
Center Towers in Shanghai and many others. Collectively, the
findings from such studies added to the growing data of wind
loading information. Individually, such findings extended some
part of the envelope of known wind action.
5.4.2. Tall structures which were not built
Many buildings and structures were studied but not built.
Development is a speculative process. Three of these warrant
mention:
(1) Menara (Tower) Jakarta—this projected 558 m high rein-
forced concrete telecommunication/observation was exten-
sively studied, with main findings reported at ICWE-10 in
Copenhagen and in the Journal Wind and Structures (Isyumov
et al., 2001). It was projected to surpass the height of the CN
Tower and become the world’s tallest free-standing structure.
It was not built as the Indonesian economy collapsed.
(2) Nippon Tower at Kukora—this 750 mþsteel structure was
also to be a telecommunication/observation tower and was
studied to evaluate concept feasibility and performance in
typhoon winds. If constructed, this would have been a major
upward push of the envelope of tall structures.
(3) Millennium Tower—designed by Norman Foster and Partners
for the Obayashi Corporation this cone-shaped tower was
projected to have a total height of about 800 m. This residen-
tial/commercial development was to be constructed offshore
near Tokyo and was to house a population of some 50,000.
 N. Isyumov / J. Wind Eng. Ind. Aerodyn. 104–106 (2012) 12–24
All 3 of these super-high structures extended beyond the
known design envelope and each required special consideration.
The foremost difficulty in this process was the choice of the
design wind speed. It should be noted that the projected Nippon
and the Millennium Towers, studied in the early 1990s, were in
the same height range as the recently completed 828 m
Burj Khalifa in Dubai.
5.4.3. Roofs
Studies for many large span roofs were made at the BLWTL
using both pressure and aeroelastic models. One of these war-
rants special mention. This is the Haj Terminal near Mecca,
Saudi Arabia designed by Skidmore Owings Merrill at the time
when Dr. Fazlur Khan was still alive. The structure of enormous
footprint consisted of interconnected tent modules open on the
sides and each with an opening at its peak. Concerns were for the
action of wind on the tensioned tent fabric and in particular the
possibility of a ‘‘flapping’’ type instability of the leading edges of
tent modules positioned along outside of the arrays which made
up the entire structure. An aeroelastic model which simulated the
tensioned roof fabric and the support cables of an outside tent
module was designed and wind tunnel tested. Results indicated
that flapping type instability of the fabric would not occur at wind
speeds of practical interest.
Also, simultaneous measurements of local pressures on
selected models were made to estimate overall wind loads. This
was one of the earliest attempts to extract instantaneous overall
loads from the spatial averaging of simultaneously measured local
pressures. This proved to be a challenge as these tests were made
prior to the availability of solid state switching. Pneumatic
switching with Scani-valves was the pressure technology of that
day and only 16 pressure taps could be examined simultaneously.
5.4.4. Use of influence lines
Alan’s knowledge of the theory of structures allowed him to
use influence line theory for describing wind loads and responses
of structures. This approach is particularly effective in situations
where wind-induced resonant vibrations are of secondary impor-
tance and where the excitation by background turbulence dom-
inates. This includes the wind loading on large span roofs, guyed
towers and low-rise buildings. Examples of this approach can be
found in Davenport and Sparling (1992).
6. Codification
Alan Davenport had a life long involvement with the improve-
ment of wind loading specification in codes and standards in
North America, Europe, the Developing Countries and interna-
tionally. In Canada he continued as an active member of the
National Building Code of Canada (NBCC), well beyond his retire-
ment from the BLWTL and from the University.
His contributions were based on his insight into the action of
wind action on buildings and structures and the growing data
base at the BLWTL, both from studies of specific commercial
projects and from basic research carried out with graduate
student involvement. One of these warrants special mention.
Professor Ted Stathopoulos, now with Concordia University, in
his graduate studies with Alan examined wind loads on generic
low-rise buildings. This pioneering work demonstrated that the
instantaneously acting wind pressures are not fully correlated
over the whole structure and developed equivalent pattern loads
for structural design (Stathopoulos, 1979). Influence lines were
used to determine how key structural quantities, such as the
maximum bending moments and the shear and vertical forces at
the frame supports were influenced by the instantaneous action
23
of local pressures on exterior surfaces of typical low-rise build-
ings. It also showed that component loads varied with the size of
their tributary area. This research was carried out with the
encouragement and support of the Metal Building Manufacturers’
Association. These pattern structural loads, as well as the corre-
sponding local pressures on cladding and components, have since
been codified and appear in leading building codes and standards,
including the NBCC and the American Society of Civil Engineers
Standard ASCE-7.
7. Concluding Remarks
This paper has touched on some of the many contributions
which Alan Davenport and his Boundary Layer Wind Tunnel
Laboratory have made to wind engineering worldwide. It is not
possible to be all inclusive in one paper and individual readers no
doubt have additional recollections, not mentioned here. Alan’s
passing, signals the end of the 1st tome of wind engineering
where, the primary goals were to lay the foundations of this new
discipline, with emphasis on the wind loading process. It required
new ideas and cutting-edge research with physical models of
buildings and structures and tests in wind tunnels which were
able to simulate the turbulent boundary layer characteristics of
natural wind. Many of the ideas and methods, which were once
novel have now, become conventional or routine. Simply said, the
wind engineering discipline has matured and is entering its next
stage with sights set on new directions and new challenges. Time
and new players will determine the next stage of this discipline.
At ICWE-13 in Amsterdam, the International Association of
Wind Engineering (IAWE) formally recognized the term ‘‘Alan G.
Davenport Wind Loading Chain’’, when referring to Alan’s
approach for estimating the action of wind, as required for wind
engineering applications. This marks an important step in
the history of wind engineering and honors the late Alan G.
Davenport for his many contributions. The recognition and the
use of the term the ‘‘Alan G. Davenport Wind Loading Chain’’ by
the wind engineering community are important and fitting
honors. This will help to keep both the man and his achievements
in our memories.
Acknowledgments
Contributions by other members of the Boundary Layer Wind
Tunnel Laboratory towards the preparation of this paper are
gratefully acknowledged. In particular, Dr. Peter King helped to
generate some of the figures and he and Dr. Eric Ho proof read the
manuscript and made valuable suggestions. Ms. Karen Norman
put the manuscript into its final form.
References
ASCE, 1999. Manuals and Reports on Engineering Practice no. 67, Wind Tunnel
Studies of Buildings and Structures.
Canadian Journal of Civil Engineering, 2011. Special Issue in Honor of Founding
Editor Professor Alan G. Davenport (1932–2009), vol. 38 (9). In: Bartlett, F.M.
(Ed.), NRC Research Press, September 2011.
Chen, P.W., Robertson, L.E., August 1972. Human perception thresholds of
horizontal motion. In: Proceedings of the American Society of Civil Engineers.
Journal of Structural Division, vol. 98.
Davenport, A.G., March 1961a. A Statistical Approach to the Treatment of Wind
Loading on Tall Masts and Suspension Bridges. Ph.D. Thesis. Department of
Civil Engineering, University of Bristol, United Kingdom.
Davenport, A.G., April 1961b. The spectrum of horizontal gustiness near the
ground in high winds. Quarterly Journal of the Royal Meteorological Society
87, 194–211.
Davenport, A.G., 1964. Note on the distribution of the largest values of a random
function with application to gust loading. In: Proceedings of the Institution of
Civil Engineers. vol. 28, p. 187.
24 N. Isyumov / J. Wind Eng. Ind. Aerodyn. 104–106 (2012) 12–24
Davenport, A.G., 1965a. The relationship of wind structure to wind loading. In:
Proceedings of the Conference on Wind Effects on Structures. National
Physical Laboratory (NPL United Kingdom), (ICWE-1), HMSO, p. 53.
Davenport, A.G., 1965b. The buffeting of structures by gusts. In: Proceedings of the
Conference on Wind Effects on Structures. National Physical Laboratory (NPL
United Kingdom), (ICWE-1), HMSO, p. 357.
Davenport, A.G., June 1967. Gust loading factors. In: Proceedings of the American
Society of Civil Engineers. Journal of Structural Division, vol. 93.
Davenport, A.G., Isyumov, N., Miyata, T., 1971. The experimental determination of
the response of suspension bridges to turbulent wind. In: Proceedings of the
Third International Conference on Wind Effects on Buildings and Structures,
Tokyo, Japan.
Davenport, A.G., September 1972. An approach to human comfort criteria for
environmental wind conditions. Colloquium on Building Climatology, Stock-
holm, Sweden.
Davenport, A.G., December 1975. Tall buildings—an anatomy of wind risk.
Construction, South Africa.
Davenport, A.G., Isyumov, N., Rothman, H., Tanaka, H., July 1979. Wind induced
response of suspension bridges—wind tunnel model and full scale observa-
tions. In: Cermak, J.E. (Ed.), Proceedings of the Fifth International Wind
Engineering Conference, vol. 1, Pergamon Press, Fort Collins, Colorado, USA.
Davenport, A.G., Sparling, B.F., 1992. Dynamic gust response factors for guyed
towers. Journal of Wind Engineering and Industrial Aerodynamics 41–44,
2237–2248.
Davenport, A.G., Grimmond, C.S.B., Oke, T.R., Wieringa, J., 2000. Estimating the
roughness of cites and sheltered country. In: Proceedings of the. 12th
American Meteorological Society Conference on Applied Climatology.
Engineering Science Data Units (ESDU), 1990. Strong Winds in the Atmospheric
Boundary Layer. Part I: Mean Hourly Wind Speeds, Item no. 82026, Amend-
ments A to C.
Gatey, D., Isyumov, N., Case, P., Ho, T.C.E., Miller, C., 2011. Design wind climates for
very tall structures. In: Proceedings of the 13th International Conference on
Wind Engineering, Amsterdam, The Netherlands.
Harris, R.I., Deaves, D.M., 1986. The structure of strong winds. In: Proceedings of
the CIRA Conference on Wind Engineering in the Eighties. Construction
Industry Research and Information Associates, London, England.
Irwin, P.A., 2007. Wind Engineering challenges of the new generation of super-tall
buildings. In: Proceedings of the 12th International Conference on Wind
Engineering, Cairns, Australia.
Isyumov, N., Davenport, A.G., Monbaliu, J., 1984. CN-Tower, Toronto: model and
full-scale response in wind. In: Proceedings of the International Association for
Bridge and Structural Engineering (IABSE), 12th Congress, Vancouver, British
Columbia, Canada.
Isyumov, N., Case, P.C., Ho, T.C.E., Soegiarso, R., 2001. Wind tunnel model studies to
predict the action of wind on the projected 558 m Jakarta Tower. International
Journal of Wind and Structures 4 (4).
Isyumov, N., 2009. Obituary of Alan Garnett Davenport (1932–2009). Journal of
Wind Engineering and Industrial Aerodynamics 97 (9–10), 409–412.
Isyumov, N., Ho, T.C.E., King, J.P.C., Surry, D., Vickery, B.J., October 2010. The Japan
connection. Journal of Wind Engineering, Japan Association for Wind Engi-
neering (JAWE) 35 (125).
Jensen, Ing. M., November 1958. The Model-Law for Phenomena in Natural Wind.
Ingeniøren, International Edition, vol. 2 (4), Copenhagen, Denmark, pp. 121–
128.
Krishna, P. (Ed.), International Association for Wind Engineering, Wiley Eastern
Limited, New Delhi, India.
Mardukhi, J., Isyumov, N., Knoll, F., 2001. CN-Tower—performance over a quarter
century. In: Proceedings of the Canadian Society of Civil Engineering (CSCE)
29th Annual Conference, Victoria, British Columbia, Canada.
National Building Code of Canada, 2005. National Research Council, Ottawa,
Canada.
Stathopoulos, T., 1979. Turbulent Wind Action on Low Rise Buildings. Ph.D. Thesis.
University of Western Ontario, Faculty of Engineering Science, London,
Ontario, Canada.
Solari, G., Cheung, J., Isyumov, N., Kareem, A., Stathopoulos, T., Surry, D., Tamura,
Y., May 2008. The Davenport medal: a tribute from the International Associa-
tion for Wind Engineering to Alan Garnett Davenport. Journal of Wind
Engineering and Industrial Aerodynamics 96 (5).
Taylor, G.I., 1935. The statistical theories of turbulence. In: Proceedings of the
Royal Society, London, England, Series A, vol. 151, pp. 421–444, 455–464.
Tschanz, T., 1982. The Base Balance Measurement Technique and Application to
Dynamic Wind Loading of Structures. Ph.D. Thesis. University of Western
Ontario, Faculty of Engineering Science, London, Ontario, Canada.
Van der Hoven, I., 1957. Power spectrum of horizontal wind speed in the
frequency range from 0.0007 to 900 cycles per hour. Journal of Meteorology
14, 160.
Wieringa, J., Davenport, A.G., Grimmond, C.S.B., Oke, T.R., 2001. New revision of
Davenport roughness classification. In: Proceedings of the International 3rd
European and African Conference on Wind Engineering.
Xu, Z., Bekele, S., Mikitiuk, M., Isyumov, N., Ho, E., 2008. Bi-modal Weibull
distribution fit for climatic wind speed histogram. In: Proceedings of the
13th Australasian Wind Engineering Society Workshop, Hobart, Australia.