l bu

nd

out

win

lysi

istic

wi

For

or l

shape both the architectural design and structural design. The wind tunnel methods used include the

force balance technique, aeroelastic modeling, high frequency pressure integration tests, as well as

the traditional pressure model and pedestrian wind studies. A super-tall building pushes the limits of the

1. Introduction

ncingction og wasBuildi

ed, arere onhe cur

to cite comprehensively in a paper of this length. Some of these

load.t alsoithin

the towers shape needs to be considered as a critical design

ARTICLE IN PRESS

Contents lists available at ScienceDirect

.el

Journal of Windand Industrial A

J. Wind Eng. Ind. Aerodyn. 97 (2009) 328334be carefully examined. Increasingly the structural designers of veryE-mail address: Peter.Irwin@rwdi.comappear in the reference lists of the references cited in this paper. parameter from the very outset.The response of the tower to wind depends not only on its shape

but also its stiffness distribution, mass distribution and damping.For optimal design the interplay between these variables needs to

0167-6105/$ - see front matter & 2009 Elsevier Ltd. All rights reserved.

doi:10.1016/j.jweia.2009.05.001topic. Nonetheless, it is important to acknowledge that manyother researchers have made major contributions, too numerous

comfortable limits is often a bigger challenge than meetingstructural strength requirements. Therefore, the aerodynamics ofengineering. These are discussed in this paper. It is primarilybased on the experience of the author and his colleagues and it isnot intended to be a comprehensive review of the literature on the

and for super-tall buildings wind is the governing lateralWind affects not only the structural integrity of the tower buits serviceability. Keeping the motions of the tower wTaipei 101 at 509m. Several are under design with heights wellover that of Taipei 101. There are a number of designs beingcontemplated in the 5001500m range. Burj Dubai, Fig. 1, whichis scheduled for completion by 2009, will be well over 700m tall.

This new generation of towers poses new challenges for wind

alone, especially when their response is so sensitive to windspeed, and sometimes, direction. More reliable information onupper level wind statistics is needed.

The aerodynamics of tall buildings can have a huge impact ontheir cost. The main structural system is a large part of the costbuildings have either been constructto use an old fashioned expression, athe height range 400m and up. TIllinois, at 440m, held the record for many years. Now, numerous engineering have traditionally been almost entirely based onIn the present day we are experieof activity in the design and construused to be that a 300m high buildinfew buildings exceeded. The Searsforce balance method due to difculties in maintaining sufcient model stiffness and in accounting for the

inuence of higher modes of vibration. Since the impact of wind on people using terraces and balconies

increases with building height, it is an issue needing particular attention for super-tall buildings.

& 2009 Elsevier Ltd. All rights reserved.

an unprecedented levelf super-tall buildings. Ita threshold that only ang, located in Chicago,

under construction, or,the drawing boards inrent worlds tallest is

Most building codes still use traditional models of theplanetary boundary layer, developed in the 1960s, that assume theboundary layer tops out between about 250 and 500m, depend-ing on exposure. The validity of these models is questionablewhen dealing with building heights above about 300 in.

The statistics of wind speed and direction used in wind

records from ground based meteorological stations, taken at about10m height. It is a long extrapolation to develop roof height windstatistics for super-tall buildings from the ground-based dataWind engineering challenges of the new

Peter A. Irwin

RWDI, 650 Woodlawn Road West, Guelph, Ontario, Canada

a r t i c l e i n f o

Article history:

Accepted 18 May 2009Available online 11 August 2009

Keywords:

Tall buildings

Wind effects

Wind tunnel testing

Wind statistics

a b s t r a c t

The new generation of tal

wind engineering. The bou

for buildings less than ab

statistics of upper level

the archived global re-ana

the upper level wind stat

buildings of non-synoptic

winds of the Middle East.

earlier in the design than f

journal homepage: wwwgeneration of super-tall buildings

ildings is going much higher than before. This poses new challenges for

ary layer models in many building codes and standards have served well

300m but more realistic models need to be used above 300m. The

ds need also to be known with better certainty. New tools such as

s data coming from weather forecast models can help shed more light on

s. There are also questions to be answered about the effects on all tall

nd proles such as occur in thunderstorm downbursts and the Shamal

the super-tall buildings wind tunnel testing is often commenced much

esser buildings. This permits the results to be used in a pro-active way to

sevier.com/locate/jweia

Engineeringerodynamics

tall towers are prepared to extend their thinking beyond the

type winds, more in the 20003000m range. Not only does thewind speed in this model continue to increase with height all theway to the tops of super-tall buildings (and beyond) but, just asimportant, the ow is turbulent up there. The traditional modelwould have the tops of these buildings in smooth uniform ow.

ARTICLE IN PRESS

stic

10.0

20.0

30.0

40.0

1000100101Return Period (years)

Pred

icte

d W

ind

Spe

100 m 200 m 400 m 600 m

Fig. 2. Estimated extreme wind speeds for Las Vegas at various heights fromballoon data.

P.A. Irwin / J. Wind Eng. Ind. Aerodyn. 97 (2009) 328334 329traditional structural variables of stiffness and mass, and to treat thedamping as a third controllable structural parameter. Supplemen-tary damping systems allow them to do this and open up a wholenew range of possibilities for optimizing the design. To date theiruse has been targeted primarily towards satisfying serviceabilitycriteria. However, they also have the potential to mitigate ultimatedesign wind loads, in a similar manner to their use in earthquakedesign.

The wind tunnel techniques used for super-tall buildings arelargely the same as for lesser towers but their extreme height canpose challenges. Typically smaller model scales become necessary,and, because of the importance of aerodynamics, more iterationsof shape may well be needed during the design optimizationprocess. Because of the time required to build super-tall towersthe design of the upper portions is often still underway duringconstruction. However, wind tunnel tests to establish base loadsmay well have to be done while there is still uncertainty as towhat the top part of the tower will nally look like. Therefore,initial testing to supply foundation loads must allow for thepossible range of shapes that the nal design might take.

Another issue that arises for super-tall towers is that windspeeds on terraces high up on the tower can be expected to bemuch higher than on normal buildings. Yet there is often a desireto have these terraces as usable space.Fig. 1. Burj Dubaifuture view, aeroela2. Wind statistics and wind proles

In North America the ASCE 7-05 standard sets the standard forwind design in the USA and in Canada the National Building Codeserves this purpose. The boundary layer models in thesedocuments are very similar to each other and were developedempirically in the 1960s. They will be referred to here astraditional models. They have boundary layer depths rangingfrom about 210m in very at open terrain to 460m in dense urbanterrain. These models appear to have served well for the vastmajority of buildings. However, they are purely empirical and notbased on much consideration of atmospheric physics. The vastmajority of buildings on which our experience is based comenowhere near high enough to test the assumptions concerningboundary layer depth in these traditional models. However, thenew generation of super-tall towers certainly does.

The Harris and Deaves (1981) model, which was adopted in the1980s by ESDU (1993), is based on more fundamental physicalconsiderations than the traditional model and at high windspeeds indicates considerably deeper boundary layers in synopticmodel and recent construction photo.

Predicted Wind Speeds by height fromBalloon Upper Air

50.0

60.0

ed (m

/s)Turbulence can have important inuences on vibration phenom-ena and aerodynamic instabilities such as vortex shedding andgalloping. The continuation of the boundary layer to much greaterheights than predicted by the traditional models is also supportedby balloon measurements and weather forecasting computermodels.

Figs. 2 and 3 show estimated extreme mean hourly wind speedsat various heights for the Las Vegas area based, respectively, on12 years of twice daily upper air balloon soundings and 20 yearsof global re-analysis data. The results were obtained usingextreme value analysis methods on the monthly extremes. The 20years (19872006) of re-analysis data were obtained from theNational Center for Atmospheric Research/National Centers forEnvironmental Prediction (NCAR/NCEP). The NCAR/NCEP datasetsare based on a worldwide meteorological observation network,including surface and upper air balloon measurements, satelliteand radar measurements, etc. The data were available at 3hintervals on a three-dimensional grid and were derived bymeteorological modeling software similar to that used forweather forecasting.

ARTICLE IN PRESS

P.A. Irwin / J. Wind Eng. Ind. Aerodyn. 97 (2009) 328334330Predicted Wind Speeds by height fromNAM\NARR

10.0

20.0

30.0

40.0

50.0

60.0

1000100101Return Period (years)

Pred

icte

d W

ind

Spee

d (m

/s)

100 m 200 m 400 m 600 m

Fig. 3. Estimated extreme wind speeds for Las Vegas at various heights fromarchived global re-analysis data.What is clear in both Figs. 2 and 3 is that when estimatingextreme wind speeds at the tops of very tall buildings, the windspeed continues to increase signicantly above 400m. Fig. 4shows the predictions of extreme wind speeds based on thetraditional boundary layer model, i.e. a power law with exponent0.14, but assuming the power law extends to at least 600m height.Comparing with 600m results in the 50100 year return periodrange in Figs. 2 and 3, the traditional boundary layer model, Fig. 4,gives slightly higher wind speeds. It is noteworthy also that athigh return periods the slope of the wind speed versus returnperiod is lower in all cases than implied in the commentary ofASCE 7 (2005).

The ratio of wind speed at 600m to that at 10m as a functionof wind speed as derived from the NCAR/NCEP re-analysis datashows interesting trends. Figs. 5 and 6 show this ratio as functionof wind speed at 10 and 600m, respectively, for the Chicago area.At high 10m level speeds the ratio asymptotes very well towardsthe value 1.77 that is derived by assuming a 0.14 power law for themean velocity. However, at high 600m level speeds the ratio ishigher indicating that 10m wind speeds are not always a goodindicator of high winds up at 600m. This emphasizes theimportance of direct measurements of winds at upper levelsrather than relying on extrapolations from ground-based data.

still affect the response at the upper end of the design speed

Predicted Wind Speeds by height fromSurface Observations

10.0

20.0

30.0

40.0

50.0

60.0

1000100101Return Period (years)

Pred

icte

d W

ind

Spee

d (m

/s)

10 m 100 m 200 m 400 m 600 m

Fig. 4. Estimated extreme wind speeds for Las Vegas at various heightsextrapolated from 10m height data.range. Fig. 8 illustrates the estimated across-wind peakacceleration response, due to both buffeting and vortexshedding, of a 600m tall tower with the modes similar to thosejust described. It can be seen that there is hump in the responsearound 17m/s due the excitation by vortex shedding in the rstmode. Then at high speed around 5060m/s there is a secondhump. Traditional criteria for maximum acceptable accelerationsin buildings, typically based on events with return period in the110 year range, need to be re-assessed for super-tall buildingswith very low rst mode frequency. Perceptible motions could befelt in these buildings on a very frequent basis unless care is takento avoid vortex excitation or to supplement the damping to reducevortex-shedding amplitudes.

While there are measures that can be taken in the structuraldesign such as stiffening, adding mass or introducing supplemen-tary damping systems, these do not attack the vortex shedding atits source. The source is the building shape and it is possible toWhile the Harris and Deaves (1981) model is an improvementon the traditional model, it does assume that the winds of interestare created by synoptic, i.e. large-scale wind systems such as anti-cyclones, where the wind blows in a roughly constant directionover large distances. However, there is evidence that a goodproportion of the strong wind events experienced at ground levelin many parts of the world are due to small-scale phenomena suchas thunderstorms and the downburst phenomenon associatedwith them. The wind proles in downbursts are very differentfrom the equilibrium proles depicted in traditional or the Harrisand Deaves boundary layer models, taking the form of a jet ofhigh-speed air near the ground as depicted in Fig. 7. Jets atsomewhat higher levels can occur in the Shamal winds that arefound in the Arabian Gulf area (Qiu et al., 2005) as also illustratedin Fig. 7. The impact of these non-standard wind proles on tallbuildings needs further research.

3. Optimization of shape

One of the critical phenomena that effect tall slender towers isvortex excitation. The well-known expression of Strouhal gives thefrequency N at which vortices are shed from the side of thebuilding, causing oscillatory across-wind forces at this frequency.

N SUb

(1)

where S Strouhal number; U wind speed; and B buildingwidth.

The Strouhal number is a constant with a value typically in therange 0.10.3. For a square cross-section it is around 0.14 and for arough circular cylinder it is about 0.20. When N matches one ofthe natural frequencies Nr of the building, resonance occurs whichresults in amplied across-wind response. From Eq. (1) this willhappen when the wind speed is given by

U NrbS

(2)

Thus for a building 50m wide, and with Strouhal number 0.20say, and with a fundamental natural frequency of 0.06Hz, which ispossible for a super-tall structure of 150 or more stories, theresonant condition of the fundamental mode will happen whenthe mean speed at the top of the building is only U 15m/s. Thisis a very common speed. A representative second mode frequencyof such a building would be around 0.2Hz, which is similar to therst mode of a typical 50-story building. The resonant conditionfor the second mode will occur at 50m/s, which is likely tovirtually eliminate the vortex shedding forces through selection

ARTICLE IN PRESS

P.A. Irwin / J. Wind Eng. Ind. Aerodyn. 97 (2009) 328334 331andtha

Fig

Figrenement of the building shape. There are several directionst one can go in developing an aerodynamically favorable shape.

Softened corners: Square or rectangular shapes are verycommon for buildings and experience relatively strong vortexshedding forces. However, it is found that if the corners can besoftened through chamfering, rounding or stepping theminwards, the excitation forces can be substantially reduced. Thesoftening should extend about 10% of the building width infrom the corner. The corners on Taipei 101 were stepped in

. 5. Ratio of mean wind speeds at 60010m as a function of mean wind speed at 10m fo

Fig. 6. Ratio of mean wind speeds at 60010m as a function

0

200

400

600

800

heig

ht, m

0 10 20 30 40 50wind velocity, m/s

SynopticShamal

Thunderstorm

. 7. Typical mean velocity proles in synoptic, thunderstorm and Shamal winds.Figandorder to reduce across-wind respond and drag, resulting in a25% reduction in base moment (Irwin, 2005).Tapering and setbacks: As indicated in Eq. (1), at a given windspeed, the vortex shedding frequency varies depending on theStrouhal number S and width b. If the width b can be varied upthe height of the building, through tapering or setbacks, thenthe vortices will try to shed at different frequencies at differentheights. They become confused and incoherent, which candramatically reduce the associated uctuating forces.Varying cross-section shape: A similar effect can be achieved byvarying the cross-section shape with height, e.g. going from

r the Chicago area. Line at 1.77 represents value obtained from 0.14 power law.

of mean wind speed at 600m for Chicago area.

0

20

40

60

80

100

Acc

eler

atio

n, m

illig

0 10 20 30 40 50 60 70 80 90Mean wind speed at top, m/s

Modes 1 and 2 Mode 1 only

f(1) = 0.058Hzf(2) = 0.200HzCL' (1) = 0.170CL' (2) = 0.120

dCLda (1) = 0.700dCLda (2) = 0.700zeta (1) = 0.010zeta (2) = 0.010

. 8. Example of estimated across-wind response of 150-story tower with rstsecond harmonic frequencies of 0.06 and 0.20Hz.

by vortex shedding is to stiffen the building sufciently (i.e. toincrease Nr enough) to force the resonant speed above the top end

recvarcosvarIrwthecon

theamtowdam

ARTICLE IN PRESS

P.A. Irwin / J. Wind Eng. Ind. Aerodyn. 97 (2009) 328334332ge extent it frees the design team from the need to constrainbuilding shape because of aerodynamics or to spend largeounts on increasing stiffness and/or mass. With the trendards taller and taller buildings the day of the supplementarylarThere are a variety of damping systems that have been used. Inent years tuned mass dampers or tuned liquid dampers ofious types have been increasingly implemented as the mostt benecial approach. It is not intended here to discuss theious types of damping system that have been used in detail.in and Breukelman (2001) describe several projects wherey were used. Fig. 6 illustrates several different dampergurations. The advantage of a damping system is that to aof the design range. However, this can be extremely expensive andcan become impractical on a super-tall tower.

Another approach is to increase the buildings mass. Theamplitude of motion caused by wind excitation tends to varyinversely with Scruton number 2md=rB2, where m generalizedmass per unit height, d logarithmic decrement of damping andr air density (Zdravkovich, 1982). From this relationship it canbe seen that increasing the mass helps. However, adding mass alsoresults in extra cost.

From the Scruton number relationship just described it is clearalso that another way to reduce the amplitude of wind-excitedmotion is to supplement the damping, i.e. increase d. This leavesthe vortex resonance within the design speed range but, withsufcient damping, suppresses the resulting motions to anacceptable level. The supplementary damping system bleedsenergy out of the motion sufciently fast to counter the energybeing put in by the vortex shedding.square to round. In this case the Strouhal number S varies withheight, which again, in accordance with Eq. (1) causes theshedding frequency to be different at different heights.

Spoilers: One can also reduce vortex shedding by addingspoilers to the outside of the building. The most well knownform of spoilers are the spiral Scruton strakes used on circularchimneystacks (Scruton, 1963). Architecturally and practically,the Scruton strake leaves something to be desired for circularbuildings, but other types of spoiler could be used that mightbe more acceptable, such as vertical ns at intervals up theheight.

Porosity or openings: Another approach is to allow air to bleedthrough the building via openings or porous sections. Theformation of the vortices becomes weakened and disrupted bythe ow of air through the structure.

While vortex shedding is the principal culprit causing undesirablyhigh across-wind motions, another cause is buffeting by turbu-lence cast off from upstream buildings. This is less easy to dealwith through the building shape since the origin of the turbulenceis not the building itself. However, some cross-sectional shapes,e.g. a lens shape, are more prone to across-wind buffeting becausetheir streamlined shape causes them to act somewhat like avertical aerofoil, generating high across-wind force variations forrelatively small changes in angle of attack of the wind caused byturbulence. Shape changes that make them less like an aerofoilcan help in this situation. Irwin et al. (1998) describe some resultswith different shapes.

4. Stiffness, mass and damping

Based on Eq. (2), one approach to tackling the problems causedping system has truly arrived.5. Wind tunnel testing

5.1. Its use as design tool

How has the advent of the massively tall building in the last fewyears altered wind tunnel testing? Since shape is so critical for thesebuildings, one of the trends has been for the wind tunnel to be usedmore proactively to optimize the aerodynamic shape in an iterativeprocess starting early on in the design process. This is in contrast tothe more conventional and passive approach of simply testing at apoint near the end of design and then presenting results when theshape is already xed with little or no consideration of aero-dynamics. To quote the structural engineer for Burj Dubai, Bill Bakerof Skidmore Owings Merrill: We practically designed the tower inthe wind tunnel (ENR, 2006). The shape of Burj Dubai was tailoredto optimize the aerodynamics through a series of iterative testsstarting at the earliest stages of design (Irwin and Baker, 2005).Through this approach a building shape was developed that wasextremely efcient from a wind loading point of view to the pointwhere the tower has no need for a supplementary damping system.The same philosophy was extended right to the top pinnaclestructure, a slender steel tube. A combination of tapering andaerodynamic ns was developed through wind tunnel tests on thepinnacle to suppress vortex shedding.

5.2. High frequency force balance method

One of the most useful tools in wind tunnel testing of high-risebuildings has been the high frequency force balance (HFFB)(Tschanz, 1982; Xie and Irwin, 1998). It is still useful for super-tallbuildings, especially for early determination of design loads at thebase of the structure, but its limitations must be understood. Oneimportant limitation is that the HFFB method only works well forthe fundamental mode of vibration. For super-tall buildings theexcitation of higher order modes of vibration can becomesignicant, particularly for the upper portions of the structure.Another challenge for the traditional HFFB method, where themodel is mounted on a fast response force balance at its base, isthat a very stiff model is needed. For extremely slender and talltowers, achieving sufcient stiffness in the model is difcult. Inorder to avoid encroachment of the model resonance frequencyinto the frequency range of interest for wind loading, it becomesnecessary to test at lower wind speeds than normal. This in turncan cause the Reynolds number of the test to stray into a lowenough range that Reynolds number effects become signicant.

A recent development that promises to make the HFFB methodeven more useful as part of a wind tunnel test program has been themove from frequency domain analysis to time domain analysis (Xieet al., 2007). The HFFB method has traditionally been applied in thefrequency domain only. However, by going to the time domain moreprecise information can be gathered on the correlations betweenpeak responses in two different directions, and on the peak factorsrelating peak responses to RMS responses. When used withsimplied aeroelastic model tests, as described by Xie et al. (2007)the time domain analysis of HFFB data can be particularly powerful.For buildings with supplementary damping systems the timedomain has further advantages in allowing the non-linear char-acteristics of the damping system to be incorporated in the analysis.

5.3. High frequency pressure integration method

A way around the limitations of the HFFB method in handlinghigher mode effects is to use what RWDI has called the HFPImethod, HFPI standing for high frequency pressure integration.

This approach, rst published by Irwin and Kochanski (1995), is in

response can be measured, including higher order modes andincluding aeroelastic feedbacks effects such as aerodynamicdamping. The tests can also be run at higher Reynolds number

ARTICLE IN PRESS

P.A. Irwin / J. Wind Eng. Ind. Aerodyn. 97 (2009) 328334 333use at a number of wind tunnel laboratories, and as the nameimplies, involves integration of point pressure measurements on aninstantaneous basis to obtain time histories of overall modal loadsin each mode of vibration. Its advantage is that it is not limited tothe fundamental mode and the test speed is not constrained bymodel frequency issues, as is the case for the HFFB method.However, the obvious limitation of the HFPI method is that it is only

Fig. 9. China Central Television Headquarters.as good as the density with which the model can be instrumentedwith pressure taps. For some designs, particularly those with manyintricate features it may simply be impractical to install a sufcientdensity of pressure taps to be condent that the overall integrationwill be accurate. A useful approach to is to cross check the HFPImethod results for the fundamental mode with those of the moredirect HFFB method. The HFPI method is particularly well suited tobuildings with complex shapes such as the China Central TelevisionHeadquarters in Beijing, Fig. 9 (Xie and To, 2005). In fact it is difcultto conceive of an alternative approach for this building other than afull aeroelastic model which would be extremely intricate andcostly to construct.

For the Burj Dubai the wind tunnel test program includedspecial high Reynolds number studies using a 1:50 scale model ina large wind tunnel capable of wind speeds up to 55m/s (Irwinand Baker, 2005), see Fig. 10. Since it was not possible to modelthe entire tower at this scale, only the top portion was modeled.The measurement technique used was the HFPI method, appliedto six rings of pressure taps at different levels, and this permitteddirect comparison with identical measurements made on the1:500 scale model at lower speeds.

The same remarks that were made above with respect to use oftime domain analysis in HFFB studies applies also to HFPI studies,only of course with HFPI studies higher order modes can beanalysed as well as the fundamental mode.

5.4. Aeroelastic models

To check the nal design of a super-tall building there aremany benets to undertaking a full aeroelastic model test. The fullFig. 10. 1:50 scale model of top portion of Burj Dubai for high Reynolds numbertesting.than the HFFB tests. For very tall slender towers the design of theaeroelastic model can often be simplied since these towersusually respond in a very similar manner to a simple cantilever.Thus the model stiffness can be incorporated into a single metalspine member, its cross-section varying with height so as toachieve the desired deection shapes for the modes of vibrationthat need to be included in the study. The central photograph inFig. 1 shows the aeroelastic model of Burj Dubai, which was of thistype, with some of the outer shell segments removed to show thespine inside. Measurement of bending moments at variousheights can be made in a very straightforward manner usingstrain gauges on the spine. Fig. 11 illustrates the power spectrumof bending moment at the base of the aeroelastic model of BurjDubai. It can be seen that the rst three modes were modeled welland even the fourth mode response can be identied. Highermode responses can contribute signicantly to the wind loadingand accelerations in the upper parts of super-tall buildings.Aeroelastic model testing is the most accurate type of test but isalso more intricate than HFFB or HFPI tests and, for buildings withsignicant torsional response can become very complex indeed.For these reasons it is usually best to reserve aeroelastic modeltesting until the design has evolved to close to its nal form.

5.5. Winds at ground and higher levels

It is normal to test for pedestrian level comfort around tallbuildings and to assess the results against comfort criteria such asdescribed in ASCE state of the art report on outdoor humancomfort (ASCE, 2004). There are a number of massing features

6. Concluding remarks

The age of the super-tall building, combined with a movetowards increasingly novel shapes, has brought interesting new

ARTICLE IN PRESS

P.A. Irwin / J. Wind Eng. Ind. Aerodyn. 97 (2009) 328334334Fig. 11. Response spectrum of base moment from an aeroelastic model of BurjDubai.that can help reduce ground level speeds, see for example(Williams et al., 1999). For super-tall buildings the ground levelwind problems can be amplied due to their height but many ofthe same solutions apply at ground level. A more challenging issueis the increasing desire to have balconies and terraces high up onthe towers that can be used by occupants. In many cases it isdifcult to satisfy the normal comfort criteria at these higherlevels. For example, in a leisure area it would normally bedesirable to have wind conditions suitable for sitting over 80% ofthe time. At upper level terraces this may simply not be achievablebut perhaps 50% is achievable and may be satisfactory as far as theoccupants are concerned. It is important for the owner andoccupants that they be informed about how frequently theseareas will be usable and be warned if unsafe conditions exist.Improved conditions on terraces and balconies can be achievedthrough testing of screens and different parapet designs such asillustrated in Fig. 12. These devices represent a compromisebetween having the open feeling that makes a terrace or balconyattractive and closing them in which, if carried too far defeats, theoriginal purpose of these design features.

Fig. 12. Model testing of solutions to improve wind comfort on terraces.primarily on the tasks of determining and controlling thestructural response to wind action. However, we are also in anage when interest in green buildings is rapidly increasing,brought on by concerns about global warming. One way a buildingcan improve its greenness is to use less material in its structuralsystems and the techniques discussed here for rening knowledgeof wind loads, and methods of reducing them, certainly helpachieve this. However, the use of natural ventilation, doublelayered wall systems, and even integrated wind turbines arebeginning to present yet further exciting challenges for windengineers.

References

ASCE, 2004. Outdoor Human Comfort and Its Assessment. In: Irwin, P.A. (Ed.), ASCEState of the Art Report. Prepared by a task group of the AerodynamicsCommittee. American Society of Civil Engineers.

ASCE 7, 2005. Minimum Design Loads on Buildings and Other Structures. AmericanSociety of Civil Engineers ASCE 7-05 Standard.

ESDU, 1993. Strong Winds in the Atmospheric Boundary Layer. Part 2: DiscreteGust Speeds, Item 83045, Issued November 1983 with Amendments A and BApril 1993. Engineering Sciences Data Unit, ESDU International, 27 CorshamStreet, London N16UA.

ENR, 2006. Engineering News Record.Harris, R.I., Deaves, D.M., 1981. The structure of strong winds, paper no. 4. In:

Proceedings of the CIRIA Conference, London, 1213 November, 1980,Construction Industry Research and Information Association, 6 Storeys Gate,London SW1P 3AU.

Irwin, P.A., Kochanski, W.W., 1995. Measurement of structural wind loads using thehigh frequency pressure integration method. In: Proceedings of ASCEStructures Congress, Boston, USA.

Irwin, P.A., Breukelman, B., Williams, C.J., Hunter, M.A., 1998. Shaping andOrienting Tall Buildings for Wind. ASCE Structures Congress, San Francisco.

Irwin, P.A., Breukelman, B., 2001. Recent applications of damping systems for windresponse. In: Proceedings of the Council on Tall Buildings and Urban Habitat,World Congress, Melbourne, Australia.

Irwin, P.A., Baker, W.F., 2005. The wind engineering of the Burj Dubai Tower. In:Proceedings of the Council on Tall Buildings and Urban Habitat Seventh WorldCongress, Renewing the Urban Landscape, New York, October 1619.

Irwin, P.A., 2005. Developing wind engineering techniques to optimize design andreduce risk. In: Biennial Scruton Lecture. UK Wind Engineering Society, TheInstitution of Civil Engineers, London UK, November 2. Written version isavailable at the UK Wind Engineering website at /www.ukwes.bham.ac.ukS.

Qiu, X., Lepage, M., Sifton, V., Tang, V., Irwin, P., 2005. Extreme wind proles inPersian Gulf region. In: Proceedings of the Sixth Asia-Pacic Conference onWind Engineering, Seoul, Korea.

Scruton, C., 1963. On the wind-excited oscillations of stacks, towers and masts. In:Proceedings of the First International Conference on Wind Effects on Buildingsand Structures, Teddington, Middlesex, UK, June 2628, pp. 798819.

Tschanz, T., 1982. The base balance technique and applications to dynamic windloading of structures. Ph.D. Thesis, University of Western Ontario, London,Ontario, Canada.

Williams, C.J., Wu, H., Waechter, W.F., Baker, H.A., 1999. Experiences with remedialsolutions to control pedestrian wind problems. In: Proceedings of the10th International Conference on Wind Engineering. Copenhagen, Balkema,2124 June, pp. 813818.

Xie, J., Irwin, P.A., 1998. Application of the force balance technique to a buildingcomplex. Journal of Wind Engineering and Industrial Aerodynamics 77&78,579590.

Xie, J., To, A., 2005. Design-oriented wind engineering studies for CCTV newbuilding. In: Proceedings of the Sixth Asia-Pacic Conference on WindEngineering, Seoul, Korea.

Xie, J., Haskett, T., Kala, S., Irwin, P., 2007. Review of rigid building model studiesand their further improvements. In: Proceedings of the 12th InternationalConference on Wind Engineering, Cairns, Australia.

Zdravkovich, M.M., 1982. Scruton number: a proposal. Journal of Wind Engineeringand Industrial Aerodynamics 10, 263265.challenges for wind engineers. This paper has concentrated

Wind engineering challenges of the new generation of super-tall buildingsIntroductionWind statistics and wind profilesOptimization of shapeStiffness, mass and dampingWind tunnel testingIts use as design toolHigh frequency force balance methodHigh frequency pressure integration methodAeroelastic modelsWinds at ground and higher levels

Concluding remarksReferences