wind and urban climate

8/6/2019 Wind and Urban Climate

1/13

J7.8 WIND AND URBAN CLIMATES

E. J. Plate, em. Prof. Dr.-Ing.*University of Karlsruhe, Germany (Formerly: Director of the Institute for Hydrology and Water Re-

sources Planning, University of Karlsruhe, Germany)

H. Kiefer, Dr.-Ing.Wacker Ingenieure, Birkenfeld/Wrttemberg, Germany (Formerly: Wind Engineering Division, Institute

for Hydrology and Water Resources Planning, University of Karlsruhe, Germany)

J. Wacker, Dipl.-Ing.Wacker Ingenieure, Birkenfeld/Wrttemberg, Germany (Formerly: Wind Engineering Division, Institute

for Hydrology and Water Resources Planning, University of Karlsruhe, Germany)

ABSTRACT

A brief overview is given of wind problems inurban areas, on which the wind engineeringteam (WET) of the Institute of the senior authorhas been working for the last 25 years. After ashort historical introduction we focus on investi-gations of wind forces on structures embeddedinto urban canopy layers and give a short surveyof diffusion work done in atmospheric wind tun-nels of the University of Karlsruhe. Wind tunnelstudies were conducted of forces and diffusionfor building arrangements in an urban environ-ment. We concluded that wind tunnel results onbuildings in different configurations of urbanbuilding complexes could be used to infer possi-ble modifications of wind force codes for build-ings in urban areas. Differences between freestanding buildings, used for obtaining load coef-

ficients for building standards, with the sametype of building in an urban environment, wereused to yield necessary corrections, if any, formodifications of design codes. We reached theconclusion that first order reliability methodsneed to be used for design peak pressures. Fur-thermore, we were able to systematize diffusionproblems for urban areas by defining 4 differentregions of diffusion fields, for which differentmodelling criteria apply, both for wind tunnel andnumerical modelling.

1. INTRODUCTION (ERICH J. PLATE)

As the American Meteorological Society haskindly decided to host a session in honor of my75

thanniversary, I may be permitted to start this

paper with some personal historical notes.

*Corresponding author address: em. Prof. Dr.-Ing. E. J. Plate Universitt Karlsruhe, Institut frWasserwirtschaft und Kulturtechnik AbteilungHydrologie, Kaiserstr. 12, D-76128 Karlsruhe, E-MaiL: [email protected]

In 1959 I joined the staff of Colorado StateUniversity to assist Prof. J.E.Cermak in his pro-gram on wind tunnel modeling of atmosphericboundary layers, and this topic has been with methroughout my professional career of more than40 years. It started with crude estimations ofatmospheric variables, as design information fora wind-tunnel for modelling the atmosphere. Thelogarithmic wind profile as a modelling criterionhad just been discovered, and J.E.Cermak haddecided that the best way of modelling a naturalboundary layer was to build a wind tunnel with avery long test section. In order to capture ther-modynamic effects, the tunnel was to have hu-midity and moisture controlled air flow, and asection of the floor should be heated or cooled toenable the creation of thermal boundary layers(Plate & Cermak, 1963). The project was sup-ported by the US Army, by means of a (in retro-

spect very limited) financial grant, and the per-mission to use huge amounts of Army surplus,delivered free of charge, of which we made ex-tensive use. It was my task to build this (and anumber of other) wind tunnels, and later on touse this facility for studies on atmosphericboundary layers. An outstanding first study usingthe capacity of the wind tunnel led to the Disser-tation of Prof. Arya (Arya & Plate, 1969) onboundary layers along a smooth, heated plate, atsmall values of z/L, where z is the vertical coor-dinate and L is the Monin - Obukhov length. Anumber of studies of a more fundamental natureon internal boundary layers, boundary layers

over water waves, effect of fences and other twodimensional building shapes on boundary layerson typical building elements followed.

In 1968, then at the Argonne National Labo-ratory, I summarized the state of knowledge onatmospheric boundary layers in a monograph(Plate, 1971) which became the guide for all thework that was done later on in my Institute at theUniversity of Karlsruhe. I created a program on


2/13

atmospheric research in wind-tunnels, with theexpress purpose of generating design informa-tion for practical applications. Much of my ownwork was concerned with proving the applicationof wind tunnel testing to a large variety of prob-lems, exploring its limitations, and extendingwind tunnel testing to different categories of

problems. The result were summarized in Plate(1982). Our work naturally involved a lot of rou-tine studies (not always really routine!) of alltypes of practical problems to which wind-tunnels are applied. I was ably supported by asuccession of excellent junior staff members - J.Loeser, Dr. A. Lohmeyer, Dr. W.Bchlin, M.Rauand Dr. P.Kastner -Klein as group leaders. Fromthese studies, a catalog of problems was derivedthat needed theoretical underpinnings leading toan extensive research program with many dis-sertations. Initially, most of the work was ondiffusion, and diffusion in urban complexes wasthe first type of work on urban environments.

Dissatisfaction with wind design information instandards, where pressure and load coefficientswere usually taken from wind tunnel studies inair flows without appropriate boundary layersalong the ground surface, led us to consider alsowind loads on buildings in boundary layers andlater to buildings embedded in the urban canopylayer.

The methods developed at the authors insti-tute were further extended and applied to nu-merous practical problems by private companiesfounded by former members of the institute,

among them Ingenieurbro Lohmeyer, specializ-

ing on environmental investigations, and WackerIngenieure, specializing on wind loads on struc-tures, who built their own multi functional windtunnel with a large cross section .

A summary of the results of these studieswill be given in the remainder of the paper. Astate of the art survey of wind tunnel explora-tions of urban environments was also presentedin Plate & Kiefer, (2001).

2. THE URBAN BOUNDARY LAYER.

a. The approach flow The structure of theurban atmospheric boundary layer is understoodas a multi-layer air flow, as schematically indi-cated in Fig.1 (see also Oke,1987). For neutralstratification conditions the open country sur-rounding the city yields a uniformly adjustedconstant stress boundary layer (an equilibriumboundary layer), in which the wind velocity pro-file can be expressed, for modest to high (4 m/sand up) wind speed conditions, by a power law:

( )u zuref

z

d=

(1)

with being the exponent of the powerlaw, which reflects the roughness conditions ofthe surface upstream of the city. This exponentusually is of the order of 0.15 to 0.3. Wind tunnelmodelling must take cognizance of this fact, andwind tunnel modelling of atmospheric flows startwith showing that within the test section of thewind tunnel the power law Eq.1 also holds andhas an exponent of 0.15 - 0.33, as is shown, forexample in Fig.2.

When the air flow reaches the city, the windprofile has to adjust to its roughness. An internalboundary layer forms above the city, which dis-places the outer boundary layer of the approachwind profile. Between internal and outer bound-ary layer a transition region is formed, but if thefetch of constant city roughness extends farenough, - for a 100 m thick boundary layer ap-proximately 1000 m - the outer layer and theinternal layer merge into the new, locally ad-justed equilibrium boundary layer correspondingto the aerodynamic properties of the city.

b.The layered flow above the city. The fullydeveloped equilibrium boundary layer above acity consists of four layers. The lowest is theurban canopy layer, in which houses, streets andother urban features are embedded. The flow inthis region is governed mainly by the form dragon individual buildings, i.e. by pressure differ-ences between front and rear of structures. Its

height is approximately equal to the averagebuilding height (but see below). For the canopylayer, no reasonable models of the velocity fieldcan be given, because the local arrangements ofstreets and trees, topographic features, high-ways and rivers lead to a complicated three di-mensional flow field.

At some distance above the canopy layer, theflow field is two dimensional and a constantstress layer exists, in which momentum istransmitted to the ground by horizontal shearstresses, and in which the velocity distribution isdescribed by the famous logarithmic law:

u

u

1ln

z d

zo=

(2)

where is v.Karmans constant, usuallytaken to be 0.4, zo is the roughness height, and

d the displacement height, and u = ,where is the horizontal shear stress, and isthe density.


3/13

do

logarithmicprofile

blendingregion

internal

boundarylayer

initialprofile

boundarylayer

outer

Fig.1: Schematic development of an urban boundary layer.

The major difference between canopy layer andlogarithmic layer is that the horizontal force onthe flow in the canopy is composed of dragforces on the individual roughness elements -i.e. by pressure forces, resulting in a highly

three-dimensional flow velocity distribution -whereas in the logarithmic region, the verticalmomentum flux is carried by a constant shear only, as is indicated by parallel streamlines ofthe mean velocity field. Consequently, betweencanopy layer and logarithmic layer there exists ablending layer in which the highly three-dimensional canopy flow pattern is convertedinto a basically two-dimensional flow field withconstant (atmospheric) pressure, which formsthe logarithmic layer. Experience has shown thatthe thickness of the blending region can be ex-pressed roughly as a fraction of the distance Dbetween elements, with an added effect coming

from the shape of the roughness elements. It isnot unreasonable to expect that the roughnessof the surface carries its effect through theblending region, and that coefficients d and zo of

the logarithmic law are quantities which are de-termined by the building configurations.

The literature reports many different expres-sions for a relationship between the parametersof Eq.2, andcharacteristics of the canopy layer. The earliestones (see Plate, 1971) assumed a simple rela-tionship between d and z0 and average rough-

ness height H, such as:

z0= 0.15 H, and d = 0.85 H (3)

However, this equation is too simple, and isbased on experiences with low growing crops,and a few early experiments. Much more de-tailed analyses have shown, that zo depends

also on the arrangements of the roughness ele-ments. Its value must be considered a measureof the drag exerted on the roughness arrange-

ment. Empirically, it seems reasonable to expectthat the more roughness elements or buildingsthere are, the higher will be the drag. Two pa-rameters are found (Theurer et al., 1992) to beimportant for classifying roughness in terms of

building arrangements. These are the coefficient ar and fa , defined as:

ar = sum of all areas covered by buildings/total urban area

fa = sum of average building areas normalto the wind/ total urban area.

Eq.3 is valid for fairly dense vegetative cov-ers, where roughness elements cover about60% of the area. Kondo and Yamazawa (1986)

found that for Japanese cities zo and ar arelinearly related. For different lower fractions ar

they show that:

z Ho ar= 0 25. (4)

fits observed data quite well for 0.4m


4/13

Fig.2: Wind profiles in a wind tunnel: mean velocity distributions for different exponents of Eq.1 caused by different roughness (left) and turbulent intensity (right).

By varying the distance between the rows,

one finds two minimum values of u*. The firstone (isolated roughness flow IRF) occurs whenthe distance between rows is large, i.e. building

density ar is small, so that the contribution ofthe few individual buildings is a small fractiononly of the total shear force on the surface area.The other one occurs when the building density

ar approaches 1, i.e. the distance between therows is very small. All roughness effects disap-pear and an elevated flat and smooth surface isgenerated.

Arrangements with building densities in be-

tween these extremes will have larger values ofu*. With increase in building density , the individ-ual buildings interfere with one another, and thecase of wake interference begins. Theurer(Theurer et al., 1992) found a linear relationshipof parameters d and zo with form parameters

ar and fa respectively, for small values of theorder of 0.1 -0.25. For larger building densities orfrontal area ratios the data show a tendency todecrease with large values, although the scatterbecomes large. The relationship between build-ing geometry and roughness parameters be-comes less well defined, and at this time nosystematic relationships has been found relating

parameters zo and d to parameters derived from

geometric configurations of urban complexes.Therefore, the best way of obtaining operationalvalues of zo and d for urban areas is obtained

from experiments on built up areas, or on theirmodels in a wind tunnel. If this information is notavailable, we prefer to use the empirical collec-tion of data from Theurer (see Theurer, et al.,1992) and Badde (Badde & Plate, 1994) who

summarized what is known about these parame-

ters in Table 1.

3. WIND FORCES ON STRUCTURES INURBAN AREAS.

As design quantities of forces in urban areasare caused by extreme winds, it can be assumedwithout loss of generality that these forces aremodelled in the wind tunnel. Disputable is thequestion of the Reynolds number necessary forgetting beyond a critical value Recrit. At Reynoldsnumbers higher than Recrit the flow field and itsturbulence become independent of Reynoldsnumber, except possibly for buildings with

shapes for which local Reynolds numbers aresignificant. In order to prove without doubt thatmodelling is indeed accomplishable, numerousstudies with parallel measurements on full sizebuildings and their wind tunnel models havebeen made by the authors team. Blohm (Blohm& Plate, 1975) and Schnabel (1981) studiedwind forces on circular light house towers underideal conditions. These towers have been builtinto the ocean many kilometers offshore. Thewind profile developed over the ocean, thereforehad practically infinite fetch, in particular whenthe wind velocity was so low that only small wa-ter surface waves were generated. These stud-

ies were important also as Schnabels resultsyielded excellent vertical and lateral coherenciesof the pressures on the towers. Maier - Erba-cher (Maier-Erbacher & Plate, 1991) extendedthe study to towers in varying topographies,finding to our surprise that even a fairly steepembankment behind the tower hardly affectedthe forces, and not at all the drag coefficient ofthe tower.


5/13

Configuration Characteristics Roof Shape z0 H H H L B L H/ ar fa

1New district,one family buildings1 - 2 storeys

Mainly gableroofs, rarelyflat roofs

0.1 -0.3

(1.3)

8 - 10 ~ 0 ~ 1 ~ 1.5 0.1 -0.2

~ 0.1

2

Residential area1 - 3 storeys

Mainly gableroofs, rarely

flat roofs

0.1 -0.3

(1.4)

8 - 12 < 0.2 ~ 1 ~ 1.5 -2.5

0.15 -0.25

~ 0.1

3

Residential blocksregularly aligned3 - 5 storeys


~ 0.3(1.5)

12 - 20 < 0.2 < 0.5 ~ 1 - 2 0.1 -0.25

0.1 -0.25

4Residential areahigh-rise buildingsand residential blocks4 - 15 storeys

Gable roofs,flat roofs

> 0.5 > 15 0 - 0.5 < 0.5 ~ 0.7 -1.5

0.1 -0.2

0.15 -0.3

5Cultural facilitieschurches, schools,etc.in residential areas

Gable roofs,flat roofs

0.3 -1.5

(2.4)

> 8 > 0.5 0.5 -2.0

~ 2 - 5 0.1 -0.3

0.05 -0.15

6Block of buildingsin City Centers3 - 6 storeys


~ 0.7(2.1)

15 - 25 < 0.3 ~ 1 ~ 0.7 -0.9

0.3 -0.7

-

7

City Center

areas including parks,,high-rise buildings andpublic facilities

Gable roofs,

flat roofs

0.3 -

0.7(>2)

>15 < 0.4 ~ 1 ~ 1.5 -

2

< 0.5 0.1 -

0.2

8Commercial andindustrial area2 - 5 storeys

Mainly flatroofs orgable roofs

~ 0.3(0.6)

5 - 15 < 0.5 < 1 ~ 2 - 5 0.3 -0.4

0.05 -0.2

9Industrial plantwith tanks

Mainly flatroof

~ 0.5(1.6)

10 - 25 < 0.5 ~ 1 ~ 0.5 -1.5

0.1 -0.4

0.1 -0.2

10Industrial area1 - 4 storeys

Mainly flatroofs, rarelygable roofs

0.3 -0.5

(1.6)

5 - 15 0.3 -0.5

~ 1 ~ 2 - 7 0.2 -0.4

0.05 -0.2

Table 1: Typical building configurations and their geometric and aerodynamic parameters

found in German cities (Badde & Plate, 1995). H is the mean building height,

H is the standard deviation of the building heights, B,L are mean lengthand mean width, resp. of buildings

Bchlin (1987) investigated the forces on atennis hall with an elliptic roof, and verified thatnot only the pressure distribution on the roof, butalso the internal pressure in the building wereproperly modelled in the wind tunnel. We thenturned to look at forces on structures in urbancomplexes. Initial studies were concerned withdifferent building shapes in smooth and uni-formly rough boundary layers.

The influence of different wind profiles and

approach flow conditions from Fig.2 on rectangu-lar shaped test buildings of different side ratiosand heights according to Table 2 was investi-gated by Wacker et al. (1990a, 1991, 1992,1993). Model tests on a family of test buildingsprovided an extensive data base for wind in-duced loads as function of approach flow condi-tions and building geometry. Furthermore, thedata sets enabled checks of the general designrules given in codes. The commonly accepted

quasi-steady approach which relates the pres-sure fluctuations to the wind velocity fluctuationsby use of gust factors ( )z(k21)z(G u+= , with

=u rms value of u(z), and k = 3.5) is compared

with results of model scale tests in Fig. 3. Theratio of the peak pressure coefficients to the timeaveraged mean values of the pressure time se-ries is plotted against the turbulence intensity

)z(u/)z()z(T uu = for the stagnation point at the

front wall (Fig. 3 A) and the roof centreline (Fig.

3 B).

The wind tunnel data showed that peak pres-sures averaged over a 0.5 sec time period in fullscale (this corresponds to small tributary areasof approximately 1 m in size, typical for smallfaade elements or roof pavers) may be under-estimated by the classical approach for elementscorresponding to this size. Wacker (1994) alsocompared standard assessment methods for


6/13

local peak wind pressures with a reliability-basedassessment method, which combined secondmoment reliability (SMR) principles with windpressure data from wind tunnel study. Peakpressure coefficients, based on the simplifiedCook-Mayne method were calculated on a con-sistent safety level (Wacker, 1994).

Furthermore, the data provided useful resultsconcerning the spatial correlation of the windinduced pressure field at the building surfaceand leads to the definition of tributary areas andcorresponding wind load coefficients. Also, theanalysis in the frequency domain gives evidenceabout vortex shedding phenomena due to differ-ent approach flow conditions and building di-mensions (Wacker, 1995)). A study by Wacker &Plate (1990a) used wind tunnel data to inferfatigue loads for cladding elements - with theresult that buffeting of cladding elements did notlead to any significant dynamic loads on the

surfaces, -fatigue loads to be expected at mostfor the fastening elements. The study, whichused experimental data in conjunction with the

classical fatigue model of Miles, and of Minerscumulative damage function, proved to be anexcellent direct application of wind tunnel resultsin a fatigue analysis, but it also yielded a niceapplication example for stochastic design, for abook on Statistics and Applied Probability The-ory for Civil Engineers (Plate, 1993).

We then addressed the problem of windforces on structures in urban areas (Kiefer andPlate (1998, 1999)). Until now design codes arebased on wind-tunnel experiments on free-standing building models. However, the free-standing building is a very rare case in reality. Indensely populated areas buildings are groupedtogether and arranged in characteristic configu-rations due to their function or location in theurban area. In some design codes the effect ofthe surrounding is taken into consideration byadjusting the wind profile, and using a modifiedreference velocity in combination with pressure

coefficients of the isolated building case.

Table 2: Model types and their dimension used by Wacker (Wacker , 1995)

0 0.05 0.1 0.15 0.2 0.25 0.3

Tu(z) [-]

1

1.5

2

2.5

3

3.5

4

cp,p

eak

/cp,m

ean

[-

]

0 0.05 0.1 0.15 0.2 0.25 0.3

Tu(z) [-]

1

1.5

2

2.5

3

3.5

4

cp,p

eak

/cp,m

ean

[-

]

1+2*5*Tu(z)

1+2*3.5*Tu(z)

1+2*3.5*Tu(z)

A)

B)

15/15

5/15

15/5

10/15

L/B [cm/cm]

Fig. 3 Local pressure gust factors and comparison with those resulting from linear quasi-steady the-

oryA) front wall; B) roof; (Wacker & Plate, 1993)


7/13

The main reason for not explicitely account-ing for the effect of the surroundings is the lackof data. The Study of Hussain & Lee (1980) pro-vided some data, but their regularly arrangedroughness configurations are not representativefor most of the building arrangements found in

reality. Ho et al. (1991) investigated the effect ofa typical North American industrial area on windloads. Their study was limited to low-rise struc-tures. Hertig & Alexandrou (1995) also pre-sented some first results of measurements intwo models of characteristic urban areas in Swit-zerland. The influence of neighbouring buildingson the wind induced loads is evident in the re-sults of these studies, but the results are notsufficient to infer general design rules for windloads on structures in built-up areas. To improvethe data base extensive investigations werecarried out at the IHW, including classification ofrepresentative building layouts, wind-tunnel ex-

periments, and full-scale measurements. Thebuilding classes of Theurer were collected intothree basic types of building arrangements whichwere built to scale for experiments for wind loadson buildings in built-up areas (see Fig. 4):

Type I: Homogeneous row pattern - constantbuilding dimensions and distances between

the buildings

Type II: Variable building dimensions anddistances just like in industrial areas

Type III: Buildings are arranged in blocksjust like in city centers

Type I consisted of buildings with dimensionsidentical to the lowest rectangular shaped testbuilding and the distance s between the buildingrows was varied according to the three flow re-gimes of Fig.1. The mean building height of theother types was nearly the same as the constantbuilding height of type I (H = 16m in full scale).The type II building arrangement is representa-tive for a large number of existing built-up areas.The pressure measurements on the test build-ings inside the industrial area (Type II) weremade at different locations in order to determinethe influence of the immediate surroundings. Thewind directions were varied for all investigated

configurations. Flat-roofed test buildings of dif-ferent heights (H = 16m, 32m, and 64m in fullscale) and aspect ratios (W = 16m, L/W = 1 andL/W = 3.3) were used for systematic measure-ments. Overall forces and local peak pressurecoefficient were determined for the test build-ings.

The study yielded a large data base for windloads on buildings of the same height or lower orexceeding the mean height of the surroundings.

Among the results was the finding that the im-mediate surroundings or the near field around anindividual building strongly reduces wind inducedforces for buildings deeply embedded into thecity canopy, whereas for taller buildings theforces increase, reaching force coefficients

nearer to those on free standing buildings. Be-cause near field conditions will not remain con-stant during the lifetime of a building due tomodifications of neighboring structures, andbecause large variability of loads caused bydifferent near field conditions we determineddesign loads by means of second moment reli-ability analysis (SMR) to obtain design windloads on a defined safety level (Kiefer, 2003).

SMR-based local peak wind load coefficientson the same level of safety for freestandingbuildings and buildings embedded in built-upareas are compared in Fig. 5. The reduction of

local wind load coefficients found for buildings ashigh or lower as the surroundings correspondsapproximately to the reduction of design windspeed and the resulting design dynamic windpressure given in the wind profile model of theEurocode (1994) for urban areas. This meansthe design procedure of the Eurocode leads to asufficient level of safety and an economic de-sign. However, the Eurocode approach for thewind model in built-up areas is not fully includedin the new formulation of the German wind loadcode.

In addition to the systematic wind tunnelstudy measurements on two test buildings on theUniversity Campus embedded in complex sur-roundings were carried out in full-scale, paral-leled by wind tunnel tests on the scaled modelsof these test buildings embedded in their sur-roundings were conducted (Kiefer.& Plate(1998b)). The comparison of the results in fulland model scale shows good agreement (seeFig. 6), if the loads are determined as function oftributary area and are based on statistically de-fined peak load coefficients (Cook-Mayne-method). The results in the frequency domain(Fig. 7) show that the fluctuating wind inducedpressure field was properly modelled in the windtunnel and represents the characteristic features

in time and space.


8/13

Type I Type II Type III

- Row pattern - Industrial area - Town blocks

5

11.5

5

13 2

8

H616

6,0

G611F601

15,8

F602

6

G606

10,8

G604A 8

G604B4,2 1,7

G619

6,6H608 6

H617

1,8

H611

11,4

H620a 5,8

H620b7,5

H621

2,5

H622 3

H636

3,3

H633

5,5

H630

2,7

H640

4,5

H612

G7207,6

10,0G722

10,4

G700

6,0F604

F605

4,5

F6085

F710C

F710B

F710A

9,5

10,310,3

3,3

F702

10,7

F712

H5194

H5305,9

H540b

2,4

H540a4

J513A

J513B

32,7

7,0

6,0

5,4

6,0

J517

J5

16

1,01,0

J514J508J507

J506

2

2

2

2,3

J502 2,6

0,3

J531

2,1

J522

4,0

J500A

4,8

4,8

4,8

J535

4,8

J536

5,2

J536AJ536B

3,4

5,0

5,5

G615

9

6,4

G612

13

5

H600

9,5

E7003,2

E703

E701

E605

C621

2,8

C611

D611A

4,8

D618

2,2 D619

5,0

D731 9,2

D732

D722

2,4

D700

7,8

9,7

D610

1,8

C630

6,0

F517

11,5

F520

G504

5,6

G529

G50813,8

G525

7,8

1

H506a6,0

H510

H517

3,8

1,6

H536

5,3

H5226,5

H5262

H537

G511

G514

G519

G515

2,4

4,0

G505

H5149,5

H501

4,1

H513

1,5

H534

4

H533

6,3

H445

3,0

H535

2,5

H538

H443

2

2,5H528

2,7

H527

2,5

H524

1,6

H520

2,6

2,2

3.2

6,6

F605

E605A

D618B

F702A

F702B

F605A

G517A

H501A

H513B

H513C

H513A

H521

H616A

G715A

H601B

H420

D610A

9

E503A8

J5421.4

J541J540

1.3

1.2

J512

J515

1

1

G605

13,0

spacing sPosition H

Position B

Position G

Fig.4 Different fundamental types of built-up areas used for the systematic wind tunnel investiga-

tions (Kiefer, 2003)

-5 -4 -3 -2 -1 0

cp,smr- isolated

-5

-4

-3

-2

-1

0

cp,smr

-built-up

-5 -4 -3 -2 -1 0

cp,smr- isolated

-5 -4 -3 -2 -1 0

cp,smr- isolated

-5

-4

-3

-2

-1

0

cp,smr

-built-up

-5

-4

-3

-2

-1

0

cp,smr

-built-up

H/H_

built-up area= 1 H/H

_

built-up area= 2 H/H

_

built-up area= 4

Side wall

Roof

Regressionf(x) = 0.8x

Fig.5 Comparison of local peak pressure coefficients based on SMR-principles for test buildings ofdifferent heights (Kiefer, 2003)


9/13

0.1 1 10 100

A[m2]

0

1

2

3

4

5

6

7

|cp,min|[]

Full scale, mean value and scatter

Modell including vegetation

Modell without vegetation (Kiefer, 2003)

Regression model scale data

Regression full scale data

Fig.6 Comparison of local peak pressure coefficients for different tributary areas at the roof corner

in full and model scale

0.001 0.01 0.1 1 10 100f *B/U

H

0.001

0.01

0.1

1

Scpcp

*f/

cp

Full scale

Model scale

0.001 0.01 0.1 1 10 100f*B/UH [-]

0

0.2

0.4

0.6

0.8

1

Coherence

[-

]

Full scaleModel scale

Fig.7 Comparison of spectra and coherence of the pressure fluctuations at the roof corner taps

in full and model scale

4. DIFFUSION IN URBAN AREAS

Pollutant loads in urban areas depend onmany factors, climate among them. Obviously,for zero wind without convection either self in-duced turbulence from the sources themselves,or from vehicles moving through city canyons ofurban areas will be the only agents for mixing ofpollutants. Critical pollution cases occur whenthe wind velocity is practically zero, and when aninversion inhibits the spreading of pollutants

beyond some minimum height. In these cases,the profiles of temperature and heat flow are asshown in Fig.8

In convective situations thermal convectioncontributes to the mixing, However, even atsmall wind velocities, the urban complex takes ahigh degree of mixing energy from the wind flowand combines with thermal mixing to produce afield of practically constant temperature, and

also of strong mixing within the urban streets. Itis not yet fully determined at which conditionsshear induced turbulence starts to overcomeconvective turbulence. We suspect that beyondsome critical conditions convective turbulence isovercome by mechanical turbulence. This condi-tions is likely to occur at some critical value ofu*/w*, where w* is the Deardorf velocity:

3/1)hQ(w = (5)

where T/1= is the buoyancy parameter for

mean temperature T, wTQ = is the turbulentheat flux, and h is the height of the mixing layer.In order to find this critical number (likely of theorder of 0.35), and more generally , to studyconvective flows by means of experiments. Webuilt a wind tunnel having a test section with aheatable floor, and a return duct consisting of tenducts on top of one another, each of which canbe heated or velocity controlled individually (Rau


10/13

et al., 1991). One of its primary purposes is tomodel convective flows in urban environments:is convection really completely suppressed bywind shear, or else, how does convection incombination with wind shear affect turbulenttransport? Unfortunately, for a variety of reasonsthese experiments have not yet been done.

Our research on diffusion in urban areas inneutrally stratified boundary layers has beenconcerned with two categories. The primaryconcern was with safety of people living inneighborhoods of chemical factories, in whichtoxic substances are stored or handled. Acciden-tal releases are of foremost concern in suchsituations, and warning systems in case of anaccident are of the utmost importance, in particu-lar in Germany with its large concentrations ofchemical factories in the Rhine - Main area andin the Ruhr district. The second category con-cerns traffic pollution. The growing traffic density

has compensated technical reductions of ex-haust contaminants. In almost all cities, extremetraffic pollution can occur. We envisioned thegeneration of traffic control systems in whichtraffic is re-routed whenever the toxicity level in astreet reaches levels critical to human health.

As a framework for studying these problems,it is useful to distinguish four regions of applica-tions of different modelling techniques, which areshown schematically in Fig.9:

Region 1:source dominated region. This isthe initial region, extending over a few tens ofmeters close to the source, involving the buildingunder consideration or its direct surroundings.Modeling in this and in the near field region isthe traditional scope of wind tunnel studies, as itis unlikely that numerical models can well coverthe complex flow field induced by the local ar-rangement of buildings of different heights,

sizes, and orientation. Numerous studies of thiskind have been made by the author and histeam, as have been in many other laboratories inthe world.

It can be argued that for such models therealways exists a need to model a large part of the

upstream surroundings, as the initial wind profileover the city very much influences exchangesbetween urban canopies and the blending re-gion. We contend that the upstream region hasto be modelled far enough to reestablish a fullydeveloped boundary layer to replace the internalboundary layer forming over the edge of buildingcomplexes (Plate, 1971, Garratt, 1990), asschematically shown in Fig.1. Of special interestwere effects of secondary turbulence created bytraffic. We developed a criterion for modellingthe turbulence introduced by moving traffic, andgenerated a method for moving traffic modelling,which was successfully used both for modelling

traffic situations, for example in the neighbor-hood of tunnel in- and outlets, (Bchlin, 1994)and for basic studies of wind and traffic combina-tions on diffusion in city canyons (Kastner - Klein& Plate, 1999)

Region 2: Near field region . A near field re-gion, extending up to a few hundred meters, inwhich the exact location of the source is notimportant, because the buildings downstream ofthe source strongly influence the wind field andcause strong mixing, which obscures the initialconditions. The outer edge of this region isformed by the "radius of homogenisation", whichobtains its name because beyond this distancethe concentration plume can be modelled as if itwere in a field of homogeneous turbulence. Thecircle with this radius denotes the border be-tween near and far field.

super - adiabatic layer

stably stratified layer

mixed layer top at time t

z

z (t ) temperature

T(t )

T(z) = T +azo

mixed layer ML

free convection layer

blending region

i

interfacial layerz

w

w (z)

o

heat flux

Ti

Fig. 8: Idealized temperature profiles in convective conditions


11/13

Region 3: Far field region . The far field re-gion extends up to 4 to 5 km downstream of thesource. It is mainly of interest for long range airpollution studies, where simple models of theGaussian plume type can be used to calculateconcentration fields from different urban sources.Concentration fields in the far field are character-

ized by a vertically constant concentration in citystreets, and by a Gaussian plume extendingfrom the ground to heights z. Consequently, thefar field concentration field resulting from a con-tinuous source located at point P(x0, y0, z0) canbe represented approximately by a concentrationfield c(x,y,z) emanating from a virtual continuoussource located at point 0),yy,xx(P 00 ++ at

ground level:

+

=

2z

2

2y

2

zy

m

2

)z(

2

)yy(exp

q)z,y,x(c

(6)

where qm is the mass of pollutant (with density

) released per unit of time. For this equation,diffusivities Dz and Dy are assumed constant.They yield spreading coefficients 2

y and2

z which are linear function of x:

)xx(u

D2and)xx(

u

D2z2

z

y2y +=+=

. (7)

where u is a (constant) reference velocity. Mass

conservation is also possible if the spreadingcoefficients are non-linear in x, as often used inempirical formulas (depending on Pasquill stabil-ity classes). Due to mixing in the near field, the

apparent source location is different from theactual source location by an offset of x parallelto the wind direction, and an offset y horizon-

tally perpendicular to the wind direction. Theapparent height of the virtual source is z = 0.

The initial conditions are determined by thecharacteristics of the urban area. The applicationof these equations requires the determination of5 parameters, coefficients 2Dz/ u and 2Dy/ u ,

and coefficients x and y and since Eq. 7 isvalid only for x > R, where R is the radius ofhomogenization, this quantity also becomes anunknown. Sets of parameters for different typesof urban complexes have been collected andreported by Theurer (Theurer et al., 1996, seealso Plate & Kiefer, 2001).

.6. CONCLUSION.

A long series of experiments, in our and otherinstitutes, interpreted in the light of existing andnewly developed theories have created a state

of the art on urban wind problems, whose firstresults were discussed in a NATO AdvancedStudy Institute in Karlsbad, Germany.(Cermak etal., 1995). The knowledge accumulated allows tosolve practical applications with great confi-dence. We have identified and studied the urbanboundary layer as the region in which buildingsand building arrangements interact with the at-mospheric wind field.

main wind direction

radius of homogenizationvirtual source

actual source

NEAR FIELD

zone of source domination

FAR FIELD

x = longitudinal

displacement

y = lateral

displacement

Fig.9: The zones of the diffusion field downwind of a point source (schematic map of an urbanarea)


12/13

A categorization is possible both through thecharacteristics of the urban canopy as it interactswith the external flow field, as well as with thescale of the problems to be considered. For windforces, the scale is determined by the buildingunder consideration and its surroundings, andfor diffusion we can extend the scale to three

regions: source area, near field, and far field.

We may increase the number of regions to beconsidered from three regions of diffusion cate-gories to a fourth one, the region of meso- scalemeteorology. Meso-scale meteorological proc-esses and Coriolis forces dominate, gaseousreleases from all sources, as calculated from farfield models, are superimposed and can only bedetermined by means of meso-scale atmos-pheric transport models. This region is beyondthe interest of local urban climate modelling.However, the meso scale meteorological fieldsets the dynamic boundary conditions for the

flow field which has to be considered. The statis-tical variability of wind field and climate is directlytransferred into design criteria for the wind loads,which obtain a large degree of uncertainty fromthe uncertainty of the atmospheric processes.Two conclusions can be drawn: the first is toconsider in more detail the stochastic nature ofdesign loads due to wind (Plate & Davenport,1995). The second is that the necessity exists ofclose cooperation of meteorologists and expertson building aerodynamic.

6. REFERENCES:

Arya, S.P.S., E.J.Plate, 1969: Modeling of thestably stratified atmospheric boundary layer.Journal of the Atmospheric Sciences Vol.26,1969, No.4, pp.656-665

Bchlin, W., 1987: Pressure inside buildings - Adominant parameter for calculating effectivenatural ventilation.In:Indoor Air '87. Proc. 4

th

Intern. Conf. on Indoor Air Quality and Cli-mate, Berlin, 17-21 Aug. 1987, Vol.3,pp.337-341

Bchlin, W., Theurer, W., Plate, E.J., 1992: Dis-persion of gases released near the ground inbuilt up areas: Experimental results com-pared to simple numerical modelling. In:Journal of Wind Engineering and Industrial

Aerodynamics Vol. 44, 1992, pp. 2721 - 2732Bchlin, W., P.Klein, 1994: Diffusion problems

near tunnel portals in built-up areas. TheScience of the Total Environment, vol.146/147, 1994, pp. 377 - 385

Badde, O., E.J.Plate, 1994: Einfluverschiedener Bebauungsmuster auf diewindinduzierte Gebudebelastung.Abschlukolloquium StrmungsmechanischeBemessungsgrundlagen fr Bauwerke",

lagen fr Bauwerke",Sonderforschungsbereich 210, Univ.Karlsruhe, 24./25. Nov. 1994Blohm, O., Plate, E. J., 1975: Wind, wave andwindpressure investigation at the lighthouse'Alte Weser' (for prototype and model). In:2nd National Conference on Wind Effects onBuildings and Structures,June 1975,Fort

Collins,Col.,USABottema, M. and P.G. Mestayer (1997) Urbanroughness mapping techniques and somefirst results. Proc. 2nd European - AfricanConference on Wind Engineering, Genova,pp.197

Cermak, J.E., A.G.Davenport, E.J.Plate, andD.X.Viegas, (eds.),1995: Wind Climate inCities. (u.a.): Kluwer Dordrecht 1995 (NATOASI Series E vol.277)

Garratt ,J.R. (1990) "The internal boundarylayer: a review" Boundary Layer Meteorology,Vol.50, pp.171-203

Hertig + Alexandrou, (1995): The influence of

surroundings on pressure distributionsaround buildings, F. Cermak et al. (eds.),Wind Climate in Cities, 293-317, Kluwer Aca-demic Publishers....

Ho, T.C.E., D.Surry, and A.G.Davenport, (1991):Variability of low building wind loads due tosurroundings. Journal of .Wind Engineeringand Industrial Aerodynamics, (Elsevier Sci-ence Publisher),Vol.38, pp.297-310

Hussain, M. & B.E.Lee (1980): A wind tunnelstudy of the mean pressure forces acting onlarge groups of low-rise buildings. Journal ofWind Engineering and Industrial Aerodynam-ics, (Elsevier Science Publisher), Vol.6,pp.207-225

Oke, T., 1987: Boundary layer climates, 2nd ed. ,Routledge, London

Kastner-Klein, P., E.J. Plate, 1999: Wind-tunnelstudy of concentration fields in street can-yons. Atmospheric Environment Vol 33,1999, No. 24 - 25, pp. 3973 - 3980

Kiefer, H., Plate, E.J., 1998a: Modelling of meanand fluctuating wind loads in built-up areas.Journal of Wind Engineering and IndustrialAerodynamics, Vol. 74-76, 1998, pp. 619 -629

Kiefer,H., E.J.Plate, 1998b: Wind loads on build-ings in built-up areas - results from full-scaleand wind-tunnel measurements. 2nd East

European Conference on Wind Engineering,7-11 Sept. 1998, Prague, Czech Rep., Proc.Vol. 1, Eds.: M. Pirner et al., pp. 263 - 269

Kiefer, H., E.J.Plate, 1998b: WinderzeugteBauwerkslasten in bebauten Gebieten.Baukonstruktionen unter Windeinwirkung, U.Peil,(ed): 5. Dreilndertagung D-A-C-H 97,WTG-Berichte Nr. 5, Aachen: Windtechnol.Ges. 1998, pp. 51 - 63


13/13

Kiefer, H., E.J.Plate, 1999: Mean and fluctuatingwind loads in built-up areas. Wind Engineer-ing into the 21st Century, : A. Larsen et al.(eds.), Balkema 1999, Vol. 3, pp. 1813 1818

Kiefer, H., 2003: Windlasten auf quaderfrmigeGebude in bebauten Gebieteten,

Dissertation, Institute for Hydrology andWater Resources Planning, University ofKarlsruhe, Germany

Kondo,J. H.Yamazawa, 1986: Aerodynamicroughness over an inhomogeneous groundsurface, Boundary Layer Meterorology, Vol.35, pp. 331 - 348

Maier-Erbacher, J., E.J.Plate, 1991: Measure-ment of velocity near and pressure on a cy-lindrical tower located in irregular terrain.Journal of Wind Engineering and IndustrialAerodynamics, Vol. 38, 1991, pp. 167 - 184

Plate, E.J., 1971: Aerodynamic Characteristicsof Atmospheric Boundary Layers. Oak Ridge,

Tenn. : USAEC 1971.Plate, E.J., 1982: Wind tunnel modelling of windeffects in engineering. In: Engineering Mete-orology, ed.:E.J.Plate, Amsterdam:Elsevier1982, Studies in Wind Engineering and In-dustrial Aerodynamics 1,pp.573-639

Plate, E.J., 1993: Statistik und angewandteWahrscheinlichkeitslehre fr Bauingenieure.Berlin: Ernst & Sohn, 1993

Plate, E.J., J.E.Cermak, 1963: Micrometeo-rological Wind Tunnel Facilities: Descriptionand Characteristics. Final Report, ContractNo. DA-36-039-SC-80371 with U.S. ArmyElectronic Research and Development Activ-ity, Fort Huachuca, Arizona, February 1963,CER63EJP-JEC9

Plate, E.J., Davenport, A.G., 1995: The risk ofwind effects in cities. J.E.Cermak, A.G. Dav-enport, E.J.Plate, D.X.Viegas (eds). WindClimate in Cities, Kluwer Dordrecht 1995(NATO ASI Series E Vol.277) pp. 1 - 20

Plate, E.J., H. Kiefer, 2001: Wind loads in urbanareas. Journal of Wind Engineering and In-dustrial Aerodynamics Vol. 89, pp. 1233-1256

Rau, M., W. Bchlin, E.J. Plate, 1991: Detaileddesign features of a new wind tunnel forstudying the effects of thermal stratification.Atmospheric Environment, Vol. 25A, 1991,No. 7, pp. 1257-1262

Schnabel, W., 1981: Field and wind-tunnelmeasurements of wind pressures acting on atower. Journal of Wind Engineering and In-dustrial Aerodynamics,Vol.8, 1981 pp.73 - 91

Theurer, W., W.Bchlin, E.J.Plate, 1992: Modelstudy of the development of boundary layersabove urban areas. Journal of Wind Engi-neering and Industrial Aerodynamics Vol. 41,1992, pp. 437 - 448

Theurer, W., E.J.Plate, K.Hoeschele, 1996:Semi-empirical models as a combination ofwind tunnel and numerical dispersion model-ling. Atmospheric Environment Vol. 30, 1996,No. 21, pp. 3583 - 3597

Wacker, J., 1994: Towards reliability-based localdesign wind pressures for simple rectangular-

shaped buildings. Journal of Wind Engineer-ing and Industrial Aerodynamics Vol. 53,No.1-2, 1994, pp. 157 - 175

Wacker, J., R.Friedrich, E.J.Plate, U.Bergdolt,1991: Fluctuating wind loads on claddingelements and roof pavers. Journal of WindEngineering and Industrial Aerodynamics,Vol.38,1991, pp.405-418

Wacker, J., E. J.Plate, 1990a: Fatigue caused bywind pressure buffeting on surface elementsof high rise buildings. Structural Dynamics,Proc. of the European Conf. on StructuralDynamics, Eurodyn '90,5-7 June1990,Bochum,Germany, Eds.: W.B.Krtzig et

al. Rotterdam: Balkema 1991, pp. 1133 -1138Wacker, J., Friedrich, R., Plate, E.J., Dorsch, F.,

1990b: Drag and lift on rectangular bluff bod-ies immersed in deep boundary layer flowover rough surfaces. Journal of Wind Engi-neering and Industrial AerodynamicsVol.33,1990,No.1/2,pp.113-122

Wacker, J., E.J.Plate, 1992: Correlation struc-ture of wind pressure buffeting on cuboidalbuildings and corresponding effective areawind loads. Journal of Wind Engineering andIndustrial Aerodynamics Vol. 43, 1992, pp.1865 - 1876

Wacker, J., E.J.Plate, 1993: Local peak windpressure coefficients for cuboidal buildingsand corresponding pressure gust factors.Journal of Wind Engineering and IndustrialAerodynamics, vol. 50, pp. 183 - 192

Wacker, J., E.J.Plate, 1994: Reliability-basedlocal design wind pressures for simple rec-tangularly-shaped buildings. Structural Safetyand Reliability, Eds.: Schueller et al. Rotter-dam: Balkema 1994, pp. 1689 - 1696

Wacker, J. (1995): "Local wind pressures forcubical buildings in turbulent boundary lay-ers", in: Wind Climate in Cities, Cermak etal., 1995, Kluwer Academic Publlisher, Lon-don

wind and urban climate

Documents