-
Proc. lnstn Cio. Engrs, Part 2, 1982,73, Sept., 653-666
8560 Assessment of wind loading on the claddings of high-rise
buildings
R. P. LAM, BSc(Eng), PhD, MICE, MIStructE* L. c. H. LAM,
BSc(Eng), PhD, MICE*
Wind damage to the cladding of high-rise buildings is not
uncommon in areas subject to typhoon winds. The consequences of
failure are not only material losses, but also possible loss of
life and injury caused by flying debris. This Paper deals with the
assessment of wind loads on the claddings of high-rise buildings,
and concludes that design based on wind tunnel model tests alone is
often inadequate; it must be supplemented by experienE gained in
full-scale tests, good engineering judgement and further research
in the response of clad- ding elements to dynamic wind forces.
Notat ion C , exposure factor C , gust factor C,, external
pressure coefficient CJi internal pressure coefficient C,, mean
external pressure coefficient G gust factor g peak factor g average
peak factor I , intensity of turbulence p peak local wind pressure
P, maximum wind pressure averaged over a period oft S
q mean velocity pressure q, maximum wind pressure averaged over
a period o f t S p air density
gust velocity pressure
I n t r o d u c t i o n For aesthetic and other reasons the use
of curtain walls has become very popular in recent years. In Hong
Kong the majority of prestige tall buildings completed in the past
five years or so are cladded with curtain walls. As this type of
construction requires large-size glass panels and high quality
metallic surface finish, its cost is often a fairly significant
portion of the total cost of the building. Being the external
surface of a building, the glass cladding of curtain walls is
subjected to adverse environmental effects such as rain-water
penetration, heat loss/absorption, and buffeting of wind gusts.
Among these effects, wind loading is the one which causes
Written discussion closes 15 November 1982, for publication in
Proceedings, Part 2. * University of Hong Kong.
653
-
L A M A N D L A M
the most severe damage. Hence appropriate assessment of loadings
and proper detailing of the cladding units are essential in order
to produce a safe, durable and economical design.
2. This Paper deals with the assessment of local wind loads on
tall buildings standing in areas where the occurrence of strong
winds or tropical cyclones is frequent. Typical current design
methods are reviewed and appraised in the light of full-scale
experimental results.
Current design methods 3. Wind turbulence causes dynamic loading
on structures. In considering fail-
ure of cladding materials under dynamic wind loads, Allen and
Dalgliesh' showed that the design wind load for metal cladding
failing primarily by yield can be determined by assuming that the
panel is a static structure which fails when the wind pressure
exceeds the standard plastic resistance. Less is known about glass
which is a brittle material, and its strength depends on the panel
size, existence of flaws and the rate of loading. By applying the
damage criterion determined from glass manufacturers' loading tests
to a sustained random wind pressure of duration T , they suggested
that glass cladding, like metal cladding, can be considered as
statistically loaded structures in which failure occurs when wind
pressure exceeds the structural capacity as determined by standard
tests. It follows that both types of cladding should be designed to
resist peak wind loads.
4. Current cladding design methods are normally on a static
basis. The net design wind pressure on cladding units is usually
taken to be the algebraic differ- ence of the external pressure or
suction and the internal pressure or suction. The latter is
estimated by considering whether or not the claddings are airtight,
and if not, by the location and size of the openings. The external
pressure or suction is more directly related to wind speed and
direction, shape of the building and location of the cladding unit
under consideration.
5. Similar to the static wind load design of the structure as a
whole, the calculation of wind loads on cladding units is based on
either the mean wind' or peak gusts3. The mean wind approach to
calculate the design local wind pressure is generally expressed
by
P = G(Cpe - cpi)q (1) in which the design mean velocity pressure
4 is calculated from the design mean speed v. The external pressure
coefficient C,,-which is the ratio of the mean external pressure to
the mean velocity pressure q, is usually determined by wind tunnel
model experiments in the absence of adequate field data. The
internal pressure coefficient depends on the permeability of the
building and whether the openings are experiencing external
pressure or suction. The gust effect is accounted for by the gust
factor, G, which is the ratio of peak to mean effects.
6. In a comparative study the Authors have shown4 that mean
pressure coef- ficients obtained in a wind tunnel with a correctly
simulated velocity profile can be used to predict quite accurately
the time-averaged wind loads on a building. Equation (1) can
therefore provide an accurate assessment of the design load
provided that a suitable gust factor G is adopted. Research work5e6
on gust effects are carried out both in wind tunnels and full-scale
experiments'-'' in order to study the peak distribution of wind
pressure or suction at critical locations on buildings of different
shapes under different wind directions.
7. It is simple to assume that wind loading is a stationary
random Gaussian 654
-
W I N D L O A D I N G O N C L A D D I N G S O F H I G H - R I S
E B U I L D I N G S
process such that the gust factor G may be written as
G = 1 +gZ, (2)
and a further assumption that the peak distribution is narrow
enables the average value g of the Deak factor g to be used for
design purposes.' However, more recent work in wind tunnel^^.^ and
full-scale experiments'* have shown that the wind pressures do not
quite follow the Gaussian distribution.
8. As an alternative to the mean wind approach, calculation of
the local exter- nal pressure or suction can be on a gust speed
basis. For instance the British code3 employs the following
expression
in which the gust velocity pressure q is defined in a similar
manner to the mean velocity pressure q , but with the design mean
velocity replaced by a design three-second gust speed V having a
certain probability of occurrence. Although it is desirable to
determine maximum wind loads on small units of cladding using gusts
with shorter averaging times, it is not at present possible because
available meteorological data can only provide maximum gust speed
averaged over about three seconds as a result of the
characteristics of the type of anemometers used.
9. The external pressure coefficient C,, is the ratio of the
local maximum gust loading to the design gust velocity pressure q.
As external pressures and suctions on building surfaces vary
considerably with wind direction, velocity profile and building
shape, pressure Coefficients are of necessity obtained from
measurements on models in wind tunnels, and the great majority of
data available have been obtained in conditions of relatively
smooth flow. It is therefore suggested3 that where full-scale
measurement data are available, values of the pressure coefficients
should be adjusted to allow for turbulence and wind tunnel model
scale effects.
10. The accuracy of the design wind load calculated from
equation (3) depends on the pressure coefficient C,, for short
duration gusts. In a study13 of force coefficients on a
multi-storey building obtained from full-scale measurements the
authors have found that the force coefficients are not constant but
vary with the averaging period. The general trend is that the force
coefficient increases with the averaging period, implying that gust
loading on the building as a whole decreases with its duration.
This phenomenon is expected because short duration gusts are
smaller in spatial extent. However, this variation does not
necessarily apply to local gust loadings, in particular where the
gust is spatially large enough to engulf a cladding element.
11. It would therefore be enlightening to study the
characteristics of the gust factor G and the gust pressure
coefficient C in full-scale buildings in order to obtain a better
understanding of the nature o&st loads on cladding.
Full-scale experiment 12. The full-scale measurements were
carried out at the experimental building
of the Centre of High Building Research in the Department of
Civil Engineering, University of Hong Kong. The experimental
building is situated on a piece of lowland in an open area where it
is exposed to winds blowing from all directions. It is a ten-storey
steel-framed building measuring 18.29 m by 9.15 m in plan and 30.48
m in height, and is enclosed on all sides to its full height with
steel-framed glass curtain walls (Fig. 1). A site plan is shown in
Fig. 2.
655
-
L A M A N D L A M
Fig. 1. Experimental building of the Centre of High Building
Research at Cape D'Aguilar, Hong Kong
(a)
Fig. 2. (a) Site plan of experimental building; (b) layout plan
of experimental station
656
-
W I N D L O A D I N G O N CLADDINGS O F H I G H - R I S E
BUILDINGS
13. Wind velocities in the vicinity of the building are measured
by ane- mometers mounted on four latticed steel masts, each 54.86 m
high and spaced equally at 15.24 m intervals in a straight line
parallel to the north-east side of the building at a distance of
approximately 60 m. Anemometers are positioned at four levels on
each mast, at 7.62, 22.86, 38.10 and 54.86 m above the common base
level of the masts, which is about 5.5 m above mean sea level. Mean
and gust velocity profiles are obtained and the roof level of the
experimental building (about 40 m above mean sea level) is selected
as the reference level.
14. Wind pressure fluctuations on the four vertical faces of the
building are measured by 72 pressure transducers flush-mounted at
selected points (Fig. 3) in the glass wall panels at 9.1, 15.2,
21.3 and 27.4 m above ground floor level. The pressure transducers
are of the strain gauge type designed by the Building Re- search
StationI4 to measure the difference between the external pressure
and that within its body cavity which serves as a reference or
datum pressure. For the measurement of local gust loading on
claddings the reference or datum pressure is ideally the internal
pressure inside the building so that the measured load is the total
load acting on the cladding, and this is achieved by verting the
body cavity of each transducer through a small opening to the
inside of the building. Openings in the curtain wall are kept
closed to create a more stable internal pressure inside the
A B
S
- Latticed towers
F Guyed ma
l
C D
/ 32
\
Experimental building
( b)
Fig. 2 continued
657
-
L A M A N D L A M
18 288 mm P- -____I 23243098 6731 30982324 !riimlmm mm
;mmimAl
T - r
91 44 mm.
mm 1 mm
N . / / / ///7 Location of pressure transducers on the
full-scale building
Front elevation Side elevation (Y
I 18 288 mm
I p2324 , 3098 , 6731 mm , 3098 , 2324 , mm ' mm ; : mm : mm j
NE1 'NE2 'NE3 NE4 N E5' NE$- -
E : E
0 m N p.
E E N W--
m p.
m
0 N
E E
0 m r.
.F* W v)
J --?SW1 ,SW2 , SW3 sw4, SW5, SW6j-r -
ib)
Fig. 3. (a) Location of pressure measuring points on full-scale
building; (b) floor plan showing location of pressure
transducers
658
-
W I N D L O A D I N G O N CLADDINGS O F H IGH-RISE BUILDINGS
building during measurement. 15. Analogue signals from the
anemometers and pressure transducers are
scanned and digitized by a 240-channel high-speed data logger at
a rate of ten data points a channel per second, and the digital
data are recorded on magnetic tape. Description of other
experimental facilities at the building are given in references 4
and 15.
Results 16. Wind pressure data obtained in six occasions of
typhoon with different
wind speeds and directions were analysed. Gust factors and local
pressure coef- ficients defined by the following expressions were
computed:
maximum instantaneous pressure mean pressure* G = (4)
where P, is the maximum pressure averaged over a period of t
seconds, q, is 1/2(pV;), and ?( is the maximum gust speed averaged
over a period o f t seconds. It is observed that there are marked
differences between the characteristics of pressures and suctions.
Typical results of the gust factor and pressure coefficients at
selected transducer positions are shown in Tables 1 and 2 for
face-on and glancing winds respectively. The selected positions
are: points of maximum windward pressureypoints near the leading
edge of windward face or side walls, and points of
Table 1 (a). Pressure coefficients and gust factors at selected
transducer posi- tions under a wind at 7" to the normal of a major
building face
Transducer
0.1 S
9NE3 -0.52 9SE1
0.95
- 0.84 3NW2 -0.52 3sw5 - 1.00 3SE1
0.92 3NE3 - 0.68 5NW2 - 0.45 5sw5 - 0.97 5SE1
1.04 5NE3 - 1.20 7NW2 -0.37 7sw5 - 0.70 7SE1
0.96 7NE3 -0.78 9NW2 -0.39 9sw5
C P C
1 s
0.97 -0.35 -0.38 -0.74
0.98 - 0.50 -0.37 - 1.13
1.04 - 0.59 -0.46 -0.64
0.94 -0.72 -0.50 -0.82
T 3 s
0.94 - 0.28 -0.37 - 0.64
0.95 - 0.46 -0.35 - 1.05
1.03 -0.53 - 0.42 - 0.54
0.9 1 -0.58 - 0.47 - 0.70
C,, 30 min.
0.92 -0.19 - 0.20 -0.25
0.96 -0.25
0.14 - 0.65
1.05 - 0.27 -0.13 - 0.22
0.90 -0.27
0.13 -0.32
G
2.02 5.36 3.82 6.12 1.96 5.49
-5.18 3.62 1.94 7.04 6.78 6.06 2.00 7.26
- 7.84 5.15
-
Mean wind speed averaged over 30 minutes at roof level = 23.7
m/s. Mean wind direction = N 38"E.
PIG
1.9 - 1.0 - 0.8 - 1.5
1.9 - 1.4 - 0.7 - 2.4
2.0 - 1.9 - 0.9 - 1.3
1.8 - 2.0 - 1.0 - 1.6
* averaged over 30 minutes unless noted otherwise.
659
-
L A M A N D L A M
maximum suction on the leeward face and side walls. 17. On
examining the gust factors, it is apparent that large values (up to
33.18)
are associated with suctions on the leeward and side walls,
whereas gust factors for pressures seldomexceed three. It is also
observed that these large values of the gust factor actually
resulted from small values of the mean local pressure/suction
averaged over 30 minutes.
18. With regard to the local pressure coefficients, the results
show that different patterns ofvariation with the averaging period
I exist. In general, the local pressure coefficient decreases with
the averaging period at locations where suctions occur,
irrespective of the angle ofincidence ofthe wind on the building.
On the other hand, variation of positive pressure coefficients with
the averaging period depends on the angle of attack.
19. For winds acting almost normal to a major building face
(aspect ratio width/depth = 2) the windward pressure coefficients
are rather constant, i.e. C,,
Table 1 (b). Pressure coefficients and gust factors at selected
transducer posi- tions under a wind at 11 to the normal of a major
building face
Transducer
9NE 1 9NE3 9NE5 9NE6 9NW3 9SE1 9 s w 3 7NE1 7NE3 7NE5 7NE6 7NW3
7SE1 7sw3 5NE1 5NE3 5NE5 5NE6 5NW3 5SE1 5sw3 3NE1 3NE3 3NE5 3NE6
3NW3 3SE1 3sw3
0.1 S
0.85 0.84 0.66 0.57
- 0.92 - 0.43 - 0.50
0.92 0.93 0.78 0.62
-0.67 - 0.49 -0.52
0.95 0.93 0.66 0.54
- 0.8 1 -0.77 - 0.49
0.86 0.84 0.55
- 0.28 -0.71
0.40 -0.36
C P C
1 s
0.91 0.89 0.71 0.61
- 0.90 -0.37 - 0.54
0.96 0.97 0.83 0.64
-0.67 - 0.40 -0.55
0.99 0.96 0.69 0.55
- 0.80 -0.60 -0.52
0.89 0.85 0.58
-0.29 -0.65
0.41 -0.37
T 3 s
0.92 0.90 0.69 0.60
- 0.90 - 0.30 - 0.54
0.93 0.96 0.86 0.63
- 0.69 -0.31 -0.52
1.03 0.95 0.70 0.56
- 0.79 - 0.40 - 0.49
0.89 0.87 0.58
- 0.27 -0.56
0.40 -0.36
30 min
0.96 1.15 0.55 0.52
-0.61 -0.12 - 0.43
1.05 1.26 0.82 0.60
-0.50 0.05
- 0.37 0.97 1.17 0.62 0.53
- 0.63 -0.11 -0.35
0.86 1.02 0.58 0.46
-0.38 0.48
- 0.41
G
2.28 1.90 3.12 2.85 3.93 9.45 3.47 2.26 1.91 2.47 2.68 3.48
-25.15 4.15 2.53 2.06 2.76 2.65 3.35
17.47 4.48 2.59 2.04 2.46
- 4.54 4.90 2.15 8.50
T
L
Mean wind speed averaged over 30 minutes at roof level = 20.5
m/s. Mean wind direction = N 34"E.
660
2.2 2.2 1.7 1.5
- 2.4 - 1.1 - 1.5
2.4 2.4 2.0 1.6
- 1.7 - 1.3 - 1.5
2.5 2.4 1.7 1.4
-2.1 - 1.9 - 1.6
2.2 2.1 1.4
- 0.7 - 1.9
1.0 - 0.9
-
W I N D L O A D I N G ON C L A D D I N G S O F H I G H - R I S E
B U I L D I N G S
Table 1 (c). Pressure coefficients and gust factors at selected
transducer posi- tions under a wind normal to a minor building
face
Transducer
9SW6 9SE1 9SE2 9SE3 7SW6 7SE1 7SE2 7SE3 5SW6 5SE1 5SE2 5SE3 3SW6
3SE1 3SE2 3SE3
0.1 S
-0.78 0.69 0.7 1 0.74
- 0.46 0.82 0.74 0.80
- 0.70 0.7 1 0.78 0.71 0.67
- 0.7 1 0.70
-0.63
c,,
1 s
0.44 0.74 0.77 0.77 0.44 0.88 0.80 0.82
-0.57 0.77 0.84 0.77 0.69
- 0.65 0.76
-0.55
3 s
0.36 0.8 1 0.83 0.82 0.41 0.87 0.87 0.82 0.55 0.82 0.90 0.8 1
0.71
- 0.60 0.79
- 0.46
l- CPe
30 min
- 0.06 1.10 1.16 I 0 4
-0.13 1.08 1.39 0.97
1.14 1.27 1.05 0.84
-0.29 1.04
- 0.06
- 0.28
G
~
33.18 1.63 1.59 1.85 8.92 1.96 1.39 2.12 6.35 1 60 1.59 1.75
2.06 6.42 1.73
25.38
pi4
- 2.0 1.8 1.8 1.9
- 1.2 2.1 1.9 2.1
- 1.8 1.8 2.0 1.8 1.7
- 1.9 1.8
- 1.5
Mean wind speed averaged over 30 minutes at roof level = 333
m/s. Mean wind direction = S 45"E.
Table 2(a). Pressure coefficients and gust factors at selected
transducer posi- tions under a wind at 22" to the normal of a major
building face
Transducer
9NE2 9NE6 9 s w 3 9SE2 7NE2 7NE6 7sw3 7SE2 5NE2 5NE6 5sw3 5SE2
3NE2 3NE6 3sw3 3SE2
0.1 S l 1 S l 3 s 1.13 I 1.19 I 1.25 0.62
-0.59 -0.58 -0.56 -0.42 -0.45 -0.45
0.63 0.63
1.29
-0.27 -0.29 -0.27 0.2 1 0.23 0.23 0.40 0.46 0.44 1.26 1.20
1.16
-0.28 -0.29 -0.30 -0.32 -0.35 -0.35
0.72 0.73 0.72 1.38 1.32 1.27 0.59 0'60 0.57
-0.35 -0.37 -0.37 0.77 0.75 0.72 1.41 1.35
30 min 1.18 0.54
-0.36 -0.57
1.32 0.66
- 0.30 0,48 1.25 0.6 1
- 0.27 -0.18
1.15 0.30 0.14
-0.19
G
2.39 2.87 3.12 2.45 2.44 2.72 3.08 2.96 2.54 2.95 3.24
. 4.16 2.52 3.66 4.10 3.55
Mean wind speed averaged over 30 minutes at roof level = 14.9
m/s. Mean wind direction = N 23"E.
661
-
L A M A N D LAM
for gusts are approximately equal to the mean pressure
coefficient c, at locations near the central portion of the
building face. Elsewhere on the windward face the coefficient tends
to increase with the averaging period. However, when the wind is
normal to a minor building face (i.e. the aspect ratio changes to
05) , the windward pressure coefficients are not constant but
rapidly increase with the averaging period. Under glancing winds
the pressure coefficients at the leading edge of the windward face
are fairly constant, but others tend to increase with the averaging
period (except at points which are very close to the trailing
edge).
20. It should be noted that windward pressures are greater in
magnitude than leeward suctions and therefore have a greater
influence on the properties of the total load on the building as a
whole. This explains why the total force coefficient increases with
the averaging period13 despite the fact that suctions decrease with
that period.
21. As the gust factor and the pressure coefficient are
dependent on the averaging time, a suitable averaging time has
therefore to be determined before using appropriate values ofthe
gust factor, the pressure coefficient and the reference velocity to
calculate a design wind load.
22. In order that the experimental results may be compared with
design wind loads given by codes of practice, it is necessary to
make adjustments for the difference in height of the reference mean
or gust velocity pressures.
23. Using a power law mean velocity profile with an exponent of
0.19 which has been determined for the site of the experimental the
mean velocity pressure q l 0 at 10 m could be related to the mean
velocity pressure q at 30 m (roof level) by
ql0 = 0.66 4 (6) 24. The three-second gust profiles obtained
simultaneously with the pressure
Table 2(b). Pressure coefficients and gust factors at selected
transducer positions under a wind at 22" to the normal of a major
building face
Transducer
9sw4 9sw5 9SE3 9NW3 7sw4 7sw5 7SE3 7NW3 5sw4 5sw5 5SE3 5NW3 3sw4
3sw5 3SE3 SNW3
0.1 S
1.17 0.83
- 0.70 -0.86
0.96 0.96
- 0.93 -0.87
0.99 0.97
- 0.99 - 1.19
0.92 0.85
- 0.45 -0.72
CPC
I s
1.17 0.83
-0.57 - 0.69
0.99 0.98
-0.71 -0.63
1 . 1 1 1 .W
-0.82 -0.90
1.01 0.95
- 0.49 - 0.64
- 3 s
1.16 0.82
- 0.47 - 0.55
0.96 0.94
- 0.56 -0.55
1 40 0.97
-0.58 - 0.69
0.93 0.88
- 0.45 - 0.49
c
1.27 0.90
- 0.41 - 0.46
1.06 1 a4
- 0.50 - 0.48
1.10 1.02
- 0.43 -0.57
1.02 0.88
- 0.34 - 0.40
G
1.84 1.85 3.42 3.73 1.82 1.83 3.72 3.62 1.80
4.65 4.18 1 .so 1.95 2.61 3.62
1.90
Mean wind speed averaged over 30 minutes at roof level = 24.9
m/s. Mean wind direction = S 23"W.
662
-
W I N D L O A D I N G O N C L A D D I N G S O F H IGH-RISE
BUILDINGS
measurements indicated that there was no significant variation
of the maximum three-second gust with height at the levels ofthe
pressure transducers. Therefore no reduction to the three-second
velocity pressure q at the respective height of the transducers was
considered necessary.
25. The nondimensional ratios of the observed local maximum wind
pressure P to the mean velocity pressure 4 based on the mean wind
velocity at roof level of the experimental building are also given
in the last columns of Tables 1 and 2. Table 2(c) reveals an
unexpectedly high local suction at transducer position 3NE6 which
is very close to the leading edge of the windward face, when the
mean wind direction was inclined at about 58 degrees to the normal
of this face. The observed peak suction of 3.6 4 is equivalent to
5.5 times the mean velocity pressure at 10 m or 1.7 times the
three-second gust velocity pressure at roof level. As other nearby
transducers recorded positive pressures (with the exception of
transducer 5SE3 which recorded a mean positive pressure but a peak
suction) it is very likely that the wind loads at both transducer
positions (3NE6 and 5SE3) are localized effects.
Table 2(c). Pressure coefficients and gust factors at selected
transducer positions under a wind at 32" from the normal of a minor
building face
Transducer
9NE2 9NE3 9NE5 9NE6 9NW1 9SE3 9SW1 7NE2 7NE3 7NE5 7NE6 7NW1 7SE3
7SW1 5NE2 5NE3 5NE5 5NE6 5NW1 5SE3 5SW1 3NE2 3NE3 3NE5 3NE6 3NW1
3SE3 3SW1
0.1s
0.48 0.68 0.87 0.8 1
-0.76 1.09
- 0.44 0.8 1 0.96
1.02 - 0.74
1.09
-
- 0.49 0.66 0.83 0.79 0.89
- 0.69 - 0.97 - 0.64
0.57 0.73 0.77
- 1.45 - 1.10
0.71 -0.32
CV
1 s
0.45 0.66 0.90 0.83
- 0.64 1.12
- 0.44 0.83 0.99 -
- 0.70 -
1.13
0.6 1 0.78 0.84 0.91
- 0.64 - 0.8 1 -0.63
- 0.46
0.54 0.68 0.74
- 0.96 -0.86
0.65 - 0.29
3 s
0.43 0.6 1 0.89 0.83
- 0.62 1.12
- 0.46 0.82 1.01 -
-
-0.61 1.19
- 0.52 0.62 0.79 0.84 0.91
- 0.49 -0.78 - 0.66
0.56 0.73 0.74
- 0.86 -0.73
0.62 -0.28
-r c,, 5 min
0.50 0.70 1.20 1.07
- 0.80 1.05
-0.35 1.14 1.47
1.65 -
- 0.65 1.18
- 0.44 0.67 0.96 1.08 1.25
- 0.45 0.26
-0.61 0.54 0.85 0.98
-0.91 - 0.87
0.78 - 0.24
G
2.40 2.43 1.82 1.88 2.37 2.58 2.49 1.77 1.63 1.45 1.54 2.83 2.3
1 2.78 2.46 2.16 1.83 1.77 3.89
-9.26 2.62 2.64 2.15 1.96 3.98 3.14 2.27 3.33
Mean wind speed averaged over 5 minutes at roof level = 18.3
m/s. Mean wind direction = S 77"E.
1.2 1.7 2.2 2.0
- 1.9 2.7
- 0.9 2.0 2.4
2.5 - 1.8
2.7 - 1.2
1.7 2.1 2.0 2.2
- 1.8 - 2.4 - 1.6
-
1.4 1.8 1.9
- 3.6 - 2.7
1.8 - 0.8
663
-
L A M A N D LAM
26. Another interesting feature is shown in Table l(c) in the
results obtained under a wind blowing in a direction normal to the
minor face. All windward transducers recorded positive pressures,
except 3SE1 and 3SE3 which recorded suctions. This phenomenon has
been confirmed in a wind tunnel model test in which the surrounding
topographic features are also modelled, and it is believed that
this localized effect is caused by the rock mass upstream of the
building (Fig. 2). The height of this rock mass is almost half that
of the building, and it produces a shielding effect on the
pressures on the windward minor face.
21. Adverse effects due to the influence of nearby objects,
buildings and structures are bound to exist in a congested built-up
environment. Such effects, including high local suctions caused by
channelling of flow and wake-induced building vibrations, are
difficult to codify. In these situations the designer has to rely
on experience gained in field observations and exercise his
engineering judgement to decide the acceptable level of safety
against wind damage.
Comparison of experimental results with design wind loads 28. In
view of the wide range of variation of the gust factor G and the
local
pressure coefficients C,, over the building surfaces, suitable
choice of these two parameters for use in codes of practice has to
rely on a substantial amount of research in full-scale as well as
wind tunnel measurements. For a building of given shape and size,
the design wind load for claddings and walls at critical locations
should be assessed under the worst possible conditions (speed and
direction) of wind attack. Although the full-scale experimental
results reported in this Paper may not have revealed the worst
conditions of loading, comparison with design wind pressures
calculated from some codes of practice could still provide a valu-
able basis for appraisal of the codes. Magnitudes of the design and
observed wind pressures can be compared in terms of a reference
velocity pressure after suitable adjustment of reference heights
has been made. Comparison with the design wind pressures calculated
from two codes of practice, the Canadian code and the British code,
is made in $2!?-32.
29. It is both stated in the codes of practice and observed
experimentally that when the wind acts at an angle to the windward
face of a building, high local suctions will occur at the leading
edge of the wall which is at a small angle to the wind where flow
separation takes place. The results in Table l(b), at pressure
transducer locations NW3, clearly show this effect and are taken
for comparison.
30. The Canadian code which adopts a mean wind approach
specifies a gust factor of 2.5 for cladding or window design, a
local pressure coefficient of - 1.5 for flat-roofbuildings ofheight
greater than twice the width, and an exposure factor of 1.3
appropriate to the dimensions, shape and site conditions of the
experimental building. The exposure factor which accounts for the
height of the structural element above the adjacent ground level is
required because the reference (design) velocity pressure is based
on hourly mean speeds measured at a height of 30 ft (about 10 m) in
an open exposure. The design local wind pressure is given by the
algebraic difference of external and internal pressures :
where C, = exposure factor = 1.3, Cb = local gust factor = 2.5,
C, = external pressure coefficient = - 1.5, Cei = factor for
internal pressure = 1.1, CPi = internal pressure coefficient for
openings uniformly distributed in all four walls =
664
-
W I N D L O A D I N G O N C L A D D I N G S O F H I G H - R I S
E B U I L D I N G S
-0.3, and ql0 = reference velocity pressure based on hourly mean
wind speeds at a height of 30 ft.
31. Referring to Table l(b), the maximum local suction at
transducer positions 9NW3, 7NW3, 5NW3 and 3NW3 are 2.4 q, 1.7 q,
2.1 q and 1.9 4 respectively. Using equation (6), these observed
peak suctions can be expressed in terms of the mean velocity
pressure at 10 m, q l 0 , as 3.6 q l 0 , 2.6 q l 0 , 3.2 q l 0 and
2.9 q l 0 respectively, which are smaller than the design wind
pressure calculated by equa- tion (7).
32. The British code3 adopts a gust-speed approach. The design
wind pressure on claddings is calculated based on a gust speed
averaged over a period of three seconds at the height of the
cladding element (use of the factor S , in calculating the design
gust speed). For the experimental building the design local wind
pressure is given by
P = (Cpe - C,& = -0.9 q (8)
where C,, = external pressure coefficient = - 1.2, CPi =
internal pressure coefficient = -0.3 and q = design velocity
pressure based on the three-second gust atthe height of the
cladding element.
33. It has been reported in $24 that there was no significant
difference between the three-second gust velocities at roof level
and at the levels of the pressure transducers. Therefore the
observed peak suctions at transducer positions NW3 of Table l(b)
can be expressed in terms of the three-second gust velocity
pressure q at roof level for comparison purpose. These observed
peak suctions are 1.1 q, 0.8 q, 1.0 q and 0.9 q respectively,
varying from the highest (9NW3) to the lowest (3NW3) level.
Apparently the observed peak suctions can be regarded as reason-
ably close to the design value given by equation (8).
34. Considering that local wind loads are momentary in nature
because they are caused by gusts of very short duration (small
eddies in the turbulent flow of air around the building), the gust
speed approach provides a direct and sound method of assessing wind
loads for cladding design in high-rise buildings. However, it has
been that this approach tends to result in unduly high total wind
loads on a building or structure as a whole, and that the mean wind
approach is more suitable for assessing the overall wind effects on
a building or structure.
Conclusion 35. In the design ofthe cladding for a multi-storey
building it would appear that
the gust-speed method of the British code is likely to provide
lower pressure values than the mean wind approach of the Canadian
code. These lower pressure values are likely to be close to the
actual average values of local pressure or suction applied to the
cladding.
36. Higher pressure values may occur at isolated positions in
the cladding caused by the influence of nearby buildings or other
obstacles. These localized higher values are of the order of
magnitude predicted by the Canadian code. For the present, a design
method based on CP33 would seem to be suitable on the understanding
that isolated positions of higher pressure represent a modest en-
croachment on the factor of safety inherent in the design of the
cladding.
Acknowledgement 37. The Authors wish to thank the Centre of High
Building Research of the
University of Hong Kong for providing the full-scale test
facilities. The Centre of
665
-
L A M A N D L A M
High Building Research is sponsored by the Hong Kong Government,
the Nuffield Foundation, the British Iron and Steel Federation, the
Ministry of Overseas De- velopment of H.M. Government, a local Hong
Kong benefactor and the Univer- sity of Hong Kong.
References 1. ALLEN D. E. and DALCLIESH W. A. Dynamic wind loads
and cladding design. Division of
2. NATIONAL BUILDING CODE OF C A N A D ~ Part 4 Design. 3.
BRITISH STANDARDS INsTIwnoN. Code of basic data for the design of
buildings. British
Standards Institution, London, 1972. CP3: Ch. V, Part 2. 4. LAM
R. P. and LAM L. C. H. Mean wind pressure distribution on a
multi-storey
building. Proc. lmtn Civ. Engrs, Part 2, 1981, 71, Mar.,
119-129. 5. PETERKA J. A. and CERMAK J. E. Peak pressure duration
in separated regions on a
structure. Proc. US-Japan research seminar on wind effects on
structures, Kyoto, Japan, 1974.
6. CERMAK J. E. Separation-induced pressurefluctuations on
buildings. Colorado State Uni- versity, 1970, Report CEP70-71
JEC19, Fluid Mechanics Programme.
l. MACKEY S. et al. Gust factors. Proc. semin. on wind loads on
structures, Hawaii, 1970. 8. NEWBERRY C. W. et al. Wind pressure on
the Post Office Tower, London. Proc. int.
con5 on wind efects on bldgs and structs, Tokyo, 1971. 9.
NEWBERRY C. W. et al. Wind loading on tall buildings-further
results from Royex
House. Building Research Station, 1973. 10. DALCLIESH W. A.
Statistical treatment of peak gusts on cladding. J. Struct Diu. Am.
Soc.
Ciu. Engrs. Sept., 1971. 1 1 . DAVENPORT A. G. Note on the
distribution of the largest value of a random function
with application to gust loading. Proc. lnstn Ciu. Engrs., 28,
June, 1964, 187-196. 12. WONC K. Y. Some aspects of wind loading
effects on the design of multistorey buildings.
University of Hong Kong, PhD Thesis, 1976. 13. MACKEY S. et al.
A full-scale and wind tunnel study of wind loading on a building.
Proc.
of ASCE-IABSE regional conference on tall buildings, Bangkok,
Jan. 1974. 14. MAYNE J. R. A wind pressure transducer. Building
Research Establishment, 1970, BRS
Current paper 17/70. 15. LAM L. C. H. and LAM R. P. An
experimental study of the dynamic behaviour of a
multistorey steel-framed building. Proc. Instn Ciu. Engrs., Part
2, 1979,67, Sept. 707- 720.
16. MACKEY S. et al. A basis for estimation of wind pressures on
high-rise buildings in areas subject to typhoon winds. The
Structural Engineer, 1973,51, Dec., No. 12, p. 453.
17. DAVENPORT A. G. and DALCLIESH W. A. Commentary No. l-wind
loads. National Building Code of Canada, 1970, Supplement No.
4.
18. TAYLOR R. G . Blow, blow, thou winter wind, thart not so
unkind as Chap. V: Part 2. The Structural Engineer, 1973.51, Dec.,
No. 12, p. 449.
19. Discussion on Blow, blow, thou winter wind, thart not so
unkind as Chap. V: Part 2. The Structural Engineer, 52, Sept.,
1974, No. 9, p. 347.
Building Research, National Research Council, Canada, Research
paper no. 61 1.
666