Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Elsevier
W i n d l o a d s o n p l a n a r c a n o p y r o o f s - P a r t 1:
Mean pressure distributions
(Received October 10, 1991; revised version accepted December 20, 1991)
Summary
Mean pressure distributions on planar canopy roofs are presented based on 1:100 scale
wind tunnel measurements obtained in a simulated suburban atmospheric boundary layer.
For a range of roof pitch angles and wind directions, point and area-averaged pressure
measurements were obtained with the worst loads experienced adjacent to lines of flow
separation. In this study, generally smaller wake or suction mean pressures were obtained
when compared with the earlier studies of Gumley. These differences are attributed to the
interference at separation by the canopy supports in Gumley's work, and the higher level of
turbulence in the present study.
1. I n t r o d u c t i o n
weight, and consequently more flexible roofs which can be wind sensitive.
Hence, there is a growing need for more detailed knowledge of the wind
loading on these structures.
This paper and a companion paper [1] seek to add to the relatively small
database of wind loads on planar canopy roofs by summarising the results
of an extensive wind tunnel investigation carried out at the University of
Queensland [2]. This paper describes the experimental arrangement and pres-
ents the mean pressure results. The companion paper presents and discusses
the measured fluctuating and peak pressure distributions and their correla-
tions. The following Section reviews wind loading investigations on canopy
roofs. The experimental arrangement for the present investigation is described
in Section 3. The results of this study together with comparisons with previous
work and code values are presented in Section 4. The conclusions are pres-
ented in Section 5.
2. W i n d l o a d o n c a n o p y r o o f s
Newberry and Eaton [3], in discussing the codification of wind loads for the
1972 British Standard [4], report t h a t for canopy roofs there were surprising
inconsistencies in the available literature. They concluded t h a t these probably
reflected differences in test conditions as well as variations in the roof and
height aspect ratios. The most detailed tests on canopy roofs at that time were
those of Jensen and Franck [5] conducted in the early 1960s.
Gumley and Wood [6], and Gumley [7,8] undertook probably the most compre-
hensive study of the wind loading of free standing canopy roofs in atmospheric
boundary layer flow. Gumley and Wood [6] conducted wind tunnel model tests
of a full scale canopy roof on which pressure measurements were being ob-
tained by Robertson et al. [10]. Gumley [7,8] went on to conduct a systematic
study of these structures. A very extensive parametric study investigating the
influence of roof pitch, wind direction, aspect ratio, eaves height, and internal
stacking arrangement was undertaken. The specific structures modelled were
typical of those used in the agricultural industry in England.
Gumley employed, at a length scale of I : 75 in a simulated rural (Zo = 0.03 m)
atmospheric boundary layer, a variable pitch, planar, 20 x 20 m canopy roof
standing on eight square sectioned legs located evenly along the eaves. Each
half of the roof was divided into five sections: four strip regions close to the
edges and ridge line, and a central region. Only area-averaged pressures over
these separate regions were measured. The results of his study were used
to update pressure coefficients on such structures in the British [4] and
Australian [9] wind loading codes.
Full scale measurements of wind loads on two agricultural canopy roof
structures were reported by Robertson et al. [10]. These structures had a width
to span ratio of approximately two and, thus, differed from the basic canopy
studied by Gumley (width to span of one). Proposals for design based on these
C.W. Letchford, J.D. Ginger/Wind loads on planar canopy roofs 27
full scale measurements differed from those proposals [11] based on the wind
tunnel tests of Gumley, in particular for the case of no blockage under the
canopy roof. The spatial extent of downward pressures on the windward half
and the magnitude of uplift pressures on the leeward half of the roof were
disputed, with the wind tunnel based proposals leading to a particularly
onerous design case for net drag on the canopy roof. These differences were
probably due to different philosophies adopted during the codification process
because reasonably good agreement is found between full scale [9] and model
scale [6] mean pressure differences across the roof centreline for a wind normal
to the ridge as shown in Fig. 1.
In their study, Gumley and Wood [6] obtained a detailed point pressure
distribution over the roof for the one pitch angle (15 ~) corresponding to the full
scale canopy roof. The subsequent studies by Gumley [7,8] involved only
area-averaged measurements, and the nature and extent of the flow mechan-
isms over the roof were not investigated further. Furthermore, the correlation
of fluctuating pressures on various sections of the roof were not obtained. This
is the data necessary for determining the overall wind load on the structure
and with this data it would be possible to obtain more realistic design load
cases. The investigation undertaken at the University of Queensland [2] was
specifically aimed at covering these areas.
In passing, various configurations of partially open or grandstand roofs have
been studied in the wind tunnel by amongst others Barnard [12], Cook [13], and
Melbourne and Cheung [14]. Apperly and Pitsis [15] have made comparisons
0 . . . . :: Q . . . . . .
&
-0.5 . . . . . . . . . . . . . . . . . .
A c~
1.5 . . . . . . . . . . . .
&
between model and full scale pressure measurements on the Belmore Oval
grandstand in Sydney. While J a n c a u s k a s and Holmes [16] studied wind loads
on attached canopies.
3. Experimental arrangement
100 •j 100 •I
------ Theoretical
l ,I Experimental
80 •I Theoretical 80
I
,I •I
I Experimental
A 6O a/ ~60
vE ,f
I E
"I
v o~
N I
4O N40
I
t
20
{ 20 %.
g
/
0 l I 0--- [ I '%%1~--I I
0.5 1.0 1.5 2.0 0.1 0.2 0.3 0.4 0.5
015~om ouIU
Fig. 2. Mean velocity and turbulence intensity profiles.
C.W. Letchford, J.D. Ginger/Wind loads on planar canopy roofs 29
The model canopy roof used in this study was similar to that used by Gumley
[7,8], except th at a modular arrangement of panels was employed. A single
interchangeable panel with closely spaced pressure taps was employed for the
detailed pressure distribution investigation. Care was taken to ensure that
panel edges were sealed to prevent leakage flow between top and bottom
surfaces. The other main difference was the placement of the supporting legs
inboard of the eaves in an effort to reduce interference at the separation lines.
A 7 mm thick, 300 x 300 mm rigid canopy roof model as shown in Fig. 3 was
constructed. The model consisted of two halves hinged along the ridge line to
allow for variable pitch. Each half was divided into three 100 x 150 mm sections
each supported by a 12 mm square hollow leg. The six 150 x 100 mm sections
were identified as A to F and the 0 ~ wind orientation was with the flow
perpendicular to the ridge line.
The tapping distribution for the area-averaged pressure measurements is
shown in Fig. 4, while the interchangeable panel employed for the detailed
pressure study is shown in Fig. 5. Tappings locations were identical on top and
bottom surfaces so that net pressure differences could be obtained. Flexible
1.5 mm diameter PVC tubes from the pressure tappings were passed through
the hollow legs via 48 port Scanivalve Corporation type D pressure switches to
Honeywell 163PC pressure transducers fixed under the turntable. Both single
and manifolded tubing networks were employed.
The frequency response of the two tubing systems, one for measuring the
pressure at a single tapping and the other for measuring the area-averaged
300
LI I
tx
Ridge line
I
I
I
Jc_
• ~ ,~ 3s .... 2~ ~.25 ,
X X X X X X
T--
i.o
X X X X
x ~2~
!
x [-~,2 .,dge .he
550
~ 0 . 3 ~b reatrlctor
16 Pressure 360
t .... ducer ~65 L -Scanivalv.
14
1.2 4~ . • 270
o &
to
¢i
0.8 o
180 <=
• &
-=
a 06
E
< 0.
04 90 ,
• &
02
J I L [ I 0
50 100 150 200 250 300
Frequency, Hz
Manifold
o 12 • 2 7 0 o~
@
• "1o
~ 1.0
~ 0.8 e• 1 8 0 ==
"6O.6
E o
< 0.4
0.2
I I I 1 i 0
5o 100 150 200 250 300
Frequency, Hz
Fig. 7. F r e q u e n c y response of six i n p u t pressure m e a s u r i n g system.
pitch angles (~) of 0 '~, 5', 10 ~, 15 '~, 22.5:, and 3 0 for wind d i r e c t i o n s (fl) in steps of
30". The second set of tests d e t e r m i n e d the a r e a - a v e r a g e d pressure on each of
the six r o o f panels (A to F). Tests were carried out on c a n o p y ' s at heights (h) of
75 mm, 100 mm, and 125 mm and r o o f pitch angles (~) of if, 5 °, 10 ~', 15 '~, 22.5, and
30 '~ for wind d i r e c t i o n s (fl) in steps of 10 °. The final set of tests d e t e r m i n e d the
a r e a - a v e r a g e d pressures on smaller areas a d j a c e n t to the r o o f edges and ridge.
4. R e s u l t s
c( = 1 5 °
?
i -~= 0 o
Fig. 8. Mean pressure difference coefficient contours (~= 15°, fl=0 °, 30°, 60°, 90°).
C.W. Letchford, J.D. Ginger/Wind loads on planar canopy roofs 33
~=15 °
2'7 !;
~='~30 o
0~=15 °
(c/
//k/ /-
~n:6o °
Fig. 8. Continued.
34 C.W. Letchford, J.D. Ginger/Wind loads on planar canopy roofs
6=15 °
(d)
:>..__J
_J
j ~ J ~
I
~=90 °
Fig. 8. Continued.
o~ = 2 2 . 5 °
(a)
o o,
-13= 0 °
Fig. 9. Mean pressure difference coefficient contours (u=22.5:, fi=ff~, 3if, 6@', 9@).
©
HI J v ~ - ~ ~ ..... ~2....~--~
=J
II
0 - -
% o
36 C. W. Letchford, J.D. Ginger/Wind loads on planar canopy roofs
=22.5 °
% o
i= 90 °
Fig. 9. Continued.
which are carried away by the freestream. This process produces a fluctuating
reattachment point and leads to high suction pressures beneath the separation
bubble [20].
As the angle of attack is changed from 0 °, a mean velocity component in the
direction of the edge discontinuity arises, enabling the vorticity diffused from
the separating shear layer to be convected away along its axis. The balance
between the vorticity diffused and convected leads to the formation of a stable
flow mechanism, namely the conical vortex [2]. The largest suction pressures
are generated under the core of the vortex. Within the conical vortex the flow
spirals, as evidenced by flow visualization and surface oil flow patterns, with
the spiral angle increasing as the angle of attack increases. This is because the
velocity component in the direction of the edge discontinuity increases with
increasing angle of attack.
The regions influenced by the formation of the separation bubbles and
conical vortices on the bottom surface of the windward half and the top surface
of the leeward half can be identified in the contour plots for fl = 0 °, 30 °, and 60 °.
The largest pressures are experienced close to the inclined edges at an attack
angle of 30° . The pressure distribution caused by the interference of the
supports can also be identified in the contour plots. For fl = 90 °, the canopy is
effectively a flat plate and sensibily would experience negligible net pressure
difference.
C. W. Letchford, J.D. Ginger/Wind loads on planar canopy roofs 37
4.2. Area-averaged p r e s s u r e s
The v a r i a t i o n of m e a n a r e a - a v e r a g e d p r e s s u r e difference coefficients with
wind d i r e c t i o n on the six r o o f panels (A to F) for the v a r i o u s r o o f pitches is
s u m m a r i s e d for two cases: a c o r n e r r o o f p a n e l (F) and a middle r o o f p a n e l (E)
in Figs. 10 and 11, respectively. Ten r e p e a t e d m e a s u r e m e n t s of a c o r n e r p a n e l
p r e s s u r e difference g a v e a s t a n d a r d e r r o r of 5% for the mean.
As expected, for the c o r n e r r o o f p a n e l (Fig. 10) the l a r g e s t positive and
n e g a t i v e p r e s s u r e s o c c u r for the h i g h e s t r o o f pitches. Indeed, for r o o f pitches
below 15 ~ the s e p a r a t i o n bubble and the conical v o r t i c e s are small in s p a t i a l
e x t e n t with m o s t of the p a n e l e x p e r i e n c i n g r e a t t a c h e d flow. This is evidenced
in the p o i n t p r e s s u r e d i s t r i b u t i o n s of Figs. 8 and 9. M u c h the s a m e observa-
tions a p p l y for the middle r o o f p a n e l (Fig. 11) with only the m a g n i t u d e s of the
m e a n p r e s s u r e s reduced s o m e w h a t . In e a c h case the m e a n p r e s s u r e difference
for a n a n g l e of a t t a c k of 9 0 , i.e. p a r a l l e l to the ridge line, is p r a c t i c a l l y zero. No
c o m p a r i s o n s w i t h the r e s u l t s of G u m l e y [7,8] c a n be m a d e at this s t a g e b e c a u s e
of the choice of different panel areas.
An effort to q u a n t i f y the discrete a p p r o x i m a t i o n e r r o r in the a r e a - a v e r a g e d
m e a s u r e m e n t s h a s been m a d e by c o m p a r i n g p a n e l p r e s s u r e s w i t h those
o b t a i n e d by i n t e g r a t i n g p o i n t p r e s s u r e distributions. F i g u r e 12 p r e s e n t s this
c o m p a r i s o n for a c o r n e r p a n e l (F) for two r o o f pitches (15 and 22.5") and for
corner panel
1
o0.8 . . . i
. . . . ~
. . . . . . . . . . . . ii
. . . . . . . . . . . . . . . . . .. .. . . . . . . . . . . . . . .. .. . ..
o.°4 ...................................
........ !i
('~ 0.2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
-0.4-. . . . . . . . . . . . . . . . . . . . . . . .
-0.6. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
-1.2 3'0 6'0 9'0 120 150 1~}0 2i0 2,~0 270 300 330 360
angle of attack
Fig. 10. Mean area averaged pressure difference coefficients on a roof corner panel (F).
38 C.W. Letchford, J.D. Ginger/Wind loads on planar canopy roofs
middle panel
1 ,
0"8 I
o6!
0.4!
0.2-..
o_ oI~
0 -0.2- -
-0.4 -
-0.6
-0.8- •
-1-
-1,2
0 3o -~o 95.... 1~o ...........~so~- ~o
angle of attack
corner panel
1
0.8 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
0,6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
0.4 -"-'--~=. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
-1 ............................................................................................. !
-1.2
0 3'0 do 90 120 i 50 i so
angle of attack
I -m- 15, point -e~ 15, area --N-- 22.5, point -~- 22.5, area ]
half roof
1-
0.8-
0.6!
0.4-E
0.2
o.. 0-'
0 0.2
0.4
0.6
-0.8.
-1 . . . . . . . . . . . . .
-1.2
30 60 90 110 1.50 180
angle of attack
Fig. 13. Mean area averaged pressure difference coefficients on a half roof.
40 C. W. Letchford, J.D. Ginger/Wind loads on planar canopy roofs
angle of attack = 0
0.8
0.6
0.4"
0.2
-0.2;
-0.4-
-0.6.
-0.8"
-1- . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I
-1.2
-1'0 i5 2S ~5 30
roof pitch
angle of attack = 30
0.8 ¸
0.6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
° 4 iiiiiiiiiiiiiiiiiiiiiiiiiiiiii
0.2
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii
ol
3 -o.2;
-0.4 . . . . . . . . . . . 4
-0.6.
-0.8
-1
-1.2
roof pitch
present study, the mean half roof pressure difference coefficient is effectively
zero, whereas it is approximately - 0 . 1 for Gumley [7]. Gumley reasons this
negative offset to be due to the finite thickness of the model in a sheared
velocity flow, causing the top surface suctions to be larger by virtue of the
increased velocity over t ha t surface. This velocity increase is however very
small over a 7 mm thick plate.
C.W. Letchford, J.D. Ginger/Wind loads on planar canopy roofs 41
angle of attack = 60
0.8-
0.6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
0.4 ......................................................................
..... I
-0.2) I---= =_ ..............................
-0.4
-0.6
-0.8-
-I i
i
-1.2 5 1'0 1'5 2'0 2'5 3~0
roof pitch
angle of attack = 90
04 t
0,2 ..............................
-0.2
~ -_-2 -222-_ -~2-22-_- . . . . . ~. . . . . . :::::::::::::::::::::::
0"4i . . . . . . . . . . . . . . . . . . . . . . . . . .
-0.6 -
0 5 10 15 20 25 30
roof pitch
Figs. 14 and 17 must, therefore, be due to the presence of the supporting legs at
the eaves edge, i.e. the separation line. Of course, all canopies must have
supports, but the scale of the legs should be in proportion, and this was not the
case in either of these studies, although more significant in Gumley's.
The above discussion goes someway to explaining the difference between the
two data sets. It is believed t hat the remaining difference is substantially due
to the different boundary layer simulations that were employed in the studies.
Gumley's simulation was essentially t hat of a rural terrain, zo = 0.03 m, with
the turbulence intensity at eaves height being ~:20%. The present study was
for a rougher suburban terrain, zo = 0.2 m and ~ 25% turbulence intensity at
eaves height. The more t ur bul e nt flow is sure to lead to earlier r e a t t a c h m e n t
and, hence, smaller spatial extent of both separation bubble and conical
vortex. This implies that, especially for the leeward half roof, the present
study's pressure difference coefficients would be reduced in magnitude in
comparison with those of Gumley, which is indeed the case.
Edge
J
IO0
1%
L ,30
Fig. 18. Edge and ridge strip area-averaged pressure locations.
C.W. Letchford, J.D. Ginger/Wind loads on planar canopy roofs 43
angle of attack = 30
2
1.5-
1
J
0.5
J
0...
© o::-2.: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
_1 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . -77:~=:
0 5 10 15 20 25 30
roof pitch
- , , - edge zone maxima --~- edge zone minima - - * - ridge zone minima 1
only mean pressure coefficients are required. However, where the quasi-steady
approach breaks down, i.e. in regions of separated flow or for large structures,
modifying factors are required. These factors account for localised effects
within separated flow regimes and the lack of spatial correlation of the gusts
(and/or pressures) over the structure.
Figure 20 compares the values presented in ASl170.2 [9] for the canopy roof
without blockage with the worst case mean half roof pressure difference
coefficients obtained from Fig. 13. The code values have been modified by an
area reduction factor K a=0.8, based on the full scale half roof area being
greater than 100 m 2. In regions of near zero mean pressure the quasi-steady
theory breaks down completely and alternative approaches are necessary.
Such approaches require the measurement of peak pressures and will be
considered in part 2 of this paper [1].
5. C o n c l u s i o n s
half roof
0.8.
0.6-
0.4-
0.2- . . . . . . . . . . . . . . . . . . ..>-:'+?: . . . . . i
O . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i
0 -0.2- ......... ~ : z S : . . . . . . . .
-0.4 + i
-0.6-
-0.8-
-1- IZZIZZIII;I;IZIII;;II;II iZi?i?iiiil!!!!!!!!!!!iiiiii I
-1.2 . . . . .
0 g 1'0 t'5 20 i5 30
roof pitch
Fig. 20. Comparison of worst mean half roof pressure difference coefficients with values in
ASl170.2 [9] for a canopy roof with no blockage underneath.
Not surprisingly, comparisons with data in the Australian wind loading code
[9] revealed similar trends to those noted for the comparisons with Gumley's
[7,8] work. This is because the code was largely based on Gumley's w o r k and
must, by necessity, be conservative.
In part 2 of this paper [1], a more sophisticated approach to determining
realistic design loads, t hat of covariance integration incorporating the correla-
tion of fluctuating pressures over the complete roof structure, will be pres-
ented. With this data, re-evaluation of codified values for determining overall
uplift and drag on canopy roof structures can be made.
Acknowledgements
References
[1] C.W. Letchford and J.D. Ginger, Wind loads on planar canopy roofs-Part 2: Fluctuating
pressure distributions and correlations, J. Wind Eng. Ind. Aerodyn., (1992)to be published.
[2] J.D. Ginger and C.W. Letchford, Wind loads on canopy roofs, Research Report CE 132,
University of Queensland, 1991.
C.W. Letchford, J.D. Ginger/Wind loads on planar canopy roofs 45
[3] C.W. Newberry and K.J. Eaton, Wind Loading Handbook, Department of Environment,
HMSO, London, 1974.
[4] Code of Practice CP3, Code of basic data for the design of buildings, 2, Wind loads,
British Standards Institution, London, 1972, Ch.V.
[5] M. Jensen and N. Franck, Model-scale tests in turbulent wind: Part 1. The Danish
Technical Press, 1963.
[6] S.J. Gumley and C.J. Wood, Mean and extreme pressure study on model dutch barns,
OUEL Report No. 1333/80, University of Oxford, 1980.
[7] S.J. Gumley, Panel loading mean pressure study for canopy roofs, OUEL Report No.
1380/81. University of Oxford, 1981.
[8] S.J. Gumley, Design extreme pressures A parametric study for canopy roofs, OUEL
Report No. 1394/82, University of Oxford, 1982.
[9] Australian Standard, SAA loading code Part 2: Wind loads ASl170.2, Standards
Australia, Sydney, 1989.
[10] A.P. Robertson, R.P. Hoxey and P. Moran, A full-scale study of wind loads on agricul-
tural ridged canopy roof structures and proposals for design, J. Wind Eng. Ind. Aero-
dyn., 21 (1985) 167-205.
[11] Building Research Establishment Digest 284, Wind loads on canopy roofs, Building
Research Station, Garston, Warlord, UK, 1984.
[12] R.H. Barnard, Wind loads on cantilevered roof structures, J. Wind Eng. Ind. Aerodyn.,
8 (1981) 21 30.
[13] N.J. Cook, Reduction of wind loads on a grandstand roof, J. Wind Eng. Ind. Aerodyn., 10
(1982) 373 380.
[14] W.H. Melbourne and J.C.K. Cheung, Reducing the wind load on large cantilevered
roofs, J. Wind Eng. Ind. Aerodyn., 28 (1988) 401 410.
[15~ L.W. Apperly and N.J. Pitsis, Model/Fullscale measurements on a grandstand, J. Wind
Eng. Ind. Aerodyn., 23 (1986) 99 111.
[16] E.D. Jancauskas and J.D. Holmes, Wind loads on attached canopies, 5th US National
Conf. on Wind Eng., Lubbock Texas, November 1985.
[17] J.D. Ginger and C.W. Letchford, A computerized data acquisition system for the bound-
ary layer wind tunnel, Research Report CE 127, University of Queensland, 1990.
[18] T. Stathopoulos and D. Surry, Scale effects in wind tunnel testing of low buildings,
J. Wind Eng. Ind. Aerodyn., 13 (1983) 313 326.
[19] C.W. Letchford, P. Sandri, M.L. Levitan and K.C. Mehta, Frequency response require-
ments for fluctuating wind pressure measurements, J. Wind Eng. Ind. Aerodyn., 40 (1992)
263 276.
[201 P.J. Saathoff and W.H. Melbourne, The generation of peak pressures in separated/re-
attaching flows, J. Wind Eng. Ind. Aerodyn., 32 (1989) 121 134.