Professional Documents
Culture Documents
Abstract
Wind loads on free-standing canopy roofs have been studied in a wind tunnel. Three types of roof
geometries, i.e. gable, troughed and mono-sloped roofs, with roof pitches ranging from 01 to 151, were
tested. Wind pressures were measured simultaneously at many points both on the top and bottom
surfaces of the roof model for various wind directions. The paper describes the characteristics of overall
wind forces and moments acting on the roof with special attention to load combination. Correlation
between wind force and moment coefcients is investigated and wind force coefcients for the design of
main wind force resisting systems are proposed. The roof is assumed rigid and simply supported by four
corner columns, whose axial forces appear as the most important load effect. Two loading patterns that
cause the maximum tension and compression of columns are considered. The proposed values are also
compared with the specications of the Australian/New Zealand Standard [Standards Australia, 2002.
Australian/New Zealand Standard, AS/NZS 1170.2].
r 2007 Elsevier Ltd. All rights reserved.
Keywords: Free-standing canopy roof; Wind tunnel experiment; Overall wind force; Main wind force resisting
system; Design wind load; Codication
1. Introduction
The characteristics of local pressures on free-standing canopy roofs are described in a
companion paper, Part 1 (Uematsu et al., 2007). This paper presents the experimental
results on overall wind forces and moments acting on these roofs.
Corresponding author. Tel./fax: +81 22 795 7875.
ARTICLE IN PRESS
Y. Uematsu et al. / J. Wind Eng. Ind. Aerodyn. 96 (2008) 10291042
1030
Gumley (1984), Letchford and Ginger (1992) and Ginger and Letchford (1994)
measured area-averaged pressures for various areas using a pneumatic averaging
technique. Furthermore, Ginger and Letchford (1994) calculated the overall forces using
the covariance integration technique. Holmes (2001) discusses the application of their
results to load estimation. Recently, Letchford et al. (2000) measured the mean wind forces
on solid and porous canopy roof models using a force balance technique. This is perhaps
the rst attempt to measure the overall forces on canopy roofs directly. More recently,
Altman (2001) has made extensive measurements of the overall forces and moments using
a high-frequency force balance developed at Clemson University, USA.
In the present study, the wind forces and moments are computed from the time history
of pressures measured simultaneously at many points both on the top and bottom surfaces.
Such simultaneous pressure measurements leading to a reliable denition of realistic windinduced forces on canopy roofs have not been made so far, to the authors best knowledge.
Correlation between the wind force and moment coefcients is also investigated. Design
wind force coefcients are based on suitable load combinations for development of the
unbalanced load distribution. The analysis assumes that the roof is rigid and supported by
the four corner columns, and the axial forces induced in the columns are taken as the most
important load effect for determining the wind force coefcients.
2. Denition of wind force and moment coefcients
The notation and sign of the wind forces and moments are schematically illustrated in
Fig. 1, together with the coordinate system. The wind force and moment coefcients are
dened as follows:
CNW
NW
,
qH b l=2
(1a)
CNL
NL
,
qH b l=2
(1b)
+ NW
+ NL
b
+L
+ My
= 0
(Duo-pitched roof)
+N
+L + My
= 0
(Mono-sloped roof)
+Mz
= 0
+ My
+Mx
= 90
Fig. 1. Denition of wind force and moment coefcients and coordinate system. (a) Cross section; (b) Plan view.
ARTICLE IN PRESS
Y. Uematsu et al. / J. Wind Eng. Ind. Aerodyn. 96 (2008) 10291042
1031
CL
L
,
qH bl
(1c)
CN
N
,
qH b l
(1d)
CMx
Mx
,
qH bl 2
(1e)
CMy
My
,
qH b2 l
(1f)
CMz
Mz
,
qH bl 2
(1g)
where Mx, My and Mz are moments about the x-, y- and z-axis, respectively; qH is the
velocity pressure at the mean roof height H; and b* the actual length of the roof ( b/
cos b, with b being the roof pitch).
ARTICLE IN PRESS
Y. Uematsu et al. / J. Wind Eng. Ind. Aerodyn. 96 (2008) 10291042
1032
1.0
0.5
Present
Gumley (1984)
Letchford et al.(1992)
Letchford et al.(2000)
Altman (2001)
0.0
CNL (mean)
CNW (mean)
1.0
-0.5
Present
Gumley (1984,)
Letchford et al. (1992)
Letchford et al. (2000)
Altman (2001)
0.5
0.0
-0.5
-1.0
10
20
30
Roof pitch (deg)
40
10
20
30
Roof pitch (deg)
40
Fig. 2. Mean normal force coefcients on the windward and leeward halves of gable roofs (y 01). (a) Windward
half; (b) Leeward half.
0.0
2.0
CN (mean)
CN (mean)
1.5
-0.5
-1.0
1.0
0.5
Present
0.0
Gumley (1984)
Letchford et al. (2000)
Altman (2001)
-1.5
-0.5
0
10
20
30
Roof pitch (deg)
40
10
20
30
Roof pitch (deg)
40
Fig. 3. Mean normal force coefcients on mono-sloped roofs for y 01 and 1801. (a) Wind direction: 01;
(b) Wind direction: 1801.
ARTICLE IN PRESS
Y. Uematsu et al. / J. Wind Eng. Ind. Aerodyn. 96 (2008) 10291042
1
CNL (max & min)
3
CNW (max & min)
1033
2
1
0
0
-1
Present (max)
Present (min)
-2
-3
-1
0
10
20
30
Roof pitch (deg)
40
10
20
30
Roof pitch (deg)
40
Fig. 4. Maximum and minimum peak wind force coefcients on the windward and leeward halves of gable roof
(y 0451). (a) Windward half; (b) Leeward half.
0.1
CNL
0.0
-0.1
-0.2
-0.3
-0.4
-0.8
-0.4
0.0
CNW
0.4
0.8
the C N W value becomes an extreme in each run. The result indicates a poor correlation
between C N W and C N L . Similar features were observed for all cases tested. Therefore, a
combination of the peak values of C N W and C N L , which is often used in code provisions,
may overestimate the design wind loads, because they are not induced simultaneously.
The variation of CL, C M x and C M y (maximum, mean and minimum values) with wind
direction y is plotted in Fig. 6 for a gable roof and in Fig. 7 for a mono-sloped roof, both
with b 101. When yE01, the peak values of CL and C M y are generally large in
magnitude, while the value of jC M x j is relatively small. With an increase in y, the
magnitude of the peak values of CL and C M y decreases, while that of jC M x j increases. In
the mono-sloped roof case, the value of |CL(min)| and C M y max are large when yE1801.
Fig. 8 shows the coefcients of correlation (R) between the wind force and moment
coefcients, plotted as a function of y. When the wind direction is nearly normal to the
eaves (yE01 or 1801), the correlation between CL and C M y is relatively high and that
between CL and C M x is rather low. The coefcient of correlation between CL and C M y is
low for oblique winds.
Fig. 9 shows the variation of the maximum and minimum axial forces induced in the
columns with wind direction y for three roof geometries, where the axial force is reduced
ARTICLE IN PRESS
Y. Uematsu et al. / J. Wind Eng. Ind. Aerodyn. 96 (2008) 10291042
1034
0.05
1.0
0.00
mean
max
0.0
mean
CMy
CL
0.5
CMx
max
0.1
max
-0.05
min
0.0
mean
-0.1
-0.10
min
min
-0.5
-0.2
-0.15
0
30
60
90
Wind direction (deg)
30
60
90
Wind direction (deg)
30
60
90
Wind direction (deg)
1.5
1.0
0.5
0.0
-0.5
-1.0
-1.5
0.2
Maximum
Mean
Minimum
0.4
Maximum
Mean
Minimum
0.0
-0.1
60
120
180
Wind direction (deg)
0.2
0.1
0.0
-0.2
0
Maximum
Mean
Minimum
0.3
CMy
0.1
CMx
CL
Fig. 6. Variation of the statistics of CL, C M x and C M y with wind direction y: gable roof (b 101). (a) CL;
(b) CMx; (c) CMy.
-0.1
0
60
120
180
Wind direction (deg)
60
120
180
Wind direction (deg)
Fig. 7. Variation of the statistics of CL, C M x and C M y with wind direction y: mono-sloped roof (b 101). (a) CL;
(b) CMx; (c) CMy.
1.0
CL-CMy
0.5
R
0.5
R
1.0
CL-CMy
0.0
0.0
CL-CMx
CL-CMx
30
60
90
Wind direction (deg)
CL-CMx
-1.0
-1.0
-1.0
0.0
-0.5
-0.5
-0.5
CL-CMy
0.5
R
1.0
90
30
60
Wind direction (deg)
60
120 180
Wind direction (deg)
Fig. 8. Coefcient of correlation between the wind force and moment coefcients. (a) Gable roof (b 51);
(b) Gable roof (b 101); (c) Mono-sloped roof (b 51).
by qH(bl/4). Note that the axial forces induced in the columns are easily computed from the
time history of pressures and their inuence coefcients for the axial forces, because the
roof is assumed rigid and supported by four corner columns, as already mentioned. In the
case of gable and troughed roofs, the maximum axial force (maximum tension) occurs at a
wind direction ranging from 01 to approximately 451 and the minimum axial force
(maximum compression) occurs at yE01. In the mono-sloped roof case, the behavior in the
wind direction range from 01 to 901 is similar to that for the gable and troughed roofs. The
ARTICLE IN PRESS
max
0.5
0
min
-0.5
-1
0
max
0.5
0
min
-0.5
30
60
90
Wind direction (deg)
-1
0
2.0
1035
max
1.0
min
0.0
-1.0
-2.0
30
60
90
Wind direction (deg)
60
120
180
Wind direction (deg)
Fig. 9. Variation of the maximum and minimum axial forces with wind direction. (a) Gable roof (b 101);
(b) Troughed roof (b 101); (c) Mono-sloped roof (b 101).
WIND
WIND
WIND
Fig. 10. Equivalent pressure distribution on the windward and leeward halves caused by CL and C M y .
(2a)
C N L 0 C L 4C M y .
(2b)
Step 2: In order to consider the effects of C M x and wind direction on the axial forces
induced in the columns, a correction factor g, which is dened as the ratio of the actual
peak force for y 01451 to that computed from C N W 0 and C N L 0 , is introduced.
ARTICLE IN PRESS
Y. Uematsu et al. / J. Wind Eng. Ind. Aerodyn. 96 (2008) 10291042
1036
Step 3: The design wind force coefcients C N W and C N L , which give equivalent static
wind loads, are provided as follows:
C N W
gC N W 0
,
Gf
(3a)
C N L
gC N L 0
,
Gf
(3b)
CMy
0.0
0.0
-0.1
-0.1
0.25
CMy
0.1
0.1
0.20
CMy
where Gf represents a gust effect factor, which should be determined based on the load
effect.
Fig. 11 shows a phase-plane representation of the C L C M y relation for low roof
pitches and y 01. The circles represent the maximum and minimum peak values of CL
during each of the nine runs. The correlation between CL and C M y is generally high for
mono-sloped roofs. The value of C M y at the instant when the maximum CL (CL max) occurs
is nearly equal to the maximum C M y (C M y max ). This feature implies that the maximum
load effect is given by a combination of these peak values, i.e. the maximum tension is
given by C L max C M y max . Similarly, the maximum compression is given by
C L min C M y min ; the sufx min represents the minimum value of the coefcient under
consideration. When the roof pitch is relatively high, such as b 151, for example,
compression is no longer induced in any column. In the case of y 1801, the value of C M y
at the instant when CL min occurs is nearly equal to C M y max . In the gable and troughed roof
cases, on the other hand, the C L C M y correlation is relatively low and becomes lower as
the roof pitch increases. The peak+peak combination does not always give the maximum
load effect. The envelope of the C L C M y trajectory is approximated by a hexagon shown
in Fig. 12. The critical condition producing the maximum load effect may be given by one
of the apexes of the hexagon. In order to investigate the load combination effects, the axial
forces induced in the columns are computed for the six combinations of CL and C M y
(Points 16) in Fig. 12. Table 1 summarizes the conditions that give the maximum load
effect for each case. Load cases A and B represent the conditions producing the
maximum tension and compression in the columns, respectively. Note that the column
subjected to the maximum tension or the maximum compression depends on the roof
geometry. Substituting the CL and C M y values (for the cases provided in Table 1) into
Eq. (2), the values of C N W 0 and C N L 0 for each case are obtained.
0.15
0.10
CL
-0.2
-0.2
-0.4 -0.2 0.0 0.2 0.4 0.6
-0.2
0.05
CL
0.0
0.2
0.4
0.00
-0.05
-0.5
0.6
CL
0.0
0.5
1.0
Fig. 11. Phase-plane representation of the CLC M y relation (y 01). (a) Gable roof (b 51); (b) Gable roof
(b 101); (c) Mono-sloped roof (b 51).
ARTICLE IN PRESS
Y. Uematsu et al. / J. Wind Eng. Ind. Aerodyn. 96 (2008) 10291042
1037
Table 1
Combination of CL and C M y values producing the maximum load effect for gable and troughed roofs
Roof
Pitch (deg)
Load case
A
C L max C M y max
C L min C M y min
Gable
5
10
15
C L max C M y max
C L max C M y mean
C L max C M y mean
C L min C M y min
C L min C M y min
C L min C M y min
Troughed
5
10
15
C L max C M y max
C L max C M y max
C L max C M y max
C L min C M y mean
C L min C M y mean
C L min C M y mean
Flat
The correction factor g in Eq. (3) is obtained by calculating the ratio of the actual
maximum or the minimum axial force to the predicted value from C N W 0 and C N L 0 .
Fig. 13 shows the gust effect factor Gf, dened as the ratio of the maximum or the
minimum axial force to the mean value induced in the columns, plotted against the
mean reduced axial force N mean for y 01451 (W.D.1) and y 13511801 (W.D.2).
When the value of jN mean j is small, Gf exhibits a large value. However, as jN mean j increases,
the values of Gf collapse into a narrow range around Gf 2.0 (dashed line), which
corresponds to a peak factor of gv 3.0, based on the quasi-steady assumption, i.e.
GfE(1+gvIuH)2 (1+3.0 0.14)2 2.0. Therefore, Gf 2.0 is used for evaluating the
wind force coefcients hereafter.
Plotted on Figs. 1416 are the estimated values of C N W and C N L for gable and monosloped roofs (triangles). For comparative purposes, the maximum and minimum peak
values of C N W and C N L are plotted by circles; these values are divided by Gf to make a
direct comparison with C N W and C N L . Note that they are not induced simultaneously.
Furthermore, the specication of the Australian/New Zealand (AS/NZ) Standard
(Standards Australia, 2002) is also shown by the dashed lines. The wind force coefcients
on the windward half for the two load cases are consistent with the maximum and
minimum peak values. When yE01 (W.D.1), the values of C N W for load cases A and B are
nearly equal to those of C N W min and C N W max and close to the AS/NZ specication.
ARTICLE IN PRESS
Y. Uematsu et al. / J. Wind Eng. Ind. Aerodyn. 96 (2008) 10291042
1038
10
Flat (W.D. 1)
Gable (W.D. 1)
Troughed (W.D. 1)
6
Gf
Mono-sloped (W.D. 1)
Mono-sloped (W.D. 2)
Gf = 2.0
2
0
0
0.1
0.2
0.3
|N*mean|
Fig. 13. Gust effect factor based on the load effect.
0.5
1.0
AS/NZ (positive)
0.0
0.0
AS/NZ (positive)
CNL*
CNW*
0.5
AS/NZ (negative)
AS/NZ (negative)
-0.5
-0.5
Load case A
Load case B
Load case A
Load case B
Maximum
Minimum
Maximum
Minimum
-1.0
-1.0
0
10
15
Roof pitch (deg)
20
10
15
Roof pitch (deg)
20
Fig. 14. Equivalent static wind force coefcients C N W and C N L (gable roof, W.D.1). (a) Windward half; (b)
Leeward half.
0.5
0.5
AS/NZ (positive)
AS/NZ (positive)
0.0
-0.5
CNL*
CNW*
0.0
AS/NZ (negative)
-1.0
AS/NZ(negative)
-0.5
-1.5
Load case A
Load case B
Load case A
Load case B
Maximum
Minimum
Maximum
Minimum
-1.0
-2.0
0
10
15
Roof pitch (deg)
20
10
15
Roof pitch (deg)
20
Fig. 15. Equivalent static wind force coefcients C N W and C N L (mono-sloped roof, W.D.1). (a) Windward half;
(b) Leeward half.
Similarly, when yE1801 (W.D.2), the values of C N L of mono-sloped roofs for load cases A
and B are nearly equal to those of C N L min and C N L max . Note that C N L , not C N W ,
represents the wind force coefcient on the windward half when yE1801. Regarding the
ARTICLE IN PRESS
Y. Uematsu et al. / J. Wind Eng. Ind. Aerodyn. 96 (2008) 10291042
0.5
1.5
1039
AS/NZ (positive)
0.0
0.5
CNW*
CNL*
1.0
AS/NZ (positive)
0.0
-0.5
AS/NZ (negative)
-0.5
AS/NZ (negative)
Load case A
Load case B
Load case A
Load case B
Maximum
Minimum
Maximum
Minimum
-1.0
-1.0
0
10
15
Roof pitch (deg)
20
10
15
Roof pitch (deg)
20
Fig. 16. Equivalent static wind force coefcients C N W and C N L (mono-sloped roof, W.D.2). (a) Windward half;
(b) Leeward half.
0.5
CNW*(Load case A)
CNW*(Load case B)
CNW_Nmax/Gf
CNW_Nmin/Gf
0.5
CNL*, CNL_Npeak/Gf
CNW*, CNW_Npeak/Gf
1.0
0.0
-0.5
CNL*(Load case A)
CNL*(Load case B)
CNL_Nmax/Gf
CNL_Nmin/Gf
0.0
-0.5
0
10
15
Roof pitch (deg)
20
10
15
Roof pitch (deg)
20
Fig. 17. Comparison of C N W and C N L with the actual wind force coefcients producing the maximum and
minimum axial forces (gable roof, W.D.1). (a) Windward half; (b) Leeward half.
leeward half, on the other hand, the wind force coefcients for the two load cases are
similar to each other and close to either of the maximum or the minimum peak value. The
values are signicantly different from the AS/NZ specication. The results for troughed
roofs were found quite similar to those of the gable roofs.
eN and C
eN producing the maximum tension and
The actual wind force coefcients C
W
L
compression in the columns are obtained from the time history analysis. The results are
compared with the C N W and C N L values for the two load cases in Fig. 17; the values of
eN and C
eN are obtained by applying ensemble average to the results of nine runs and
C
W
L
then dividing by Gf. Results are consistent with each other. Therefore, the proposed values
of C N W and C N L correspond approximately to the pressure distribution that causes the
maximum load effect.
The maximum shear force induced in the columns is obtained from the time history
analysis and compared with the predicted values from C N W and C N L together with
Gf 2.0. The calculation of the shear force considers the torsional moment Mz as well as
the drag. The results for gable and troughed roofs are shown in Fig. 18, which indicates
that the predicted values from C N W and C N L capture the actual maximum shear force
ARTICLE IN PRESS
Y. Uematsu et al. / J. Wind Eng. Ind. Aerodyn. 96 (2008) 10291042
1040
0.4
Reduced shear force
0.4
Load case A
0.3
Load case B
0.2
Time history
0.1
0
-0.1
Load case A
0.3
Load case B
0.2
Time history
0.1
0
-0.1
10
15
Roof pitch (deg)
20
10
15
Roof pitch (deg)
20
Fig. 18. Maximum reduced shear forces induced in the columns (W.D.1). (a) Gable roof; (b) Troughed roof.
1.5
3.0
Present (Max)
Present (Max)
1.0
2.0
0.5
N*/Gf
N*/Gf
Present (Min)
AS/NZ (max)
0.0
AS/NZ (min)
AS/NZ (max)
Present (Min)
1.0
0.0
-0.5
AS/NZ (min)
-1.0
-1.0
0
10
15
20
Roof pitch (deg)
25
30
10
15
20
Roof pitch (deg)
25
30
Fig. 19. Comparison for axial forces between the predicted values from C N W and C N L and from the AS/NZ
specications (W.D.1). (a) Troughed roof; (b) Mono-sloped roof.
reasonably well. In fact, the maximum shear force obtained from the time history analysis
lies between the two predicted values corresponding to the load cases A and B.
Finally, the axial forces induced in the columns are computed by using C N W and C N L for
the two load cases and compared with those predicted from the AS/NZ Standard. Sample
results are shown in Fig. 19. The AS/NZ Standard generally provides two values of the
wind force coefcients for each of the windward and leeward halves and this results in four
combinations of C N W and C N L . The maximum and minimum axial forces among the four
values from the AS/NZ Standard are generally consistent with the results for the load cases
A and B, respectively, in spite of the difference in the wind force coefcients (see Figs.
1416). Consequently, the proposed values of C N W and C N L can be used for design
purposes. Table 2 summarizes the proposed wind force coefcients (tentative), which are
obtained from a simplied model of the variation of the wind force coefcients with roof
pitch b.
In the present study, it is assumed that the roof is supported by four corner columns, and
the axial forces induced in the columns are taken as the load effect for discussing the wind
force coefcients. These coefcients can also be used for the cases where the roof is
supported by more columns, say six, because the critical condition may be given when
ARTICLE IN PRESS
Y. Uematsu et al. / J. Wind Eng. Ind. Aerodyn. 96 (2008) 10291042
1041
Table 2
Wind force coefcients C N W and C N L
Roof
Pitch (deg)
Wind direction
Load case A
C N W
Flat
Load case B
C N L
C N W
C N L
W.D.1
0.40
0.10
0.20
0.10
Gable
5
10
15
W.D.1
W.D.1
W.D.1
0.40
0.20
0
0.12
0.33
0.55
0.35
0.50
0.65
0.08
0.27
0.45
Troughed
5
10
15
W.D.1
W.D.1
W.D.1
0.49
0.58
0.68
0.21
0.33
0.44
0.13
0.07
0
0.27
0.43
0.60
Mono-sloped
5
10
15
W.D.1
W.D.1
W.D.1
0.65
0.90
1.15
0.07
0.03
0
0.13
0.07
0
0.07
0.03
0
Mono-sloped
5
10
15
W.D.2
W.D.2
W.D.2
0.07
0.03
0
0.27
0.13
0
0.13
0.17
0.20
0.47
0.73
1.00
yE01 or 1801 irrespective of the number of columns, and the values of C N W 0 and C N L 0 are
determined based on a combination of CL and C M y when y 01 or 1801. Furthermore, the
effect of C M x on the axial forces, which is considered by the correction factor g, becomes
largest when the roof is supported by four columns. Therefore, the wind force coefcients
proposed in the present study will be conservative in so far as they may somewhat
overestimate the wind-induced loads in the columns when the number of columns is more
than four.
5. Concluding remarks
The overall wind forces and moments acting on free-standing canopy roofs have been
investigated based on a series of wind tunnel experiments. The study provides basic data
for the establishment of design wind loads for this type of roofs. Two load cases yielding
maximum tension and compression in the columns are considered. Based on the
combination of CL and C M y , design wind force coefcients C N W and C N L , which give
equivalent static wind loads, are proposed as a function of roof pitch from 01 to 151. Axial
forces produced by using these coefcients are consistent with those obtained from the
AS/NZ Standard.
Further research is necessary to establish design wind force coefcients for higher roof
pitches.
Acknowledgments
This study was made during the rst authors appointment at the Department of
Building, Civil and Environmental Engineering, Concordia University, Montreal, Quebec,
Canada, as a visiting professor, supported by the Ministry of Education, Culture, Sports,
ARTICLE IN PRESS
1042
Science and Technology, Japan, for the period May 2002February 2003. The authors are
much indebted to Mr. Kai Wang, graduate student of Concordia University, for his
assistance with the experiments.
References
Altman, D.R., 2001. Wind uplift forces on roof canopies. M.Sc. Thesis, Clemson University, NC, USA.
Ginger, J.D., Letchford, C.W., 1994. Wind loads on planar canopy roofs, Part 2: uctuating pressure distributions
and correlations. J. Wind Eng. Ind. Aerodyn. 51, 353370.
Gumley, S.J., 1984. A parametric study of extreme pressures for the static design of canopy structures. J. Wind
Eng. Ind. Aerodyn. 16, 4356.
Holmes, J.D., 2001. Wind Loading of Structures. Spon Press.
Letchford, C.W., Ginger, J.D., 1992. Wind loads on planar canopy roofs, Part 1: mean pressure distributions.
J. Wind Eng. Ind. Aerodyn. 45, 2545.
Letchford, C.W., Row, A., Vitale, A., Wolbers, J., 2000. Mean wind loads on porous canopy roofs. J. Wind Eng.
Ind. Aerodyn. 84, 197213.
Standards Australia, 2002. Australian/New Zealand Standard, AS/NZS 1170.2.
Uematsu, Y., Stathopoulos, T., Iizumi, E., 2007. Wind loads on free-standing canopy roofs, Part 1: local
pressures, J. Wind Eng. Ind. Aerodyn., to appear.