Professional Documents
Culture Documents
Abstract: Wind tunnel tests of open-frame, low-rise buildings were carried out to determine the drag 共base shear兲 and bracing loads in
the direction normal to the frames 共parallel to the ridge兲. In total, 18 configurations were examined in an open country terrain at a scale
of 1:100. The worst wind angles for all configurations are between 0°–40° with 20°–30° typically yielding slightly higher loads, 0° being
Downloaded from ascelibrary.org by MARRIOTT LIB-UNIV OF UT on 12/01/14. Copyright ASCE. For personal use only; all rights reserved.
parallel to the ridge. The largest load coefficients are observed for the smallest frame buildings, consistent with observations for enclosed
buildings, which is due to three-dimensional 共edge兲 effects. The solidity ratio has a clear effect on the load coefficients with higher
coefficients for lower solidity, similar to the behavior observed on lattice frames or trussed towers. However, when these coefficients are
multiplied by the solidity ratio, so that they can be directly compared to enclosed building coefficients, it is clear that the total load
increases monotonically with solid area. Bracing was observed to take up to 75% of the total drag load. An empirical model was
developed for obtaining design loads.
DOI: 10.1061/共ASCE兲ST.1943-541X.0000082
CE Database subject headings: Wind loads; Buildings, low-rise; Aerodynamics; Standards and codes; Frames.
Author keywords: Wind loads; Buildings, low-rise; Aerodynamics; Standards and codes.
90°
End Elevation
Frame Width:
0° 180°
Sx, Bx
6
1
(a)
Downloaded from ascelibrary.org by MARRIOTT LIB-UNIV OF UT on 12/01/14. Copyright ASCE. For personal use only; all rights reserved.
Roof Plan
3.7, 6.1 and 9.1 m
Side Elevation
lowest value of solidity ratio is due to the bare frames 共with though the integral scale is slightly small compared to the target
roofs兲, but additional values were obtained with 共1兲 a wider frame values, as is usual in wind tunnel testing of low-rise buildings.
member 共with a slightly higher solidity兲; 共2兲 the gable ends filled
in; and 共3兲 the entire end walls filled in. These additions were
Data Acquisition Parameters
made with modular parts which could be added to the frames as
required. Photographs of the model details can be found in Fig. 2, The load cell measurements were repeated for 19 wind angles
with further details in Kopp et al. 共2008兲. from 0°–180°, as defined in Fig. 1. Horizontal loads from the
No testing was performed with obstructions underneath the balance Sx and Sy as well as the horizontal bracing loads Bx were
building since the worst case for the horizontal loads is expected
to be with the building open. It should be noted that this is not the
case for uplift on the roof, but this is not a subject considered in 16
the current study. 14
12
Terrain Simulation 10
Elevation (m)
Wind angles 共degrees兲a 0°–180° in 10° increments where VH = the mean 共1 h兲 wind speed at the eave height and Ao
Sampling frequency 400 Hz = the blocked area. Table 1 lists Ao for each configuration. One
Low-pass filter cut-off frequency 10–20 Hz could also define the base shear coefficient with the actual total
Sampling time 100–250 s area enclosed by the end frame 共i.e., the total area of the end wall
Reference wind tunnel speedb 7.6–18.3 m/s if it were present兲; this coefficient is CSx · , where = Ao / A is the
Model scale 1:100 solidity ratio.
Downloaded from ascelibrary.org by MARRIOTT LIB-UNIV OF UT on 12/01/14. Copyright ASCE. For personal use only; all rights reserved.
Velocity scale 1:7 to 1:3 The horizontal component of the bracing force is normalized
Time scale 1: 0.030 to 1: 0.070 in the same way such that
a
See Fig. 1 for the definition of wind angle. Bx
b
Actual wind tunnel speed varied within 5% of the nominal reference CBx = 1 共3兲
2 VHAo
2
level speed. Load coefficients were normalized using the actual wind
speed measured during the run. These force coefficients can be used with the ASCE 7-05 once
they have been converted to that format. Following the analysis of
measured. The sign convention is depicted in Fig. 1. The data St. Pierre et al. 共2005兲 and Kopp et al. 共2005兲, the wind tunnel
were sampled at a rate of 400 Hz, but digitally low-pass filtered in load coefficients can be converted to equivalent GC p values as
the range of 10–20 Hz, depending on the configuration. As well, A
q HC Sx o
the model resonant frequencies fell within measurement range 共GCp兲eq = = FwtC Sx 共4兲
and were band-pass filtered. The load cell experiments were chal- q10 m,3 s ztKhKdIAo
K
lenging since the peak loads depend on the configuration, requir- where 共GC p兲eq=equivalent wind tunnel pressure coefficient;
ing each configuration to be run at a different wind tunnel speed qH=wind pressure at the eave height; q10 m,3 s=basic wind pres-
to maximize the range used on the transducers to minimize ex- sure in ASCE 7-05; Kzt=topographic factor; Kh converts the wind
perimental, measurement uncertainty. Thus, in order to ensure pressure to eave height; Kd=directionality factor; and
accurate measurements, the wind tunnel speed was altered for I=importance factor. The factors Kzt and I were set to unity. Kd
each configuration. This leads to changes in the velocity scale for was also set to unity since worst coefficients are considered in this
each model, which of course alters the time scale, since T analysis. Thus, all peak load coefficients in this report can be
= L / V, where T is the time scale, L is the length scale, and V converted to 共GC p兲eq by multiplying by Fwt. One can then use
is the velocity scale. The ranges of velocity and time scales are ASCE 7-05 formulation together with the current measured coef-
given in Table 2, with the time scales varying in such a way that ficients for design.
the equivalent full-scale sampling period was held to 1 h.
The maximum blockage ratio 共i.e., ratio of blocked area to
total wind tunnel cross-sectional area兲 for the models was in con- Results and Discussion
figurations 10, 14, and 18 and was about 0.2%. The minimum
Reynolds number, based on the eave height H was 20,000. Full-
scale Reynolds numbers would be larger by the length scale mul- Basic Aerodynamics
tiplied by the velocity scale, so the present experiments are about
Fig. 4 shows the mean CSx and standard deviation C of the base
Sx
three orders of magnitude low. This is consistent with wind tunnel
shear in the x-direction 共normal to main axis of the frames兲 for
test procedures for low-rise buildings 共ASCE 1999兲 and is not
nine configurations versus wind angle in order to get a basic
expected to have a significant impact on overall 共structural兲 loads.
understanding of the aerodynamic load coefficients. Recall that
The peak load coefficients presented in the study are not the
Table 1 lists the configuration details, but that Configurations 1–3
absolute worst coefficients recorded within the sample time, but
are for the W = 12.2-m wide frame, Configurations 4–6 are for the
are formed peaks using a peak factor such that
W = 30.5-m wide frame and Configurations 8, 12, and 16 are for
= C + gC
the 21.3-m wide frame. These results are for the minimum solid-
C Sx Sx Sx 共1兲
ity tested for each of these frames. Several observations can be
where the “^” indicates the peak value; the overbar is the mean; made.
the “ ⬃ ” = standard deviation; and g = suitable peak factor. Given First, the worst wind angle for all configurations is between
the variable velocity scales that were required to obtain the data, 0°–40° with 20°–30° typically yielding slightly higher load coef-
this is deemed a more reliable and consistent way to handle the ficients. The reason for this is the combination of the mean load
data. Throughout this work, a value of g = 3.8 has been assumed. coefficients 共which contribute roughly 1/2 of the peak load coef-
The peak factor of 3.8 used in the formed coefficients is de- ficients兲 and the fluctuations, both of which tend to be highest for
rived from the theoretical result first developed by Davenport winds from 20°–30°. The mean values go through values of 0 at
共1964兲. Using a peak factor provides an average value of the peak 90°, as they should, while the RMS 关we use RMS 共for root-mean-
forces, which is appropriate for design coefficients. The theoreti- square兲 to denote the standard deviation throughout兴 values are
cal value of the peak factor ranges from 3.5 to 4.0 depending on also minimum at 90°. There is a clear similarity in the variation of
the rate of fluctuations in the load. The value used in this study is the mean and RMS particularly for the critical wind angles.
judged to be appropriate for this class of structure and wind flow Second, for all frame sizes, the number of frames is an impor-
field. tant factor since the load is observed to increase monotonically
0 0.0
0 15 30 45 60 75 90
(a) win d an g le (d e g re e s )
Downloaded from ascelibrary.org by MARRIOTT LIB-UNIV OF UT on 12/01/14. Copyright ASCE. For personal use only; all rights reserved.
-0.5
2.5 0 30 60 90 120 150 180 210 240 270 300 330 360
w ind
nd ang
ngle (degr ees)
2.0
Fig. 6. Peak base shear coefficients from the NIST aerodynamic
1.5 database for enclosed, low-rise buildings
1.0
0.5
tested, only two frame sizes have the same solidity ratio so that
0.0
one has to infer the effects of frame size at slightly different
0 15 30 45 60 75 90 solidity ratios. Nevertheless, the effects of the frame size are
(b) win d an g le (d e g re e s ) clearly observed and the largest load coefficients are observed for
W=12.2m; 9 frames W=12.2m; 6 frames the smallest frames. This is an important observation since it im-
W=12.2m; 3 frames W=30.5m; 9 frames plies that there would be significant challenges in modeling the
W=30.5m; 6 frames W=30.5m, 3 frames loads for this type of building as an open lattice structure for
W=21.3m; 3 frames W=21.3m; 6 frames which it is assumed that horizontal forces are added up over the
W=21.3m; 9 frames
blocked areas using drag coefficients and shielding factors.
Fig. 4. Effects of wind angle on base shear coefficients in the There does not appear to be an easy explanation for this be-
x-direction for low solidity frames: 共a兲 mean; 共b兲 RMS coefficients havior, but it does appear to be consistent with drag on enclosed
buildings 共e.g., Nywening 2006兲 which may be due to the rela-
tively larger end effects on the surface pressures for smaller build-
with the number of frames. Third, for a constant number of ings. This immediately suggests that an empirical load model
frames, the size of the frame is also an important factor, with the based on enclosed low-rise buildings holds some promise. To
smallest frames having the largest load coefficients 共although this examine this further, Fig. 6 shows an analysis of C for enclosed Sx
effect is skewed slightly by the fact that the solidity ratio is not buildings from the NIST aerodynamic database 共Ho et al. 2005;
held constant in this data兲. These points will be examined in Nywening 2006兲. These data exhibit the same characteristics as
greater detail below.
the open-frame buildings in a few ways. First, the worst wind
direction is in the range of 0° to ⫾ 30°. Second, there is a clear
Effects of Building „Frame… Size trend of the largest buildings having the smallest load coefficients.
for models with For example, the largest C for the W = 12.2 m and H = 3.7 m
Fig. 5 shows the peak, formed coefficients C Sx
Sx
six frames as a function of frame width. For the configurations enclosed building is 2.0, while for the W = 24.4 m and H
= 7.3 m building it is 1.7. From the pressure data analyzed in
Nywening 共2006兲, it appears that this is due to relatively larger
14 three-dimensional effects around the edges of the smaller build-
W= 12.2m
ings. These measured values will be compared to the current
12
W= 21.3m
共open frame兲 coefficients in greater detail in the following sub-
0 degrees
10 10 degrees section.
W= 30.5m
W= 21.3m
20 degrees It is interesting to compare the ratios of
C for the different
Sx
8 30 degrees
兩
sized buildings 共i.e., C
Sx W=12.2 m / CSx 兩W=21.3 m兲, as they are rela-
6 tively constant for any two building sizes, regardless of the num-
ber of frames. In fact, the ratio C 兩 兩
/C is
Sx W=12.2 m Sx W=21.3 m
4
0.0 0.1 0.2 0.3
between 1.3 and 1.4 for three, six, and nine frames. In addition,
兩
so
solidit y rat io the ratio C Sx W=30.5 m / CSx 兩W=21.3 m is between 0.7 and 0.8 for
three, six, and nine frames. This has significant consequences for
Fig. 5. Effects of building width W on the peak base shear for a a load model because it means that individual parameters for
building length of six frames as a function of solidity ratio 共Configu- frame size and number of frames 共building length兲 are indepen-
rations 2, 5, 11, and 12兲 dent.
2.0
1.5
1.0
0.5
0.0
0.0 0.2 0.4 0.6 0.8 1.0 (a)
solidit y rat io
so
Downloaded from ascelibrary.org by MARRIOTT LIB-UNIV OF UT on 12/01/14. Copyright ASCE. For personal use only; all rights reserved.
W 共m兲 兩 /C
C 兩
Sx W Sx 21.3 m
12.2 1.4
21.3 1.0
30.5 0.8
and
Downloaded from ascelibrary.org by MARRIOTT LIB-UNIV OF UT on 12/01/14. Copyright ASCE. For personal use only; all rights reserved.
Fig. 10. Solidity factor Ks as a function of solidity ratio and num- Ks = 0.2 + 0.073共n − 3兲 + 0.4 exp共1.5兲 共9兲
ber of frames, for W = 21.3 m. The lines are obtained via Eq. 共9兲. where n = number of frames and = solidity ratio. These design
loads apply only to the base shear parallel to the ridge and per-
pendicular to the frames for winds in the range 0°–45°. Using the
size W = 21.3 m, for which most of the testing was done, Bx /
Sx is loads defined by Eqs. 共7兲–共9兲, it is recommended that 75% of the
in the range of 0.5 to 0.75, with an overall average value for all
configurations over the critical wind angle range of 0.6. Thus, the total base shear 共drag兲
Sx be used as the basis for the design of the
bracing takes roughly 1/2 to 3/4 of the total horizontal load in the bracing members. The justification for Eqs. 共8兲 and 共9兲 are given
x-direction 共perpendicular to the pinned frames兲. below.
The frame width factor Kw is defined to account for the effects
of frame or building width that are observed in the current experi-
mental data 共e.g., Fig. 5兲. This term is needed since such effects
Empirical Model for Base Shear Perpendicular
are not currently accounted for in the ASCE 7 for enclosed build-
to Frames
ings. For enclosed buildings, the edge effects are accounted for by
The major parameters which affect the wind-induced base shear the length of the edge zone a which has the same proportion of
loads on low-rise, open-frame buildings in the direction perpen- wall area for all three frame widths so that the net shear in the
dicular to the frames are frame size 共width and height兲, number of x-direction is the same. Interestingly, there are building width
frames, and solidity ratio. Analysis of the variations of the load effects noted for the enclosed buildings in the NIST database, as
coefficients holding any two of these parameters constant and can be seen in Fig. 6, but these are not to the same extent as
varying the third suggests that multiplicative factors accounting observed for the open buildings and they are not accounted for in
for the load effects of these parameters would lead to a reasonable the code either. Table 3 indicates the ratio of the peak base shears
loading model. In addition, both the magnitude of the loads and for the three widths in the current study, relative to the measure-
the aerodynamic behavior appear to have more in common with ments for W = 21.3 m, which was the case for which most of our
enclosed low-rise buildings than with open lattice structures such measurements were taken. Given that this variation is not ac-
as electrical power transmission towers. Therefore, using en- counted for in ASCE 7-05, choosing the magnitude of Kw is not
closed building drag loads is a reasonable starting point for con- straightforward since it involves making a detailed comparison of
structing a load/design model for open-frame, low-rise buildings. the loads for different sized enclosed structures and then compar-
The ASCE 7-05 defines the pressure loads due to wind as ing this with the code provisions. Such an analysis was performed
by St. Pierre et al. 共2005兲 using a portion of the NIST database
p = 共GCp兲q10KztKdKhI 共5兲 and comparing with the original source data for the current low-
rise wind load provisions 关e.g., Stathopoulos 共1979兲兴, as well as
where GCp = pressure coefficient appropriate to the building type
with the current ASCE 7 coefficients. St. Pierre et al. found that,
and size; q10 = design wind pressure; Kzt = topographic factor; Kd
in general, the coefficients in the current provisions are low com-
= wind directionality factor; Kh = velocity pressure exposure factor
pared to current measurements and analysis. To deal with this, we
evaluated at the mean roof height; and I = importance factor. For
made the choice of making Kw a relative coefficient which ac-
the current building shapes and sizes, Fig. 6–10 in ASCE 7-05
counted for the relative variation over the range of buildings
defines GCp values, with coefficients from the end walls, parallel
tested. Since the data that went into the code for low-rise build-
to the ridge being appropriate. Then, the total base shear 共parallel
ings was based largely around building widths of 20 to 30 m, we
to the ridge兲 for the enclosed building is
chose to use W = 21.3 m as our reference point, so that Kw = 1 for
Sx = 共pwindward − pleeward兲A 共6兲 W = 21.3 m. Eq. 共8兲 gives the value of the factor as a function of
frame width W where an upper bound is given since it is expected
where A = area of the entire end wall; pwindward = average pressure that for much larger buildings, the effects of the edge will not
on the windward end wall; and pleeward = average pressure on the further reduce the loads compared to smaller buildings. In the
leeward end wall. To modify this for open-frame buildings, we range of W from ⬃6 to ⬃30 m, a linear variation in Kw matches
introduce two multiplicative factors: Kw, which accounts for the data reasonably well, given the resolution of building widths
building width effects; and Ks, which accounts for both the solid- that we have. Thus, the magnitude of the overall loading is made
ity ratio and number of frames. These coefficients are applied to to match the experimental observations using the second coeffi-
Eq. 共5兲 such that cient Ks.
The values of the factor Ks defined by Eq. 共9兲, account for the
p = 共GCp兲q10共KztKdKhI兲共KwKs兲 共7兲
variation of loads due to solidity ratio and the number of frames,
with as well as enforcing the magnitude to the correct value relative to
References
Conclusions ASCE. 共1999兲. “Wind tunnel studies of buildings and structures.” ASCE
manuals and reports on engineering practice No. 67, ASCE, Reston,
The objectives of the present study were to determine the total Va.
ASCE. 共2006兲. “Minimum design loads for buildings and other struc-
base shear and bracing loads in the direction normal to the frames
tures.” ASCE 7-05, Reston, Va.
共parallel to the ridge兲 for open-frame buildings and develop an Bayar, D. C. 共1986兲. “Drag coefficients on latticed towers.” J. Struct.
appropriate empirical model for use in design. In total, 18 con- Eng., 112, 417–430.
figurations consisting of three different frame sizes in the range of Davenport, A. G. 共1964兲. “Note on the distribution of the largest value of
widths from 12 to 30 m and heights from 3.6 to 9.1 m, with a random fluctuation with application to gust loading.” Proc.- Inst.
buildings consisting of three, six, and nine frames for several Civ. Eng., 28, 187–196.
solidity ratios were examined in an open country terrain at a scale Engineering Sciences Data Unit 共ESDU兲. 共1974兲. “Characteristics of at-
of 1:100. mospheric turbulence near the ground.” ESDU data item 74031,
The main conclusions of the work are: ESDU.
1. The worst wind angle for all configurations is between Engineering Sciences Data Unit 共ESDU兲. 共1982兲. “Strong winds in the
0°–40° with 20°–30° typically yielding slightly higher load atmosphere boundary layer. Part 1: Mean-hourly wind speeds.” ESDU
coefficients. data item 82026, ESDU.
Engineering Sciences Data Unit 共ESDU兲. 共1983兲. “Strong winds in the
2. The largest load coefficients are observed for the smallest
atmosphere boundary layer. Part 2: Discrete gust speeds.” ESDU data
frames. This is consistent with observations on enclosed
item 83045, ESDU.
buildings where it is due to the relatively larger influence of
Georgiou, P. N. 共1979兲. “A study of the wind loads on building frames.”
edge effects for the smaller buildings. This contrasts with MESc thesis, Univ. of Western Ontario, London, Ont.
observations for open, lattice structures and indicates that a Georgiou, P. N., and Vickery, B. J. 共1979兲. “Wind loads on building
load model based on low-rise enclosed buildings is appropri- frames.” Proc., 5th Int. Conf. on Wind Engineering, Fort Collins,
ate for design. Colo., 421–433.
3. The solidity ratio has a clear effect on the load coefficients Ho, T. C. E., Surry, D., Morrish, D., and Kopp, G. A. 共2005兲. “The UWO
with higher coefficients for lower solidity 共with the drag contribution to the NIST aerodynamic database for wind loads on low
forces normalized by the blocked area of the end frame兲. buildings. Part 1: Archiving format and basic aerodynamic data.” J.
These results are similar in behavior to lattice tower coeffi- Wind Eng. Ind. Aerodyn., 93共1兲, 1–30.
cients, although the magnitudes are different. When these Kopp, G. A., Galsworthy, J., and Oh, J. H. 共2008兲. “Wind loads on open-
coefficients are multiplied by the solidity ratio so that they frame buildings.” Rep. No. BLWT-SS4-2008, Boundary Layer Wind
can be directly compared to enclosed building coefficients, it Tunnel Laboratory 共BLWTL兲, London, Ont.
Kopp, G. A., Mans, C., and Surry, D. 共2005兲. “Wind effects of parapets
is evident that the total load increases monotonically with
on low buildings. Part 2: Structural loads.” J. Wind Eng. Ind. Aero-
solidity and are of similar order of magnitude as enclosed
dyn., 93共11兲, 843–855.
buildings. Mara, T. 共2007兲, “The effects of multi-directional winds on lattice sec-
4. For open-frame buildings with entirely filled-in end walls, tions.” MESc thesis, Univ. of Western Ontario, London, Ont.
the total base shear is about twice that observed for a similar Nywening, M. 共2006兲, “The effects of geometry on wind loads of low-
enclosed building due to the additional exposed surfaces 共end rise buildings.” MESc thesis, Univ. of Western Ontario, London, Ont.,
walls兲 for the open building. Buildings with solidity ratios of Canada.
about 0.16 have similar base shears 共perpendicular to frames兲 St. Pierre, L. M., Kopp, G. A., Surry, D., and Ho, T. C. E. 共2005兲. “The
as enclosed buildings. UWO contribution to the NIST aerodynamic database for wind loads
5. Loads increase linearly with number of frames 共except per- on low buildings. Part 2: Comparison of data with wind load provi-
haps for the case of the filled in end walls兲. sions for low buildings.” J. Wind Eng. Ind. Aerodyn., 93共1兲, 31–59.
Stathopoulos, T. G. 共1979兲, “Turbulent wind action on low rise build-
6. The ratio of the horizontal component of the bracing force to
ings.” Ph.D. thesis, Univ. of Western Ontario, London, Ont., Canada.
the base shear is in the range of 50 to 75% of the total Uematsu, Y., Iizumi, E., and Stathopoulos, T. 共2007兲. “Wind force coef-
horizontal load. ficients for designing free-standing canopy roofs.” J. Wind Eng. Ind.
An empirical load model for use in design was developed Aerodyn., 95共9–11兲, 1486–1510.
based on the ASCE 7-05 model for loads on enclosed, low-rise, Uematsu, Y., and Stathopoulos, T. 共2003兲. “Wind loads on free-standing
gable roof buildings. Two additional coefficients were required canopy roofs: A review.” J. Wind Eng., 95, 245–256.