AT REFERENCE
LIBRARY
A project of Volunteers in Asia
Building To Resist
The Effect Of WInd
In five volumes
Sponsored by:
::.;':
i~hl:~;,;~ ::, f;' : ,
Library of Congress Catalog Card Number: 77600013
National Bureau of Standards Building Science Series 1002
!\lat. Bur. Stand. (U.S.), Bldg. Sci. Ser. 1002, .~9 .pages (May 1977)
CODEN: BSSNBV
REFERENCES ........................................................................... 9
APPENDIX A
Illustrative Example ................................................................... 22
Comment ............................................................................ 23
FIGURES
Fig. 1 Ratio, r, of Maximum Probable Wind Speeds
Averaged over t seconds to those Averaged over 2 sec ............................... 11
Fig. 2 Quantity B.... ....... " ..................... , .. " ................................ 111
Fig. 3 Probability Plots;
(a) Type II Distribution, 'Y = 2 ..................................................... '12
v
••••••
t.
'i:c..... '
1
I
Tables
Table 1 Suggested Values of Zo for Various Types of Exposures ................................ 3
Table 2 Maximum Annual Winds (I minute average) ......................................... 7
Table 3 Station Descriptions and Estimated Extreme Wind Speeds ............................. 7
Table 4 Mean Recurrence Interval ......................................................... 18
Table 5 Relationships Between Risk of Occurrence, Mean Recurrence Interval and Expected
Life of Building ..................................... '. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
Table 6 Pressure Coefficients for Walls of Rectangular Buildings ............................ , .. 19
Table 7 Pressure Coefficients for Roofs of Rectangular Buildings ......................... , .. 20
Table 8 Internal Pressure Coefficients for Rectangular Buildings ............................... 21
Table 9 Correction Factors (R)for Height of Building ...................................... , .. 21
1.1 INTRODUCTION
In modern building COlit'S ilnd ... t,lJ1dard~ II, ~J ba~i( wind speed, Itthl'~tdti(ln l"l'l1l' Illr \\'hllh wind
design wind "~'ll'l'ds are spl'ciiil'd in explicitlv pro rl'cord" ,'\'l'r a numlwr lli (OI1"l'cutl\'l' \'l'.1r" ,11'l'
babilistic terms, At any givcn station a random \'.H1,1 ilvailabl!.', ,hen thl' ClIl1lul,1ti\'l' di~tJ Ibutilln tUI1t'tf('11
ble can be defined, which consists of thl' I.Hgl'~t \'Carlv (CDI') p; \hi~ ral1dlllll \ dri,1bll' 111.1\', ,It le.l"t II'i tlll'llr\
be estimated to characterize the probabilistic behavior monthly wind speed recorded at the station investi
of the largest yearly wind fpeeds. The basic design gated. The question of selecting the most appropriate
wind speed is then defined as the speed correspond distribution is one that deserves close attention: in
ing to a specified value FO of the CDF or, equivalently deed, as indicated in References 23 and 22, the mag
(in view of the relation N = 1/ (I Fa> in which N = nitude of the basic design wind speed may depend
mean recurrence intervaI), as the speed correspond strongly upon the probabilistic model used.
ing to a specified mean recurrence interval. For exam
A!>Suming that the type of probability distribution best
ple, the American National Sta,ndard ASS.l [1]
suited for modeling the behavior of the extreme
specifies that a basic design wind speed corresponding
winds is known, a second important question arises,
to a SOyear mean recurrence interval (i.e, to a value
viz., that of the errors associated with the probabilistic
Foofthe CDF equal to 0.9S, or to a probability of ex
approach to the definition of design wind speeds.
ceedance of the basic wind speed in anyone year
Such errors dep~nd primarily upon the quality of the
equal to 0.02) be used in designing all permanent
data and upon the length of the record (Le., the sam
structures, except those structures with an unusually
pie size) available for analysis.
high degree of hazard to life and property in case of
faHure, for which a lOOyear mean recurrence inter These questions wi II be dealt with in this work, which
val (FO = 0.99) must be used, and structures having no wiII also present results of statistical ana lyses of wind
human occupants or where there is negligible risk to speed data recorded in the Philippines. In the liglit of
human life, for which a 25year mean recurrence the material presented herein, possible approaches
(fO = 0.96) may bP. used. A wind speed corresponding will be examined to the definition of extreme wind
to a Nyear recurrence interval is commonly referred speeds for purposes of structural design in the Philip
to as the Nyear wind. pines.
The mean recurrence intervals specified by building 1.2 WIND SPEED OAT A
codes, rather than being based on a formal risk
analysiswhich is in practice not feasible in the pre For the statistical analysis of extremt' wind speeds to
sent state of the artare seiected in such a manner as be meaningful, the data used in the analysis must be
to yield basic wind speeds which, by professional con reliable and must constitute an homogeneous set. The
~nsus, are judged to be adequate from a structural
data may be considered to be reliable if:
safety viewpoint. Nevertheless, it is generally • The performance of the instrumentation used
a!"".,umed that adequate probabilistic definitions of for obtaining the data (i.e., the sensor and the record
design wind speeds offer, at least in theory, the ad ing system> can be determined to have been adequate.
vantage of insuring a certain d~ee of consistency • The sensor was exposed in such a way that it
with regard to the effect of the wind loads upon struc was not influenced by local flow variations due to the
tural safety. This is true in the sen.o;e that, all relevant proximity of an obstruction (e.g., building top, ridge
factors being equal, if appropriate mean recurrence or instrument support).
intervals are used in design, the probabilities of failure
of buildings in different wind climates will, on the A set of wind speed data is referred to herein as
average, be the same. homogeneous if all the data belonging to the set may
be considered to have been obtained under identical
In the practical application of the probabilistic ap or c?quivalent conditions. These conditions are deter
proach to the definition of design wind speeds, cer mined by the following factors, which will be briefly
tain important questions arise. One such question per discussed below:
tains to the type of probability distribl tion best suited
for modeling the probabilistic behavior of the extreme • type of instrumentation used
winds. The provisions of the National Building Code • averaging time (i.e, whether highest gust, fastest
of Canada [2] are based upon the assumption that this mile, oneminute average, fiveminute average, etc.
behavior is best modeled by a Type I (Gumbel) dis was recorded).
tribution. The American National Standard ASS.l [1], • height above ground
on the other hand, assumes that the appropriate • roughness of surrounding terrain (exposure)
models are Type II (Frechet) distributions with loca
tion parameters equal to zero and with tail length • in the case of tropical cyclone winds, distance
parameters dependent only upon type of storm. inland from the coastline.
Finally, Thorn [29J has proposed a model consisting of 1.2.1 Type of instrumentation
a mixed probability distri17ution, the parameters of
which are functions of (a) the frequency of occurrence
of tropical cyclones in the 5° longitudelatitude square If, during the period of record, mOre than one' type of
under consideration and (b) the maximum average instrument has been employed for obtaining the data,
2
the various instrument characteristics (anemometer The zero pI ;me displacement Zd may in all cases be
and recorder) must be carefully taken into account assumed to be zero, except that in cities (or in wooded
and the data must be adjusted accordingly. terrain) Za = 0.75 h, where h = average height of
buildings in the surrounding area (or of trees) [10, 16].
1.2.2 Averaging Time Thus, for example, if in open terrain with Zo = 0.05 m,
If various averaging times have been used during the UC23} = 30 mIs, then adjustment of this value to the
period of record, the data must be adjusted to a com height Z. = 10 m, using Equation 1, gives
mon averaging time. This can be done using graphs
such as those presented in Reference 19 and included
in figure 1 in which Z. is a parameter deJning the o:os
0_
<UI
10
5.3~ = 25 .9
terrain roughness (see, for example, Ref. 10). UOO} = U(23) 
0_23
= 30 6.13 I
In S.
<UlifOS
1.2.3 Height Above Ground
If, during the period of record, the elevation of the
It is noted that, in most cases, the roughness
anemometer had been changed, the data must be ad
justed to a cbmmon elevation as follows: Let the
parameters Zo, Zd must be estimated subjectively,
roughness length and the zero plane displacement be rather than being determined from measurements.
Good judgment and experience are reqUired to keep
denoted by Z. and Zd, respectively (Zo' Zd' are
parameters which define the roughness of terrain. see the errors inherent in such estimates within ~easona
Ref. 10). The relation between the mean wind speeds ble bounds. In conducHng statistical studies of the ex
U(Z1) and U(ZJ over horizontal terrain of uniform
treme winds, it is advisable that for any particular set
roughness at elevation Z1 and Z. above ground, of data, an analysis be made of the sensitivity of the
respectively, can be written as results to possible errors in the estimation of Z and
Zd' 0

U(Z.)
U(Z)
tratropical storms, the mean wind U(Z) at height Z in
terrain of roughness ZQ' Zd is related ao; follows to the
mean wind U1(ZI) at height ZI in terrain of roughness
ZOJI Zdl [21]:
Table 1. Sugested Values of Z. for Various Types Let, for example, U(ZI) = 29 mIs, Zi = 10 m, ZOi = 0.05
of Exposure m, Zdl = O. The corresponding speed U(Z) at Z = 40 m,
say, in open terrain of roughness Zo = 0.25 m, Zd = 0 is
Type of Exposure Z.(meters)
Coastal 0.0050.01 Rn 40
Open 0.03 0.10
VOO) = 1.12 x 29 ~ = 31.1 m/s.
Outskirts of towns, suburbs 0.20 0.30
<:;enters of towns 0.35 0.45
.en
0.05
Centers of large cities 0.60~ 0.80
where 1.12 is the value of fJ for ZOI = 0.05 m, Zo = 0.25
m,obtained from figure 2.
3
It is pc ~nted out that, just as in the case of Equation 1, various authors as follows:
errors are inherent in Equation 2 that are associated
(a) In Reference 23, estimates are made of all three
with the subjective estinl.'!',lon of the roughness
parameters, p., u and 'Y in Equation 3, no specific
parameters. Also, recent research suggests that in the
value being assigned a priori to any of these
case of tropical cyclone winds Equation 2 underesti
parameters.
mates wind speeds over builtup terrain, calculated as
(b) In References 27 and 28, the location parameter is
functions of speeds over open terrain, by amounts of
assumed to have the value p. = O. Estimates are then
the order of 15% or morel17].
made of the remaining parameters, u and p.. The ar
1.2.4 Distance InI.tnd from the Coastline bitrary assumption that p. = 0 entails a sacrifice in
goodness of fit; the justification for using this assump
The intensity of hurricane or typhoon winds is a tion is that it makes possible the application of
decreasing function of the distance inland from the Lieblein';, wellknown estimation procedure [13] to
coastline. Hurricane wind speeds may be adjusted to a obtain values of u andy [27]. However, in view of the
common distance from the coastline by applying recent development of alternative estimation pro
suitable reduction factors. Such reduction fadors have cedures applicable to type II distributions with p.FO
been proposed by Malkin, according tc ... hom the [23], the assumption that p. = 0 becomes unneces
ratios of peak gusts at 48, 96 and 144 km from the sary.
coastline to peak gusts at the coastline are 0.88, 0.82
and 0.78, respectively [8, 141. (c) Court in the United States (3), Davenport in
Canada (5] and Kintanar in the Philippines [12) have
1.3 PROBABILISTIC MODELS OF EXTREME assumed the universality of the type I distribution,
WIND SPEEDS i.e., that the tail length parameter is always y = 00,
The nature of the variate suggests that an appropriate Estimates are then made of the parameters #L and u,
model of extreme wind behavior is provided by prob The second procedure assumes the universal validity
ability distributions of the largest values, the general of the mixed distribution
expression for which is 1II]:
4
large values of N, such as are of interest in structural see table 2), estimated by using the technique
safety calculationsif the following conditions are described in Reference 23. are fl = 38.89 km/hr, fT =
satisfied. First, the value of opt ('Y) for that record is 9.40 km/hr. It follows from Equation 8 that v (SO) = 75
large, say 'Y .? 40 (opt(y) = value of 'Y [see eq. 3] for km/hr and from Equations 6. 7, and 10 that SD[ v (SO)]
which the best distribution fit of the largest values is > 5.18 km/hr. Subsidiary calculations not reported
obtained). Second, meteorological information ob here have shown that Equation to provides a good
tained at the station in question, as well as at nearby indication of the order of magnitude of the sampling
stations at which the wind climate is similar, indi errors.
cates that winds considerably in excess of those
reflected in the record cannot be expected to occur ex 1.4.1 Wind Climates Characterized by Small
cept at intervals many times larger than the record Values of opt (y).
length. Wind climates which satisfy these two condi
Occasionally, a record obtained in wellbehaved
tions will be referred to as wellbehaved.
wind climates may exhibit small values of opt (,.); this
Assuming that the wind speed data are reliable, lower will occur if that record contains a wind speed that
bounds for the sampling error in the estimation of the corresponds to a large mean recurrence interval.
Nyear winds in a wellbehaved climate may be There are regions, however, in which, as a rule, the
calculated on the basis of a mathematical result, the statistical analysiS of extreme wind records taken at
CramerRao relation, which states that for the type I anyone station yields small values of opt (,.). This is
distribution (see ref. 11. p. 282) the case if. in the region considered, winds occur that
are meteorologically distinct from, and considerably
"') 1.10867 ,
var (1£ ?u stronger than the usual annual extremes. Thus. in the
n regions where tropical cyclones occur, opt (,.) will in
general be small, unless most annual extremes are
"') 0.60793 ,
V ar (u >u associated with tropical cyclone winds. An example
n (7) oi a record for which,. (opt) is small is given in figure
3a, which represents the probability plot with 'Y = opt
where var (,1), va; (u) are the variances of,1 andu, (,.) = 2 for the annual extreme fastest milespeeds
where,1 and t't are the estimated values of 1£ and u, recorded in 194973 at the Corpus Christi, Texas, air
respectively. obtained by using any appropriate port. For purposes of comparison. the same data have
estimator consistent with basic statistical theory re been fitted to a type I distribution (opt ( ,.) = 00, or Eq.
quirements; u is the actual value of the scale 4); the fit in this case is seen to be exceedingly poor,
parameter and n is the sample size. Using Equations 6 i.e., the plot deviates strongly from a straight line (fig.
and 7, lower bounds for the standard deviation of the 3b). As shown in Reference 23. a measure of the good
sampling error in the estimation of the Nyear wind, ness of fit is given by the extent to which the pro
SD[v(N»). can be approximated as follows. Equation 4. bability plot correlation coefficient is close to unity;
in which the parameters 1£, u are replaced by their this coefficient is printed out in figures 3a and 3b.
estimates 1£, u, is inverted to yield
To small values of the tail length parameter there fre
quently correspond implausibly high values of the
(8)
estimated speeds for large recurrence intervals. In the
case of the 191248 record at Corpus Christi, for exam
where
ple, opt (,.) = 2 and the estimated 5minute average is
327 mph (155 m/s) for a 1000year wind, which is
(9)
highly unlikely on meteorological grounds. For 20
year records, the situation may be even worse: thus,
Then for the 191736 Corpus Christi record, which contains
an exceptionally high wind speed due to the 1919
hurricane [3, 25], opt (,.) = 1 and the calculated 1000
year wind is 1952 mph (873 m/s) [23], a ridiculous
Equation 10 is based on the assumption that the error result. Also, the situation is not likely to improve sig
involved in neglecting the correlation between 1£ and nificantly if the record length increases. From a 74
CT is small. The validity of this assumption was year record, a plot quite similar to figure 3 would
verified by using Monte Carlo simulation techniques. presumably be obtained, with twice as many points
similarly dispersed, to which there would correspond
Since the actual value of CT is not known, in practical a similar least squares line on probability paper.
calculations the estimated value u is used in Equations
6 and 7. For example, the distribution parameters cor It may be stated, consequently, that while in the case
respondin& to the wind speed data at Davao (n = 24. of wellbehaved climates it appears reasonable to in
5
fer from a good fit of the probability curve to the data in which 4.5 < y (v) < 9. If the mean rate of arrival of
that the tail of the curve adequately describes the ex tropical cyclones in the region considered is high,
treme winds, such an inference is not always justified then y ( til will be closer to 4.5. Otherwise, y (til will be
ifopt(y) is small. closer to 9; in re3ions where hurricanes cannot he ex
pected to occur, y (t) = 9.1n order that estimates not
It may be argued that one could avoid obtaining
be based upon possibly unrepresentative annual ex
unreasonable extreme values by postulating that the
treme data, Thom's model does not make use of an
annual largest winds are described by a probability
nual extreme speeds. Rather, the parameter f7 is esti
distribution of the type I, i.e., by assigning the value y
mated from the maximum of the average monthly
= 00 to the tail length parameter. This has been done
wind speeds on record at the location considered,
by Court [3J and Kintanar [12]. As can be seen in
presumably a quantity for which the variability is
figure 4, the corresponding fit may be quite poor. small.
However, the estimated extremes at the distribution
tails will be reduced. The drawback of this approach While the quasiuniversal climatological distribution
is that unreasonably low estimated extremes may be proposed by Thon is tentative, it will yield results
obtained. For example, at Key West, Florida, if all which, for a first approximation, may in certain cases
three parameters of Equation 3 are estimated as in be regarded as acceptable. This model has recently
Reference 23, to the 191248 record there corresponds been used by Evans l6J as a basis for obtaining design
v (tOO) == 99 mph (44.2m/s) and v (1000) = 188 mph (84 wind speeds for Jamaica. Estimates of extreme speeds
m/s~ Reference 23. If it is postulated that y = 00, obtained by Evans are substantially higher than the
then v (SO) = 70 mph (31.7 m/s), v (I 00) = 77 mph (34.4 results obtained by SheJlard [20J in his 1971 analysis
m/s) and v (t000) = 97 mph (38.8 m) (23), an unlikely of Commonwealth Caribbean wind data.
result in view of the high frequency of occurrence of
It was shown in the preceding section that the ap
hurricanes (about 1 in 7 years) at Key West.
proach which utilizes the series of annual largest
It may also be argued that since the estimated ex speeds may fail in regions in which hurricanes occur.
tremes resulting from small values of y (say y < 4) For such regions, therefore, it may be that alternative
may be too large, and those corresponding to y = 00 approaches need to be deve!oped. Among such ap
may be too small, a probability distribution that might proaches is one in which esti mates of extreme winds
yield reasonable results is one in which y has an in are based upon the follOWing information:
termediate value, say 4< y < 9. Such an approach
• average number of hurricanes affecting the
has been proposed by Thom and will now be ex
coastal sector considered (per year)
amined.
• probability distribution of hurricane intensities
1.5 ASSESSMENT OF PROCEDURE BASED • radial dimensions of hurricanes
ON THE HIGHEST AVERAGE
MONTHLY SPEED • dependence of wind speeds upon central
pressure and distance from hurricane center.
The procedure for estimating extreme winds in hur This approach appears to provide useful estimates of
ricaneprone regions on the basis of annual highest
extreme winds corresponding to large recurrence in
winds at a station was seen to have the following
tervalswhich are of interest in ultimate strength
shortcomings. First, because hurricane winds are
calculationsand is currently under study at the Na
relatively rare events, the available data may not con
tional Bureau of Standards.
tain wind speeds associated with major hurriCane oc
, currences and are therefore not representative of the 1.6 STATISTICAL ANALYSIS OF EXTREME
wind climate at the station considered (see the case of WIND DATA IN THE PHILIPPINES
Calapan in Section 1.7 of this report). Second, in
regions subjected to winds that are meteorologically Through the courtesy of the Philippine Atmospheric,
distinct from, and considerably stronger than the Geophysical and Astronomical Services Administra
usual annual extJ'emes, implausible estimates may be tion (PAGASA), 16 sets of data were obtained consist
obtained. ing of maximum yearly wind speeds recorded during
at least 14 consecutive years. The data for each of the
The model proposed by Thom lEq. 5J in Reference 29 16 stations are listed in table 2. Table 3 includes sub
represents an attempt to eliminate these shortcom jective station descriptions provided by P AGASA
ings. It can be easily shown by applying the inter personnel and the results of the analysis. In Table 3
mediate value theorem, that if this model is assumed, are listed VN"pt(oy)= Nyear wind based on the dis
the estimated extreme winds may be obtained by in tribution for which the best fit of the largest values is
verting an expression of the form: obtained and VN°O = Nyear wind based on the type I
distribution, N = mean recurrence interval in years.
6
TABLE 2. MAXIMUM ANNUAL WINDS (ONE MINUTE AVERAGES)
  .  
NO. Station Period of Record Maximum Annual Winds for Each Year of Record Ckm/hourl
h Manila n 1"'124(1 W h" Op,mm"kl ~5 1Il~ 112 144 lUI 1111 1:111
Central
Mlraaor II 1~1~4fl "7 II~ I Muunt.lin hlp 00 IW 151 192
BagUill (I:;UU m .ItWt.·
sea J~Vl.·1)
~ Baguiu II ,q5fI7J 2~ 111 Mllut.lin top 00 1M IHO 230
;, (l:;UU m .lblwt.,
*a level)
9 Cal.pan II ,q~7J 15 iO Top of hill; 00 2(lq 234 31h
town on one
!iide.
to I ""rill"O I 19. 2~ .
" 7 L.I~Cily
Top uf 40rn
40 170 191 26. 167 IR7 250
hUI
II I..loaa n 194973 2., 12 o...n 00 174 192 252
19~73 1,2
;':;'~::',.o
12 Ilioio II 24 I.own 2 1110 z.l~ MIl 1311 192
194973
19
Z,  o""vo
~.,m
Ai!Port
!!OPO .m
9(1
14
2.15
,,
2M
,.~
3""
J". 2114
M
22M
,>411
JOh
'1m bldg.l
~~~~IiIOry
l'b n anla '_'1. 14 1111 own; ""'don·
Hal area
• 24" 2..., 4l1li 214 242 '3:
a3 CUp anemometer; mean speed averaged over one minute. dNorth and East: sea exposure.
"Mean speed averaged Over one minute. e .:>mitted if opt (y) = 00.
cTrees at East side of anemometer_ lOne minute averages,
1.7 INTERPRETATION OF RESULTS It is noted that fO,r all three Zone III stations listed in
Table 3, opt (y) = 00. It is convenient to adjust the
The results will be grouped into three classes, accord speeds at Davao and Cagayan de Oro to open terrain
ing to the wind zone (as defined in Ref. 15) in which exposure. On the basis of the terrain descriptions of
the stations considered are located (table 3). Table 3, if it is assumed Zo = 0.30 m, Zd = 0, ZOI = 0.08
m, Zdl = 0, it follows from Equation 2 that
1.7.1 Zone III
7
as obtained if the data covering the period 19591973
UOO}== 0.8
U (10) are used (see table 3). Since wind loads are propor
tional to the square of the wind speeds, the ratio be
tween the respective estimated winds loads is
where U(10}, U 1(10} are mean speeds above ground in (209/141)2 = 2.2.
town and open exposure, respectively. Thus, in
Davao and Cagayan de Oro the calculated 50y~ar Although the record at Pasay City is best fitted by a
mean speeds at 10 m above ground in open terrain are type II distribution with opt (y) == 2. it is unlikely, as
58 mph and 47 mph (94 km/hr and 76 km/hr) respec noted previously, that such a distribution correctly
tively, versus 55 mph, (88 km/hr) in Zamboanga. If the describes the behavior of the extreme winds. This is
corresponding highest gusts are obtained by multiply obvious, particularly in the case of the 1000yr wind,
ing.the oneminute means by a factor of, ~y, 1.20 (see which, on physical grounds, could not possibly attain
fig. H, the estimated highest 50yr gusts at Davao, 509 mph (820 km/hr) (see table 3).
Cagayan de Oro and Zamboanga are at most 94 x 1.20
The National Structural Code of the Philippines
== 113 km/hr, (70mph), i.e., conSiderably lower than
specifies, for Zone II and elevations under 9.15 m, a
the value specified for design purposes by the Na
design wind of 109 mph (175 km/hr). In the light of
tiona 1 Structural Code ofthe Philippines [I5] for Zone
the data shown in table 2, the value appears to be
flI, viz., 95 mph (153 km/hr). This suggests that the re
reasonable. It will be noted that tables 2 and 3, and
quirements of Reference 15 regarding wind loading in
figure 13 of Reference 15 indicate that the extreme
the Zone III portion of Mindanao are conservative
speeds and the frequency of occurrence of tropical
and might be somewhat reduced. (It can be easily
cyclones, are considerably higher at Laoag than at
shown, on the basis of Eq. 2 and figure 1, that this
Cebu. This suggests that Zone II could be divided, ac
statement holds even if it is assumed that the errors in
cordingly, into two subzones. with wind load require
the estimation of the parameter values Zo == 0.30 m
ments higher in the northern than in the southern
and Z 01 = 0.08 m are of the order of as much as 50%.)
subzone.
To validate such a conclusion it would however be
necessary to determine, from longterm records of 1.7.3 Zone I.
tropical cyclone occurences, that the 195073 data at
the three stations analyzed are indeed representative As indicated previously. if y(opt) is small, i.c., if the
for southern Mindanao. differences am0ng maximum wind speeds rL' 'vrded
in various years are large, the probability di~" , ,:.
1.7.2 Zone II. tions that best fit the dnta may not describe em .
Several difficulties arise in interpreting the results for the extreme wind speeds for large recurrence in'.,>;
the Zone II stations in table 3. It is noted, first, that the vals. The minimum and the maximum winds for the
results obtained at stations in and near Manila (sta period of record are, at Legaspi, 25 mph (40 km/hr)
tions 4, 5, 6 in table 3) are widely divergent. The dis and 127 mph (204 km/hr), respectively, and, at
crepancies between the results for Pasay City and Tac1oban, 26 mph (42 km/hr) and 120 mph (194
Manila may be due to the different elevations of the km/hr), respectively. In the writer's opinion, the
respective anemometers. It may also be conjectured reliability of the Nyear wind estimates obtained at
that the discrepancies between these results and those these stations for N==50, 100 and 1000 is therefore
obtained from the 19021940 Manila Central record doubtful. The same comment applies to the estimates
are due to differences in the averaging times and in for Infanta, where the record length is quite insuffi
the exposure, elevation and calibration of the instru cient (14 yearsl. The writer therefore believes that the
ments, as well as to possibly inaccurate estimates of results of table 3 should not be used to assess the ade
the maximum speed in Manila and Pasay City in 1970 quacy of the design wind speed requirement for Zone
(200 km/hr, see table 2). I specified in Reference 15. Rather, it is reasonable to
base such an assessment on a comparison between
The estimated wind speeds at Baguio based upon the wind speeds in Zone I and in areas affected by hur
195073 record are higher than those obtained from ricanes in the United States. In the light of U.S. ex
the 191440 data. No explanation is offered for these perience, it is the opinion of the writer that from such
differences; an investigation into their causes seems a comparison it follows that the 124 mph (200 km/hr)
warranted. wind speed requirement for Zone I and elevations
The record at Calapan illustrates the limitations of the under 30 ft. (9.15 m) is adequate for structural design
approach to the definition of design wind speeds purposes.
based on the statistical analysis of the highest annual
winds. From the data covering the period 196172, the
estimated 50yr wind based on a Type I distribution is
88 mph (141 km/hr) [12], versus 131 mph (209 km/hd,
8
1.8 CONCLUSIONS ACKNOWLEDGMENTS
From the analysis of available extreme speed data in The writer wishes to express his indebtedness and ap
the Philippines, the following conclusions may be preciation to Dr. Roman L. Kintanar, Mr. Manuel
drawn: Bonjoc, Mr. Bayani S. Lomotan, Mr. Jesus E. Calooy,
1. The design wind speeds specified by the National Mr. Leonicio A. Amadore, Mr. Samuel B. Landet, and
Structural Code of the Philippines for the Zone III Mr. Daniel Dimagiba, of the Philippine Atmospheric,
part of Mindanao appear to be conservative and Geophysical Astronomical Services Administration
might be somewhat reduced. For this conclusion to (PAGASA), for kindly permitting him to use the
be validated, it would be necessary to determine, PAGASA records and facilities and for their effective
from longterm records of tropical cyclone occur and generous help. He also wishes to thank Dr. R. D.
rences, that the data analyzed herein are represen Marshall of the Center for Building Technology, In·
tative for southern Mindanao. stitute for Applied Technology, National Bureau of
2. Methodological difficulties and uncertainties with Standards, for useful comments and criticism of this
regard to the reliability of the data preclude, at this work. The computer program used here was
time, the estimation for Zones II and I of Nyear ex developed by Dr. J.J. Filliben, of the Statistical
treme winds that could be used, with a sufficient Engineering Laboratory, National Bureau of Stan
degree of confidence, as design values within the dards.
. framework of an explicitly probabilistic code.
3. According to the data included herein, Zone II can
REFERENCES
be divided into two subzones, with wind load re (I) Building Code Requirements for Minimum Design
quirements higher in the northern than in the Loads in Buildings and Other Structures,
southern subzone. A58.11972 (New York: American National
4. The data included herein suggest that the wind Standards Institute, 1972).
speed requirement specified by the National Struc (2) Canadian Structural Design Manual (Supplement
tural Code of the Philippines for Zone I is adequate No.4 to the National Building Code of
for purposes of structural design, except as noted Canada)(National Research Council of
below. Canada, 1970).
5. Higher wind speed values than those specified by (3) Court, A., "Wind Extremes as Design Factors,"
the National Structural Code of the Philippines Journal of the Franklin Institute, vol. 256 (July
should be usedexcept perhaps in the Zone III part 1953), pp. 3955.
of Mindanaoin open, and in coastal exposure. (4) Csanady, G. T., "On the Resistance Law of a Tur
6. Improved design criteria for Zones II and I. includ bulent Ekman Layer," Journal of the At
ing possible redefinitions of these zones, could in mospheric Sciences, vol. 24 (September 1967),
the future be achieved by applying the pp.467471.
methodology briefly described at the end of the (5) Davenport, A. G., "The Dependence of Wind
section" Assessment of Procedure Based on the Loads Upon Meteorological Parameters," Pro
Highest Average Monthly Speed." This would re ceedings, Vol. 1 (International Research Semi
quire, in addition to data on the frequency of oc nar on Wind Effects on Buildings and Struc
currence of tropical cyclones at various locations in tures) (Toronto: University of Toronto Press,
the Philippines, that the following data be availa 1968).
ble: (6) Evans, C. J., "Design Valuesof Extreme Winds in
a. Reliable wind speeds, carefully defined with Jamaica" (Paper presented at Caribbean
respect to terrain roughness, averaging time and Regional Conference, Kingston, Jamaica,
distance from shore line. November 67,1975).
b. ApprOXimate radial dimensions of tropical (7) Fichtl, G., and McVehil, G., Longitudinal and
cyclones. Lateral Spectra of Turbulence in the A tomospheric
c. Approximate dependence of tropical cyclone Boundary Layer, Technical Note D5584
speeds upon minimum central pressure and dis (Washington, D.C.: National Aeronautics and
tance from storm center. Space Administration, 1970).
(8) Goldman, J.L., and Ushijima, T., "Decrease in
Maxinum Hurricane Winds after Landfall,"
Journal of the Structura I Division, vol. 100, no.
STl, proc. paper 10295 (New York: American
Society of Civil Engineers, January 1974), pp.
129141. .
9
(9) Harris, R.I., "Measurements of Wind Structure At (22) Simiu, E., and Filliben, J.J., "Probabilistic Models
Heights Up to 585 ft Above Ground Leve\," of Extreme Wind Speeds: Uncertainties and
Proceedings {Symposium on Wind Effects on Limitations," Procl'l'dillgs(Fourth International
Buildings and Structures) (Leicestershire: Conference on Wind Effects on Buildings and
Loughborough University of Technology, Structures)(London,1975).
1968). (23) Simiu, E., and Filliben, J.J., Statistical Analysis of
(10) Helliwell, N.C., "Wind Over London," Proceed Extreme Winds, Tech. Note 868 (Washington,
ings (Third International Cc,ference of Wind D.C.: National Bureau of Standards, 1975). SD
Effects on Building and Structures) (Tokyo, Catalog No. C13.46:868
1971) . (24) Simiu, E., and Lozier, D. W., The Buffeting at Tall
(11) Johnson, N.L., and Kotz, 5., Continous Univl,riate Structures by Strong Winds, BSS 74
Distributions, vol. 1 (Wiley, 1970) (Washington, D.C.: National Bureau of Stan
(12) Kintanar, R. L., "Climatology and WindRelated dards, 1975). sDCatalog No. 13.29/2:74.
Problems in the Philippines," Development of (25) sugg, A. L., Pardue, L. C., and Carrodus, R. L.,
Improved Design Criteria to Better Resist the Memorable Hurricanes of the United States,
Effects of Extreme Winds for LowRise Buildings NOAA Technical Memorandum NWs sR56
in Developing Countries, BSS 56 (Washington, (Fort Worth: National Weather Service, 1971).
D.<;.: National Bureau of Standards, 1974). SD (26) Tennekes, H.~ and Lumley, J.L., A First Course ill
. Catalog No. C13.29/2:56 Turbulellce(Cambridge: The MIT Press, 1972).
(13) Lieblein, J., Method of Analyzing Extreme Value (27) Thom, H. C. 5.. "Distributions of Extreme Winds
Data, Tech. Nute3053(Washington, D.C.: Na in the United States," Joumal of the Structural
tional Advisory Committee for Aeronautics, Dillision, vol. 86, no. sT4, proc. paper 2433
1(54). (New York: American Society of Civil
(14) Malkin, W., "Fillirig and Intensity Changes in Engineers, April 1960), pp. 1124.
Hurricanes Over Land," National Hurricane (28) Thom, H.C.S., "New Dis~ributions of Extreme
Reseszrch Project, vol. 34(1959). Winds in the United States," Journal of the
(15) .National Structural Code of the Philippines, (Sec Structural Division. vol. 94, no. ST7, proc.
tion 206 in BSS 56: Development of Improved paper 6038 (New York: American Society of
Design Criteria to Better Resist the Effects of Ex Civil Engineers, July 1968), pp. 17871801.
treme Winds for LowRise Buildings in Develop (29) Thorn, RCs., "Toward a Universal Climatologi
ing Countries) (Washington, D.C.: National cal Extreme Wind Distrihution," ProL·(t'dings,
Bureau of Standards, 1974). SD Catalog No. vol. 1 (International Research Seminal on
CI3.29/2:56 Wind Effects on Buildings and Structurt:s)
(16) Oliver, H. R., "Wind Profiles In and Above a (Toronto: University of Toronto Press, 1968.1.
Forest Canopy," Quarterly Journal of tile Royal
Meterorological Society, vol. 97 (I971), pp.
548553.
(17) Patel, V.c., and Nash, J.F., Numerical Study of the
Hurricane Boundary Layer Mean Wind Profile
(Report prepared for the National Bureitu of
Standards) (Sybucon, Inc., June 1974).
(18) Peterson, E. W., "MCldification of Mean Flow and •
Turbulent Energy by a CharLge in Surface
Roughness Under Conditions of Neutral
Stability," Quarterly Journal of the Royal
Meteorological SoCiety, vol. 95(1969), pp.
•
569575.
(19) Sachs, P., Wind Forces in Engineering (Pergamon
Press, 1972).
(20) Shellard, H.C., "Extreme Wind Speeds in the
Commonwealth Caribbean," Meteorological
Magazine, no. 100(1971), pp. 144149.
(21) Simiu, E., "Logarithmic Profiles and Design
Wind Speeds," Journal of the Engineering
Mechanics Dillision, vol. 99, no. EMS, proc.
paper 10100 (New York: American Society of
Civil Engineers, October 1973), pp. 1073 1083.
10
r I I
I
10 I min. 10 min.  fI hour
oa
.::::::: ~ ~

..........
.......... ~
:::::::::: :!
~ ..........

.r...
Zo ;r 0.030.08 m
06
r r r Zo ;rO.l50.30m
0·4
r 
Zo;r I.OOm
I 2 4 6 10 20 40 60 100 200 400600 1000 2000 3600
Time, tt seconds
1.50 ,.,.....r...,....,....,....,.......
lAO t+++""':
1.30 HHC+~+__+'~+a:liotfE:I
95.0000000=MAX x
I
I 1
I
I x I
19.0000000
I
I
1
83.0000000
77.0000000
71.0000000
I
I
I
65.0000000=1410
I
I
I x
59.0000000
x
l!
53.0000000
x X
x X
Xx
1t7.0000000 XX X
X
X
XXXX
111.0000010 XX
XXX
XX
J XU
35.0000000=MIN X
1____________ 1___________ 1____________ 1___________ 1 ____________ 1___________ 1__ ··t
.SOO~l81 2.2021567 3.90~1!\~ 5.60111139 7.!061112S
EXTREME VALUE nPE 2 (CAUCHY TYPEI PliO!!. PLOT WITH EXP. PAQ. = 2.000000000' SAII'LE SIZE N :: 37
PROBABILITY PLOT CORRELATION COEFFICIEOlT = .97191 ESTIMATED INTERCEPT = :51.0711109 ESTIMATED SLOP£ = 9."757"7
FIGURE la. TYPE 1II_DISTRIBUTION, "y = 2.
_ _ _ 1_ _ _ 1_______ 1 ______ 1_ _ _ __
95.000COOO=MAX
1
1
_ _ _ _ _ '_ _ _ _ 1_ _ _ _ 1
1
. I
I
19.0000000
I
I • I
13.0000000
I
I
I
77.0000000
I
1
I
71.0000000
I
I
I
65.0000000:M10
I
I
I lC
59.00000dO
lC
lC
53.0000000
x X
x X
X X
"'.0000000 X X X
X
X
xx XX X'
111.0000010 xxx
I XX XX"
I X XX
I x X X X
35.0000000=MIN X
1____________ 1___________ 1____________ 1___________ 1____________ 1___________ 1__...__ 1___________ 1
1.312981t7 .0"2~107 1.29722311 2.637327.. 3.977113111
EXTREME VALUE TYPE 1 (EXPONENTIAL TYPEI PRORAIlILITY PLOT THE: SAMPLE SIZE N = 37
PROBABILITY PLOT CORIIELATlON CO£FFI·CIENT .MIO~ =
ESTltIIlT£O INTERCEPT II •• 033329 = ESTIMATED SLOPE = ,. ""21209
FIGURE lb. TYPE I DISTRIBUTION. Facing Page: This wind tunnel at the University 0; the
Phi11il,ines is used to study wind effects on scalemodel
buildings. Shown is a model of the CARE, Inc., test house.
Thl! rows of Mocks on the floor offlle tunnl!l xenerate tur
bulence or gustiness similar to that observed in full scale.
12
2. A GUIDE TO THE
DETERMINATION OF
WIND FORCES
2;2;3R()C}fQverhangs
In caIculatingthe total uplift load on a roof, the
pressure acting on the underside of roof overhangs
must also be included. These pressures are usually pected life of a building. For example the probability
positive and the resultant force acts in the same direc that the basic wind speed associated with a 50year
tion as the uplift force due to suction on the top sur mean recurrence interval will be exceeded at least
face of the roof. Pressures acting on the inside of the once in 50 years is 0.63. The relationship between risk
buiding (to be discussed later) can also contribute to of occurrence during the expected building life and
the total uplift force and must likewise be accounted the mean recurrence interval is given in table 5. It
for. should be noted that the risk of exceeding the basic
wind speed is, in general, not equal to the risk of
failure.
15
(a) The average or mean pressure is negative (suc pressure, give the wind loads acting in a direction
tion) normal to those surfaces or elements. The total resul
tant forces and moments acting on a building can then
(b) Pressure fluctuations tl'nd to occur in bursts be determined by considering the appropriate compo
nents of these loads acting on each of the surfaces or
(d Maximum departures from the mean are in the elements.
negative (suction) direction As discussed in the previous section. the instan
taneous peak pressure can be expressed in terms of the
(d) The peak values far exceed the mean value mean pressure and a fluctuating component. Since
pressure fluctuations are limited in spatial extent, it is
:r20~~~~tf~~~~~~~~~:fti~!i~~~
necessary to consider the size of the building surface
or element when selecting the pressure coefficient.
Pressures on Extended Areas: For the purpose of
determining wind loads acting on sizeable surface
o .+.• ~~ •••• 29 r' . .~~_~~·lQr+~ • .• _. ~ areas such as the walls and roof of a building, the
tDIE  SECOlIIlS r
pressute coefficients listed in "tables 6 and 7 should be
100
200 _ _
used. These coefficients have been determined experi
o mentally from measurements taken on fullscale
100 buildings and from windtunnel tests and they repre
200 sent an upper limit of conditions likely to occur on the
indicated building surfaces.
FIGURE 1. TYPICAL RECORD OF WIND
SPEED AND SURFACE
Pressures on Localized Areas: It is to be expected that
PRESSLTRE.
the smaller the area considered, the larger the effec
tive peak pressure will be. In addition, there are cer
To quantify these pressures, it is essential that a suffi tain surface 'areas where intense suctions occur as
ciently long time interval be used to obtain a stable pointed out in sections 2.2.1 and 2.2.2. To provide for
mean. p. The fluctuations an' described by their stan these cases, pressure coefficients for localized areas
dard deviation or rootmeansquare. Prms' taken are included in tables 6 and 7. These coefficients are
about the mean. Finally. the peak pressure f1uchla for the purpose of assessing wind loads on local clad
tions arl' dl'scribed by a peak factor. g. which indi ding and roofing elements and should not be used to
cates the numbt.'r of standard deviations that the peak calculate overall loads on buildings. They should be
prl'SSllfl' dl'viatl's from thl' ml'an. Thus, the peak used in conjunction with the internal pressure coeffi
PTl'S.<;Ufl.' can be expressed as cients (where appropriate) as described in the follow
ing.
"max =p + g I'rms or "min = pg Prrns (3) Internal Pressures: As indicated in section 2.2.3 the
net load or force acting on the roof or walls of a build
It should bt.' noted that thl' peak factor, g, is a random ing depends not only on the external surface
variable and has a prob,'biIity distribution function pressures, but on the internal pressure as well. The
that depends on thl' !1;eometry of the building and tur magnitude of the internal pressure depends upon the
bulent structurl' of the wind. The values of g are building geometry, size and location of openings, and
selected so that thl' associated probabilities of being wind speed and direction. As with external pressures,
l'xceeded are in line with the expected building life. it is convenient to express internal pressures in terms
of the dynamic pressure and a pressure coefficient
2.4.3 Pressure Coefficients
Cpi' These coefficients can be positive or negative as
It is convenient to express pressures acting on the sur indicated in table 8. The net pressure acting on a
faces of a building in terms of the dynamic pressure as building element is the algebraic sum of the external
follows and internal pressures
(5)
p=cpll (4)
where Cp is a pressure coefficient whose value de Thus a positive internal pressure will increase the
pends upon the geometry of the building and local loading on those areas of roofs and walls subjected to
flow conditions. Pressure coefficients are specified for external suction.
particular surfaces or elements of a building and,
when multiplied by the surface area and dynamic
16
2.4.4 Correction Factor for Height of Building from table 4
2. Check the associated factor of risk in table 5 and
The pressure coefficil'nts described above are based on select a longer mean recurrence interval if ap
building heights of 33 ft 00 m) and peak wind speeds propriate.
at 33 ft (10 m) above ground, averaged over 2seconds. 3. Determine the basic wind speed for this mean
Overall loads calculated for buildings appreciabl y less recurrence interval and the appropriate terrain
than 33 ft (10 m) in height (measured to eaves or roughness and type of exposure as outlined in sec
parapet) will thus be overestimated if these coeffi tion 1.
cients are used without modification. On the other 4. Convert the resulting basic wind speed to a 2se
hand, tributary areas such as doors, windows, cIad cond mean speed using the procedure described in
ding and roofing elements will respond to pressure section 1.
fluctuations with duration times considerably less 5. Calculate the dynamic pressure q using the ex
than 2 seconds. To account for this, the pressures must pression
be multiplied by the correction factors, R, in table 9.
Thus the expression for the net pressure acting on a '1=0.613 U 2
buiiding surface becomes
6. Select the appropriate pressure coefficients from
1'= q(CpR C,I;R;) (6) tables 6,7 and 8.
7 Select the appropriate correction factors from ta
and the force acting normal to a surface of area A i~ ble9.
8. Calculate the pressures from the expressions
p= qCpR
where Rand R; are correction factors for external and
internal pressures, respectively. or
p = q(CpR  Cp;R;)
2.5 PROCEDURE FOR CALCULATING
WIND FORCES 9. Multiply these pressures by the respective surface
The procedure for calculating wind forces on a build areas to obtain the wind forces.
ing is summarized in the following steps. 10. Sum appropriate components of these forces to
obtain net uplift and drag loads.
1. Select the appropriate mean recurrence interval
knots
o 10 20 30 40 SO 60 70 80 90 100 110
I" " ' " " !. . " ' , , , ,r " ' " , ," I " " , , " ,r , .., ! , " ,r , ,!1 ! • , 'I !. " " , , , ,I , " ,r " , , /, , , .r " , ,I, .r , " 1 , ! I, " , I , " , I
mp.h.
o 10 20 30 40 50 60 70 80 90 100 110 120 130
VELOCITY V 1""111,,1"''''11.1'''''11''''11.1,,1.1',,11,,,.1'11,1111.1.,,,,,,"1II!,I,!"!,!!,!!",!.,!,!,, !,I,!,,!,!,.!, !"I!I"!.".!.,,.1
m/sec,
0 5 10 15 20 25 30 35 40 45 50 55 60
I I I I I II I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I ..I I I I I I I I I I I I I I I II I I
Ibf/ft 2
0
, , , , , , ,,,,
2 3 4 5 6 7 8 910
I I
15
I I
20
' , , J I
25
/ ' , JI , , ,
30
I ! I ! !
35
I II II
40
I
N/m2
I ,
0 600 1000 1200 1400 1600 1800 2000
DYNAMIC I ' , I ,
100
rI
200
Ih I
I I
400
I I I I
,800
I ' I II I I I I I I I I I I
PRESSURE q
kgf/m2
0 10 20 30 40 50 60 70 80 90 100 120 140 160 180 200
I I I III lid I I I I I I I I I I II I II I I I I I I I I I
Structures presenting a lo~ degree o( hazard to life and other property in the case of
failure. 20
10 10 15 20 29 45 95
20 20 29 39 56 90 190
50 50 72 98 140 224 475
100 100 144 196 280 448 949
Note: From this table it will be seen that there isa IO'}/ risk that the wind speed corresponding toa mean recurrence interval of 475
years will be exceeded in a lifetime of 50 years.
18
TABLE 6. PRESSURE COEFFICIENTS FOR WALLS OF RECTANGULAR BUILDINGS
C p for Face Local
Building Building Wind Angle
Height/Width Length/width u Cp
Ratio Ratio tDegrees, A B C D

I <1iw<1.5
91l lUI O.H II.K 115
1.5
I5:s,h/w<4
0 II.M O.n 0.9 0.9
15<1iw<4 1.5
 90 0.4 0.4 lI.H 11.3

Notes: (t)h isthe height to eaves or parapet, I is the greater plan dimension of the building and w is the lesser plan dimension.
(2) Local C p valuesOast column) should be used in conjunction with correction factors for overall areas in Table 9 .
..
LESSER OF h OR 0.2 w ~
""=f:::t::
~
1~ h
:A
C
T
1
B I
. ~
1
:
8 D
~w~ ~w~
ELEVATION PLAN
19
TABLE 7. PRESSURE COEFFIOENTS FOR ROOFS OF RECTANGULAR BUILDINGS
BuDding WinclAngle Area Roof Slope IJ
Height/Width a Designation De, rees
Ratio (Degrees) 0 10 2if 25
Notes: (I) The pressure coefficient on the underside of roof overhangs should be taken as that on the adjoining wall surface.
tv Local Cp values (J and K) should be used in conjunction with correction factors for overall areas in table 9.
E G I
K
~ F
H K ~ b ~
f   f+t
b = lesser of h or 0.15 w
1~.I
141:
.... "~~I W
1<1r
... w .:>:..11
Extended areas
1 ~~
.... W
Local areas
~I
ELEVATION PLAN
20
TABLE 8. INTERNAL PRESSURE COEFFICIENTS FOR RECTANGULAR BUILDINGS
Condition Internal pressure coefficient Cpi
· Four walls equally permeable 0.3 or +0.2 whichever is the more severe for
combined loadings
"
h or more
(b) Dominant opening .10 leeward wall
+0.8
value of e'l for leeward external wall surface
value of e for side external wall surface
(e) Dominant opening on side wall
(d) Dominant opening in a roof segment value of C: for external surface of roof segment
Notes: (I) Internal pressures developed within an enclosed structure may be positive or negative depending on the position and
size of the openings.
(2) In the context of table 8 the permeability of a surface is measured by the total area of openings in the surface under con
sideration.
(3) The value of ~'J;can be limited or controlled to advantage by deliberate distribution of permeability in the wall or roof.
or by the deliberate provision of a venting device which can serve as a dominant opening at a position having a suitable
exter'nal pressure coefficient.An example of such is a ridge ventilator on a lowpitch roof. and this. under all directions
of wind. can reduce the uplift force on the roof.
Notes: (l)The term "Overall" refers to the entire area of a given wall or roof slope.
(2) The term "Elements" refers to roof and cladding elements. doors. windows. etc.
(3) The terrain roughness parameter Z. must be estimated subjectively. The following values are suggested for various
types of exposure.
TYPE OF EXPOSURE Z. (meters)
Coastal 0.005·0.01
Open country 0.02·0.12
Outskirts of towns. suburbs o.l3.;:}.'30'
Centers of towns 0.40
21
For wall A,
APPENDIX A
ILLUSTRATIVE EXAMPLE p= (544) 10.8(0.3)](0.85)
= 509N/m 2
A housing development is to be located in flat, open
country on the outskirts of Zamboanga, Philippines, For wall C,
and will ultimately consist of several hundred single
family dwellings of quite similar geomei:ry. The p= (544) i 0.6(0.2) J(O.85)
period of construction is anticipated to be from 10 to = 370 N/m2
15 years. The basic plan dimensions are 6.2 x 7.5 m
and the height to the eaves is 2.7 m. The gable roof has For cladding elements, the worst cases are
an overhang of 0.7 m on all sides and a slope of 10
degrees .. Openings for doors and windows are evenly p= (544) [0.8  (0.3) (0.85)]
distributed on the exterior walls. = 574 N/m 2
The dynamic presure is calculated from equation 2 of p = (544) [1.1  (O.2)J (0.85)
section 2.4.1 = 601 N/m2
Wind pressures are next calculated using equations p = (544) 10.6  (0.2)] (0.85)
46 and the coefficients presented in table 69. Note = 370 N/m2
that
Pressures acting on roofing elements in sections E and
h/w = 2.7/6.2 = 0.44 G ar.e ot;tained as follows:
and
l/w = 7.5/6.2 = 1.21 p = (544) [(1.1)( 1.05)  (0.2)(0.85)]
= 727 N/m2
WALLS
Inspection of tables 6 and 8 reveals that the worst and for sections F and H,
cases are walls A and C with the wind blowing nor
mal to the ridge. For wall A, Cl! = 0.8 and for wall C, p = (544) [(0.6)(1.05)  (0.2)(0.85)]
Cp = 0.6. The local Cp is 1.2. The internal pressure = 438 N/m2
coefficients can range fr.om 0.2 to 0.3. Table 9 indi
cates that the reduction factor is 0.6.5 for walls and in Localized pressures act on ".trips of width 0.15 w =
ternal pressures and 1.00 for cladding elements, doors, 0.93 m as shown in table 7. The worst case occurs for
windows, etc. area J with the wind bicwmg normal to thE' ridge.
Note that the uplift pressure under the eaves must
also be included.
22
p =(544) L1.9  (0.8)] (0.85) COMMENT
=1.2k N/ml
The loads calculated above are the loads that can
For area K in section F. this negative pressure or suc reasonably be expected to occur under the conditions
tion is slightly less stated in the example. They should be considered as
the minimum suitable loads for use with stresses and
p = (544) [1.4  (0.8)] (0.85) load factors appropriate for the type of structural
= l.Ok N/ml material used.
Along the ridge (area K). the localized pressure is For geographical areas exhibiting large variations in
annual extreme wind speeds. the basic wind speed
p = (544) [1.4  (0.2)] (0.85) should be selected. with caution. The application of
=740N/m2 probabilistic models of extreme wind speeds and
some of their limitations are discussed in section 1.0.
TOTAL UPLIFf FORCE
The total uplift force on the building is calculated for
the wind blowing normal to the ridge as follows:
Area of one roofslope = [7.5 + (2)(0.7)] [6.2/ (2Cos 10°)
+0.71
=(8.9)(3.85)
= 34.2m2
Note that areas E. F. G and H include areas J and K
when calculating overall loads.
Uplift =(544) (t.0 + 0.6) (34.2) (Cos 100 ) (0.85)
+ (544)(6.2)(7.5) (0.2)(0.85)
= 29.2kN
23
ANNOUNCEMENT OF NEW PUBLICATIONS IN
BUILDING SCIENCE SERIES
Superintendent of Documents,
Government Printing Office,
Washington, D.C., 20402
Dear Sir:
Please add my name to the announcement list of new pUblications to be
issued in the series: National Bureau of Standards Building Science Series.
~ame ___________________________________________________ 