SERI/TR-253-2169

UC Category: 62e
DE85012151

"

Wind Loading on Solar

Collectors

S. Bhadurl

(University of Texas at El Paso)

L. M. Murphy

(Solar Energy Research Institute)

June 1985

Prepared under Task Nos. 1384.30 and 5102.31

FTP No. 463

Solar Energy Research Institute

A Division of Midwest Research Institute

1617 Cole Boulevard

Golden, Colorado 80401
Prepared for the

U.S. Department of Energy

Contract No. DE-AC02-83CH10093

s=t

TR- 2 1 69

PREFACE

The research and development described in this document was conducted with in
the U . S . Department of Energy's Solar Thermal Technology Program . The goal of
the Sola r Thermal Technology Program is to advance the engineering and
scient i f ic understanding of solar thermal technology and to es tablish the
technology base from which private industry can develop solar thermal power
p roduct ion opt ions for introduct ion into the compet i t ive energy market .
Solar thermal technology concentrates the solar flux by means of tracking
mirrors or lenses ont o a receiver where the solar energy ' is absorbed as heat
and converted into electric ity or incorporated into products as process
heat .
The two primary solar thermal technologies , central receivers and
d i s t r ibuted receivers ,
employ var ious point and l ine-focus optics to
concentrat e sunligh t .
Current central rece iver systems use fields of
helios tats ( two-axis t racking mirrors ) to focus the sun' s radiant energy onto
a s ingle tower-mounted receive r . Parabolic dishes up to 17 meters in diameter
t rack the sun in two axes and use mirrors or Fresnel lenses to focus radiant
energy ont o a rece iver . Troughs and bowls are l ine-focus tracking ref lectors
that concentrate sunlight onto rece iver tubes along their focal l ines .
Concentrated collector modules can be used alone or in a mul t imodule system .
The concentrated radiant energy absorbed by the solar thermal receiver is
t ranspoted to the convers ion process by a circulat ing working f lu id .
Receiver temperatures range from 100 C in low-temperature t roughs to over
1 500 C in dish and central receiver systems .
The Solar Thermal Technology Program is d i recting efforts to advance and
improve each sys tem concept through the research and development of solar
thermal materials , components , and subsystems , and the test ing and performance
evaluat ion of subsystems and syst ems .
These efforts are carried out through
the technical d i rection of DOE and its network of nat ional laboratories that
work with private indus t ry .
Together they have established a comprehensive ,
goal-di rected program to improve performance and provide technically proven
o p t i ons for eventual incorporat ion into the nat ion' s energy supply .
To be succes s ful in contribut ing to an adequate nat ional energy supply at
reasonable cos t ,
solar thermal energy must eventually be economically
compet i t ive with a variety of other energy s ources .
Components and system
level performance target s have been developed as quant itat ive program goals .
The performance targets are used in planning research and development
act ivi t ie s , measuring progress , assess ing alternat ive technology opt ions , and
making optimal component developments .
These targets will be pursued
vigorous ly to insure a succes s ful program .
Th is speci f ic report addre sses wind loading on solar collectors s ince the
determinat ion of wind loading is one of the maj or de s ign cons iderations in
designing t racking and f i eld-mounted solar collectors . The main obj ect ive of
the report is to review and assess the present des ign methodology for wind
loading on collectors for solar thermal appl icat ions and to recommend areas of
further inve s t igat ion for developing rea l i s t ic cri teria to determine rel iable
and adequate wind loads . The fea s i b i l i ty of us ing innovat ive des ign consider
ations to reduce the magnitude of wind loads on the solar collectors is also
iii

TR- 2 1 69

cons idered .
This report contains valuable informat ion regarding the compara
t ive studies of various aspects of des ign methodology and should be of
interest to des ign engineers and developers of solar collectors .
The authors would like to thank the numerous individuals at SER I who provided
valuable d i scus s ions and support during the course of th is s tudy .
In
addi tion , the authors would like to thank those technologi s t s out side of SERI
who provided valuable comments and sugge s t i ons in their reviews of earl ier
drafts of th is document; these are Bill Delemeter , Clayton Mavi s , Jon Peterka ,
Jim Leonard , Martin S cheve , Frank Wilkins , Kenell Touryan , Jeff Shingleton ,
and Aharon Roy .
The first author extends his thanks and appreciation to the U . S . Department of
Energy ( DOE!) and the American Society for Engineering Education (ASEE) or the
opportunity to do th is research at SER I .
Th is area of research is supported
by the DOE Solar Thermal Program under the d i rect ion of Frank Wilkins and
Martin Scheve .

S . Bhaduri
Univers ity of Texas at El Paso

Approved for
SOLAR ENERGY RESEARCH INSTITUTE

Thermal Systems and Eng ineering Branch

Shannon , Director
Heat Research Divis ion

iv

SUMMARY

Objective

The 9bject ive is to review and assess the present methodology for predict ing
and designing for wind loading on solar collectors for t rough , d i sh , and
heli o s tat applicat ions and poss ibly for large f ield-mounted photovoltaic
arrays .
Another obj ect ive is to recommend areas of further invest igat ion for
develop ing rea listic criteria to adequately determine the wind loads on these
s t ructures .
,.f

Discussion

The
present
method
of
determining
the
wind
load ,
us ing
the
code
ANSI A58 . 1- 1982 of the American Nat ional Standard Ins t itute , is overly con
servat ive .
The recent model studies of solar col lectors in boundary layer
wind tunnels under s imulated atmospheric flow cond i t i ons yielded valuable
informat ion regarding the wind forces on the collectors and established var
ious dimensionless force and moment coef f ic ients corresponding to the mean
wind veloc it ies and the ir direct ion of approach .
A significant amount of
model test ing was devoted to s tudies involving both individual collectors and
fields of col lectors . The effects of porous fences or barriers at the edge of
the collector field and the sh ielding effects of the adj acent collectors in
the field in reducing the effect ive wind loads on the collectors were also
s tudied in the wind tunnels at various test facil ities . The results of these
s tudies indicate that the wind load can be reduced cons iderably even beyond
the levels that trough designers have already used .

Conclusions and Recommendations

In th is study , the following were accomp l i shed :

Development of elevation-dependent
various exposure condi t ions.

Comparat ive study of the total force coefficient of a flat plate of

various aspect rat ios and at various angles of attack .
Values obtained
for the ANS I A58 . 1- 19 8 2 code , ASCE Task Comm . 19 6 1 [ 23 ] , and full-scale
heli os tat tests were used for the comparison .

Comparat ive study of the spect ra of longitud inal , ver t ical , and lateral
Spectra for the veloc i ty f luctuat ions for fre
veloci ty fluctuat ions .
quency range 0 . 001-10 . 0 Hz were calculated and compared graph ically . The
spectra used for the Bui l d ing Code ANSI A58 . 1- 1982 and the spectra in
Kaimal et al . [ 26 ] were cons idered .

Est imat ion of the resultant angle of attack ( relat ive to the horizontal )
of the turbulent wind field for var ious exposures such as open field and
suburban location .
Th is is particularly important when cons idering
heliostats in the stowed pos i t ion . The angle of attack due to turbulent
fluctuat ing veloc ity is s trongly dependent on the roughness parameter of
the flow field .
For a roughness element of 0 . 12 3 m ( 0 . 4 ft ) in a
v

longitudinal

velocity

p rofiles

for

TR - 2 1 69

suburban area , and wind speeds o f 30 mph , 50 mph , and 9 0 mph at an eleva
t ion of 10m (3 2 . 8 ft) , the angles of attack due to fluctuating vert i cal
veloc i ty are 4 . 68 , 4.60, and 4.30, respectively .
Recommendations for further studies that might reduce the wind loading on the
collector are to

Mod i fy the flow locally by using a turbulence stimulator in the form of

porous eaves around the heliostat to discourage the flow separation at
the 9dges of the heliostat

Determine experimentally the appropriate air gap between the modules to

encourage ventila t ion around the module
.1

S tudy the effects of perimeter fen ce s and fences in the field around a
group 'of heliostat zones to reduce the overall wind load

Determine the magnitude of wind speed and its direc t i on by developing a

wind rose for the helios tat f ield .
A physical scale model o this
hel iostat field may be te s ted in the atmospheric wind tunnel to obtain
informat i on regarding the wind speed , its direc t i on , and turbulent
character i s t ics .

vi

TR-2 1 69

TABLE OF OOBTENTS

1 . 0 Introdue: t ion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2 . 0 Est imates of Wind Forces

2.1
2.2

,, .

.. . . . .. . .

Atmospheric Turbulence S pectra

Variat ion of Wind Angle from the Horizontal .

3 . 0 Reduct i on of Wi.1J Loading

19
23

............................... . ....

29

4 . 0 Conclus i on and Recommendati ons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

32

5. 0 Refe rences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

34

vii

TR-2169

LIST OF FIGURES

2-1

Dis tribut ion of Gradient Wind for Var i ous Exposures

2-2

Var iat ion of Wind Speed with Elevat ion ,

2-3

Var iat ion of Wind Speed with Elevat ion ,

24

Var iat ion of Wind Speed with Elevation ,

2-5

Var iat ion of Wind Speed with Elevat ion ,

2-6

Var iat ion of Veloci ty Pressure with Wind Veloci ty

2-7

A Geometry for Flow past a Flat Plate

2-8

Var iat ion of Force Coeffic ient with Angle of At tack ,

Aspect Ratio , = 1 .0

(u30 ) c
(u30>c
(u30 ) c
(u30 )c

27

mph . . . . . . . . . . . . .

14

50

mph

14

70

mph

90

mph

15
15
16
17
18

2-9

Var iat ion of Force Coef f i cient with Angl e of At tack ,

Aspect R at io , A = 3 .0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2-10

Longitudinal Turbulence Spectra , U =

2-11

T urbulent Wind S pectra , U =

44 .73

mph . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

22

2-12

Turbulent Wind Spectra , U =

90 .0

mph . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . .

23

2-13

Strong Wind Veloc ity Spectra of Daniels

2-14

Peak Angle of Attack due to Turbulent Lateral Velocity

2-15

Peak Angle of Attack due to Turbulent Ver t ical Veloci ty

vi i i

44 .73

mph

19
22

24
27
28

TR-2 1 69

LIST OF TABLES

2-1

Gradi ent Wind

2-2

Variat ion of Wind Velocity with Heigh t , 03o

27 . 0 mph
=

2-3

Variat i on of Wind Velocity with Heigh t , 0 3o

50 . 0 mph
=

2-4

Variat ion of Wind Velocity with Heigh t , 03o

70 . 0 mpb
=

2-5

variat i on of Wind Veloci ty with Heigh t , 0 3o

2-6

Veloci ty Pres sure . ... . . ............... . ... ... ... . . .... . . . ...........

16

Force Coefficient C f for Flow past a Flat Plate , Aspect R at i o ,

1 . 0 . . . . . . . . . . .... . .. . .... . ....... . . ... .... . .. .. . . . ... .. . . .......

17

Force Coefficient C f for Flow past a Flat Plate , Aspect Ratio ,

3 . 0 . . . .. . . . . . . . ............. . . ...... ... . ... ..... ... . . .. ........ .

18

Dimensionless Constants for Longi tudinal and Lateral Turbulence

Spec t ra . . . . . . . . . . . . .. . . ...... . .......... . . . . ... . . .. . . . . . .. . . .... . .. .

25

2-10

Variation of Angle of Attack due to Fluctuat ing Lateral Veloci ty

26

2- 1 1

Variat ion of Angle of Attack due to Fluctuat ing Vert ical

Veloci ty . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2-7

A.

2-8

A.

2-9

9 0 . 0 mph

ix

8
10
11
12
13

27

:'fR-2169

NOMENCLATURE

A , B , C ,D

basic exposure categories

characteris t ic surface area of collector , m 2 ( f t 2 )
force coefficient , d imens i onless
moment coefficient , dimens ionless
wind force , N (lbf )
drag force , N (lbf )
l ift force , N (lbf)
Monin coordinate , dimens ionless
\

gust response factor at height z = h , dimens i onless

veloci ty pre s sure exposure coefficient at height z
moment arm length, m (ft )
moment of a force , N m (lbf ft )
n

frequency , H z

dynamic pressure , N/m2 (lb/ft 2 )

qz

velocity pressure evaluated at height z , N/m 2 (lb / f t 2 )

S(z ,n)

spectrum of longitudinal veloci ty fluctuations , N m/n (lbf f t s )

Sw ( z ,n)

spec t rum o f vert ical veloc i ty fluctuati ons , N m/n (lbf ft s )

S v ( z ,n)

spectrum of lateral velocity fluctuat ions , N m/n (lbf f t s )

U(z)

mean veloc ity a t height z

ug
uz

longitudinal component of veloci ty at gradient height zg

long i tud inal component of veloci ty at height z

u*

frict ion veloci ty , m/s (ft / s )

basic wind velocity, m/s (mph)

vert ical elevation , m (ft )

reference height of ANSI code , m ( f t )
gradient height , m (ft )
reference height , m ( f t )
angle of attack , degree
coefficient ( z / z r ) c 5
aspect rat io
mass dens ity , kg/m 3 (slug/f t 3 )
s tabi l i ty parameter

f requency , Hz

TR-2 1 69

SECTION 1. 0
INTRODUCTION

During the last two decades, there have been a number of s ignif i cant develop
ments in the meteorological aspects of the wind-loading problem. Advances in
s t ructural designs, particularly for high-rise s t ructures, requi red accurate
and more reliable meteorological information about the wind-loading pro
cesses. Vari ous research results and histories of fai lures of structures [1 ]
indi cate the damaging effects of repeated wind-induced loads on structures.
Efforts have been made to design s tructures that resi s t the repeated loading
act i on of wind [ 2, 3 ] . This contrad i cts the more conventi onal design approach,
whi ch i s mafnly concerned with s tatic app l i cation of a s ingle large load
derived from the guidelines furnished by the National Bui ld ing Codes.
Experience suggests that unserviceabi l i ty due to repeated loading ef fects is a
more likely occurrence than unserviceab i l i ty or collapse from a single appli
cati on of an except i onally large load [ 4 ].
The common types of structural
unserviceabi l i ty caused by repeated wind loading are ( 1 ) fatigue failure,
( 2 ) foundati on s e t tlement, ( 3 ) excess ive deflecti on, and (4 ) induced mot i on of
unacceptable level. Fatigue failure is an important des i gn consi derati on even
for the design of low-rise structures.
Examples of fat igue failure due to
wind are found in structures such as towers, lamp standards, chimney stacks,
and even bridges [ 5 ]. As innovat ive concentrator des igns become less robu s t
to reduce the amount o f materials and cos ts requi red, a l l o f these i s sues
become more of a concern. An example of such a lightweight concentrator i s
the stret ched-membrane heliostat.
W ind is one of the principal loads acting on above-ground engineering struc
tures. An accurate determinati on of wind loading is a fundamental desi gn con
s iderati on in deciding what degree of safety and economy can be achieved in
such a s tructure.
It is common practice to design s tructures to resi st the
highes t wind to whi ch they may be subjected. The highest win d is determined
from the analysi s of extreme value s tat is t i cs of recorded wind velocities for
a long period of t ime for a parti cu lar region.
The wind loads are consi dered in terms of a s teady applied force, independent
of the struc tu re's shape, s i z e, and dynami c character i s tics, and a coefficient
denot ing the amp l i f i cation .that can arise through the interaction of the
superimposed pressure fluctuati ons and the structure's dynamic response. The
reli able estimate of this coef f i ci ent is qui te involved and requires detailed
knowledge of the s ite-speci f i c wind climatology. The wind speeds used in cur
rent desi gn s pecif i cation are based on the u.s. Weather Bureau observat ion of
the "fast e s t mile"* of wind (from distri bu t i on of extreme wind in the Uni ted
States ) mul t i pl ied by a gust factor to allow for the f luctuat i ons in the w ind
The gus t response factor is a measure of the effective dynami c load
speed.
produced by the gus ts and is intended to translate the dynamic res ponses pro
duced by the gus t loading into s impler stati c des ign cri te r ia. Althugh thi s

*"Fastest mile" corresponds to the highest (extreme ) wind speed in miles per
hour, measured at a s tandard hei ght of 33 ft above the ground level in an
open-country location.
1

TR-2169

approach has often re sulted in a safe , conservative est imate of wind loading ,
the neglect of the dynamic properties and si z e of the structure , the surface
roughnes s , local obstruct ions , and the wind directions and their frequencies
inherent in applying the fas test mile of wind could result in unsafe struc
tures or in costly overdes igns .
Vello zzi and Cohen [ 6 ] ieveloped a des ign methodology based on a power spec
t rum analys is of the dynamic s t ructural response of a s imple l inear s ingle
degree of freedom system wth v iscous damp ing and reported measurements of
wind gust spectra and gust corre lat ion coefficients .
The results of the
analysis are presented in equat ions and chart s that permit us to determine the
gust response factors. for buildings and conventional s t ru tures . However , for
s tructures like solar collectors , the method is not sat isfactory s ince it does
not account for the effects of surface roughness , terrain characterist ics , and
d i rect ional distribut ion of wind .
We st ill do not understand the mechani sm
whereby gust pressure is induced on ground-based s t ructures .
Moreove r , these
des ign guidelines are app l i cable only to the s ituat ions where the principal
wind loading is drag .
Lateral and vert ical wind components normal to the
gross flow d i rect ion and gust iness are not cons idered .
Davenport [ 7] used a stat i s t ical approach to estimate the mean gradient wind
speed that has a speci f ied return period or probab i l ity of exceeding on the
bas is of the mode and d ispersion factor for the extreme gradient wind speed
f ield at the s i te of the s tructure .
This method may be used to determine the
mean wind speed at some spec i f ic height above the ground if a reliable
e s t imat e of ground roughnes s and exposure of the site is ava ilable .
The
ground roughness is the p rinc ipal parameter governing the mean wind speed
profile and gus t ines s .
Knowing the mean wind speed at the structure height
corresponding to the grad ient speed and the ground roughness , we can define
the vert ical and hori z ontal gust spectra . Davenport based his est imate of the
gust force superimposed on the mean wind forces on the lateral gust spectra .
In his pioneering paper , Davenport [ 7 ] outlined the rat ionale for determining
the des ign wind veloc i t ie s .
Brief ly , he cons idered the extreme value
s tat i s t ics of the site , speci f ic wind veloci ty , local ground roughness and
exposure
characteristics ,
wind
d i rect ion
roset tes ,
and
a
weight ing
parameter .
The methodology outlines the determinat ion of the bas ic des ign
wind veloc i ty that corresponds to the extreme sustained wind velocity giving
rise to the steady component of the pressure . This veloci ty is essent ially an
average veloci t y , and the determinat ion of a suitable averaging interval is an
important feature of the method .
Davenport attempts to process the records
from a wide variety of anemometers of d i ffering exposures and periods of
records and to determine stat i s t ically how the records may be related to one
anothe r .
This minimi zes the systemat ic errors that may arise in the records
because of uncontrolled anemometer s i t ing and improves the records' overall
rel iability . The influence of the local surface roughness on both the magni
tude of surface veloc i t ies and the increase of veloc i ty with height is very
important . Combining these results with extreme value s tat i s t ics [8 ] leads t o
a qualitat ive method through which bas ic des ign wind velocit ies of a given
p robabi l i ty of occurrence may be predicted for locat ions of d i ffering surface
roughnes s .

TR-2 169

Thorn [ 9] introduced the concept of the annual fastest mile wind speed as the
best available measure of wind speed for design purposes. The physical model
for the boundary shear layer was assumed to have a standard level of 9.14 m
( 30 ft ) and a velocity profile us1ng a 1/7 power law. Computational methods
were developed for fi t ting the Fisher-Tippett Type II extreme value dis tribu
tion [ 1 0] , which fit ted the data wel l. The methods developed [ 9] were applied
to the airport and open-country data series for 1 38 s tations; maps were drawn
for the 2-year , 50-year , and 1 00-year mean recurrence intervals. These maps
received general acceptance as the u.s. standards. The methods of determining
other quantities , such as probabilities for a fixed period and confidence
limits for the estimates obtained from the maps , were given

.J

In his subsequent paper , Thorn [ 1 1] observed that the distribution of the

logarithm of a Fisher-Tippett Type II ( Frechet ) distribution [5] is a
logarithmic transformation of the Type I distribution. Although this fact was
not used in the previous paper , further examination of the extensive non
extreme wind data showed that such data follow a log normal distribution
closely , which would indicate a Type I distribution for the extreme of the
logarithm. This also substantiates the validity of the choice of the Frechet
distribu tion for extreme winds. The maps developed by Thorn [ 1 1] in which the
Type I distribution is fit by order statistics to 150 stations were used
extensively as the ANSI standard for building codes of the Uni t ed States. The
maps depicted the isot ech lines in miles per hour of the annual extreme mile ,
1 0.0 m ( 33 f t ) above the ground, for 2-year , 1 0-year , 25 -year , 5 0-year , and
1 00-year mean recurrence intervals.
The basic wind speeds used to determine the design wind loads on buildings and
other structures are given in Figure 1 of the ANSI A5 8. 1 - 19 82 code [ 1 2] for
the contiguous United States and Alaska and in Table 7 for Hawaii and Puer to
Rico. Special considerations are allowed for those regions for which records
or experience indicate that the wind speeds are lighter than those indic ated
in Figure 1 and Table 7 of the code [ 1 2].
The ANSI A5 8. 1 -1982 code also
allows estimation of the basic wind speed based on regional climatic data arrd
extreme value s tatistical analysis; however , the basic wind speed must be at
least 70 mph.
All investigators recognize the qualitative nature of the idealized isotech
lines based on data co llected under diverse conditions and suggest that they
be used with great care in determining the wind load for a specific structure
in a specific site.
Often, the probability of exceeding a particular wind
velocity for a certain recurrence interval is related to the degree of risk
associated with the lifetime of the structure. The probability that a struc
ture wil l be subj ected to a magnitude of wind loading is not site specific
since wind data do not ref lect the true nature of wind distribution at the
site in question.
The ground roughness , the exposure features , wind direc
tion , and terrain conditions are not included in the methodo logy used by the
codes.
The American National Standard A58. 1 -1982 [ 1 2] contains a definition of design
wind speed in terms of the mean recurrence interval or return period. For a
mean recurrence interval , the corresponding design wind speed is calculated
from the expression of the cumulative dis tribution function of the extreme
wind.
The parameters of this func tion are estimated from the data of the

TR-2 1 69

largest wind speed for every year on record. The mean recurrence interval to
be used in the design is selected as a func tion of "intended operational
usage , anticipated life of the s t ructure , degree of sensi tivity to wind and
the risk of loss of human life and property in case of failure" [ 1 2 ] .
Implicit in the selec tion of the mean recurrence interval is the assumption
that a given recurrence interval will ensure the same level of safety for any
two s t ructures subj ected to wind loads provid ed that the reliability of the
two structures is equal. Simiu [ 1 3 ] examined the validity of this simplified
assumption and showed conclusively that the use of design wind maps based on
the standard mean recu rrence interval does not ensure a consistent level of
Simiu emphasized that exten
safety for s t ructures subj ected to wind loads.
sive research is required to develop regional and site-specific wind speed
provisions based on a reliability consideration.
\

Because of the uncertainty and qualitative nature of the methodology used to

estimate wind loading on s t ructures , it seems reasonable to obtain the design
parameters from simulation studies of sites under various wind speeds and wind
directions.
Various simulated experimental studies in the wind tunnels have
already yie lded valuable information.
Murphy [ 1 4 ] reviewed the exis ting methodologies for es timating the wind
loading on solar collectors.
He summarized 33 s tudies relevant to wind
loading on solar collectors , heliost at s , and photovoltaic arrayse The report ,
which contains valuable information regarding the p resent s tate of the art ,
has been used as a guide for this study.
Roschke [ 15] examined the problems and comp lications arising from wind loading
on solar concentrators.
He emphasized the site-specific nature of wind
loading with associated important bearing on the design, cos t , performance ,
operation and maintenance , safety , survival , and replacement of solar col
lectors , particularly the paraboloidal concentrators.

i!
i

li
I

The primary obj ective of this study is to review and assess the present design
methodology on wind loading on solar collectors , particularly th heliostat s ,
and to recommend areas of further inves tigation to define realistic criteria
for determining adequate wind loads.
The model tes ting of the solar collec
tors in the environmental wind tunnels [ 1 6 , 1 7 ] indicates that it is possible
In view of the feasibility of load
to reduce the wind load considerably.
reduction , we compare studies of the total force coefficient of a flat plate
of various as pect ratios and at various angles of attack and studies of the
spectra of longitudinal , vertical , and lateral velocity fluctuations.
The
angles of a t t ac k of the turbulent wind to the horizontally placed f lat plate
are important because of their effects on the net moment of the structure. We
at tempt to e stimate the magnitude of angles from turbulent spectrum analysis.
This study does not consider the wind-loading effects caused by turbulence;
i.e. , ( 1 ) buffeting , ( 2 ) vortex shedding forces produced by the s t ructure's
own wake ac ting on the afterbody of the structure behind the separation point ,
and (3 ) aeroelas tic forces induced by the oscillating motion of the structure.

il

!Il
!I
1'1I

TR-2169

SECTION 2.0
ESTIMATES OF VIND FORCES

The wind load on the low-height s t ructures , such as hel i o s tats extending over
a large land area , is a major des ign cons ideration. Properly determining the
wind loads on the hel i o s tats is essential for rea l i s t ic s t ructural des igns and
cost est imates.
The ANSI A58.1-198 2 code [12 ] is commonly used as a des ign
guide.
The code is primarily developed for bui ldings and other structures.
Generally , est imates are made , in the absence of more rel iable experimental
values obtained from model s tudy , by cons idering the heliostats as other
s tructure s.
The wind force F on a s t ructure is given by
(2-1 )
where
qz

= velocity pressure evaluated at height z above ground ( lb/ft 2 ) *

Gh

= gust response factor at height z = h

Cf

= force coe f f i cient

Ap

= projected area normal to wind.

The velocity pres sure q z i s calculated from the s tandard formula [ 1 2 ]

q z = 0.00256Kz V2 ,

(2-2)

where
Kz

= veloc i ty pres sure exposure coef f ic i ent at height z

= basic wind speed (mph).

The values of the various coef ficients and factors are taken f rom the tables
p rovided in the code for var ious expQsures of the s ite of the structure. An
exposure category is determined for the s ite at which the structure i s
s ituated.
It reflects the characterist ics of ground
surface i rregularities .
' I
Generally , four bas ic exposure categories are considered. Exposures A , B , C,
and D represent the large c i ty cente r , urban and suburban areas , open t errain ,
and flat , uobstructed coas tal areas , respectively.

*S ince the standards and codes used for comparat ive assessments are described
in their original form us ing Engli sh uni t s , SI units will follow the Engli sh
equivalents.

TR-2169

The structure is designed for the environmental conditions with specified wind
speed and mean recurrence interval.
Generally, operating environmental con
ditions and extreme environmental conditions with assumed mean recurrence
interval and expected life of the structure are considered to estimate the
probability of exceeding the maximum design wind velocity. The probability is
derived from the consideration of the extreme value ordered statistics of Thorn
[9]. The probability of exceeding the design velocity, or risk of occurrence
as it is called, is not based on site-specific extreme wind records.
The
recorded wind data at the airport of a relatively large city may not be
applicable to a smaller suburban city 50 miles away with different terrain
features.
In fact Simiu [13], Davenport [7], and McDonald [18] have ques
tioned the validity of the basic assumptions and simplified methodology.
The variation\ of longitudinal component of wind velocity with elevation in the
earth's lower atmosphere is expressed by
Uz = Ug

( )
:

(2-3)

where
Uz

= longitudinal component of velocity at height z

Ug

= longitudinal component of velocity at gradient height z

= an exponent.

The following analytical relations of the velocity profiles relative to the

velocity u30 at a height equal to 30 ft have been derived for ready reference.
Exposure C:

zg = 900.0 ft

n = 0.1429

Exposure B:

zg = 1200.0 ft

n = 0.2222

(2-5)

z g = 1500.0 ft
n = 0.3333
)
(z 0.3333
Uz = u3o
30

(2-6)

Uz = u3o
Exposure A :

(2-4)

() 0.1429
Uz - U3o
30

( z )0.2222
30

The longitudinal components of wind velocity at elevation z = 30 ft for the

exposure conditions A, B , and C are related by
and

(2-7b)

TR-2169

Table 2-1 shows the distribution of gradient wind for various exposures. Fig
ure 2-1 graphically shows the dimensionless gradient wind distribution with
elevation for the exposures of A, B, and C. Tables 2-2 through 2-5 give the
values of the velocity variation with height for reference velocities of
27.0 mph, 50.0 mph, 70.0 mph, and 90. 0 mph, respectively. Figures 2-2 through
2-5 show the variations of wind speed with elevation for the exposures.
( u30) c is the velocity of wind at an elevation z = 30 ft for exposure category
C. Table 2-6 gives the computed values of velocity pressure obtained from the
formula qz = 0.00256 v2
Figure 2-6 shows the variation of velocity pressure
with wind velocity.
In recent years, the model studies of structures in simulaed flow fields have
yielded more reliable wind-load design values expressed in terms of
dimensionless\ force and moment coefficients. The aerodynamics of structures
in turbulent shear flows of the earth 1 s lower atmosphere involve nonlinear
interaction of the turbulent approach--flow with highly complex turbulent
boundary layers and subsequent boundary layer separation and turbulent wake
formation.
Cermak [19] and Bhaduri [20] showed the feasibility of simulating
the
earth 1 s
boundary
layers
in the
properly
designed wind tunnel.
Lindsay [21], Randall et al. [22], Peglow [16], and Xerikos et al. [17] tested
the models of the solar collectors, heliostats, and heliostat arrays in the
environmental wind tunnels and presented useful experimental results that
could be used for determining design wind loads.
Based on the principle of similitude in model studies, the equality of the
force and moment coefficients for the model and for the prototype, respec
tively, is well established. The force coefficient C is defined by
f
F
(2-8)
Cf =
qAp'
where
F

= resultant force acting on the structure

= dynamic pressure, often called velocity pressure

A p = projected area normal to wind.

The moment coefficient CM is defined by

M
CM =
qApL '

(2-9)

where
M

= moment of the total force about a reference point

= a characteristic moment arm length.

The dynamic pressure q is expressed by

q =

2
-k
u
2

(2-10)

TR-2169

55,11*'
Table 2-1.

Gradient Vind

c
z (ft)

z/z

= 900 ft
(274.32 m )
8

= 1200 ft
(365.76 m)

U/Ug

z / zg

U/Ug

Z g = 1500 ft
(457.20 m )
z/z

U/Ug
0

15

0.017

0.557

0.013

0.378

0.010

0.215

30

0.033

0.615

0.025

0.441

0.020

0.271

40

0.044

0.641

0.033

0.470

0.027

0.299

50

0.056

0.662

0.042

0.494

0.033

0.322

75

0.083

0.701

0.063

0.540

0.050

0.368

100

0.111

0.731

0.083

0.576

0.067

0.406

200

0.222

0.807

0.167

0.672

0.133

0.511

300

0.333

0.855

0.250

0.735

0.200

0.585

400

0.444

0.891

0.333

0.783

0.267

0.644

500

0.555

0.919

Oo417

0.823

0.333

0.693

600

0.667

0.944

0.500

0.857

0.400

0.737

700

0.778

0.965

0.583

0.887

0.467

0.776

800

0.889

0.983

0.667

0.914

0.533

0.811

900

1.000

1.000

0.750

0.938

0.600

0.843

1000

0.833

0.960

0.667

0.874

1100

0.917

0.981

0.733

0.902

1200

1.000

1.000

0.800

0.928

1300

0.867

0.953

1400

0.933

0.977

1500

1.000

1.000

,f

TR-2169
1;0

r---

Exposure categories:

A large city center

B. urban and suburban area
C. open terrain

0.4

0.2

0.2

0.4

0.6

0.8

1.0

U!Ug
Figure 2-1.

Distribution of Gradient Vind for Various Exposures

where
p

= mass density of air

= mean approach velocity.

The wind effects can be resolved into two force components--drag force F and
D
lift force F1. They are expressed as

(2-11)
where

CD = drag coefficient
Ap = projected area normal to the wind;
and

(2-12)
where
c

1 = lift coefficient.

From the vector nature of the force and its components, it can be shown that

2
F = (F
D

1
2) /2 ,
F
L

( 2-13)

TR-2I69

s-
-

II
II
:':!!

Table 2-2.

Variation of Vind Velocity with Height,

u30 = 27 mph (I2 m/s)

Exposure C

Elevation
(ft) 8

Wind Speed
v

(mph) b
0

Exposure A

Exposure B

Velocity
Pressure
q (lb/ft2) c

Wind Speed
v

(mph)

Velocity
Pressur

q (lb/ft )

Wind Speed
v

(mph)
0

Velocity
Pres sur

q (lb/ft )

23.I

1.37

IS.2

O.S9

8.3

O.I8

24.6

l.SS

I6.6

0.71

9.S

0.23

20

2S.6

1.68

I7.7

0.80

I0.4

0.28

2S

26.4

1.78

I8.6

0.89

II.2

0.32

30

27.0

1.87

I9.4

0.96

I2.0

0.37

40

28.20

2.04

20.7

I.IO

I3.2

0.4S

so

29.2

2.I8

21.7

1.2I

I4.2

O.S2

7S

30.86

2.44

23.8

1.46

I6.2

0.67

IOO

32.2

2.6S

25.3

1.64

I7.9

0.82

ISO

34.1

2.98

27.7

1.96

20.4

1.07

200

3S.4

3.2I

29.6

2.24

22.S

1.30

300

37.6

3.62

32.3

2.67

2S.7

1.69

soo

40.2

4.I4

36.2

3.3S

30.SO

2.38

700

42.4

4.60

39.0

3.89

34.I

2.98

900

44.0

4.96

4I.28

4.36

37.I

3.S2

1000

42.3

4.S8

38.S

3.79

llOO

43.2

4.78

39.7

4.03

I200

44.0

4.96

40.9

4.28

1300

42.0

4.S2

I400

43.0

4.73

lSOO

44.0

4.96

IO
IS

8 To convert to m, multiply by 0.3048.

bTo convert to m/s, multiply by 0.447.
cTo convert to Pa, multiply by 47.88.

IO

TR-2169

Table 2-3.

Elevation
(ft) a

u30 = 50 mph (22 m/s)

Exposure C
Wind Speed
v

(mph) b
0

Variation of Wind Velocity with Height,

Exposure B

Velocity
Pressure
p (lb/ft2) c

Wind Speed

Velocity
Pressur
q (lb/ft )

(mph)

Exposure A
Wind Speed
v

(mph)
0

Velocity
Pressure
q (lb/ft2)
0

10

42.8

4.69

28.1

2.02 ,

15.3

0.60

15

45.4

5.28

30.8

2.43

17.6

0.79

20

47.3

5.73

32.8

2.75

19.3

0.95

25

48.8

6.10

34.5

3.05

20.8

1.11

30

50.0

6.40

35.9

3.30

22.1

1.25

40

52.2

6.9
, 8

38.30

3.76

24.40

1.52

50

53.9

7.44

40.2

4.14

26.2

1.76

75

57.1

8.35

44.0

4.96

30.0

2.30

100

59.5

9.06

46.9

5.63

33.1

2.80

200

65.7

11.05

54.7

7.66

41.6

4.43

500

74.9

14.36

67.1

11.53

56.5

8.17

700

78.6

15.82

72.3

13.38

63.2

10.23

900

81.5

17.00

76.5

14.90

68.7

12.08

81.5

17 .oo

75.7

14.67

81.5

11.00

1200
1500
aTo convert to m, multiply by 0.3048.
bTo convert to m/s, multiply by 0.447.
cTo convert to Pa, multiply by
47.88.
and
c

= (c

2
D

1/z
c 2)
L

(2-14)

To compare the values of Cf obtained by using the guidelines of the

ANS I A58.1-1982 code with the values obtained by using the methodology of the
ASCE Task Committee of 1961 [ 23], flows past vertical plates with aspect
ratios, A = (width/height) , of 3 and 1 are considered (see Figure 2-7). The
experimental results of Peglow's [16] full-scale test of a heliostat of aspect
ratio A = 1.0 are also used for the comparison.
For angles of attack of
0-30, the corresponding values of force coefficient Cf are taken from
Table 11 of ANS I A58.1-1982 code [12] and Figure 5(d) of Paper 3269 of the

11

55fl'*'

TR'-1"2169

Table 2-4.

Elevtion
a
(ft )

Variation of Wind Velocity with Height,

30

= 70 mph (31 m/s)

Exposure c
Wind Speed
v

(mph)
0

10

Velocity
Pressu
c
p (lb/ft )

Exposure B
Wind Speed
v

(mph)
0

Exposure A

Velocity
Pressur2
q (lb/ft )

Wind Speed
v

(mph)
0

Velocity
Pressur2
q (lb/ft )
0

21.4

1.17

4.73

24.5

1.56

45.8

5.37

27.0

1.87

12.54

50.1

6.43

30.90

2.44

75.3

14.52

56.2

8.09

36.6

3.43

100

83.1

17.68

65.5

10.98

46.2

5.46

200

91.8

21.57

76.4

14.94

58.2

8.67

300

97.3

24.24

83.6

17.89

66.60

11.36

400

101.4

26.32

89.2

20.37

73.3

13.75

500

104.6

28.01

93.7

22.48

79.0

15.98

600

107.4

29.53

97.6

24.39

83.9

18.02

700

109.8

30.86

100.9

26.06

88.3

19.96

800

ll1.9

32.06

104.0

27.69

92.3

21.81

900

113.8

33.15

106.8

29.20

96.0

23.59

1000

109.3

30.58

99.4

25,.29

llOO

111.6

31.88

102.6

26.95

1200

ll3.8

33.15

105.7

28.60

1300

108.5

30.14

1400

1ll.2

31.66

1500

113.8

33.15

59.8

9.15

39.3

3.95

63.4

10.29

43.0

66.1

ll.19

30

70.0

so

15
20

To

convert to m, multiply by 0.3048.

To convert to m/s, multiply by 0.447.
c To convert to Pa, multiply by 47.88.

12

,,

TR-2169

5511!11
Table 2-5.

Elevation
(ft) a

u30 = 90 mph (40 m/s)

Exposure c
Wind Speed
v

(mph)b
0

Variation of Vind Velocity with Height,

Velocity
Pressure
q (lb/ft 2) c

Exposure B
Wind Speed
v

(mph)

Exposure A

Velocity
Pres sur
q (lb/ft )

77 .o

15.18

50.5

6.53

15

81.6

17.05

55.3

20

85.0

18.05

25

87.7

30

(mph)
0

10

Wind Speed

Velocity
Pressur
q (lb/ft )
0

27.5

1.94

7.83

31.5

2.54

58.9

8.88

34.7

3.08

19.69

61.9

9.81

37.4

3.58

90.0

20.74

64.5

10.65

39.7

4.03

40

93.8

22.52

68.7

12.08

43.7

4.89

50

96.1

23.64

72.2

13.34

47.1

5.68

75

102.6

26.95

79.0

15.98

53.9

7.44

100

106.9

29.25

84.2

18.15

59.3

9.00

200

118.0

35.65

98.3

24.74

74.8

14.32

500

134.5

46.31

120.5

37.17

101.5

26.37

700

141.2

51.04

129.8

43.25

113.5

32.98

900

146.3

54.79

137.3

48.26

123.4

38.98

146.3

54.79

135.8

47.21

146.3

54.}9

1200
1500

"

aTo convert to m, multiply by 0.3048.

bTo convert to m/s, multiply by 0.447.
cTo convert to Pa, multiply by 47.88.
Task Committee (TC) [23]. The results are shown in Tables 2-7 and 2-8.
are also presented graphically in Figures 2-8 and 2-9.

They

The ratios of Cf (code)/Cf (TC) and Cf (code)/Cf (heliostat) indicate that the
code values,. within the range of the angles of attack, are generally 10% to
25% higher for A = 1.0. For A = 3.0, they are 22% to 25% higher. Thus the
code gives an overestimate of the wind loads for the flow past an inclined
flat plate. The values of Cf estimated by using the Task Committee guidelines
compare favorably with the experimental results of the full -scale heliostat
testing. Figures 2-8 and 2-9 show that the ANSI coefficient Cf is moderately
conservative :l.f the angle of attack is less than 15 and that it is somewhat
more conservative at higher angles of attack.
For longer plates, the code
values are generally more conservative.
13

TR-2169

5=11"' 1
-

=
.

c:
0
:.;::::;
co
>
Q)

[jj
Exposure categories:
A. larg e c i ty center
B. urban and suburban area
C. open terra i n

10

20

W i n d speed ( m p h )

Figure 2-2.

30

40

Variation of Vind Speed with Elevation,

(u30>c

= 27 mph

c:
0
.;:::;
co
>
Q)

[jj

20
0====------L---
80
60
0
40
20
W i n d speed ( m p h )

Figure 2-3.

Variation of Vind Speed with Elevation,

14

(u30>c

50 mph

s-
-

TR-2 1 69

II
II

200
180
160
140
120
100
iii 80
60
40
20
0

:t:..
c:
0
:;::;
ctS
>
Q)

Exposure categories:
A. large c i ty center

B. urban and suburban area

C. open terrai n

Figure 2-4.

20

40

60

Wind speed ( m p h )

80

100

Variation of Wind Speed with Elevation,

(u30)c

70 mph

Exposure categories:
A. large c i ty center

B. urban and suburban area

C. open terrain

:t:..
c:
0
:;::;
ctS
>
Q)

iii

Wind speed ( m p h )

Figure 2-5.

Variation of Wind Speed with Elevation,

15

(u30>c = 90

mph

Table 2-6.

Wind Velocity
V (mph}8

q =

10
20

Dynamic Pressure

o.00256 v2 (lb/ft2}b

Wind Velocity

v (mph}

q =

Dynamic Pressure

0.00256 v2 (lb/ft2}
124
147
173
201
230

220
240
260
280
300

0.26
1.0
2.5
4.1
6.4
9.2
12.6
16.4
20.7
25.6
37.0
50.2
65.5
83.0
102.4

30

40
50
60
70
80
90
100
120
140
160
180
200

Velocity Pressure

,J

8To convert to m/s, multiply by 0.447.

bTo convert to Pa, multiply by 47.88.

40

Wind velocity (m/s)

---------

Velocity pressure, qz
-

0.00256V2,

where V

wind velocity, mph

::;,
(/)

a.

>.

"(3
0

Q)
>

Wind velocity (mp h )

Figure 2-6.

Variation of Velocity Pressure with Wind Velocity

16

TR- 2 1 69

T
1 I

0
"'
<D
"'

0
0

Figure 2-7.

.i\

alb

A Geometry for Flow past a Flat Plate

Table 2-7.

Cf for now past a nat Plate,

Aspect R at i o , A = 1 . 0

Angle of A t tack
( deg )
0
10
15
20
25
30

o:

cf
( Code )

cf
Task Comm .

0
0 . 45
0 . 70
0 . 90
1 . 15
1 . 30

Code values are g iven only up to

bsee Peglow

0
0 . 39
0 . 62
0 . 84
0 . 96
1 . 04
o:

cf

code

cf
Tc
0
1 . 15
1 . 13
1 . 07
1 . 20
1 . 25

= 30 .

[ 1 6] .

17

cf
Heliostat
Full-S ale
Tes t
0
0 . 39
0 . 64
0 . 84
0 .98
1 . 07

cf
code
cf

hel iostat
0
1 . 15
1 . 09
1 . 07
1 . 17
1 .21

S= !!Ill .
-

TR.,...2169

II."' I
-
-,

Table 2-8 .

Cf for Flow past a Flat Plate,

Aspect Rat io , A = 3 . 0

Angle of Attack
( deg )
0
10
15
20
25
30

a:

cf
( Code )
0

cf
Task Comm .
0
0 . 57
0 . 72
o. 78
0 . 84
0 . 88

o. 70
0 . 90
0 . 95
1 .05
1 . 10

cf

code

cf
0
1 .2 3
1 .2 5
1 :22
1 .2 5
1 .2 5

An important feature of the ANSI code is that the minimum angle of attack i s
1 0 . The code does not consider any angle of attack less than 10 .
The select ion of the angle of at tack , part icularly in the stowed horizontal
pos it ion of the hel ios tat , must be made with careful cons ideration of the
meteorological and turbulence characteris t ics of the high speed air f low .
Normally the air flows horizontally ove r the surface of the earth with a rela
t iv ely small vertical component . The vert i cal component arises from two bas ic
sources :
surface roughness and ins tab i l i t ies in the air caused by tempera
ture , dens i t y , and moisture differences .
The f i rst source is e s sent ially
mechanical and the second is thermodynamic . In the present study , our primary
concern is mainly for the s i tuat ion when survival requirements ( 9 0-mph winds )
dictate that the hel iostats be placed in the hori zontal stow pos i t i on . Such
Aspect rat io, i\ = 1 . 0

N
"'
<D
,.,
0
0

ANSI A58.1 { 1 982)

u
;;:::
:
Q)
0
(,)
Q)
(,)
......
0

ASCE task com m ittee { 1 96 1 )

(ij
0
I-

35
A n g l e of attack, a {deg)

Figure 2-8.

Variation of Force Coefficient with Angle of Attack,

Aspect Rat i o , A = 1 . 0
18
'I
'i

: i

TR- 2 1 69

Aspect ratio, i\ =

3.0

ANS I A58.1 (1982)

c5
c:
Q)
(.)

Q)
0
(.)
Q)
(.)
,_
0

(ij
0
1-

35
A n g le of attack, a (deg )

Figure 2-9.

Variation of Force Coefficient with Angle of Attack,

Aspect Ratio , A. = 3 . 0

high wind generally occurs during neutral atmospheric stab i lity , and there i s
l i t t le vertical movement o f a i r caused by thermodynamic effects .
I t appears
that the dominant causes of the vert ical component of air velocity are tur
bulence effects .
The force of turbulence-induced vert ical fluctuat ing air
act ing on the horizontal heliostat can be est imated from the knowledge of the
turbulent energy spect ra of the approach wind .

2.1

ATMOSPHERIC TURBULENCE SPECTRA

The random fluctuat ions of velocity in a turbulent flow pass ing a point in the
flow field may be caused by the superposit ion of the local vort ices or eddies ,
each with a random mot ion of f requency n ( alternat ively of wave number k) . We
can approximate the t otal kinet ic energy of the turbulent mot i on by summing
the contribut ions of each eddy in the flow. When the equat ions of mot ion of
turbulent f low are appropriately t ransformed and analyzed , it can be shown
that the inert ial transfer of energy from the larger eddies to the smaller
eddies takes place; the viscous d i s s ipat ion takes place primarily in the smal
lest eddies. through shear deformat ions . The cascade proces s of kine t ic energy
trans fer from the larger eddies to the smaller ones , and ult imately energy
d i s s ipat i on into heat , is the bas ic postulate of Kolmogorov's hypothesis .
Measurements of turbulent spect ra in the surface layer of the atmosphere sub
s tant iate the validity of the assumpt ions regard ing the turbulent energy
t ransfer mechanism .
In an ideal ized horizontally homogeneous , neutrally
s table flow , the energy product i on rate is approximately balanced by the
energy d i s s ipat ion rate .
19

TR-2 169

The wind v elocity varies randomly with t ime .

As a resul t , the wind p ressures
on the surface of the structure are not steady; they also vary with time. The
mean square dynamic pres sure is related to the mean square long itudinal
v elocity fluctuat ions, which imp l i e s that the pressure spectrum is propor
t i onal to the sp ectrum of the longi tudinal velocity . The spectrum of dynamic
p re s sure can be predicted from the known long i tud inal v eloci ty spectrum.
The est imat ion of "gust response factor" used to determine the wind force on
the structure is bas ed on the des ign methodology dev eloped by Vellozzi and
Cohen [ 6 ] . The methodology uses the results of the power spectrum analys is of
the dynamic structural respon s e of s imple linear s ingle one degree of freedom
system with v iscous damp ing and measured wind gust spectra and gust correla
tion coeffic ient.
The various turbulent wind spect ra , commonly used in the
study of lower tmosphere , are compared to assess the energy spectra used in
the ANS I building code .
The spectrum S ( z , n) of the longi tud inal velocity fluctuat ions in the inert ial
subrange can be expressed in the following normalized form [ 24 ] :

z
u*

n S( z n)
where
n Z
' n
U( z )

= 0.26 f

u*

= frict ion velocity

= elev at ion

U( z )

= mean velocity at height z.

-2/ 3

( 2 -1 5 )

'

= frequency

The left-hand member of Eq . 27 1 5 is called the reduced spectrum of the

longi tudinal v eloci ty fluctuat ions and is a function of height .
S imiu [ 2 5 ]
cons iders Eq. 2 - 1 5 a good repre s entation o f spectra in the h igh-frequency
range and suggests us ing it for f > 0.2 .
The longitudinal energy spectra currently employed in the ANSI A58. 1- 1982
building code can be expressed by the following empi rical relat ion [ 2 5 ] :
n S ( z ,n )
=
(1
u2
'"'

4x2
'
+ x2 ) 4 /3

(2 - 1 6 )

where
12 0 0n
11 ( 1 0 )

U( 1 0 )

= mean wind speed at height 10 m ( mph , m/ s ).

Equation 2 - 16 was obtained by av eraging the results of the m easurements at

various he ights and does not , therefore , show the dependence of spectra on
he ight .

20

55 '1'*'

TR-2169

In the low-frequency range the energy spectra cannot be described by a

universal relation .
Kaimal et al . [26 ] considered the limiting value of the
spectra as f approaches zero, and the product n S (z , n) reaches a maximum value
Equation 2-15 is valid beyond the value of
at f = f m such that 0 < f < f8
f = f8 . Kaimal et al . [2
proposed the following empirical relation for the
longitudinal spectra in the low frequenl:.Y range :

n S ( z , n)
-u"2

200f

Tl+5oi)513

(2-17)

Equation 2-17 approximates closely the spectrum in the higher frequency range
also, and Simiu [25] suggested using it for the entire spectrum .
The spectra that Lumly and Ponofsky [27 ] developed for the vertical and
lateral velocity fluctuations in the lower surface layer of the atmosphere are
given, respectively, by the empirical relations :
n S w (z, n)
=
u2

and

n S v (n)
:- =
--:-

u;

3 . 36f

I _;

lO f) S/3

- -

!Sf

(I

9 . 5f) 5/3

'

(2-18)

(2-19)

Equation 2-16 was obtained by averaging results of measurements obtained at

various heights above the ground and does not reflect the dependence of
spectra on height .
However, the dependence of spectra on height is clearly
suggested by Davenport [28 ] . The spectral distribution in the lower frequency
range has little influence on the structural response . However, the magnitude
of the turbulent fluctuation components at the natural frequencies of the
structure (or of one or more of its maj or elements) may affect its response
significantly . Simiu [25] considered a wind velocity of 30 m/s and a surface
roughness z0 = 0 . 08 m and computed the longitudinal spectra ln the frequency
range of 0 . 001-0 . 5 Hz by using Eqs . 2-16 and 2-17 .
He considered heights
z = 100 . 0 m, 200 . 0 m, and 300 . 0 m . The comparison showed that the expression
currently used in building codes overestimated the longitudinal spectra of
turbulence in the higher frequency range by as much as 100%-400% .
To compare the spectra obtained using Eqs . 2-15 through 2-19, computations
were made for two wind velocities , namely U (z) = 44 . 73 mph and U (z) = 90 . 0 mph
This height was chosen to conform to the
at a height of z = 30 . 0 ft .
requirement of the ANS I code .
Figure 2-10 shows the longitudinal spectra
computed by Eqs . 2-15, 2-16, and 2-17 for wind velocity U = 44 . 73 mph (20 . 0
m/s) .
The building code spectra overestimate the Kaimal spectra in the
frequency range of 0 . 007-0 . 08 Hz and underestimate beyond the frequency
n = 0 . 08 Hz .
Figure 2-11 shows the longitudinal, lateral, and vertical spec
tra of turbulence of 44 . 73-mph wind at a . height of 30 . 0 ft .
At frequency
n = 1 . 0 Hz, the local isotropy is realized .
For frequency less than 1 . 0 Hz,
the vertical spectral energy content is less than these for the longitudinal
and lateral components .
Figure 2-12 shows the energy spectra of turbulence
for a 90 . 0-mph wind at an elevation of 30 . 0 ft .
21

TR-2169

e I nert i a l subrange
Kai m a l et al.

A ANSI A 58 . 1 ( 1 982)

:::J
.....
()
Q)
0..
(/)

44.73 m p h
( 2 0 0 m/s)
30.0 ft
(9 1 4 m )
.

Z! =

i
c::
"'0
:::J
;t::
Ol
c::
0
.....1

0. 0 1 L-....1.-.L.L.L.LlJl.J..L_____L__j___L.Ll..LllJ._J..__[_L.J...LJ.J..Ll___J___...L....L..J...LL..l..I.L--l--L...L.L.LJ.L.LI
0.001

0.01

Figure 2-10 .

1 .00
0. 1 0
Freq uency, n (Hz)

Longitudinal Turbulence Spectra ,

U = 44 . 73 mph ( 20 . 0 m/s )

c::

N .
:::J
(/)
c::

( Eq . 2 - 1 7 )

(/)
c::
0
ell
:::J

tl

n S" ( Z , n )

:::J
;:;::

()
0
(j)

n S, ( z, n )
U

>

( Eq . 2 - 1 8)

( Eq . 2- 1 9 )
U = 44.73 m p h
( 2 0 . 0 m/s)

0
E
2
()
Q)

z =

30.0 ft
(9. 1 4 m )

0.00 1 L-.....--LllllL-....1.-lilL-
0.001
Freq u e n c y , n ( Hz)

Figure 2-1 1 .

Turbulent Wind Spectra ,

U = 44 . 73 mph ( 20 . 0 m/ s )

22

"'
"'
<D
"'

s=!!!t
-
.

TR-2 1 6 9

n Su (z, n )

(f)
c

u:

( 1 982)
(.)
0
Q)

>
-

,,

n Sw ( z , n )

u:

n Sv (z, n )

U = 90.0 m p h

u:

::l
....

(40.24 m/s)

Q)
a.
(f)

F req u e n c y , n ( Hz)

Figure 2-1 2 .

2. 2

Turbulent Wind Spectra ,

U = 90 . 0 mph ( 40 .24 m/ s )

VARIATION IN WIND ANGLE FROM THE HORIZONTAL

The vert ical wind components fluctuate randomly a t various rates .

When the
hel iostats are stowed in a horizontal pos it ion , the effect of the longitud inal
component of the wind velocity on the hel ios tats is relatively smal l .
How
ever , the peak angle of at tack of the vert ical turbulent component of wind
velocity may have a s igni ficant e ffect on the structure . The angle of attack
s tipulated in the ANS I code is rather large .
According to the code , the
minimum angle of attack on a f lat plate must be 1 0 .
The methodology developed by Daniels [2 9 } is used to e s t imate the probable
value of the peak angle of attack on the helios tat in the horizontal pos i
t i on . According to Daniel s , the lateral spectra are the bas is for determining
the lateral angle of attack .
S ince the vertical spectral energy content is
generally less than that of the lateral spectra , Daniels sugge s t s that the
angle of attack will give correct values at frequencies beyond n = 3 . 0 Hz i f
The angle o f at tack
it is determined on the bas is o f the lateral spectra .
will be somewhat ove restimated f or values of f requency less than n = 3 . 0 Hz .
Accordingly , the angles of at tack for wind veloc i t ies of 30 . 0 , 50 . 0 , and
9 0 . 0 mph at an elevation of 30 . 0 ft are e s timated .
First , we determine the
angle of at tack for the lateral component of turbulent velocity .
Then , by
comparing the magnitude of the lateral and vert ical spectra at the ir peak fre
quenc ie s , we use a correct ion factor to est imate the angle of attack for the
vert ical component of the turbulent veloc i t y .

23

s= '*'

TR-2169

'

The along-wind spectra are represented by the curve marked "longitudinal , " and
across-wind spectra are represented by the curve marked "lateral. "
The
longitudinal and lateral spectral values S (w) at a frequency w and height z
are given in the form of a dimensionless formula :
c 1 f/fm
w S(w)
(2-20)
u;
{ I 1 + 1. s (f/fm )c2 J 5/3 }c2
---

where
f
fm

and

wz
U(
z) '
=

c3
=

( z/z

r )C 4

2 cs
(z )
r

0.4
Zr
1n zo

U( z )

- '

Daniels ' velocity spectra for strong wind conditions are shown in Figure 2-13 .
In the above equat ions z r is the reference height equal to 60 ft (18 .3 m) for
the sites and the values of the constants suggested (shown in Table 2-9) ; z0

0 . 0 1 --------------------------_._____________________________
0.01
0. 1
1 .0
0.001
1 0. 0

Figure 2-13 .

Strong Vind Velocity Spectra

Source : Ref. 29 .
24

TR-2 1 69

Table 2-9 . Dimensionless Constants for Longitudinal and

Lateral TUrbulence Spectra

Constant

Cond it ion
c1

c2

--- - ----

c4

c5

1 . 00

-0 . 63

0 . 58

-0 . 3 5

c3

S trong wind
( Longitudinal )

6 . 198

0 . 845

0 .03

Strong wind
( Lateral )

3 . 954

0 . 78 1

0 . 10

------

- - - - -- - - ------- ------- -- -- - - ----

Sou ce :

R e f . 29 .

t s th surface roughnes s at the s i t e ; is a stabil i ty parameter depend ing on

the atmospheric stabil ity; U ( z ) is the mean longitud inal wind speed at a
height z; and c 1 , c 2 , c 3 , c4 , and c 5 are dimens ionless s i te-spec i f ic con
The parameters
stant s .
and f m account for height e ffects and are used in
norma l i z ing the d imens ionless quant i t ies
wS ( w )
c 3f

and
fm

The following parameters are required to est imate the angle of at tack for the
lateral component of the turbulent f luctuat ing veloci ty : the elevat ion Z , the
mean longitud inal wind speed U ( z ) at that height , and the character i s t ic
roughness length z 0 of the s it e .
Us ing the relat ions given in Eq . 2-20 , the
values of the parameters
U* ' and fm are computed . To est imate the ampli
tude of the lateral component of turbulent velocity v0 , a s imple s inusoidal
model is used . It can be assumed that

'

v ( t ) = vo s in 2nwt '

( 2-2 1 }

where
vo

= amp l itude of lateral veloci ty

= frequency in Hz

= t ime ( s ) .

The frequencies w for low-f requency spectrum analys i s are in the range of
0 . 0 1 - 1 . 0 Hz .
The peak frequency is approximately w = 0 . 1 Hz and is used to
calculate the normali zed frequency f .
The normali zed frequency parameter
0 . 1 f /f m on the abscissa of Figure 2-13 is then computed , and Figure 2-13 is
wS
used to get the value of
from the ordinate . For strong wind condit ions ,
the stabil i ty parameter
of velocity U* .

_L

u
vai shes

and is neglected in evaluat ing the frict ion

It can be shown [ 18 ] that at a part icular frequency w ,

25

5-1 .
-

TR-:-2 16 9

"' ..,
II

II"

wS ( w ) =

2'1t

I0

r v < t > 1 2d t .

( 2-22 )

Us ing the relat ion of v ( t ) from Eq . 2-2 1 , Eq . 2-22 can be reduced to

v 2
wS ( w ) = -
2

( 2-23 )

Equat i on 2-23 is used to estimate the magni tude of peak lateral ve lo c i ty :

( 2-24 )
The angle o f a ttack a v for lateral fluctuat ing ve loc i ty ls obtained f rom the
following relat ionship :
v
-1 o
\
( 2-25 )
a = tan
v
U( z )

,,

S ince the magnitude of lateral spec tra is approximately 1 . 2 1 t imes larger than
that of the vert ical spectra , the amp l itude of vert ical fluctuat ing veloc i ty
w0 is approximately equal to 0 . 9 lv0
The angle of attack for the vert i cal
component of fluctuation is given by
-1
aw = tan

0 . 9 1v0
U( z )

( 2-26 )

The angles of at tack for lateral and vert ical components of fluctuat ing
velo c i t ies have been computed for each of three mean longitud inal velocit ies
of 30 . 0 , 50 . 0 , and 90 . 0 mph at the standard elevation of 30 . 0 ft for various
surface roughness lengths . The results are tabulated in Tables 2- 1 0 and 2- 1 1 ,
When the surface roughness
and shown graphically i n Figures 2-14 and 2- 1 5 .
length z0 of a site is small , the values of the angle of attack are relat ively
les s . The values increase as the roughness length increases . One interest ing
feature of the present analys i s ls that when the mean wind speed increases ;
the values of the angle o f attack decrease . The design angle of attack.

Table 2-10 .

Variation of Angle of Attack due to Fluctuating Lateral Velocity

Roughnes s
He ight , z0

Angle of At tack , a
( deg )
U ( 3 0 ) = 30 . 0 mph

U ( 30 ) = 50 . 0 mph

U ( 30 ) = 90 . 0 mph

0 . 10 ft ( 0 . 0 3 1 m)

4 . 53

4 . 44

4 .15

0 . 20 ft ( 0 . 06 1 m )

4 . 0 39

3 . 96

3 . 70

0 . 40 ft ( 0 . 1 23 m )

5 . 14

5 .05

4 . 72

1 . 3 12 ft ( 0 . 4 m )

6 . 73

6 . 61

6 . 19

1 . 9 68 ft ( 0 . 6 0 m)

7 . 52

7 . 37

6 . 90

2 . 6 24 ft ( 0 . 80 m)

7 . 53
26

TR- 2 1 69

Table 2-1 1 .

Variation of Angle of Attack due to Fluctuating Vertical Velocity

---- -- - - - - - - - - - - - - - --- -

--

- - - - - - - ----- - - - -

- - - ----- - - ----------

Angle of Attack , aw
( deg )

Roughness
Hei ght , z0

-- - - - -

U ( 30 )

- -

. - - - - - - -- - - -

30 . 0 mph

- - --

U ( 30 )

- - - - - ---- --- - - - -

- - - - - ---- - -- - - - - - --

U ( 30 ) = 90 . 0 mph

50 . 0 mph

-------- --- - -- - - - ----- - - -- - - ---- - - - ---- --------------- - - --- - ------ -

- -- - --- ---- - -----

0 . 10 ft ( 0 . 03 1 m)

3 . 68

3 . 60

3 . 37

0 . 20 ft ( 0 . 062 m)

4 . 12

4 . 04

3 . 78

0 . 4 0 f t ( 0 . 1 2 3 m)

4 . 68

4 . 60

1 . 3 12 ft ( 0 . m )

6 . 13

6 . 02

5 . 64

1 . 968 ft ( 0 . 06 m)

6 . 85

6.71

6 . 28

,,

4 . 30

6 . 86

2 . 6 24 ft ( 0 . 80 m)

suggested b y the ANSI A 58 . 1-1982 code is 1 0 . The r.ecent

the angle o f attack suggested by Murphy [ 1 4 ] i s 6 .

lmp1:oved

value o f

The McDonnell Douglas Astronaut ics Company ( MDAC) [ 30 ] reported that a maximum
o f a 6 . 5 angle of t tack co u l d be expected for high winds in gent le , roll ing
terrain wi th a roughness length of 4 . 0 em .
Vann [ 3 1 ] used Daniel ' s approach and est tmated the angle of at tack due to
lateral fluctuat ions for wind 11eloc i ty U = 30 . 0 mph ( 13 . 41 m/s ) , at he ight
z = 30 ft , for surface roughness z0
3 . 0 ft . He cons idered the lateral spec
tra as representative of vert ical spectra . The e s t i mated peak angle of at tack
=

- - - - - - - - - - - - - - - - - - - - - - - --
"'

1 0 .0

"'
"'
"'

U (30.0 ft)

Ol
<l.l
"0

7.0

.:;l
Cii
0

6.0

5.0

Ol
c:

<(

30.0 m p h

8.0

ts

:X
()

U (30.0 ft )

90.0 m p h

4.0
3.0
2.0
0

Figure 2-14 .

0.1

0.3
0.4
0.5
0.6
0. 7
R o u g h n ess h e i g h t ( m )

0.8

0.9

1 .0

Peak Angle of Attack due to Turbulent Lateral Velocity

27

TR-2 169

U (30.0 ft)

Ol
Q)
'0

50.0 m p h

Ol
\c

90. 0 m p h

<(

Figure 2-15.

Peak Angle of Attack due to Turbulent Vertical Velocity

at w = 0 . 1 Hz was 8 . 9 . The surface roughness z 0 of 3 . 0 ft ( 0 . 9 1 m ) is rather

too large .
The surface roughnes s value is a s ign i f i cant parameter in the
determinat ion of the peak angle of attack due to turbulent veloc it ies .
The
values o f the peak angle o f attack increase as the surface roughnes s
increases .
One interest ing feature o f the present analysis i s that the value
of the angle of attack decreases with the increase of the mean wind speed .
F igure 2-15 shows that for 0 . 2 m < z 0 < 0 . 4 m the est imated peak angle of
attack due to turbulent vert ical veloc ity for a 9 0-mph wind and heliostat
stored in hor i z ontal pos iti on is in the range of 4 . 8 -5 . 6 .
Est imat ing the
peak angle of attack is quite involved .
To obtain more rel iable values for
the angle of attack, recourse should be taken to e ither full-scale f ield study
or model study in a s imulated flow fiel d in the environmental wind tunnel .
S imiu and Scanlan [ 25 , Table 2 . 2 . 1 ] give typical values of surface roughnes s
z0 f o r var i ous types of surfaces and corresponding values of the surface drag
coe f f i cient s . The table also includes the sugges ted values of z 0 for built-up
terrains such as outskirts of towns and suburbs , centers of towns , centers of
large c i t ies , etc .
The suggested value of z 0 for outsk irts of t owns and
suburbs i s in the range of 0 . 20-0 . 40 m ; for the open fields , z0 is in the
range of 0 0 1-0 . 1 m depending on the phys ical characterist ics .
Figure 2-15
shows that for 0 . 2 m < z < 0 . 4 m, the est imated angle of attack for a 9 0-mph
0
wind and heliostat stowed in a horizontal pos ition is in -the range o f
4 . 8-5 . 6 .
I f the lower z0 values are used for the heli o s tat f ield , the
values of the angle of attack are s malle r .
Figure 2-14 and Figure 2-15 show
clearly that the ANSI code values for the angle of at tack are too high .
.

28

TR-2 1 6 9

SECTION 3 . 0
REDUCTION OF WIND LOADING

Recent s tudies [ 14 , 15 , 16 , 2 1 , 22 , 3 2 ] of scale models of solar collectors in

environmental wind tunnels show that the present method of determining wind
loads on s olar collect ors , photovoltaic arrays , and hel !os tats us ing the
guidelines outlined in the ANS I A58 . 1- 1 9 8 2 building code requi rement s for
minimum design loads in buildings and o ther structures [ 1 2 ] is conservat ive .
Much of the available informat ion indicates that the degre of conserva t i sm i s

s ignificant , but the actual extent o f that conservatism must be assessed i n

more detail
. . The code was not developed for the low-height structures like
.
\
solar collectors .
The pert inent s i te-speci fic met eorologi cal data such as
wind veloci ty d i s t ribution in the lower atmosphere , stab i l i ty features , and
turbulent characterist ics of the wind f ield are not rel iably known . S ince the
wind profile changes [ 2 5 ] with the changes of the roughness length z0 of the
loca l i ty of the structure , values depend ing solely on the a i rport data should
not be used in opt imum des ign without care fully cons i dering the terrain con
d i t ions , local surface roughness , orientation of the hel iostat s , and direction
of the wind .
Ravindra et aL [ 3 3 ] developed an intere s t ing methodology to determine the
design wind loads on s t ructures based on the s tat ist ical parameters of the
lifetime maximum wind speed , the annual maximum wind speed , and the daily max
imum wind s peed . Ravindra et al . use the general equation for determin ins the
wind pre s sure and cons ider all the parameters ( p res sure coefficient , effect ive
velocity p ressure , velocity p res sure coeffic ient that depends on the type of
exposure and height above the ground , gust factors that depend on the response
characterist ics of the s t ructure , and the bas ic wind speed measured at the
reference height ) as random variables . The mean wind pressure and coefficient
of variation of the wind p ressure are expres sed as functions of the mean and
the coefficient of variat ion of the random variables . Laboratory tes t results
are used to obtain values of coefficients of variat i on .
Generally , the s ta t i s t ics of wind direct ions are not considered in the code .
Davenport [ 34 ] considered two cases of loadings : ( 1 ) the d i rect ion of maximum
wind is uni formly d i s tributed between and -n , corresponding to the
assumpt ion that the worst response occurs no mat ter what the wind direct ion
is ; and ( 2 ) the direct ion of maximum wind is distributed according to a cos ine
law . The rat io of the mode in case 2 to the mode value in case 1 , called the
d i rection reduction factor , has been est imated to be 0 . 7 2 .
Ravindra et
al . [ 3 3 ] cons idered o ther fact ors , such as serviceabi l i ty and ult imate l imit
s tate , and recommend that the mean des ign wind pressure be reduced by a des ign
res i stance factor of 0 . 75 . Reducing the wind load by 25% for the variabi l i ty
of wind direct ion and other cons idera t ions should be viewed with cau t i on .
Results from s imulated laboratory or f ield s tudies should be used to estimate
the reduction of wind loads caused by the variat ion in wind direct ion .
Modification of the shear layer that develops over the helios tat surface is an
area of s tudy that may lead to the reduct ion of drag experienced by the
hel ios tat s . Melbourne [ 35 ] and Gartshore [ 36 ] observed that the small-scale
29

5- /.=<\>
-

II
II
'=

TR-2169
#

turbulence inc reases s igni ficantly in the reat taching shear layers on the sur
face and conse quently increases the magnit ude of the high nega t i ve surface
pres sure .
It is reasonable to cons i der that venting the region of separated
flow and its subsequent reat tachment may be useful in reducing the magnitude
of negative pressure . Melbourne [ 35 ] observed that venting the separated flow
reg i on reduced the negat ive pre s s u re thereby leading to increased flow
s t ab i li t y .
It is well known that a pressure redu c t i on ( below ambient ) on the
s t reamwise surfaces near the lead ing edge of the flat plate occurs in the
reg i on of f low separat i on .
Vent ing or introduc ing f luid with a higher level
of moment um in the reg ion ( the bubble under the reat t aching shear layer ) is an
old technique to delay flow separat ion and decrease the reduced pressure
,
e ffect . Thi s can be accompli shed by us ing a slotted eave along the periphery
of the surface .
,

The hel iostat modules have sharp edges , and under regular and horizontal stow
pos it ions , f low separa t i ons near the leading edge and subsequent reattachment
occu r . A porous eave around the edges of the scale model of a module would be
u s e ful for t e s t ing in an env i ronmental w ind tunnel under s imulated flow con
d i t ions .
Streamlining the windward and leeward edges may be help ful in
reducing and even el iminat ing flow separat ion and subsequent reat tachment .
The air gap between the modules allows high-momentum air to flow through the
array and is expected to reduce the windward pressure as well as the negat ive
pre ssure on the back s ide .
The appropriate amount of air gap for total drag
reduc t i on needs to be determined f rom laboratory tests in the s imulated
Invest igat ions of the e f fects of high poros ity in a flat plate on the
f lows .
reduct i on of total drag may lead to bas ic knowledge of d rag control by
allowing high-momentum flui d in the wakes of a flat plate . The test results
of Xer ikos et al . [ 17 ] did not show s igni f i cant d rag reduct i on because the
slot openings were rather small .
A pre l iminary study of the effects of the
so l i d i ty ratio of a f lat plate placed normal to the f low in the boundary layer
by Rangwala and Handy [ 37 ] showed th considerable reduct ion in drag is pos
s ible when the solidity rat io is less than 70% .
Experimental results from the wind tunnel study [ 17 ] indicated that plac ing
hel i ostats in the wakes of the hel ios tats in front helped to reduce wind loads
on the hel iostats in the wakes .
However , wind d i re c t i ons , turbulent
intens ity , and dynamic response of the s t ructure due to increased turbulence
in the wake regi ons may influence the net outcome .
In the s imulated flow
f ields in the wind tunnel , we need to s tudy the development of wakes behind
the hel i o s tats in the inner part of the f ield to gain knowledge of the rela
t i ve magni tude of the force and moment coef f ic i ents that are needed to
adequately est imate the wind load on a hel i o s tat .

1.'
I
Ilj
li
/1

l,,

; \'

'I

Perimeter fences [ 38 ] are ef fect ive in reduc ing w ind loads on the photovoltaic
arrays .
Fence poros i ty of 30%-50% and fence hei ght approximately equal to
three-quarters of the array height enhance the reduction of w ind loads s ig
nif icantly .
However , the effect is only limited to the first few rows of
arrays along the perimeter .
Fences ins ide the hel l o s tat f ield , p laced at
experimentally determined s t ra tegic pos it ions , may be help ful to reduce the
base-bending moment of the hel iostat .
30

s=!!ill
"' '
- u.

TR- 2 1 69

-
'

The experimental works of Randall et al . [ 39 , 40 ] with parabolic trough solar

collectors are of great pract ical importance .
The s igni f icant inter ference
e f fect provided by the ups t ream collector t:o"ls on the lateral and l i ft forces
indicates that mos t collector modules within an array experience force reduc
t i on .
Randall et al . have s tudi ed the influence of the height of a fence
placed three aperture widths ups t ream of the first row of collectors .
The
t r >3 t
r:-e s ul ts ind icate that a fence height somewhat less than the full col
lr e t o r height provides the maj or part of the force reduct ion .
S imilar resul ts
are also obta i ned for the influence of the berm hei ght .
Shelter bel ts made up of trees , shrubs , and other small struct ures act ing as

.
wmd b reakers along the pe rime ter of the hel ios tat f ield may be helpful in
reducing the e f fect ive wind veloc i ty in the field .
Sturrock [ 4 1 ] observed
l
that proper y des igned she l t e r b1lts and wind barriers could be s ign i f icantly
e f fect i ve in reduc ing wind load on struetl t:es .
It is poss ible to study and
evaluate the e ffec t i veness of the barrier in the wind tunnel .
The packing dens ity [ 4 2 ] of hel i o s tats in the f ield had d i scernible effect on
the result ing wind loads . Higher packing dens :tty w l ll reduce net wind load on
the hel i os tats .
It might be fefts ible to develop an opt lmum des ign of the
hel los tat fleld with relati vely higher dens i ty in the outer part of the f ield .
The hel i ostats fit th( 0 1J t r:- i>' t:tphery of the f ield are subj ected to maximum
wind l oad ; therefore , they should be des igned stru c turally s t ronger and more
rigi d .
It is pos s ible to use les s rigid and relatively l ighter support struc
ture for the hel ios tats i n the l n.terior of the f ield .

31

TR-2 1 6 9

SECTION 4 . 0
OONCLUSION AND REOOMMENDATIONS

This s tudy was limited to reviewing and assess ing the present des ign
methodology for wind loading on solar collectors , particularly for heliostat s ,
and recommend ing areas of further invest igat ion to develop realis t ic criteria
for determining adequat e wind loads . The result s and conclu s i ons of the s tudy
can be summarized as follows :

The present methodology for des igning wind load usirlg the outlines given
by the NSI A58 . 1- 1982 Code is conservat ive .

The history of the s ite-speci f ic veloc i ty data is necessary to apply

extreme value order statist ics for determining des ign wind speed . Local
climatalog ical data including wind directions and surface roughness
'
should be used to develop the veloci ty prof ile .

Comparat ive study of the total force coe f f i cient of a flat plate of
various aspect rat ios and at various angles of attack shows that the
values obtained by us ing the ANS I Code are always higher .

The energy spectra obtained by the formulas sugges ted by various

author i t ies were compared with the energy spec t ra for the code and show
that in the flow-frequency domain , the code overest imates spectral values
for lower frequencies but underestimates in the region of higher fre
quencies , including the natural frequencies of heliostats and o ther solar
collectors .

The ngle of attack that result s from the fluctuat ing ver tical wind i s
cons iderably smaller than the value sugges ted by the code .

Development of locally modi f ied flow over the heliostat has the potent ial
to reduce w ind load on the hel iostat .

Appropriate vent ilat ion in the hel ios tat modules has the potent ial t o
reduce dynamic w ind load .

Recommendat ions for further studies that might reduce wind loading on the col
lector are s tated briefly :

Mod i fying the flow locally by us ing a turbulence st imulator in the form
of porous eaves around the hel ios tat will d iscourage the flow separation
at the edges of the hel iostats and help to reduce dynami c wind load .

Investigat ing the effect of appropriate gaps between the modules will
encourage vent ilat ion around the module .

S tudying the effects of the perimeter fences and fences in the field
around a group of heliostat zones will be useful in overall wind-load
reduct ion .

Developing a wind rose for the heliostat field to est imate the
of variat i on of wind d i rect ion . The terrain and topographical
the heliostat field play important local roles in determining
tudes of wind speed and d i rect ion .
The wind d i rection is not
32

s tat i s t ics
aspects of
the magni
cons idered

TR-2 1 6 9

i n the building code .

The stat i s tics o f the d i rect ional variab i l i ty of
the wind are essent ial in determining wind load reliably .

33

5
' .
-
-

II
II

TR;-2 169
f

SECTION

5 .0

REFERENCES

1.

Pugsley , A . ,
1966 .

"The Safety of Structures , " Balt imore , MD :

Edward Arnold ,

2.

Davenport , A . G . , "Est imat ion of Repeated Loads on Structures with Appli

cat ion to Wind Induced Fat igue and Overload , " Proceed ings of RILEM Inter
nat ional Sympos ium :
Effects of Repeated Loading of Materials and Struc'
tures , Vol . 38 , 1967 .

3.

Robertsoil , L . E . , "On Tall Bui ldings , " Proceedings of a Sympos ium on Tall

. --
Buildings , University of Southhampt on , edited by A . Coull and B .
Stafford-Smith , Oxford , U . K . : Pergamon Press , 1967 .

4.

Davenport , A . G . , "The Dependence of Wind Loads on Meteorological Para

meters , " Proceedings of Internat ional Research Seminar on Wind Effects on
Bui l d ings and Structures , Ottawa :
Nat ional Research Council of Canada ,
1967 .

5.

Davenport , A . G . , "The Nature of Disturbing Forces , " Proceedings of RILEM

Effects of Repeated Load ing of Materials and
Internat ional Sympos ium :
S t ructures , Mexico , 196 6 .

6.

Vello z z i , J . , and E . Cohen , "Gu s t Response Factors , " Journal o f the

Structural Divis ion , Ame rican Society of Civil Engineers , Vol . 94 , ST 6 ,
June 1 968 .

7.

Davenport , A . G . ,
"A Rat ionale for Determination of Design Wind
Velocit ies , " J . Structural Divis ion , American Society of Civil Engineers ,
Vol . 86 , No . ST 5 , Proc . Paper 2476 , May 1960 .

8.

Gumbel , E . J . , "Statist ical Theory of Ext reme Values and Some Pract ical
Applicat ions , "
Applied Mathemat ics Series No .
33 ,
Washington , DC :
Nat ional Bureau of S tandards , 1 9 54 .

9.

Thorn , H . C . S . , "Frequency of Maximum Wind Speeds , " Proc . Separate No .

539 , American Society of Civil Engineers , Vol . 8 0 , November 1954 .

10 .

Lieblein , J . , "Method o f Analyz ing Extreme Value Data , " Technical Note
3053 , Washington , DC : Nat i onal Advisory Committee for Aeronautics , 1954 .

11.

Thorn , H . C . S . , "New Di stribut ion of Extreme Winds in the Uni ted States , "
J . American Society of Civil Engineers , Vol . 94 , No . ST 7 , July 1968 .

12 .

American Nat ional Standard A58 . !-Bui lding Code Requirements for Minimum
Des ign Loads in Build ings and Other Structures , NY :
American Nat ional
S tandard Ins t itute , 1982 .

13.

Simi u , E . , "Wind Climate and Failure Risks , " J . Structural Divi s ion ,
American Society of Civil Engineers , Vol . 102 , ST 9 , September 1976 .
34

TR-2169

14 .

Murphy , L. M . , An
Assessment of Exis t ing S tudies of Wind Loading on
Solar Collectors , SERI/TR-632-8 1 2 , Golden , CO :
Solar Energy Research
Ins t itute , February 198 1 .

15.

Roschke , E . J . , Wind Loading on Solar Concentra tors : Some General Con

s idera t ions , DOE/JPL- 10 6 0-66 , JPL Publicat ion 83- 1 0 1 , Pasadena , CA : Jet
Propulsion Laboratory , May 1984 .

16 .

Peglow , S . G . , Wind Tunnel Tes t of a Full Scale Hel iostat , SAND 79-8034 ,
Livermo re , CA : Sandia Laboratories , June 1 9 7 9 .

17.

Xerikos , J . et al . , ..The Aerodynamics of Heliostats for Solar Power Plant

Appli cat i ons , .. Proceedings of the 5 th Internat ional Conference on Wind
Engineers , 1 9 79 , 2 vol s . , edi ted by J . E . Cermak , Oxford , U . K . : Pergamon
Pres s , 1 98 0 .

18 .

McDonald J . R . , Wind Effects on Solar Tower Genera t o r , Lubbock ,

Inst itute for Disas ter Research , Texas Tech Univers i ty , 1 9 78 .

19 .

Cermak , J . E . , .. Applicat ions of Flui d Mechanics to Wind Engineering--A

Freeman Scholar Lecture , .. J . Fluid Engineering , Vol . 9 7 , March 1 9 7 5 .

20 .

Bhaduri , S . , .. S imulat ion of Urban

Engineer, Vol . VI , No . 2 , 1 9 7 3 .

21 .

Lindsay , J . E . , Force and Pressure Tes t s of Solar Collector Models in the

Vought Corporation Sys tems Divis ion Low Speed Wind Tunnel , SAND 76-7007 ,
Albuquerque , NM : Sandia Laboratorie s , May 1979 .

22 .

Randall , D . E . , D . D . McBride , and R . E .

Collector
Wind
Loading ,
SAND 79-8 0 34 ,
Laboratories , June 1 9 79 .

23 .

Task Committee on Wind Forces , ..Wind Forces on Structure s , .. Transactions ,

American Society of Civil Engineers , Vol . 126 , No . 3269 , 1 96 1 .

24 .

S imiu , E . , .. Wind Spectra and Dynamic along Wind Response , .. J . Structural

Divis ion , American Soc iety of Civil Engineers , Vol . 100 , No . ST9 ,
September 1974 .

25 .

S imiu , E . , and R . H . Scanlan , Wind Effects on Structures , NY : John Wiley

and S ons , 1978 .

26 .

Kaimal , J . C . , et al . , .. Spect ral Characteris t ics of Surface-Layer Turbu

lence , .. J . Royal Meteorological Society , Vol . 8 7 , 1 96 1 .

27 .

Lumley , J . L . , and H . A . Panofsky , The Structure of Atmo spheric Turbu

lence , NY : John Wiley & Sons , 1964 .

28 .

Davenport , A . G . , ..The Spectrum of Hori z ontal Gus t ines s Near the Ground
in High Winds , .. J . Royal Meteorological Society, Vol . 8 7 , 196 1 .

,,

35

Atmosphere

for

Diffus ion

TX :

Study , ..

Tate , Paraboli c Trough Solar

Albuquerque ,
NM :
Sandia

TR-2169

29 .

Daniels , G. E . , ed . , Terre s t r ial Environment ( Climatic) Criteria Guide

l ines for Use in Aerospace Vehicle Development , 1 9 7 3 Revis ion, NASA
TM X-64 7 5 7 , Huntsville , AL :
Marshall Space Flight Center , 1 97 3 .

30 .

McDonnell Douglas Ast ronautics Company , Optimi zat ion of the

Generat ion Helios tat and Spec i f icat i on , SAND 82-8 18 1 , May 1982 .

31 .

Vann , W . P . , "Wind Effects on Solar Tower Generators , " Final Report ,

Subcontract No . 77-TT-1 ( Sub-Task b . ) , Lubbock , TX : Energy Foundat ion of
Texas , Texas Tech Univers i ty , 1 9 7 8 .

32 .

Bechtel Nat ional , Inc . , Wind Designs on Flat Panel ''Photovoltaic Array
Structurs , SAND 79-7057 , Livermore , CA : Sandia Laboratories , 198 0 .

33 .

Ravindra , M . K . , et al . , "Wind and Snow Load Factors for Use in Load and
Res i .stance Factor Des ign , "
J . Structural Divis ion , American Society of
C ivil Engineers , Vol . 104 , ST 9 , 1 9 7 8 .

34 .

Davenport , A . G . , "The Prediction of Risk under Wind Loading , " Proc . o f

the Second International Conference on S t ructural Safety and Reliab i lity ,
Technical Univers i ty in Munich , Munich , West Germany , Dus se ldor f , Wes t
Germany : Werner-Verlag , September 19 7 7 .

35 .

Melbourne , W . H . , "Turbulence Effects on Maximum Surface Pressures--A

Mechan i sm and Pos s ib i l i ty of Load Reduct ion , "
Proceedings of the Fifth
Interna t i onal Conference on Wind Engineering , F t . Collins , CO :
Colorado
S tate Universi t y , July 1979 .

36 .

Gart shore , I . S . , "The Effect s of Free-Stream Turbulence on Drag of

Rectangular Two-Dimensional Prisms , " Boundary Layer Wind Tunnel , Report
4-73 , Canada : Unive rs ity of Wes tern Ontario , 197 3 .

37 .

Rangwala , S . , and E . Handy , Drag of a Flat Plate of Variable Solidity

Rat io Placed Normal to Flow in a Boundary Layer , Proj ect Report for
ME 3506 , Advanced Fluid Mechanics I , El Paso, TX: Mechanical and Indus
trial Engineering Department , December 1983 .

38 .

Martin Marietta Corp . , Hel i o stat Field Wind Effects Test s , SAN/ 20422-2 ,
San Franc i sco , CA : Department of Energy , February 1979 .

39 .

Randall , D . E . , D . D . McBride , and R . E . Tate , S teady-State Wind Loading

on Parabol ic Trough Solar Collectors , SAND 79-2 1 34 , Albuquerque , NM :
Sandia National Laboratorie s , August 1980 .

40 .

Randall , D . E . , R . E . Tate , and D . A . Powers , Experimental Results of

Pitching Moment Tests on Parabolic-Trough Solar-Col lector Array Con
f igura t i ons ,
SAND
8 2- 1569 ,
Albuquerque ,
NM:
S andia
Nat ional
Laboratories , December 198 2 .

41 .

Sturrock , J . W . , "Aerodynami c Studies of Shelterbel t s in New Zealand , "

New Zealand Journal of Science , Vol . 1 2 , No . 4 , 1969 .
36

Second

TR-2 1 6 9

42 .

Pe t e rka , J . A . , B . Bienkiewic z , and J . E . Cermak , Technical Feas ibility

o f Innovat ive Concepts for Avoiding or Reducing Wind Loads on Concen
t rator Collectors - Task I Report for Solar Energy Research Ins t i tute ,
Fort Collins , CO : Colorado State Univers i ty , March 1 98 4 .

37

TR-2 169

DISTRIBUTION LIST

Kos ta Achin
Sandia Nat ional Laboratories
Divis ion 847 3
P . O . Box 969
Livermore , CA 94550

Robert Gervai s
McDonnel Douglas Company
5301 Bolsa Avenue
Huntington Beach, CA 9 2647
H . Gerwin , Ph . D .
Gerwin Research
14305 Anstead Road
S ilver Spring , MD 20904

R . E . Akins , Ph . D .
Profes sor , Engineering Science and
Mechanics
Virginia Polytechnic Ins t itute and
S tate Univers ity
Blacksburg , VA 240 6 1

Roger Gillette
Boeing Engineering and Cons t ruction
Company
P . O . Box 3707
S eatt le , WA 9 8 1 2 5

Don Bartlett
Boe ing Engineering and Construct ion
Company
Mail Stop 9A-42
P . O . Box 3707
Seatt le , WA 98 125

Jim Leonard , Ph . D .
Sandia Nat ional Laboratories
Divi s ion 6227
P . O . Box 5800
Albuquerque , NM 8 7 1 1 5

David Blake , Ph . D .
Engineering Manager
Acurex Corporation
485 Clyde Avenue
Mountain View , CA 94042

John Lucas , Ph . D .
Jet Propuls ion Laboratory
4800 Oak Grove Drive
Pasadena , CA 9 4 1 0 3

Floyd Blake
ARCO Ventures Company
7061 South Univers ity Blvd .
Lit t leton , CO 8 0 1 2 2

Monte McGlaum
LaJet Company
3 13 1 Ant il ly Road
Abilene , TX 79605

John Cermak, Ph . D .
Professor , Civi l Engineering
Dep t . of Civil Engineering
Fluid Mech . and Wind Eng . Program
Colorado S tate Univers i ty
Fort Col l ins , CO 80523

Clayton Mavis
Sandia Nat i onal Laboratories
Divis ion 8473
P . O . Box 969
Livermore , CA 94550
Jon Peterka , Ph . D .
Professor , Civil Engineering
Dep t . of Civil Engineering
Fluid Mech. and Wind Eng. Program
Colorado State Univers i ty
Fort Collins , CO 80523

Bill Delemeter, Ph . D .
Sandia Nat ional Laboratories
Divis ion 8453
P . O . Box 96 9
Livermore , CA 94550
William Dickhart
As s i s tant to the Director
The Budd Company
Fort Wa shington, PA 19034

J. T. Pearson , Ph . D .
Purdue Univers i ty
Mechanical Engineering Building
West Lafayette , IN 47907
38

TR-2169

Tim Reinhold , Ph . D .
Center for Building Technology
Nat ional Bureau of S tandards
Washington , DC 202034

Mart in Scheve
Department of Energy
CE-3 14 , Room 5H-04 1
Forres tal Bui lding
1000 Independence Avenue , S . W .
Washington , DC 20585

Bob Reuter , Ph . D .
Sandia Nat ional Laboratories
Divi s ion 1 5 23
P . O . Box 5800
Albuquerque , NM 8 7 1 1 5

Jeff Shingleton
49 Elm Avenue
Takoma Park , MD

209 1 2

Jack Swearingen, Ph . D .
Sandia Nat ional Laboratories
Divis ion 8453
P . O . Box 969
Livermore , CA 94550

Jim Rogan , Ph . D .
McDonnel Douglas Company
5 3 0 1 Bolsa Avenue
Huntington Beach , CA 9 2 647
T . W . Rogers
United Space Boosters
2-EO
P . O . Box 1626
Huntsvi lle , AL 35807

H. Tielman , Ph . D .
Professor , Engineering Science and
Mechanics
Virginia Polytechnic Ins t itute and
S tate Univers ity
Blacksburg , VA 2406 1

E . J. Roshke , Ph . D .
The Jet Propulsion Laboratory
Cali fornia Inst i tute of Technology
4800 Oak Grove Drive
Pasadena , CA 9 1 1 09

A . Thomas
Inst rumentation and Services Unit
Ind ian Ins t i tute of Science
Bangalore - 560-0 1
INDIA

Aharon Roy
Head , Solar Power Laboratory
Department of Chemical Engineering
Ben-Gurion Univers i ty of Negev
P . O . Box 6 5 3
Beer-Sheva 84105 , ISRAEL

Kenell J . Touryan , Ph . D .
Vice President , R&D
Mt . Moriah Trus t
6200 Plateau Drive
Englewood , CO 801 1 1

Dan Sal l i s
Dan-Ka Products
790 Umati lla
Suite 2 1 2
Denver , CO 80204

Lee Weinstein
McDonnel Douglas Company
5 30 1 Bolsa Avenue
Hunt ington Beach , CA 9 2647

John Schaef fe r , Ph . D .
Manager o f Proj ects
Proj ects Department
Acurex Corporation
485 Clyde Avenue
Mountain View , CA 94042

Frank Wilkins
Department of Energy
CE-3 1 4 , Room 5H-04 1
Forrestal Bui lding
1000 Independence Avenue , S . W .
Washington , DC 20585

39

,
,

Document Control
Page
4. Title and Subtitle

1 . SEAl Report No.

S ERIITR- 2 53_- 2169

2. NTIS Accession No.

3. Recif:>ient's Accession No.

5. Publication Date

Wi n d loa d i ng on Sol a r Col l ectors

June 1 98 5

6.
7. Author(s)

S . Bhaduri and

- r . Murphy

9. Performing Organization Name and Address

Sol a r Energy Resea rch I n sti tu te

1 61 7 Col e Bou l eva rd
Gol den , Col ora do 80401

8. Performing Organization Rept. No.

10. Project/Task/Work Unit No.

1 384 . 30 and 51 0 2 . 31

1 1 . Contract (C) or Grant {G) No.

{C)
(G)

.,

1 2. Sponsoring Organizat''ln Name a n d Address

13. Type of Report & Period Covered

1 4.

Techn i ca l Reoort

15. Supplementary Notes

1 6. Abstract (Limit: 200 words)

The p re s ent d es i gn methodol ogy for the determi nati on of w i n d l oa d i ng on the

va r i ou s sol a r col l ectors h a s been rev i ewed a n d a s s e s s e d .
The total force co
eff i c i ents of fl a t p l a te s of a s pect ra t i os 1 . 0 a nd 3 . 0 , res p ec t i el y , at var
i ou s ang l e s of a tta ck ob ta i ned by u s i ng the g u i del i nes of the ANS I A58 . 1 - 1 98 2 ,
have been compared wi th those obta i ned by u s i ng the methodol ogy of the ASCE
Ta s k Commi ttee , 1 9 6 1 , and the experi menta l resu l t s of the f u l l - sc a l e tes t of
The turbu l ent energy s pec tra , currentl y empl oyed i n the
hel i o s tats . by Peg l ow .
bu i l d i n g code , a re compa re d wi th thos e of Ka i ma l et a l . , Luml ey , and Ponof s ky
for w i n d vel oc i t i e s of 20 . 0 m/ s a n d 40 . 24 m/ s a t an el eva ti on of 9 . 1 5 m . The
l on g i tu da l s pec tra of the bui l d i ng code overe s t i mates the Ka i ma l s pectra i n
the frequency range of 0 . 007 H z - 0 . 08 H z a n d un dere s t i mates beyo n d th e fre
quency of 0 . 08 Hz .
The peak a ng l es of a tta c k , on the hel i os ta t , s towed i n
hori zontal po s i t i on , due to turbul ent vert i c a l and l a teral compo nents of wi nd :
vel oci ty , have been e s t i ma ted by u s i n g 'Da n i el s methodol ogy for three wind
vel oci ti e s a n d compared w i th the v a l u e sug g e s te d by the code .
The experi mental
resu l ts of a s i mpl e te s t in th e l abora tory i n d i cate the fea s i b i l i ty of decrea s
i ng the d ra g forces o f the fl a t p l ate by reduc i ng the sol i d i ty ra t i o .

1 7 . Document Analysis

Dynami c loa d s ; Hel i osta ts ; t1echan i cal S tructu res ; : Sol a r Col l ector s ;
Stre s se s ; W i n d ; W i n d loa d s

a. Descriptors

b . Identifiers/Open-Ended Terms

c. UC Categories

62e
1 8. Availability Statement

Nati on a l Techn i ca l I nforma t i on Serv i ce

U . S . Depa rtment of Commerce
5285 Port Royal Roa d
Spri ngfi el d , V i rg i n i a 221 61

1 9. No. of Pages

50
20. Price

A03
Form No. 0069 (325-82)

1< U.S. GOVERNMENT PRINTING OFFICE:1985-576-106 / 20044