Vous êtes sur la page 1sur 12

Journal of Wind Engineering and Industrial Aerodynamics, 41-44 (1992) 2237-2248

Elsevier

2237

Dynamic gust response factors for guyed towers


A.G. Davenport~ and B.F. Sparlingb
~Boundary Layer Wind Tunnel Laboratory, University of Western Ontario, London, Ontario,
Canada, N6A 5B9
bDepartment of Civil Engineering, University of Saskatchewan, Saskatoon, Saskatchewan,
Canada, S7N 0W0 (Formerly with the BLWT-Laboratory)
Abstract
The dynamic component of guyed tower response is generally larger than the mean
component and has significantly different characteristics. Predictions of peak response based
on the static behaviour under steady winds can give misleading and often unconservative
results. A simplified method for estimating dynamic response is presented that uses a series
of static load patterns to approximate the effects of gusting winds. The response to these
patch k~ads is modified by factors that depend on the physical properties of the tower and the
characteristics of the wind. Comparisons are made between the patch load method, the gust
factor method, and the more rigorous statistical method.

1. INTRODUCTION
in the past, considerable emphasis has been placed on the non-linear behaviour of guyed
masts due to vm'iatkms in cable stiffness and axial loads in the mast. While efficient iterative
solution methods techniques have been used successfully for the static loading case, dynamic
anaiysis methods have been slower to develop. Because of this, most guyed tower design is
still based on static response to steady wind k ads even though it has been shown that this not
a reliable indicator of dynamic behaviour [1]. Without a more realistic representation of
dynamic response, then, the effort expended on elabol'ate solution methods may be misplaced.
While non-linear dynamic models are available, they remain impractical for design purposes
due to relatively sophisticated programming requirements and prohibitive computer execution
time. Generally these models involve step-by-step integration of the equations of motion in
response to excitation provided by a simulated or measured wind field I2,3].
A useful compromise is to calculate the mean or time average response including non-linear
effects, but to assume that the system vibrates in a linear fashion about the static equilibrium
position. This approach is suggested in the IASS design recommendations for guyed masts
141 and has been adopted for this study. In order for the linear vibration model to be valid,
the cables must be reasonably taut and the axial load in the mast should be well below its
buckling load.
~ r

",

0167-6105/92/$05.(X) 1992 Elsevier Science Publishers B.V. All rights reserved.

2238
Historically, dynamic wind response has been estimated by increasing the mean or steady
wind loads by a constant gust factor. A more rigorous approach utilizes the statistical
properties of the wind and considers the tower's response in each significant vibration mode
[ 1,5,6]. While this statistical method gives a realistic estimate of response, it is difficult to
implement on conventional structural analysis software. To overcome this difficulty, Gerstoft
[7] and others have proposed the use of static load patterns, or patch loads, to model'the
effects of wind gusts. This paper presents a revised patch load method that incorporates
structural properties and wind characteristics-in a more systematic manner than similar
previous methods.

2. RESPONSE TO GUSTING WIND


A typical form of response for guyed towers subjected to gusting wind is illustrated in Fig.
I a. This response could represent a wide range of structural actions including resultant forces,
bending moments, cable tensions, as well as deflections and accelerations.

fS,
p2
-

\-'

rB
t
(a)

L_
(b)

Inf

Figure I. Response of a guyed mast to wind; (a) Time history, (b) Power spectrum.

For design proposes, the peak dynamic t~sponse, i~, may be expressed as

f',,F+gr

(1)

in which P is the mean or time average response, i' is the rms of the fluctuating response, and
g is a statistical peak factor, generally in the range of 3 to 4,
Figure l b shows the manner in which the energy of the fluctuating response is distributed
with fl'equency, The mean square response for a specified range of frequencies is represented
by the area under the spectrum con'esponding to that range, it is evident from Fig. lb that
the fluctuating response may be subdivided into two distinct components:
Background response whose energy is spread over a broad band in the low frequency
range, and
Resonant response which consists of a series of highly concentrated energy peaks centred
on the natural frequencies of the tower,
The total mean square fluctuating response can be calculated as the sum of the background

2239
response plus the contribution from each significant vibration mode, hence

(2)
Here, [~ is the rms background response and rRj is the rms resonant response in the jth mode
of vibration.
The background response, by definition, occurs at frequencies below those at which
dynamic amplification effects are significant. It can therefore be treated as a quasi-static
action in response to slowly varying wind loads.
Unlike most structures which exhibit one or two dominant vibration modes, the resonant
response of guyed towers can include significant contributions from more than 15 modes. The
lowest 8 mode shapes for a 300 m guyed mast are illustrated in Fig. 2. Also included are the
relative contributions of structural and aerodynamic damping forces for these modes.
As indicated on Fig. 2, the vibrations modes tend to be closely spaced, increasing the
importance of modal coupling particularly at lower frequencies. In addition, response
calculations are very sensitive to uncertainties in the exact nature of the wind spectrum in the
low frequency range. Both of these complications are avoided by the separate treatment of
background and resonant response.

3. DYNAMIC RESPONSE CALCULATIONS


3.1 Gust Factor Method
In the gust factor method, an equivalent static wind pressure is used to estimate peak
dynamic response. This equivalent pressure, P~, contains an allowance for dynamic effects
and can be expressed
Po(Z)

- qo

Co(Z) cg

(3)

in which qo is the reference wind pressure (usually defined at 10 meters above the ground) and
Cd~z) is an exposure factor equal to the square of the ratio between the mean windspeed at
elevation z and the mean windspeed corresponding to the reference wind pressure. The gust
factor, C s, is intended to account for fluctuations in the wind speed but must also allow for
resonant amplification effects. Typically, Cs is taken to be 1.5 to 2.0.
The gust factor approach implicitly assumes that the dynamic response at every point is a
simple multiple of its static response to steady winds. While this assumption is valid for
structures with one or two dominant vibration modes, it does not accurately predict the
response of guyed masts which, as we have seen, can have many active vibration modes.

3.2 Statistical Method using Influence Lines


The method presented in this section has been developed and extensively repo~ted by
Davenport and his co-workers [1,5,6]. Since it serves as the basis of comparison for the
proposed patch loading method, the main features of the statistical method are briefly
described below.
A unified approach for a variety of structural responses is possible through the use of
influence lines which describe the effects of moving loads acting on the structure. Examples

2240

300mt/

I SI

200-

It

i /

100-

-4
I

ii

"(

.
i

MODE
FREQUENCY (Hz)

1
0.22

2
0.27

3
0,37

4
0.44

5
0.48

6
0.60

7
0.70

8
0.86

DAMPING (%)
STRUCTURAL
AERO. MAST
AERO. GUYS
TOTAL

0.5
0.6
6.1
7.2

0.5
0.5
3.9
4.9

0.5
0.5
2.8
3.8

0,5
1.4
0,7

0.5
0.4
1.6

0.5
1,4
0,3

0.5
1.0
0.2

0,5
0.8
0.2

2.6

2.5

2.2

1.7

1,5

Figure 2. Vibration modes and modal damping ratios for a 300 m tower
in strong wind (lst 8 modes only).

BENDING
MOMENT
AT B

SHEAR FORCE
BELOW B

Figure 3. Example influence lines h)r a 300 m tower.

GUY REACTION
AT B

2241
of influence lines for a guyed tower are shown in Fig. 3. The instantaneous response at some
point on the tower, r(t), due to along wind forces is given by
H

r( t) =

F(z, t) i r d z

(4)

where F(z,t) is the instantaneous lateral force at some elevation, z, and some time, t, and
is the value of the influence line for that response.
(i) M e a n R e s p o n s e :

iR(Z)

The mean response of the structure to steady winds is then

F "f F(Z) i r ~Z) dz


0

(5)

in which [~(z) is the time averaged lateral wind load defined by


F(z)
- -3.r p C . ( z ) D(z) u(z) z

(6)

Z - -

and p is the air density, CD(z) is the drag coefficient, D(z) is a characteristic width, and fi(z)
is the mean windspeed at elevation z. The mean windspeed profile can be described by any
suitable logarithmic or power law expression.
The static equilibrium position was determined using an iterative modified Newton-Raphson
solution technique. Non-linear guy stiffness based on a catenary model and P-A effects in the
mast were incorporated as were the effects of mean wind loads acting on the cables.
(ii) Background Response: The quasi-static mean square response to partially correlated
fluctuating loads c'ua be calculated by
HH
(7)
~ a . ff RF(ZI, Z~ ) F(zl)F(z 2) ir~(zl) irD(Z2) dzl d z 2
00

where Rl~(Zl,Z2) is the correlation coefficient between the fluctuating forces at the two
elevations, zt and z2, and i,D(Z) iS the influence line based on structural properties at the mean
displaced equilibrium position. The rms fluctuating wind load, F, is given by

- 0 CD(Z) D(Z) V(Z) ~

(8)

in which fi is the rms fluctuating windspeed in the along wind direction.


(iii) Resonant Response: For the jth vibration mode, the rms resonant response can be
approximated by the expression

2242
H

f m (z) p .i(z) i~odz


0
H

(9)

fro(z> p(z) dz
0

In this, m(z) is the mass of the tower per unit height at z including the mass of guys and pj(z)
is the jth mode shape. For locally correlated excitation at the jth natural frequency, fi, the
spectrum of the generalized force, denoted fjS~Fj(fj), can be written
H

H2

fo fJSe(fJ'z) p}(z) dz

1101

in which fjSrj(fj,z) is power spectrum of the wind force at fj and elevation z, and L~<fj) is the
spanwise scale of correlation at ti.
The total damping in the jth vibration mode is comprised of the sum ol~ the structural
damping, ~s, and the aerodynamic damping from the guys and the mast, ~. If the
aerodynamic force at elevation z can be expressed in the form of [Velocity of motion .A(z)]
then the modal aerodynamic damping ratio is
/4

f A z,
o

aj

az
Ill)

4 ejfm(zl

(z) dz

(iv) Response Components: The bending moment response for a 300 m guyed mast in open
cmmtry is summarized in Fig. 4. Several typical features of dynamic guyed tower respo.se
can be, s e e l l :

The background response is often the dominant component,


The fi.'m of the static or mean response is quite different from both of the dynamic
response components, and
. The form of the background and the resonant components is fairly similar with the ratio
between the two remaining nearly constant along the entire height of the tower.
These observations form the basis of the proposed simplified approach to dynamic response
based on patch loading.
3.2 Patch Load Method
The proposed patch h,ad method is all extension of the approach outlined in References 7
and 8. The method's flexibility is enhanced by the introduction of scaling factors that depend
on the physical properties of the tower and the nature of the wind load.
Rewriting Eq. I, the peak dynamic response may be approximated by

2243
300 m

200

I00

STATIC

BACKGROUND RESONANT

TOTAL

Figure 4. Bending moment response components for a 300 m tower.

PpL I = 2 i0 q, CeD(Zl)

- Z) = qo Ce(Z)

ZI

Step 1:
Mean Wind

i=

3
4
5
Step 2: Patch Loads

Figure 5. Definition of design pressures for the Patch Load Method.

2244
where the only new term, ~PL,is the peak fluctuating response as predicted by the patch load
method.
Mean Response: The mean response to steady winds can be determined by any suitable

static analysis method, preferably including non-linear cable and mast stiffness and the effects
of mean wind loads on guys. For this study, the influence line method outlined in the
previous section was used.
In the patch load method, a series of
static load patterns is used to recreate the effects of gusting wind. The specified load patterns
are shown in Fig. 5 and consist of lateral loads applied in tum to each span and then from
midpoint to midpoint of adjacent spans. The patch loads should be applied to the tower in
its static equilibrium position.
For each load pattern, an equivalent static pressure, Pt,L, is defined by the expression
Dynamic Response by the Patch Load Method:

P~,r." 2io qo CeD(Z~)

(13)

Here, qo is the reference wind pressure and C~D is a dynamic exposure factor equal to the
square root of Co used with the Gust Factor Approach calculated at mid height of the load
patch, z~. The turbulence intensity (calculated at the reference height), io, depends on the
reference windspeed and the surface roughness at the site.
When combined in an appropriate fashion, the effects of the load patterns are very similar
to the background component of dynamic response. Since the wind gusts producing the
background response are not fully correlated along the mast, the effects of the load patterns
must be added as the root sum of squares, or
"

:#L,

(14)

In this, i~,~ is the effective patterned load response, t'~,~ is the ~esponse t'rom the ith load
pattern, and n is the total number of load patterns required.
Finally, the effective patterned load response must be scaled to reflect the influence of the
physical properties of the guyed mast (such as guy and mast stiffness, mass, span lengths, and
drag areas) and of characteristics of the wind (such as mean windspeed profile, turbulence
intensity, and average gust size). The peak fluctuating response, i~pL,is then the product of i~pL
and modification factors given in the expression

fl'L " 'fJ'~. X'e ~'.e LTL 9'

(15)

where kx~ is the background scaling factor, ~.r is the resonant magnification factor, gn. is the
turbulent length scale filctor, and g is the statistical peak factor.
The scaling factors were calibrated using tile statistical analysis results fi'om 8 guyed towers
ranging ill height fi'om 123 m. to 622 m. The variability of each scaling factor was found to
be small for a wide range of towel" sizes and properties. For the s',uke of simplicity, therefore,
they could be assigned conservative numerical values as follows; ~,~ = 0.75, kr = 1.2, and LrL
= 1.05, Assuming that g = 4.0, Eq. 15 becomes

2245
fpr. " 3 . 7 8 f,,r.

(16)

A more detailed method for determining the scaling factors is outlined below.

(i) Background Scaling Factor:

The effective patch loading response, iPL, tends to


overestimate the magnitude of the background response and must be scaled down accordingly.
The amount of reduction depends on the type and location of the response being considered.
Response influence lines are affected to some degree by the relative stiffness of the guys
and the mast. Since the shape of the influence line determines the effect of a wind gust on
the tower, this ratio of stiffness also influences the background scaling factor for certain types
of responses. An approximate measure of the stiffness ratio, [~s, may be defined as

(17)

q )
in which m is the number of guy levels, l~i is the total guy stiffness at the ith guy level, HGi
is the height of the ith guy level, EM is Young's Modulus for the mast, IM is the second
moment of the cross sectional area of the mast for bending in the along wind direction, and
Ls is the average span length between guys. For the 8 towers tested, 13s varied from 4 to 37.
Empirical expressions for the background scaling factor, kB, are given below:
Midspan Moments:

XB = 0.76 l~s"'7

Deflections:

kn = 0.90 13s"'s

Shear, Support Moments:

~,~ = 0.70

(ii) Resonant Magnification Factor: The ~sonant magnification factor, ~ , is the ratio
between the total rms fluctuating response and the background response. As such it reflects
the participation of all inertial and damping forces.
As a measure of the inertial and damping forces acting on the mast, Davenport and Allsop
[1] have defined a non-dimensional inertial resistance.factor, Q, which is determined by the
expression
1
6
1

p4

2246
where qn is the mean wind pressure at the top of the mast, m Mis the average unit mass of the
mast, and AD is the average effective drag area of the mast per unit height.
For the 8 towers included in this study (Q = 0.65 to 1.75), the resonant magnification
factor and the inertial resistance factor appeared to be linearly related by the equations
Moments:

~'R = 0.17 Q + 0.91

Shear:

;~R = 0.15 Q + 0.92

Deflection:

~'R

= 0.08 Q + 0.95

The definition of Q assumes a reasonably uniform distribution of mass, stiffness, and


aerodynamic damping. Guyed masts with unusual concentrations of any of these properties
should be analyzed using more rigorous techniques.
(iii) Turbulent Length Scale Factor: The turbulent length scale factor, grL, predicts how
effectively an average sized gust will interact with the tower. It depends on the ratio between
the average span length, Ls, and the vertical length scale of along wind turbulence, Lu.
Depending on the exposure conditions at the tower site, the turbulent length scale can vary
greatly but is generally assumed to be between 40 - 70 m. The approximate expressions for
LrL given below were derived for ratios of (Ls/Lu) ranging from 0.25 to 1.25.
Moments - Midspan:

Shear

- Support:

~'rL = 1.0

- Midspan:

LT~ = !.0

- Support:

Lrl, = (Ls / Lij)"'~'"

Deflection:

4.

LrL = (Ls / Ltj)'13s

LrL = (Ls/L~) T M

COMPARISON OF ANALYSIS METHODS

The peak dynamic responses for three towers (95 m, 295 m, and 622 m high) were
estimated using three methods; the gust factor method (with C~ = 2.0), the statistical method,
and the patch load method. Results are summarized in Fig. 6. Since the dynamic response
can either add to or subtract from the mean response, only the absolute value of the worst case
is shown. Although not included in this report, the analysis of six other tower,,; yielded very
simil~u' results.
The three analysis methods generally predicted similar maximum moment and shear values
at the guy support levels. It is evident, however, that the gust factor method tends to seriously
underestimate these responses away from the supports. This feature of the gust factor method
arises from the assumption of fully correlated loading on all spans. In an analogous fashion
to live loading on a continuous frame, the most severe loading case often results from a

224"/

~ o "r"

/..

/.,"

II

jI

~, o ~ : / , , -

statistical

~[

i
O"

04,,

.... I

...~t~..h..~..?#

O"
O*

2.5

zooo

6000

loo

200

(a) 622 m. lower; 9 guy levels (GlO = 22.4 m/s; z o = 0.028 m)


0

0"

/I

/"
~,,"

bJ

'

0 0

'

0.5

'

1.0

1.5

500

1000

1500

50

100

150

(b) 295 m. tower; 4 guy levels (1~10 = 25.0 m/s; z 0 = 0.100 m)


0

/I I

Oo

E//o
I

0.00

0.25

0.50

10

20

'

'

2.5

MOMENT (kN-m)
SHEAR (kN)
DEFLECTION (m)
(c) 95 m. tower; 6 guy levels (ulo = 28.4 m/s; z o = 0.010 m)
(Note - Patch Load and Statistical curves may coincide)
Figure 6. Comparison of dynamic analysis methods

2248
reduction of wind load on adjacent spans due to uncorrelated wind gusts.
In contrast, the gust factor method tends to overestimate peak deflections. Unlike those for
shear and moment response, deflection influence lines tend to exhibit the same sign, or sense,
over most of the mast length. This means that deflections are more sensitive to the total load
acting on the tower rather than the distribution of loading. For the case of deflections, then,
an assumption of fully correlated loads would overestimate the response.
The patch loading method was able to reproduce the statistical moment, shear, and
deflection response within a few percent over the entire tower height. In particular, maximum
moment and shear values were accurately predicted for the support and midspan locations.

5.

CONCLUSIONS

A simplified method for estimating peak dynamic responses of guyed masts has been
outlined. The simplified patch load method uses a series of static load patterns and specified
modification factors to approximate the fluctuating response. As only static loading is
considered, the patch load method can be readily implemented on existing analytical packages.
The gust factor method, currently used by many design codes, appears to adequately predict
maximum moment and shear response at guy support levels bt~t to seriously underestimate
these responses in midspan regions. The patch load method, on the other hand, seems to
consistently predict accurate dynamic responses at all locations on the tower.

Acknowledgements
This research at the Boundary Layer Wind Tunnel Laboratory was made possible through
grants fi'om the Natural Sciences and Engineering Research Council of Canada and to the
support to Sparling fi'orn the Ken Anthony Memorial Felh~wship, co-sponsored by Ove Arup
of London. England, and the BLWTL.

REFERENCES
1
2
3
4
5
6
7
8

Davenport, A.G., and Allsop, A., The dynamic response of a guyed mast to wind, lASS
Meeting on Tower Shaped Structures, Milan, Italy, 1983.
lannuzzi, A., Aerodynamic response of guyed masts: A deterministic approach, lASS
Bulletin, No. 89, 1986, pp. 47-59.
lannuzzi, A., and Spinelli, P., Response of a guyed mast to real and simulated wind, IASS
Bulletin, No. 99, 1989, pp. 38-45.
lASS, Recommendations for guyed masts, International Association for Shell and Spatial
Structures, Madrid, 1981.
Addie, R., Guys, guy systems and guyed towers, Ph.D. Thesis, University of Western
Ontario, London, Canada, 1978.
Allsop, A.C., Dynamic wind analysis of guyed masts: Simplified Methods, M.Sc. Thesis,
University of Western Ontario, London, Canada, BLWT-7-1984, 1984.
Gerstoft, P., Simplified methods for dynamic analysis of a guyed mast, M.Sc. Thesis,
University of Western Ontario, l,ondon, Canada, 1984.
Gerstoft, P., and Davenport, A.G., A si~nplified method for dynamic analysis of a guyed
mast, Journ. of Wind Engineer. and lndustr. Aerodynamics, Vol. 23, 1986, pp. 487-499.

Vous aimerez peut-être aussi