You are on page 1of 153

RESPONSE OF ELECTRICAL TRANSMISSION LINE CONDUCTORS

TO EXTREi^E WIND USING FIELD DATA


by
RADHAKRISHNA R. KADABA, B.E., M.E.
A DISSERTATION
IN
CIVIL ENGINEERING
Submitted to the Graduate Faculty
of Texas Tech University in
Partial Fulfillment of
the Requirements for
the Degree of
DOCTOR OF PHILOSOPHY

Approved

May, 1988

ACKNOWLEDGEMENTS
I would like to express my special thanks and sincere
gratitude to the chairman of my committee. Dr. Kishor C.
Mehta, for his guidance, everlasting inspiration, constant
encouragement and patience in teaching me throughout my
graduate study program.

I am also grateful to Drs. Eric L.

Blair, H. Scott Norville, William Pennington Vann, and Y.C.


Das, the members of my advisory committee, for their helpful
review and constructive criticism of this dissertation.
The research was accomplished under the financial
support of the Bonneville Power Administration (BPA) and the
Institute for Disaster Research (IDR) at Texas Tech
University.

Support of these organizations is appreciated.

Special thanks are extended to Mr. Leon Kempner, Jr. for his
assistance in providing the field data details of the BPA
project and encouragement.

Also, I would like to express my

deep appreciation to the Chairman of the Civil Engineering


Department, Dr. Ernst W. Kiesling, for his encouragement and
support during the course of my graduate study.
I shall ever remain indebted to my sister Vijaya
Bhanuprakash, brother Udaya Kumar, sister-in-law Latha
Ananth, and my other family members for their kind blessings

11

and the moral support during my stay away from home.


Special thanks are given for the assistance and friendship
provided by my colleagues Suresh Jonnagadla, Marc Levitan,
Saranga Kidambi and Pankaja Kidambi.
I must acknowledge with deep appreciation and
encouragement of my brother, Sri. Anantharamu, who instilled
in me the value of education and stood by my side giving all
the moral support he could and most of all his kind prayers
and love.
Finally, I would like to express 'Thanks' to my
parents, to whom I dedicate this dissertation.

Ill

CONTENTS
ACKNOWLEDGEMENTS

11

ABSTRACT

vi

LIST OF TABLES

viii

LIST OF FIGURES

CHAPTER
I. INTRODUCTION

Objectives

' 5

Contents of the Dissertation


II. STATE OF KNOWLEDGE

6
7

Response
Mean Response
Fluctuating Response
Aerodynamic Admittance Function
Mechanical Admittance Function
Peak Factor
Wind Characteristics
Mean Wind Speed
Mean Wind Profile
Turbulence Characteristics
Turbulence Intensity
Gust Spectrum
Davenport Analytical Model
Conductor Damping Ratio
Conductor Fundamental Frequency
III. FIELD DATA
Description of Test Site
Instrumentation
Data Accjuisition
Recording Procedure
Description of Recordings
IV

8
10
12
16
19
20
24
24
25
28
30
31
34
37
38
39
41
43
47
48
50

IV. FIELD DATA ANALYSIS

52

Validity of Wind Data


Power-Law Exponent for Wind Profile
Turbulence Intensity
Kaimal's Gust Spectrum Constants
Validity of Conductor Response Data
Effective Conductor Force Coefficient
Response Spectrum

53
60
65
68
74
80
85

V. COMPARISON AND REFINEMENT OF THE ANALYTICAL


MODEL

90

Comparison of Analytically Predicted Mean


Scjuare Response With Field Measured
Values
Field Measured Mean Square Response
Analytical Model Predicted Mean Scjuare
Response
Refinement of the Analytical Model
Background Response
Determining the JAF Coefficients
Resonant Response
Determining the Aerodynamic Damping Ratio
Resonant Response with Suggested
Damping Ratio
Peak Factors
Probabilistic Peak Factors from Field
Data
VI. CONCLUSIONS
REFERENCES

91
92
94
104
104
108
115
115
123
127
130
134
138

ABSTRACT
Conductors are long, slender, flexible, and wind
sensitive structures.

In most cases of transmission lines,

60-80% of wind loads coming to the support structures are


transferred from conductors.

Thus, assessment of conductor

response due to extreme wind is an important part of the


overall prediction of wind loads on transmission structures,
Extreme winds not only contain high wind speeds but also
randomly fluctuating gusts.

These gusts cause fluctuating

dynamic responses of conductors.

Since the responses

fluctuate randomly, they need to be assessed in


probabilistic terms.
An analytical model for estimating dynamic response of
transmission line structures to wind loads is developed by
Davenport.

The model can be verified using field data to

determine its effectiveness.

Bonneville Power

Administration (BPA) has conducted experimental studies in


the field at the Moro site to collect wind and electric
transmission structure response data.

During 1981-1982 BPA

collected 23 separate 12-minute duration records that


included wind speed, wind direction and conductor response
data.

The conductor response records include load cell.

VI

transverse swing angle, and longitudinal swing angle


recordings.
The BPA data of wind and conductor response are
analyzed in detail to gain as much information as possible.
The analysis of wind data include determination of wind
characteristics of mean wind profile, turbulence
intensities, and gust spectra.

The conductor response in

terms of peak responses, effective force coefficients, peak


factors, and response spectra are obtained.

The response

spectra are further analyzed to obtain contributions of


background response and resonant response.

Comparison of

the analytical model with field data reveals that the model
underestimated background response and overestimated
resonant response.
Results of these data analyses are used to improve the
analytical model to predict conductor response in extreme
wind.

The significant improvement includes determination of

peak factors from upcrossing rates, refinement in the


expression for background response and determination of
conductor aerodynamic damping ratios from field data.

VI1

LIST OF TABLES
1. Typical Values for Gradient Height and Power-Law
Exponent (ANSI, 1982)

28

2. Typical Values for Turbulence Intensity and


Surface Drag Coefficient (Kempner, 1982)

31

3. Field Data Records During 1981-1982

40

4. File Description of Mode 22

49

5. Mean Azimuth and RMS in Degrees of Wind

55

6. Mean Wind Speed in mps

57

7. Power-Law Exponent and Kaimal's Gust Spectrum


Constants

64

8. Turbulence Intensity

66

9. Mean and RMS Values of Conductor Response


(Transverse Load Component)
10. Field Measured Conductor Effective Force
Coefficients

78
82

11. West Conductor Response Spectrum Data Analysis

95

12. East Conductor Response Spectrum Data Analysis

96

13. Central Conductor Response Spectrum Data Analysis

97

14. Fixed and Assumed Parameters Used in the


Analytical Model

99

15. Variable Parameters Used in the Analytical Model

100

16. Background Response of West Conductor

116

17. Background Response of East Conductor

117

18. Background Response of Central Conductor

118

Vlll

19. Estimated Aerodynamic Damping Ratios in


Percentages

122

20. West Conductor Resonant Response With 40% Damping

124

21. East Conductor Resonant Response With 40% Damping

125

22. Central Conductor Resonant Response With 40%


Damping
23. Peak Factors for Conductor Response

126
129

IX'

LIST OF FIGURES
1. Fluctuations of Wind Speed

2. Fluctuations of Conductor Response

3. Conductor Force Coefficients Based on Wind Tunnel


and Full-Scale Tests (Davenport, 1980)

13

4. Elements of Response Spectrum Analysis

15

5. Example of a Random Variable Showing Upcrossings


of a Given Threshold
6. Idealization of Gust Spectrum Plot Over an

22

Extended Range (Davenport, 1972)

26

7. Topography of Site and Orientation of Power Lines

42

8. Schematic of Tower 16/4

44

9. Elevation Along the Test Line (Vertical Scale


Exaggerated)
10. Time History Plot of Wind Speed for Record NOl at
34.7 m on Tower 16/4
11. Mean Wind Speed and Direction Recorded at 34.7 m
on Tower 16/4 of 23 Records
12. Power-Law Plot for Record N15
13. Gust Spectrum Plot for Record NOl Recorded at
34.7 m on Tower 16/4
14. Time History Plot of West Conductor Response for
Record NOl
15. Response Spectrum Plot for West Conductor
Response for Record NOl
16. West Conductor Response Spectrum Plot for Record
NOl

45
59
61
63
70
76
86
93

17. Analytical Model Background Response Versus Field


Measured Background Response

102

18. Analytical Model Resonant Response Versus Field


Measured Resonant Response

103

19. Frequency Transfer Function of West Conductor


Response for Record NOl

110

20. West Conductor JAF Coefficients Contour Plot for


Record N15

112

21. Joint Acceptance Function Plot


22. Refined Model Background Response Versus Field
Measured Values
23. Analytical Model Resonant Response With 40%
Damping Versus Field Measured Values
24. Cumulative Probability Distribution of
Upcrossings for Conductor Response

XI

114
119
128
133

CHAPTER I
INTRODUCTION

Electrical transmission line systems are engineered


structures that traverse over all types of terrain.

Wind

loading is an important factor in the design of these


transmission line systems, consisting of towers, conductors,
and ground wires.

Transmission line conductors are long,

flexible, and wind sensitive structures.

Probably no other

structure has as much of its mass in highly flexible form,


and so continuously exposed to the forces of wind, as do
transmission line conductors.

The loads due to the effect

of wind acting on the conductors, which in turn, transmit


loads to the supporting tower, are more than the loads due
to the wind acting directly on the tower itself.

Wind loads

on conductors with spans of around 300 m account for 60 to


80% of the total wind load effect on the support tower
structure.

Accurate and reliable prediction of wind loads

that are transferred from conductors to the towers are


desirable to produce an economical and safe design of
support tower structures.
Transmission line tower structures are usually designed
for five different types of loads (Kempner, 1985): (1)

extreme wind, (2) wind on ice, (3) National Electric Safety


Code (NESC, 1984), (4) broken conductor, and (5)
construction loading.

Records show that more than 50% of

tower structure failures are due to extreme winds.

Under

these conditions, any improvement in the understanding of


conductor behavior under extreme winds which leads to better
definition of loads and better accompanying design of the
tower structure is desirable.
Extreme winds not only contain high wind speeds but
also randomly fluctuating gusts (see Figure 1 ) . The gusts
cause fluctuating wind loads on transmission line
conductors.

Transmission line conductors respond to random

gust loading in a randomly fluctuating manner (see Figure


2).

Response of conductors due to wind can be considered as

a combination of mean response (static) associated with the


mean wind and fluctuating response (dynamic) associated with
wind gusts.

Fluctuating response is conveniently expressed

as the product of a peak factor, g, and root mean scjuare


(RMS) value of fluctuations about the mean response.
The response fluctuations about the mean response can
be represented in the frecjuency domain by a response
spectrum.

A response spectrum is a plot of spectral density

values versus frecjuency.

The area under the response

spectrum is ecjual to the mean square value of the

Mean + Sigma
a

Mean

a
a>
o

Mean - Sigma

Q.

en

Time (minutes)

Figure 1:

Fluctuations of Wind Speed

c
o
a.
a>

o
t5
a

c
o
O

Time (minutes)
Figure 2:

Fluctuations of Conductor Response

fluctuating response.

Fluctuating response can be viewed as

background response due to low frequency wind turbulence and


resonant response near the natural frecjuencies of conductor
vibration.

The fluctuating response is discussed in detail

in subsecjuent chapters.

The use of a peak factor to

establish an equivalent static design load for conductors


due to wind is convenient.

The peak factor is defined as

the number of standard deviations by which the peak value


exceeds the mean value.
There are a variety of traditional methods for
computing wind loads on transmission line structures that
estimate ecjuivalent static response of the structure to
these loads.

These traditional methods are simple,

empirical, and usually conservative; therefore the resulting


designs are adecjuate in resisting the design wind.
An analytical model for estimating dynamic response of
transmission line structures to wind loads has been
developed by Davenport (1980).

This analytical model

obtains the gust response factor in a single equation using


the frecjuency domain approach.

The model considers all the

major wind characteristics and structural properties to


estimate dynamic response.

Natural frecjuencies of the

towers are generally much higher than natural frecjuencies of


the conductors.

Hence it is assumed that conductor response

is not influenced by the motion of the supporting tower


structure.

With the above assumption, the response due to

wind on the conductor and on the tower structure can be


analyzed separately.
Wind tunnel experiments and full-scale field tests have
been conducted on transmission line structures.

Because of

their slenderness and flexibility full-scale tests are


especially significant to assess dynamic response of these
structures.

Full-scale tests are of great value to compare

and refine the analytical model.

The most comprehensive

source of full-scale data is the experiments conducted on


John Day-Grizzly transmission line 2 located in northern
Oregon.

These field data, collected by the Bonneville Power

Administration, are used in this study.

The data include

simultaneous recordings of wind and transmission line


response during extreme winds.

Analysis of field data of

wind and transmission line conductor response can assist in


substantiating and improving the analytical model.
Objectives
The general objective of this research is to determine
the dynamic response of conductors due to extreme wind using
field data.

The Specific objectives are:

(1) to assess

wind parameters from the field data, (2) to develop

probabilistic peak factors for conductor response using


upcrossing rates, (3) to determine the aerodynamic damping
and the joint acceptance function for conductors from the
field response data, and (4) to compare dynamic response of
conductors measured in the field with a refined analytical
model.
Content of the Dissertation
A brief description of the contents of this
dissertation is given here.

The next chapter contains an

overview of the state of knowledge concerning wind


characteristics, and details of the analytical model to be
used to predict dynamic response of conductors.

description of the measurements, site characteristics, and


instrumentation for the field data is given in Chapter III.
Analyses of wind and conductor response field data are
described in Chapter IV.

Comparisons of field measured

conductor response data with responses calculated using the


analytical model are part of Chapter V.

Determination of

probabilistic peak factors for conductor response using the


upcrossing rate principle, refinement of the background
response expression, and determination of conductor
aerodynamic damping from the field response data are also
presented in this chapter.

Conclusions reached in this

study are presented in Chapter VI.

CHAPTER II
STATE OF KNOWLEDGE

Wind loads on transmission line conductors depend on


wind characteristics and on interaction phenomena of the
wind with conductors.

Wind speed fluctuates randomly; it

can be considered to consist of mean and fluctuating (or


gust) components.

A knowledge of both the mean wind speed

and the random fluctuations are recjuired to evaluate wind


loading.

In addition, structural properties (natural

frecjuencies, damping, size, shape,..etc) play an important


role in prediction of response of the conductors in extreme
winds.
The difficulty of proper simulation of the natural wind
characteristics and scaling of transmission line structures
in wind tunnels leads researchers to depend on full-scale
experiments.

There have been only a few full-scale

experiments for wind loads on transmission line structures.


Field measurement programs have been conducted in the United
States, Canada, Europe, and Japan to monitor the wind
response of transmission line systems.

A review of each of

the field test programs is given in a report by GAI


consultants (1981).

These field tests measured wind and

8
response data for a specific design objective such as
determination of the span factor or the gust response
factor.

Detailed analyses of field data are not reported in

the open literature.

Bonneville Power Administration (BPA)

has conducted field experiments on transmission line


structures since 1976.

Results of analysis of BPA data are

available in several reports and papers (Kempner and


Laursen, 1977, 1979, 1981; Kempner and Thorkildson, 1982;
Ferraro, 1983; Norville, 1985).

These reports and the

present study indicate that conductor response to extreme


winds is governed by wind turbulence characteristics,
aerodynamic characteristics of wind structure interaction
and structural dynamics of the conductors.

The wind and

conductor response data collected by BPA during 1981-1982


are used in the present study.

The literature on wind,

aerodynamics, and structural dynamics is very extensive.


Only the items that are obtainable through analysis of field
data and are pertinent to this study are discussed in this
section.
Response
The design of a transmission line tower structure is
generally based on the peak transverse load component of the
conductors when subjected to extreme winds.

The transverse

load component transferred to the tower is the direct result


of the transverse response of the conductor.

The the

transverse load component is considered as conductor


response in the present study.

The prediction of this peak

response value, rather than a mean response value, is needed


for design purposes.

Peak response is predicted by the

summation of mean and fluctuating responses.

For a time

period, T, the peak response can be estimated by

ft = R + g aj^

(2.1)

where ft = peak response,


R = mean response,
a = root mean scjuare (RMS) of the fluctuating
response about the mean response, and
g = statistical peak factor.
In ecjuation 2.1, the mean response is based on the mean
wind speed.

The fluctuating response is a product of peak

factor and RMS value of response.

The dimensionless peak

factor, g, is probabilistic because of the random nature of


the fluctuating response.

The peak factor is determined

from the probability of the upcrossing rate or,


ecjuivalently, a specified number of occurrences in a given
interval of time.

The RMS of the fluctuating response

10
depends on wind characteristics such as the turbulence
intensity and structural characteristics such as the
damping, frecjuency, shape,., etc.
Mean Response
The mean response of conductors is obtained from the
mean wind pressure acting at the effective height of the
conductors.

The effective height of the conductor is

considered as the average height of the conductor above the


ground level.

The ecjuation for mean wind pressure is


1
-2
P = y P V C^

(2.2)

where P = mean wind pressure,


p = mass density of air, 1.226 kg m' , at 60

F,

at sea level,
V = mean wind speed at the effective conductor
height, and
C^ = conductor force coefficient.

The mean response of conductors, R, can now be expressed as:


R = P L d
where

L =

effective conductor span, and

d =

conductor diameter.

(2.3)

11
The mean response of a conductor depends on the mean
wind speed and the aerodynamic relationship in terms of
conductor force coefficient.

The force coefficient converts

the stagnation pressure term (-^pV )


transverse force on the conductor.

in ecjuation 2.2 to a
In most cases, the force

coefficient is determined from wind tunnel tests and, in


general, it is a function of Reynolds Number, the angle of
incidence of the wind, and the shape and roughness of the
conductor.

Published results of wind tunnel measurements

show wide variability in force coefficient values, because


of difficulty in simulation of Reynolds Number and varying
tests conditions (Potter, 1981).
The force coefficient of a cylindrical shape is
strongly influenced by the Reynolds Number, Nj^, which is
given as
Np = - B ^
R

where

(2.4)

\x

= wind speed,

d = conductor diameter,
-5
-2
^1 = dynamic viscosity of air, 1.79x10
N-sec m
at 60F, at sea level, and
p = mass density of air.

12
A plot of force coefficient versus Reynolds Number is
shown in Figure 3.

The region of the curve where the force

coefficient decreases sharply with Reynolds Number is called


the critical flow range.

This decrease in force coefficient

is related to the transition from laminar flow to turbulent


flow.

For a typical conductor diameter and design wind

speed, the Reynolds Number is usually above the critical


4
range (N > 5 x 10 ) . A constant value for the conductor
force coefficient is usually given in transmission line
design recommendations (ASCE, 1984).

As indicated in Figure

3, full-scale measurements tend to give lower force


coefficient values than the wind tunnel experiments.

These

discrepancies are not yet resolved in the published


literature.

Additional data on force coefficients from

field measurements are desirable.


Fluctuating Response
Conductor response to fluctuating wind depends upon the
dynamic characteristics of the conductor as well as
turbulence in the wind.

To determine the response of a

conductor subjected to fluctuating wind, frecjuency domain


methods are usually used.

Frecjuency domain methods are

popular for computation because they are cost effective and


efficient.

In frecjuency domain analysis, fluctuations in

13
Conductor diameter (in)

Curve

Source

1
2
3
4
5

Wind Tunnel Tests


Wind Tunnel Tests
Wind Tunnel Tests
Full-Scale Tests
Full-Scale Tests

1.125
0.770
1.695
1.108
1.602

1.2

1
1

>

o
o
o
o

to

(D\
\

w
0.9

o
P

^^-

\^

(1)
0.8

\
\

B
o

/
0.7

\ \

c
o
O

0.6

0.5

Reynolds Number *10^ NR

Figure 3:

Conductor Force Coefficients Based on Wind Tunnel


and Full-Scale Tests (Davenport, 1980)

14
the wind and conductor response are represented by a
spectrum.

A spectrum is a plot of energy at each frequency

versus the frecjuency.

Therefore, it represents a

distribution of energy over the entire frecjuency range.

The

area under the spectrum is ecjual to the mean scjuare value of


the fluctuations.

The frecjuency domain approach to compute

the peak response is briefly described below.


Several steps to obtain the mean scjuare value of
conductor response from the wind gust spectrum are shown in
Figure 4.

The first step in the analysis involves the

transformation of the gust spectral density function, S (f),


into the force spectral density function, Sp(f), by
2
multiplying by the aerodynamic admittance function, x (f)The second step involves the determination of the response
spectral density function, S(f), by multiplying the force
spectral density function by the mechanical admittance
2
function, H (f). The aerodynamic admittance function and
the mechanical admittance function are frecjuency response
functions.

The third step is to calculate the mean square

value of the response, CT^, from the area under the response
spectrum.

Once the RMS value of response, cjj^, is obtained,

a peak value of the fluctuating response is determined by

15

Gust Spectrum

-2/3

Aerodynamic Admittance

Force Spectrum
logl

CO

Mechanical Admittance

Response Spectrum

Figure 4:

Elements of Response Spectrum Analysis

16
multiplying the RMS value by a statistical peak factor, g.
The peak response ft, is the addition of the mean response,
R, and the fluctuating response, <3 ^r^, as indicated in
ecjuation 2.1.
Field measurements of wind and conductor response
provide gust and conductor response spectra.

Appropriate

analysis of field data leads to estimation of the frecjuency


response functions, as indicated in Chapter V.

In addition,

tJie field data provide the peak factor in probabilistic


terms using the upcrossing rate principle.

Theoretical

expressions for the aerodynamic admittance function,


mechanical admittance function and peak factor are presented
below.
Aerodynamic Admittance Function
For a given body immersed in a flow, wind fluctuations
can be used to determine the information on resultant forces
by empirical coefficients.

The time-varying transverse

force on a body completely enveloped by wind is given by the


formula

F = [i. p (V + u ) ^ C^i A

where

F = transverse force,
p = mass density of air.

(2.5)

17
V = mean wind speed,
u

= fluctuating component of wind speed,

C^ = force coefficient, and


A

= area of exposure.

The time-varying fluctuating force is divided into two


components, mean force, F, and fluctuating force, F'.

Then

ecjuation 2.5 can be expanded as


F -f F' = ^ p (V +2VU + U'') C^ A.

(2.6)

2
If the term of the order u is neglected, the mean and
fluctuating forces can be separated as
1 p V
-2 C^ A
F = -i.

(2.7)

F' = p V u C^ A.

(2.8)

and

The power spectrum for fluctuating transverse force,


F', is then related to the gust spectrum as follows:
2
Sp(f) = (p V A C^)

S^(f)

(2.9)

or
_2
Sp(f) = ^ S^(f).
V

(2.10)

18
Ecjuation 2.10, is valid over the range of frecjuencies
contained in the gust spectrum provided all effects remain
perfectly correlated.

In practical conditions where gust

effects over the entire length of the conductor may not be


correlated, an adjustment factor or aerodynamic factor is
included in the ecjuation.

This factor is called the

aerodynamic admittance function, x (f)/ and equation 2.10


becomes:

Sp(f) = ^

X^(f) S^(f)-

(2.11)

The aerodynamic admittance function is a frecjuency


transfer function which transfers the gust spectral density
function to a force spectral density function.

It accounts

for the correlation of gusts over the structure.

The

distribution of gusts over the structure depends on the


relative size of the structure and the gusts.

A large gust

totally enveloping the structure will be well correlated


over the structure, while small gusts acting over only a
portion of the structure are uncorrelated.

In general,

low-frecjuency gusts are assumed to be correlated over the


2
structure; that is x (f) is assumed close to unity. The
2
value of X (f) fall below unity at frequencies in the range

19
of interest for the effects of winds on conductors.

This

variation in aerodynamic admittance function as a function


of frecjuency is illustrated in Figure 4.

The aerodynamic

admittance function for a structure is generally obtained


from wind tunnel tests (Blevins, 1977).

In this study

coefficients of this function are obtained from the field


data (see Chapter V)-.
Mechanical Admittance Function
After the force spectral density function, Sp(f), is
obtained by means of ecjuation 2.11, the response spectral
density function, S(f), is obtained by multiplying Sp(f) by
2
the mechanical admittance function, |H(f)| :
Sj^(f) = |H(f)|2 Sp(f).

(2.12)
2

The mechanical admittance function, |H(f)| , is


determined from an analysis using the stiffness, mass, and
damping characteristics of the structure.

For a single

degree of freedom system, the mechanical admittance function


is the scjuare of the structural dynamic amplification
function; it is expressed as (Bendat, 1980):
H(f)|2 = J- ^
k'

^
2

(2.13)
2

f
o

20
where f^ = fundamental frecjuency,
C = damping ratio, and
k = spring constant.
The form of the mechanical admittance function is
illustrated in Figure 4.

The resulting spectrum of the

response, shown in Figure 4, is peaked at the fundamental


frecjuency of the structure.

This peak is the resonant

response of the structure at that frecjuency.

One of the

major unknowns in the mechanical admittance function is the


dcunping ratio, C,. Damping can be due to structural and
material properties, and for wind response, aerodynamic
interaction.

Structural damping can be assessed only from

experiments.

A theoretical expression for aerodynamic

damping is presented in a subsecjuent section of this


chapter.

In this study damping of conductors is determined

from field data; this is presented in Chapter V.

Peak Factor
Another important component in ecjuation 2.1 is the
statistical peak factor, g.

Davenport (1977) has shown that

for a stationary random process, the statistics of the peak


response values may be represented by a Type I extreme-value
probability distribution.

For this case the peak factor

21
corresponding to the peak response occurring in time period,
T, can be approximated as:
0.577
^-^''

g = V2 In yT +

(2.14)

V2 In yT
Where

is the cycling rate of the process; that is, the

number of times the mean response value is crossed per unit


time.
The peak factor has also been determined from the rate
of upcrossing.

Melbourne (1975) used this principle of rate

of upcrossing as developed by Rice (1945) to analyze wind


tunnel aero-elastic model data.
Consider a continuous random process that can be
differentiated at least once.

A sample function of the

random process is shown in Figure 5.

The crossings of the

level x(t)=Ti, with a positive slope (upcrossings) are shown


in the figure.

The number of crossings of the level in the

time interval T is a random variable.

For a long period of

time the expected or mean number of crossings will approach


some fixed value.

Based on this average value, the average

crossing rate can be determined.


Rice (1945) showed that the average crossing rate can
be computed for any stationary random process x(t), if the
joint density distribution is known for x(t) and x(t) (the
sample functions of x(t) being dx(t)/dt).

The average

22

Up-Crossings

time, t

Figure 5:

Example of a Random Variable Showing Upcrossings


of a Given Threshold

number of upcrossings of the value x per unit time is


expressed as:
N (x) = j X p(x,x) dx

(2.15)

where p(x,x) = the joint density of x and x,and


N (x) = the average number of upcrossings.
For a linear single degree of freedom system excited by
a stationary Gaussian load, the joint probability density
can be written as (Nigam, 1983)

23

P^""'^) =

?nn\

.2

exp[-JL--^l
X

(2.16)

2
where a

= mean scjuare of x,

and

2
a = mean scjuare of x.
There is no covariance term a

in the density equation

2.16 because x and x are assumed to be uncorrelated.


Substituting ecjuation 2.16 into ecjuation 2.15 and performing
the indicated integration yields

1 "^x
x^
N (^) = ^ ^ e x p { - ^ l .
2" ^x
2al

(2.17)

For a narrow band random process, the spectral energy


is centered close to the fundamental frecjuency, f , of the
structure.

Thus ecjuation 2.17 can be written as (Reelect,

1969):
N

= f

x^
expl-^l-

(2.18)

2^x
The cumulative probability distribution in terms of
upcrossings can be stated as
-*
P(>x) = EJ2LL

^o

2
= exp{--^l.

2ol

(2.19)

24
The upcrossing rate formulation is for a narrow band
vibration process.

The upcrossing rate technicjue is applied

to the data of conductor response in Chapter V to obtain


probabilistic peak factor, assuming that the conductor
vibrates at its fundamental frecjuency (as indicated by data
in Chapter IV).
Wind Characteristics
Wind fluctuates randomly both in time and space.

Wind

speed over a given time interval can be considered as


consisting of a mean wind speed and a fluctuating component.
Knowledge of both the mean wind speed and the fluctuating
component assists in evaluating wind loads on transmission
line conductors.

The mean wind speed, wind profile,

turbulence intensity, and gust spectra are presented as part


of this chapter.
Mean Wind Speed
Mean wind speed is defined as an average wind speed for
a specified time interval.

The numerical value of the mean

wind speed can have large variations depending on the


interval used for averaging the wind speed.

A shorter

averaging time leads to a higher mean wind speed value,


while a longer averaging time leads to a smaller mean wind
speed value.

This is primarily due to short gusts of high

wind speed that last for short periods of time.

25
The length of the record for which the mean value and
the RMS value of wind speed are determined is somewhat
arbitrary.

The record should be long enough to reflect the

effects of low frecjuency components of mechanical turbulence


generated by the terrain roughness, but short enough so that
a reasonably stationary time history, free of significant
trends is obtained.

Analysis of the power spectral

densities of wind speed provides the background for an


appropriate selection of the averaging time interval for
mean wind speed.

The gust spectrum reveals that wind is

made up of two distinct types of air flow: (a)


macrometeorological or climate fluctuations, and (b)
micrometeorological fluctuations or gusts.

These

fluctuations are separated by a stationary stable interval


(spectral gap) between 10 minutes and 1 hour, as indicated
in Figure 6.

Based on this spectral gap, mean values

averaged over 10 minutes to 1 hour are optimum for stability


(Davenport, 1972).

In this study, the wind speeds are

averaged over record length of 12 minutes.


Mean Wind Profile
An important characteristic of wind is the variation of
wind speed with height.

The surface friction effects of the

ground retard the movement of air close to the ground

26

I*

;!

xtrum

Energy

2S.

1.
1

1
ii

Afui V M Dcr llovcn


- - Spck:uUli JIICI .\ ti. Oacnp<Ml

1 t
1 *

ii

!;

u.
Ivctevh
Ikiun

n
1

to-
lOUUO
1
t
1

11 yt \uHK.I i;yck

ArniM.!

*-fi*r
M;ruiii*irar<tlu|Hal ruifc
r ' . . . t l l . ( m.^ ' llMlVtlMIHI

Figure 6:

surface.

Scnutliufn.!
MkiunwicafolofivJ ruifc
IfKMtl

Idealization of Gust Spectrum Plot Over an


Extended Range (Davenport, 1972)

This retardation causes a reduction in wind speed

near the ground.

At some height above the ground, the

movement of air is independent of the ground obstructions.


This unobstructed wind speed is termed the 'gradient wind
speed,' and the corresponding height at which the air
movement is not retarded is termed the 'gradient height.'
Mean wind profiles near ground level are currently
represented by either power-law or logarithmic-law profiles
The logarithmic law profile is based on the assumption of
physical phenomena and is valid particularly up to 30 m

27
above ground (Simiu, 1984).

The power-law profile is

empirical and is assumed to be valid up to gradient height


(approximately 500 m ) .

The power-law profile is primarily

used in structural analysis and design, because of its


simplicity.

It is essential to determine the wind profile

at a particular site, so that the mean wind speed at


effective height of structure can be determined.

Power-law

is used in this study, and is briefly described below.


The power-law profile was developed by Davenport
(1960).

He modifiecj the exponential profile developed by

Brunt (1952) to obtain a mean wind speed profile.

In

horizontally homogeneous terrain, it is assumed that the


power-law is valid with a constant exponent (a) up to the
gradient height, Z .

Both gradient height and power-law

exponent are functions of the terrain roughness.

The mean

wind profile is expressed as

Iil = (-?-)"

(2.20)

where V(z) = mean wind speed at height, z,


V

= mean wind speed at gradient height, Z , and


a

= power-law exponent.

The power-law is used in both the American National


Standard ANSI A 58.1 (1982) and in the National Building

28
Code of Canada (NRCC, 1980).

Typical values of the gradient

height, Z , and the power-law exponent, a, for different


terrains, as specified by ANSI (1982), are summarized in
Table 1.

TABLE 1
Typical Values for Gradient Height and Power-Law Exponent
(ANSI, 1982)

Terrain
Category

Gradient
He ight(ft)
Z
g
700

Coastal Areas

Power Law
Exponent
a
0.10

900

0.14

Forest/Suburban

1200

0.22

City Centers

1500

0.33

Open Farmland

Turbulence Characteristics
The fluctuating part of wind is termed as the
turbulence.

The turbulence present in the wind flow is due

to the ground roughness characteristics of the terrain over


which it is passing or due to thermally-induced convection
or both.

The turbulence due to ground roughness is known as

mechanical turbulence and that due to heat convection is

29
known as convective turbulence.

Depending on the relative

importance of convective to mechanical turbulence, the


stability conditions of the atmosphere are classified as
stable, neutral, and unstable (Simiu, 1985).

The extreme

winds in which structural engineers are interested are


categorized as neutral stability conditions.

In a neutrally

stable condition, the temperature related buoyancy forces


and resulting vertical air motions are minimum.

For

engineering purposes, it is generally assumed that neutral


atmospheric stability conditions can be assumed for wind
speeds higher than 20 mph.

Details of atmospheric neutral

conditions are discussed in detail by Kancharla (1987).


Analysis of turbulence includes determination of the
turbulence intensity and the gust spectrum.

Of these two,

the turbulence intensity expression is simpler.

It

indicates relative amplitudes of the fluctuations compared


to the mean wind speed.

A complete representation of

fluctuating components of wind is the gust spectrum, which


gives the distribution of the mean scjuare over the frecjuency
domain.

The gust spectrum is useful in assessing dynamic

response of structures.

Both representations of turbulence

characteristics are discussed below.

30
Turbulence Intensity
Turbulence intensity is a measure of the gustyness of
the wind.

It is expressed as

T^ = - ^
V
where T

(2.21)

= turbulence intensity,

a^ = RMS of wind speed fluctuations, and


V = mean wind speed.
In statistical terminology this number is often called
the coefficient of variation.

Turbulence intensities are

higher for records which have lower mean wind speeds than
for records that have high wind speeds for the same terrain.
The turbulence intensity is strongly related to the terrain
roughness; a greater turbulence is caused by a rougher
terrain (refer to Table 2 ) .

A decrease in turbulence

intensity with height is also expected; at greater heights


both mean and RMS values of wind speed increase but the RMS
value increases less because the shearing action of the
ground surface is less (Jan, 1982).
Parametric study of the Davenport analytical model
(Twu, 1983; GAI, 1981) shows that the turbulence intensity
is the most influential parameter in predicting the response
of a transmission line structure to extreme wind.

31
TABLE 2
Typical Values for Turbulence Intensity and Surface Drag
Coefficient (Kempner, 1982)

Terrain
Category

Surface Drag
Coefficient

Turbulence
Intensity

Coastal Areas

0. 001

T
L
u
0. 07

Open Farmland

0. 005

0. 12

Forest/Suburban

0. 015

0. 22

City Centers

0. 050

0. 39

Therefore, emphasis is given to the computation of


turbulence intensity in the wind data analysis.

Gust Spectrum
A randomly fluctuating phenomenon such as wind speed
can be conceived of as the superposition of a large number
of harmonic fluctuations with frecjuencies ranging from zero
to infinity.

The spectral representation of turbulence is

related to this concept, and it provides information on the


contributions of fluctuating components (energy) with
various frecjuencies.

The energy of any random process, like

wind, is usually expressed in terms of a cjuantity called the

32
'Power Spectral Density (PSD).' The PSD at any particular
frecjuency, f, may be considered as the average fluctuating
wind power passing a fixed point when the wind as a random
process is filtered by a narrow band filter centered at f.
In the dynamic analysis of structures subjected to gust
loading, significant dynamic amplification of the response
may occur at a resonant frequency, i.e., when the natural
frecjuencies of vibration of the structure and of the wind
match (Simiu, 1985).

Flexible structures like conductors

can have dynamic amplification of the response because the


fluctuating component of wind has a fair amount of power at
frecjuencies of structural vibration.

On the other hand, if

the natural frecjuencies of vibration of the tower structures


are higher than 1 Hz, the dynamic amplification of the
response of the tower will be small because power in the
gust spectrum at those frecjuencies (see Figure 13, page 72)
is very small.
There are more than a dozen specific wind speed
spectrum ecjuations developed for meteorological and
engineering purposes.

Some of these spectral ecjuations are

discussed by Kim (1977).

For neutral atmospheric

conditions, wind turbulence is generated by the surface


shear stress.

It follows that the magnitude of the PSD

should be proportional to the scjuare of the frictional

33
velocity.

The analytical model for the gust spectrum used

in this study was developed by Kaimal (1978).


In general, the spectral energy of gusts is a function
of the wave-length, , of the wind fluctuations and the
height, h, above the ground.

The analytical form suggested

by Kaimal for the horizontal gust spectrum for height to


wave length ratios greater than one-half (

v>

>0.5) is given

V
as:
f S(f)
^
= A f h '^
V
^*

(2.22)

where S (f) = spectral density of gusts at frecjuency f.

u^= friction velocity, (^KV^Q)


K = surface drag coefficient,
(typical values are shown in Table 2)
V..Q= mean wind speed at 10 m height,
V = mean wind speed at height h,
h = height above ground, and
A, n = constants.
Constants A and n represent the amplitude and exponent
values of Kaimal's gust spectrum.

For neutral atmospheric

34
conditions, A=0.3 and n=2/3 are suggested by Kaimal.
Ecjuation 2.22 is useful for describing the gust
spectrum in the high frecjuency range (low wave lengths) and
for heights limited to the first few tens of meters.

Kaimal

also presents other forms of the gust spectrum which are


valid for lower frecjuencies and for unstable atmospheric
conditions.
Davenport Analytical Model
On the basis of Davenport's analytical model (1980),
the peak response, ft, of a conductor due to fluctuating
winds is represented by ecjuation 2.1.

Davenport's model

provides an analytical expression for the RMS, cjp, and


suggests a peak factor value, g, in the range 3.5-4.0.
Validation and refinement of the expression for the RMS are
part of this study.
The mean scjuare fluctuating response of a conductor is
the area under the response spectrum.

The area under the

response spectrum can be considered as the summation of the


background response, B , and the resonant response, R . The
c

background response is caused by gust with various


durations, whereas resonant response is caused by gust
frecjuencies at the natural frecjuencies of the conductor.
The total mean scjuare fluctuating response of the conductor
is given as:

35
(2.23)

^R = ^c ^ ^c
where B = the mean scjuare background response, and
R = the mean scjuare resonant response.

The expressions for background and resonant responses


consider wind properties such as mean wind speed, turbulence
intensity and gust spectrum and structural properties such
as frecjuency of vibration and damping.

Ecjuations for

background and resonant responses of the conductor response


have been developed by Davenport (1980) and are given below.
The ecjuations are:

B_ = e^
c P

E^

2
Rc = 6^
c P

2
E^

where

(2.24)
1 + 0.81(-ii-l
^s

0.323A h. 1 , ^o ^,-(n-H)

(2.25)

f7

P = mean wind pressure; pV C^


e

= the influence coefficient which translates


the force to response; for conductor transverse
response is the product of L and d,

= exposure factor at the effective height of

36
the conductor; which is twice the turbulence
intensity,
Lg = transverse scale of turbulence,
C^ = conductor force coefficient,
p

= mass density of air,

= conductor aerodynamic damping as a ratio


to the critical damping,

V = mean wind speed at the effective height


of the conductor,
f^ = fundamental frequency of the conductor
(horizontal sway),
A, n
c

= Kaimal's gust spectrum constants,


= narrow band correlation coefficient of
turbulence, with a typical value of 8,

= effective conductor span,

= conductor diameter, and

= effective conductor height.

The background and resonant terms for conductor


response are calculated using ecjuations 2.24 and 2.25.

The

mean scjuare value of fluctuating response is obtained using


ecjuation 2.23.

Some of the assumptions made by Davenport in

deriving the simplified expressions for conductor response


are discussed by Mehta (Criswell, 1987).

37
Conductor Damping Ratio
The energy gained by the conductors from the
fluctuating wind is dissipated by the conductor damping.

In

general, three sources of damping can be identified for


conductors, which are material damping, structural damping
and aerodynamic damping.

Material damping is due to

internal energy dissipation by the material of the


conductor.

Structural damping is due to friction,

impacting, and rubbing of any two surfaces of the


conductors.

Both material and structural dampings are very

small for conductors as compared to aerodynamic damping.


Aerodynamic damping is due to the retarding force which is
developed from the relative motion between the conductor and
the air.

In the analytical model the value of damping

ratio, C/ which is defined as the ratio of damping


coefficient to the critical damping coefficient, is
determined using a theoretical expression.

This expression

is based on the inertial force principle as conductor


movement displaces an ecjual volume of air (Davenport, 1980).

C = 0.000048 ( ^
o
where

) C^

V = mean wind speed,


f

= fundamental frequency of the conductor.

(2.26)

38
C^ = conductor force coefficient, and
d = diameter of the conductor.
Calculation of the aerodynamic damping ratio using field
response data is evaluated in Chapter V.

Conductor Fundamental Frecjuency


Conductor frecjuencies of vibration are analytically
estimated by modelling conductor as a conductor oscillating
from side to side and using the principles of dynamic
ecjuilibrium (Symon, 1961).

Fundamental transverse frequency

of the conductor, f , in Hz, for a parabolic profile can be


obtained using the following ecjuation:

JG
where

f =
^
o
32 S
= acceleration due to gravity, and

= conductor sag.

Ecjuation 2.27 is used in Chapter IV to calculate the


conductor fundamental frecjuency.

(2.27)
^
'

CHAPTER III
FIELD DATA

Full-scale data used in this study were collected by


the Bonneville Power Administration (BPA).

Since 1976, BPA

has conducted several projects to study wind load response


of transmission line systems by collecting and analyzing
wind and response related data on test lines in the field.
Transmission tower and conductor wind response data were
collected on an energized 500 kV single circuit transmission
line.

An instrumentation system was used to measure wind

speed, wind direction, insulator swing, insulator load, and


tower member stresses.
During the period of December 1981 through May 1982,
BPA collected twenty-three separate recordings of wind and
the transmission line response with twelve-minute duration.
Dates, times, and mean wind speeds and direction of these
twenty-three records are shown in Table 3.
are used in the study presented here.

These recordings

Each record is

numbered by Nxx, where xx is the secjuence number of


occurrence of high winds.

Data utilized in this study are

limited to wind and conductor response data.

The site

characteristics, the instruments and the data accjuisition

39

40
systems used for the collection of wind and conductor
response data are described in this chapter.

TABLE 3
Field Data Records During 1981-1982

Record
Number

Mean Wind
Direction*
(azimuth)

Mean Wind
Speed*
(mps)

Date

Time

NOl
N02
N03

12/02/81
12/05/81
12/15/81

01.31.57
06.42.45
16.11.48

N04

12/16/81

08.30.32

N05
N06
N07

12/16/81
01/14/82
01/16/82

16.10.28
10.57.52
19.04.51

93
260
215

15.8
15.5
22.3

N08
N09
NIC

01/31/82
02/03/82
02/14/82

01.36.32
14.11.35
13.05.30

263
53
228

21.4
9.6
18.5

Nil
N12
N13

02/15/82
02/15/82
02/16/82

23.26.40
10.29.05
00.38.56

227
220
237

21.0
14.0
20.3

N14

03/08/82

16.03.42

N15
N16
N17

03/11/82
03/12/82
04/12/82

14.52.13
15.13.50
01.10.26

259
220
235

15.7
10.0
18.8

N18
N19
N20

04/13/82
04/17/82
04/20/82

15.29.13
17.57.11
22.03.35

246
276
90

21.8
15.8
18.5

N21
N22
N23

04/20/82
04/28/82
05/07/82

11.50.38
12.28.43
14.19.51

274
280
278

18.4
19.3
15.2

* measured at 34.7 m on Tower 16/4

227
209
240

18.8
16.4
14.8
zero wind

zero wind

41
Description of Test Site
The test site is located at Moro, Oregon, 56 kilometers
southeast of Dalles, Oregon (east of the Cascade Mountains).
The general topography in the vicinity of the test lines is
shown in Figure 7.

As observed from the contours in the

figure, the deep Deschutes river canyon is just west of the


site.

The test line is essentially located on the rim of

the canyon.

To the east of the test lines is a flat terrain

which is uncultivated land containing grass with some


shrubs.

The test lines are approximately 1500 m to the east

and 500 m above the elevation of the Deschutes river.


As indicated in Figure 7, the test site includes three
lines running almost north-south and approximately 8 degrees
west of true north (Kempner, 1981).

The first 2 lines east

of the Deschutes river are 500 kV energized lines, called


John-Day Grizzly (JD-G) 1 and 2.

A non-energized mechanical

test line (also referred to as the Moro test line) parallels


the other lines.

JD-G lines 1 and 2 are 45.7 m apart,

whereas the distance between JD-G line 2 and the mechanical


test line is 38.1 m.

Towers at the site (on JD-G line 2)

are numbered from 1 to 5, Tower 1 being the tower at the


northern end of the site.

The instrumented tower is Tower

4, part of the John Day-Grizzly (line 2) system.

It is

referred to as Tower JD-G 16/4 or simply Tower 16/4 because


it is located on mile 16 of JD-G line 2.

42

Figure 7:

Topography of Site and Orientation of Power Lines

43
Tower 16/4 is a delta configuration lattice tower
structure as illustrated in Figure 8.

The 33.4 m tall tower

supports three twin Chukar conductors (west, east, and


central) and two overhead groundwires (west and east).

Each

Chukar conductor has an outer diameter of 40.7 mm and weighs


3.1 kg m~ .

The Chukar conductors have 84 aluminum and 19

steel strands having ciiameter of 3.7 mm and 2.2 mm,


respectively.

In order to reduce subconductor oscillation,

one conductor of each twin conductor is 229 mm lower than


the other.
The conductor span to the north of Tower 16/4 is 252 m
to a similar delta configuration suspension tower.

The span

to the south is 450 m (refer to Figure 9) to a horizontal


configuration suspension tower.

The change in tower

configuration causes both east and west twin conductors to


hang in a non-vertical (outward) position at Tower 16/4.
The west twin conductor is ecjuipped with dampers which make
it heavier than the east twin conductor.
Instrumentation
Three different types of instruments were used to
measure wind speed and wind direction.

A climatronics mark

III model anemometer was located on top of Tower 3.

Two

three-blade propeller-vane anemometers were mounted on Tower

44
Anemomeler (34.7 m)

Load cell and


Swing angle
indicators

41 mm diameter

West conductor

Figure 8:

Schematic of Tower 16/4

45

N -^

effective half span lengths of conductors

Figure 9:

Elevation Along the Test Line (Vertical Scale


Exaggerated)

46
16/4, as shown in Figure 8.

One was located on top of the

tower at a height of 34.7 m.

The second unit was located on

the northwest tower leg at a height of 10 m and it projected


out 2.3 m north of the tower.

Two four-blade propeller-vane

anemometers were installed on top of Towers 4 and 5.

The

three-blade propeller-vane anemometer has a threshold speed


of 1.7 mps and a distance constant of approximately 4.6 m.
The threshold value is the stall speed of the unit.

The

distance constant is the wind passage recjuired for 63%


recovery from a step change in wind speed.

The wind

instruments ecjuipped with internal heaters to allow for


winter operation.

Wind direction is indicated by azimuth

readings in degrees referenced to true north.

The wind

direction is such that a zero degree reading corresponds to


true north and a clockwise rotation represents an increase
in the wind direction reading.
The load cells and swing angle indicators measured the
magnitude and direction of the conductor and overhead ground
wire loads that transfer to the tower structure.

The

instruments were installed in the linkage between the


insulator string and the tower.

All conductors and ground

wires were instrumented with one axial load cell and two
swing angle indicators.

The swing angle indicators measured

longitudinal and transverse swings of the insulators.

47
Baldwin-Lima-Hamilton (BLH) strain gage load cells were
used to measure axial loads.

BLH Type T3P1 load cells,

rated at 5000 pounds, were used for the overhead ground


wires and BLH Type T2P1 load cells, rated at 20,000 pounds,
were used for the conductors.

Humphrey pendulum swing angle

indicators, model CP17-0601-1, were used to measure the


longitudinal (along the line) and transverse (perpendicular
to tihe line) swings of the insulator string.

These units

measure up to 45 degrees of swing from the zero (vertical)


position with a resolution of 0.2 degrees (Kempner, 1977).
Valid data from these units are restricted to 2 Hz or less
because of the unit natural frecjuency of 3.2 Hz, as
suggested by Kempner (1980).

The load cells and swing angle

indicators were calibrated in the laboratory and checked in


the field after installation.
Data Acquisition
Data was collected by the Moro UHV mechanical test
program data accjuisition system (Kempner, 1979).

The data

accjuisition system consisted of a PDP-11/10 mini-computer


with 12 K memory, an ADAC Model 600-11 Data Accjuisition
System, a Digi Data Controller/Formater, a 7-track magnetic
tape unit, and a teletype.

The data accjuisition system was

housed in an instrument trailer located 30 m southwest of

48
the Tower 16/4.

The system was set up to record from 256

channels of instrumentation.

RecordincT Procedure
Several selected channels constituted a recording mode,
which were selected to capture a static or dynamic
phenomenon of interest.

The data used in the present study

were recorded in Mode 22.

Mode 22 was designed for

recording wind and the response of Tower 16/4.


of 38 channels of instrumentation.

It consisted

The instrumentation on

Tower 16/4 included 12 strain gages, 5 load cells, 10 swing


angle indicators, and 2 wind instruments.

The remaining

channels provide readings from the anemometers mounted on or


adjacent to Towers 2, 3, 4, and 5 on the Moro mechanical
test line.

A complete list of the channels with a

description of each is given in Table 4 (Norville, 1985).


The PDP-11/10 computer was used to monitor wind
conditions and to initiate recordings when prescribed
conditions were met.

Prescribed information was entered

into and stored in the computer prior to placing the data


accjuisition system on-line.

This information included

channel identification, date and time of recording, number


of samples, mode number, calibration and offset factors for
each channel (Kempner, 1979).

This information was written

49

TABLE 4
File Description of Mode 22

File

Channel

LCOl
LC02
LC03
LC04
LC05
SAOl
SA02
SA03
SA04
SA05
SAO 6
SA07
SA08
SA09
SAIO

78*
79*
80
81
82
83*
84*
85*
86*
87
88
89
90
91
92

SGOl
SG02
SG03
SG04
SG05
SG06
SG07
SG08
SG09
SGIO
SGll
SG12

66*
67*
68*
69*
70*
71*
72*
73*
74*
. 75*
76*
77*

WDOl
WD02
WD03
WD04
WD05
WSOl
WS02
WS03
WS04
WS05
WS06

159
163
168
179
181
156*
158
161
167
178
180

Instrument, Location, Height


Load Cell 1
Load Cell 2
Load Cell 3
Load Cell 4
Load Cell 5
Swing Angle 1
Swing Angle 2
Swing Angle 3
Swing Angle 4
Swing Angle 5
Swing Angle 6
Swing Angle 7
Swing Angle 8
Swing Angle 9
Swing Angle 10
Strain
Strain
Strain
Strain
Strain
Strain
Strain
Strain
Strain
Strain
Strain
Strain
Wind
Wind
Wind
Wind
Wind
Wind
Wind
Wind
Wind
Wind
Wind

Tower
Tower
Tower
Tower
Tower
Tower
Tower
Tower
Tower
Tower
Tower
Tower
Tower
Tower
Tower

16/4
16/4
16/4
16/4
16/4
16/4
16/4
16/4
16/4
16/4
16/4
16/4
16/4
16/4
16/4

Gage 1 Tower 16/4


Gage 2 Tower 16/4
Gage 3 Tower 16/4
Gage 4 Tower 16/4
Gage 5 Tower 16/4
Gage 6 Tower 16/4
Gage 7 Tower 16/4
Gage 8 Tower 16/4
Gage 9 Tower 16/4
Gage 10 Tower 16/4
Gage 11 Tower 16/4
Gage 12 Tower 16/4

East
West
West
East
Cent.
East
East
West
West
West
West
East
East
Cent.
Cent.
NW
NE
NW
SE
SE
SE
NW
SW
NW
NW
NW
NW

1
2
2
1
2
3
2
2
2
1
2
3

OHGW
OHGW
Cond.
Cond.
Cond.
OHGW
OHGW
OHGW
OHGW
Cond.
Cond.
Cond.
Cond.
Cond.
Cond.
Dia.
Main
Main
Dia.
Main
Dia.
Main
Main
Main
Dia.
Main
Dia.

47.4
Tower 3
Direction Anem.
Direction Anem.
Tower 4
41.5
Directi on Anem.
Tower 5
39.3
Direction Anem.
Tower 16/4 34.7
Direction Anem.
Tower 16/4 10.0
Speed Hot Wire Anem. Near Tower 2
Speed E'rop. Anem. Tower 3
47.4
Speed Erop. Anem. Tower 4
41.5
Speed Erop. Anem. Tower 5
39.3
Speed Prop. Anem. Tower 16/4 34.7
Speed Erop. Anem. Tower 16/4 10.0

recordings of channels not used in this study

m
m
m
m
m
m
m
m
m
m

50
as a heading on the magnetic tape preceding a strong wind
recording.
Triggering of this recording mode was automatic when
the wind speed was equal to or greater than 18 mps for one
minute and the temperature was ecjual to or greater than 4
degrees Celsius.

Once triggered, the recording mode sampled

the data for 10 or 12 continuous minutes, depending upon the


sampling rate.

After a recording period, the mode had one

hour waiting period before another trigger was allowed.


Two data sampling rates were used in recording the data
for mode 22.

The sampling and recording rates were limited

by the size of the computer memory and the speed of the


magnetic tape unit.

Sampling rates of 10 samples per second

(sps) and 20 sps were used in collecting data.

Channels 66

through 79 (LCOl, LC02 and SG01-SG12) were monitored at 20


sps; all other channels were sampled at 10 sps.

All the

data utilized in the study presented here had sampling rates


of 10 sps and were collected for 12-minute durations.
Description of Recordings
The recordings that were collected in the field are
summarized in Table 3.

The winds at the test site during

the period of data collection were predominantly from the


west, with only three records of east winds.

Two zero wind

51
records were collected for initializing conductor and tower
response data.
22.3 mps.

Mean wind speed values ranged from 9.6 to

Mean wind directions varied from almost normal

(transverse) to transmission line to 55 degrees from the


normal.

The terrain over which the wind traversed in each

recording segment depended on the wind direction.

These

variations in wind speed, wind direction and terrain caused


inherent variability in the collected data.

Field

experiments depend on the vagaries of nature; they cannot be


duplicated or repeated.

The inherent variability in field

data and the inability to repeat the experiment suggest that


results from the analysis should be based on an ensemble of
data and that some scattering of results is to be expected.

CHAPTER IV
FIELD DATA ANALYSIS

Analysis of field data recjuires that the validity and


accuracy of the wind and conductor response data be checked.
It is expected that results of field data will have a
certain amount of scattering.

This scattering can be due to

an inherent variation in field data as well as a variation


in the data measuring system.

To the extent possible, wind

data and conductor response data are checked for consistency


and accuracy.
The analytical procedure to predict the response of
conductors recjuires knowledge of wind characteristics such
as the mean wind speed, power-law exponent of wind profile,
turbulence intensity, and Kaimal's gust spectrum constants.
These characteristics are determined from the field data and
are subsecjuently used in the analytical procedure.
Field recorded conductor response can be compared with
predicted values in terms of mean response and fluctuating
response.

Effective force coefficients that relate to mean

response are determined for each record of the field data.


Fluctuating response, as indicated in Chapter II, is a
combination of background response and resonant response.

52

53
These responses can be assessed from response spectra of the
field measured values.

The conductor response spectrum of

transverse loads recorded by load cells is discussed in this


chapter.

The discussion relates to background response and

resonant response.

Comparisons of recorded fluctuating

responses with predicted values from the analytical


procedure are presented in the next chapter.

In addition,

an assessments of the peak factor, admittance function and


damping from the recorded data are presented in next
chapter.

Validity of Wind Data


Wind speeds and directions were recorded at five
locations, namely on top of Towers 3, 4, 5, and 16/4, and at
10.0 m on Tower 16/4.

Tower 4 is within 38 m of Tower 16/4,

while Towers 3 and 5 are several hundred meters from the


other two towers (see Figure 9 ) .
The mean azimuth and root mean scjuare (RMS) of wind
direction for each record for these five locations are shown
in Table 5.

Data were recorded at 10 sps for 12 minutes in

each record.

One observation in Table 5 is the consistency

in mean azimuth value for the three instruments located at


the tops of Towers 4, 5, and 16/4.

The differences in mean

values for these three instruments are less than 12 degrees.

54
and in most cases less than 6 degrees.

The two instruments,

one at 34.7 m height on Tower 16/4 and one at 41.5 m on top


of Tower 4, give mean wind directions within 4 degrees for
winds from several different azimuths.

This consistency in

mean azimuth values provides credibility to the wind


direction instruments and the data accjuisition system.
The mean azimuth values for the instrument at 10.0 m
height on Tower 16/4 are erratic when compared with the
values from the other instruments for records NOl through
N13 (see Table 5 ) . The difference in mean azimuth value is
as high as 44 degrees.

Such a large difference in wind

direction between records for instruments on the same tower


located at 10.0 m and at 34.7 m cannot be explained from
physical phenomena.

Ekman's spiral suggests that a

deviation in wind direction occurs close to ground, but the


deviation in a 25 m difference in height should be less than
6 degrees (Simiu, 1985).

This large difference in wind

direction for the instrument at 10.0 m on Tower 16/4 casts


doubt in its wind data for records NOl through N13; these
records are not used in further analysis.

The remaining

records (N15 through N23) for this instrument appear to give


reasonable results.

The wind instrument on top of Tower 3

also gives mean azimuth values higher by 10 to 20 degrees


than the other instruments on a consistent basis.
from Tower 3 are not needed in the analysis.

Wind data

55
TABLE 5
Mean Azimuth and RMS in Degrees of Wind

Instrument WDOl
Location Twr 3
Height
47.4 m
Mean RMS

WD04
WD03
Twr 5
Twr 16/4
39.3 m
34.7 m
Mean RMS Mean RMS

224
205
239

229
207
246

8
12
6

227
209
240

9
12
5

212
199
224

9
12
5

NOl*
N02
N03

197
216
246

N04

Zero Wind record

N05
N06
N07

101

NOB
N09
NIC

276
64
243

Nil
N12
N13

241
234
250

N14

Refe rence Zero Wind Speed Record

N15
N16
N17

274
232
246

5
9
5

260
218
231

7
8
8

267
228
235

8
8
8

N18
N19
N20

259
292
100

5
12
1

244
277
87

5
12
1

256
276
90

N21
N22
N23

285
295
296

13
9
9

273
279
279

10
11
12

283
283
282

22
7
4

WD05
Twr 16/4
10. 0 m
MearL RMS

WD02
Twr 4
41.5 m
Mean RMS
9
12
5

94
259
211

2
7
9

94
266
219

2
8
10

93
260
215

2
7
9

81
300
258

4
7
9

6
9
7

263
54
225

8
7
8

268
56
228

6
8
7

263
53
228

8
8
8

300
97
268

9
9
9

7
9
8

225
218
235

7
9
9

230
222
236

9
12
8

227
220
237

7
10
9

266
258
278

8
10
9

259
220
235

8
9
8

267
230
244

8
9
8

10
15
1

246
276
90

5
12
1

252
283
101

7
11
2

10
13
14

274
280
278

11
10
12

280
286
283

12
10
12

**

**

k-k

*-k

* Record Number
** error in data

56
The mean wind speed value for each record for the five
wind speed instruments is tabulated in Table 6.

One

observation in Table 6 is the consistency in mean wind speed


value for the four instruments located on the tops of Towers
3, 4, 5, and 16/4.
within 15%.

The values among these instruments are

The variation in mean speed values may be due

to a combination of two reasons.

First, the instruments are

at slightly different heights above the ground; wind speed


is expected to be higher as height increases.

The second

reason may be that the distances between the towers are


several hundred meters; so the terrain over which the wind
travels can be different.

This combination of variation in

terrain and height of instrument can account for variations


in mean wind speed values.

The instrument at 41.5 m height

on Tower 4 recorded 4% higher mean wind speed than that at


34.7 m height on Tower 16/4 on an average (see Table 6 ) .
Since Towers 4 and 16/4 are only 38 m apart with no terrain
modulations, it is reasonable to assume that a height
difference accounts for the small difference in mean wind
speed values.

This consistency in mean speed values

provides credibility to the wind speed instruments and the


measurements.
The difference between the mean wind speeds at 10.0 m
and at 34.7 m on Tower 16/4 is very small for records NOl

57
TABLE 6
Mean Wind Speed in mps

Instrument WS02
Location
Twr 3
Height
47.4 m
NOl*
N02
N03
N04

18.7
18.9
15.2

WS03
Twr 4
41.5 m
19.4
16.7
16.0

WS04
WS05
Twr 5
Twr 16/4
39.3 m
34.7 m

WS06
Twr 16/4
10.0 m

19.8
16.4
14.7

18.8
16.4
14.8

18.5?
17.5?
14.8?

15.3
14.9
19.1

15.8
15.5
22.3

15.6?
15.3?
23.3?

Zero Wind record

N05
NO 6
N07

15.0
**

15.3
15.4
21.8

NOB
N09
NIC

21.8
9.6
19.0

21.5
9.8
18.8

22.6
9.7
17.3

21.4
9.6
18.5

20.4?
8.8?
18.3?

Nil
N12
N13

21.2
13.5
20.2

21.4
13.9
20.7

19.5
13.0
21.0

21.0
14.0
20.3

21.0?
14.3?
20.0?

N14

**

Reference Zero Wind Speed Record

N15
N16
N17

16.4
9.0
20.4

15.4
9.9
19.1

15.5
8.6
17.3

15.7
10.0
18.8

12.1
8.0
15.5

N18
N19
N20

22.1
16.1
18.3

22 1
15 5
18 4

17.4
16.2
17.7

21.8
15.8
18.5

18.2
13.0
14.4

N21
N22
N23

16, 0
19, 1
16. 6

18.4
19.4
15.2

18 0
18 0
16 6

18.4
19.3
15.2

15.5
16.5
13.2

* Record Number
** error in data
? these values are cjuestionable along with wind direction;
they are not used in analysis

58
through N13.

In a few records, the instrument at 10.0 m

height shows higher mean wind speed values than that at 34.7
m height.

Wind speed at 34.7 m on Tower 16/4 is more

consistent with the other instruments on top of Towers 3, 4,


and 5.

This casts some doubt on the wind speeds of records

NOl through N13 measured at 10.0 m height on Tower 16/4;


these records are not used for further analysis.
To further validate the field data, time histories are
plotted for all wind speed records.

A typical time history

plot for wind speed recorded at 34.7 m on Tower 16/4 for


record NOl is shown in Figure 10.

Each record has 7200

points, which corresponds to a recording rate of 10 sps for


the duration of 12 minutes.

The time history plot in Figure

10 uses averages of 10 consecutive points; thus the points


are at one second intervals.

Visual observation of time

history plots of wind speed records for any discontinuities,


trends, and noise shows the records to be good.
An additional check of stationarity was performed to
assess consistency of the wind data.

Stationarity checks

are done to verify that the statistical properties are time


invariant.

Stationary tests were accomplished by Levitan

(1987) as part of an earlier BPA contract.

Most of the wind

records were found to be stationary with 95% confidence


limits as checked by the run and trend tests suggested by

59

Time (minutes)

Figure 10:

Time History Plot of Wind Speed for Record NOl


at 34.7 m on Tower 16/4

Bendat and Piersol

(1966).

discussed in this study.

These stationary test are not


Finally, as a part of the

validation of the data, a power spectrum plot for each


record was generated.

These plots would reveal, in the

frecjuency domain, any electrical or other sources of noise


in the data.

Details of these power spectrum plots are

discussed in subsecjuent sections.

In general, most of the

wind speed and wind direction records appear to be valid.

60
The mean wind speed and mean wind direction for each
record at 34.7 m height on Tower 16/4 are shown in Figure
11.

There are eighteen west wind records, three east wind

records and two zero wind records.

Wind directions vary

from almost normal to the transmission line to as high as 55


degrees of yaw angle (see Figure 11). The mean wind speed
values are 10 mps or higher for all records, except for the
zero wind records.
Two zero wind records (N04 and N14) were collected in
near calm or zero wind conditions.

Based on a review of the

data of these two records, N14 is selected as a reference


record to assess conductor response.
Wind speed on top of Tower 16/4 (at 34.7 m height) is
used as the reference wind speed record.

The reference wind

speed record is important to determine the wind speed at the


effective height of the conductor.

The wind speeds at the

effective heights are determined using the reference wind


speed and vertical wind profile.
Power-Law Exponent for Wind Profile
The wind speed profile is assessed using mean wind
speed records of instruments located at 10.0 m and 34.7 m
heights on Tower 16/4 and at 41.5 m height on Tower 4.
Since Towers 16/4 and 4 are only 38 m apart with no terrain

61

Record Number N_
Mean Wind Speed and Direction

Figure 11:

Mean Wind Speed and Direction Recorded at 34.7 m


on Tower 16/4 of 23 Records

62
modulations between the towers, it is reasonable to assume
that the wind environment is the same at both the towers.
The other wind speed instruments are located far away; it
would be inappropriate to use their data to assess the wind
speed profile.

As mentioned earlier, wind speed records NOl

through N13 at 10.0 m height on Tower 16/4 are cjuestionable.


Hence the wind speed profile is assessed using records N15
through N23 only.

The power-law expression given in

ecjuation 2.20 can be written as

V^
Z,
ln(-f.) = a ln(-^)

(4.1)

Vl
where V, = wind speed at height Z^
V2 = wind speed at height Z2

and

a = power-law exponent.
Ratios of wind speeds and ratios of heights for record
N15 are plotted in Figure 12. Three data points in the
figure represent data collected at heights of 10.0 m, 34.7
m, and 41.5 m.

A straight line using the least squares

method is fitted to the three data points.

The slope of the

straight line is the value of the power-law exponent.


Power-law exponent values obtained for records N15
through N23 are given in Table 7.

The flat land on the east

63

CVi

>

ln(Z2/Zi)
Figure 12:

Power-Law Plot for Record N15

and the valley on the west give similar power-law exponent


values for the limited data collected in this phase of the
project.

These values are consistent with results reported

for this site by Kempner (1982).


Calculated power-law exponent values range between 0.11
and 0.18; this wide range corresponds to open farmland and

64
TABLE 7
Power-Law Exponent and Kaimal's Gust Spectrum Constants

Record
Number
NOl
N02
N03
N04

Power 1 aw
exponent
_
-

n**

0.327
0.591
0.269

1.355
1.721
1.571

0.009
0.133
0.254

1.686
1.494
1.463

Zero Wind record

N05
NO 6
NO 7

N08
N09
NIO

Nil
N12
N13

N14

A*

-.

0.151
0.182
0.242
0.209
0.294
0.296

1.579
1.925
1.786
1.200
1.696
1.420

Reference Zero Wind Speed Record

N15
N16
N17

0.18
0.16
0.15

0.209
0.157
0.267

1.516
2.052
1.573

N18
N19
N20

0.14
0.13
0.18

0.151
0.304
0.005

1.383
1.681
1.565

N21
N22
N23

0.13
0.12
0.11

0.152
0.265
0.342

1.375
1.628
1.695

* constant representing amplitude


** constant representing exponent
Note: Kaimal's gust spectrum constants are calculated for
wind speed recorded at 34.7 m on Tower 16/4.

65
suburban terrains (see Table 1 in Chapter II). An average
power-law exponent value of 0.14 is used to determine the
wind speed at the effective height of the conductors (points
of effective pressures).

A small change in exponent value

does not drastically change the mean wind speeds at the


effective heights in the present study.

Turbulence Intensity
Turbulence intensity represents the level of turbulence
present in the wind.

It was defined in Chapter II as a

ratio of RMS to mean wind speed.

Values of turbulence

intensity calculated using data from all five instruments,


are shown in Table 8 for all records considered.
The turbulence intensity values among the instruments
for each record are fairly consistent, though there is some
scatter.

The instruments located on top of Towers 4 and

16/4, which are at similar heights and are 38 m apart, give


turbulence intensity values almost the same in most of the
records.

Seventeen records for these two instruments give

turbulence intensity values within 0.02; only four records


show a higher difference.

Turbulence intensity values for

the instrument located at 10.0 m on Tower 16/4 are


consistently higher than those at 34.7 m on Tower 16/4.
Winds near the ground are expected to contain more

66
TABLE 8
Turbulence Intensity

Instrument
Location
Height
NOl*
N02
N03
N04

WS06
WS05
Twr
16/4
Twr 16/4
10.0 m
34.7 m

WS02
Twr 3
47.4 m

WS03
Twr 4
41.5 m

WS04
Twr 5
39.3 m

0.14
0.13
0.08

0.16
0.22
0.07

0.15
0.24
0.11

0.18
0.17
0.11

0.03
0.12
0.22

0.03
0.11
0.16

***
***

Zero Wind record

N05
NO 6
N07

0.03
**

0.03
0.11
0.16

NOB
N09
NIO

0.11
0.10
0.11

0.13
0.09
0.15

0.11
0.09
0.15

0.14
0.13
0.17

***

Nil
N12
N13

0.12
0.18
0.12

0.14
0.16
0.13

0.13
0.22
0.14

0.18
0.17
0.15

***

N14

**

***

***
***

***
***

Reference Zero Wind Speed Record

N15
N16
N17

0.10
0.12
0.12

0.13
0.11
0.14

0.14
0.14
0.16

0.12
0.12
0.14

0.14
0.18
0.14

N18
N19
N20

0.09
0.17
0.02

0.08
0.17
0.02

0.16
0.17
0.02

0.11
0.17
0.02

0.13
0.21
0.04

N21
N22
N23

0.18
0.17
0.18

0.13
0.15
0.21

0.15
0.13
0.16

0.11
0.15
0.21

0.17
0.19
0.23

* Record Number
** error in data

*** cjuestionable d a t a ; see Tables 5 and 6

67
turbulence than the winds at higher elevation.

Therefore,

the fluctuating wind data can be considered valid.


The turbulence intensities of records for the
instrument at 34.7 m on Tower 16/4 (reference wind
instrument) vary between 0.02 and 0.21.

The turbulence

intensities for winds from west and southwest show cjuite a


bit of scatter; they range between 0.11 and 0.21 for the
reference wind instrument.

The west winds traverse over

hills and valleys (see Figure 7) before approaching the


transmission line.
are random.

The variations in turbulence intensity

Little correlation between turbulence intensity

and mean wind speed and direction (for west winds) was
found.

Turbulence intensities of winds from the east, where

the winds traverse flat terrain (see Figure 7), are low.
Two east wind records N05 and N20 show very low turbulence
intensities, 0.03 and 0.02, respectively, compared to the
third east wind record N09 (turbulence intensity of 0.13).
The turbulence intensity can be related to a terrain
classification.

In general, low values of turbulence

intensity are related to flat terrain and high values to


rough terrain.

The west wind records, show a wide variation

in turbulence intensity.

The turbulence intensity values

range from terrain classification of open farmland to


forest/suburban terrain (refer to Table 2).

This suggests

68
that a terrain of canyons and hills is unpredictable for
characterization of wind.

The power-law exponent values for

the profile, as well as turbulence intensity values, vary


from flat to suburban terrain, but there is no trend between
the two.
Turbulence intensity is one of the major parameters in
the analytical procedure to predict response.

The exposure

factor in ecjuations 2.24 and 2.25 is twice the turbulence


intensity.

A parametric study of the analytical procedure

conducted by GAI consultants (1981) and by Twu (1983) showed


that the turbulence intensity is the most significant wind
characteristic in predicting the response.

Turbulence

intensity values recorded at 34.7 m on Tower 16/4 (reference


wind instrument) are used in predicting response of the
conductors in the next chapter.
Kaimal's Gust Spectrum Constants
Gust spectra were plotted for all wind speed records
collected at 34.7 m on Tower 16/4 (reference wind
instrument).
Figure 13.

The gust spectrum for record NOl is shown in


Gust spectra were obtained utilizing the

International Mathematics and Statistical Libraries (IMSL,


1982) program FTFREQ.

The program uses an autocorrelation

function to calculate the spectral density function

69
estimates, S(f). Calculation of spectral density function
involve standard expansions (Jenkins, 1968) not involving
fast Fourier transformation.

The highest frecjuency obtained

is the Nycjuist frecjuency, 5 Hz, which is one half the


sampling rate of 10 sps.

The lowest frecjuency is restricted

by the number of lags selected for correlation.

Here the

lowest frecjuency is 0.0025 Hz because of the selection of


2000 lags (200 seconds) from autocorrelation plot.
program applied the Hamming window for smoothing.

The
The plot

2
in Figure 13 uses (f S(f)/a ) on the ordinate to give a
2
normalized linear scale, where a is the mean scjuare of the
time series, and f is frecjuency in Hz.

The abscissa in

Figure 13 is the log of the frecjuency, f.


The gust spectrum shown in Figure 13 is typical of the
gust spectra of other wind speed records.

Fluctuations in

the spectral density values are partly due to statistical


methods employed in the program.

A general trend of

reduction in ordinate (spectral energy) with increase in


frecjuency is noticeable in the figure.

The ordinate becomes

negligible for frecjuencies greater than 1 Hz.

Calculation

of the area under the gust spectrum (which is ecjual to the


mean scjuare value) indicates that only 1% of the spectral
energy is in the frecjuency range above 1 Hz.

The gust

spectrum in the figure shows a reasonable amount of spectral

70

O
P^

S (f) - spectral density value at f


f
- frequency
SIGSQ - mean square value

CO

in

00

C/1

CO

LJ
3*

U.

u
3

o
en
o

rt

>
^^
C3
^

o
CM

1)
Q.
CO

o
o
o
C.OOl

Frequency (Hertz)

Figure 13:

Gust Spectrum Plot for Record NOl Recorded at


34.7 m on Tower 16/4

energy in the frequency range of 0.1 to 0.4 Hz; natural


frecjuencies of the conductors are in this range.
The analytical procedure to predict response of the
conductor developed by Davenport (1980) utilizes an
analytical form of the gust spectrum.

The specific

71
analytical form used in the procedure is the one proposed by
Kaimal, as described in Chapter II (ecjuation 2.22).

Gust

spectra obtained from the field data are used to determine


Kaimal's gust spectrum constants A and n.

Constants A and n

represents the amplitude and slope of the analytical gust


spectrum.
The curvilinear regression procedure (Miller, 1977) is
used to obtain suitable values for constants A and n from
the field wind data.

Ecjuation 2.22 is rewritten to separate

dependent and independent variables:

A u^

f h '^.

(4.2)

Ecjuation 4.2 is nonlinear. It is made linear by taking


natural logarithms on both sides.

In S^(f) = ln(A nl)+n

ln(^)-(n+l) ln(f).

(4.3)

Ecjuation 4.3 can be written in the form


Y = C - (n+1) X
where
C = ln(Au2) ^ ^ ln(-^), and
X = ln(f).

(4.4)

72
In e q u a t i o n 4 . 4 , Y i s dependent on X.

To solve for A

and n, normal ecjuations of t h i s ecjuation a r e u t i l i z e d


(Miller,1977).

The normal e q u a t i o n s a r e ,
^Y = N C -

(n+1) j ; x

^X Y = C (X) -

(4.5)

(n+1) p 2

^4 gj

where N = total number of data points.


Values of n and C are found by solving ecjuations 4.5
and 4.6.

Substitution of n and C in ecjuation 4.3 gives the

value for A for each record.

Table 7 shows the calculated A

and n values considering gust spectrum values between the


frecjuencies of (
'-

) and 1 Hz. As indicated in Chapter


h

II,

ecjuation 2.22 i s v a l i d for f g r e a t e r than (

f^ ^ ) . The
h

gust spectrum values for frecjuencies greater than 1 Hz are


neglected because spectral energy (the ordinate of the gust
spectrum) is very small.

The value of cionstant A is least

affected by the selection of maximum frecjuency range,


specially beyond 1 Hz. Values of constant A range between
0.005 and 0.591 (refer to Table 7).

The gust spectrum (see

Figure 13) shows that the slope of the spectrum curve


decreases between 0.1 and 1 Hz, and remains almost constant
beyond 1 Hz. Hence, n values are sensitive to the maximum

73
frequency range selected.

Values of n range between 1.200

and 2.052.
The suggested range of values for A and n are 0.15-0.60
and 0.33-0.67, respectively (GAI, 1981).

Field measured

values of A are generally within the suggested range.


However, field measured n values are much higher than the
suggested range.

The parametric study of the Davenport

model for gust response factor of transmission line


structures conducted by (Twu, 1983) showed that the response
is not sensitive to the value of n.
Kaimal's analytical form of the gust spectrum (ecjuation
2.22) is valid for values of f greater than (-^LLL_Y.).

For

h
the field measured data obtained at 34.7 m, this low end of
frecjuency f would be 0.3 Hz, if the mean wind speed were 20
mps.

Since the conductor natural frecjuencies are in the

range of 0.1 to 0.4 Hz, near or below the low end of this
frecjuency range, Kaimal's gust spectrum constants obtained
from the field data are not appropriate for use in obtaining
response.

For this reason, values of A and n of 0.3 and

0.67, respectively, as suggested by Davenport, are used to


predict the response utilizing the analytical procedure in
the next chapter.

74
Validity of Conductor Response Data
The conductor response data comprise measurements from
the transverse and longitudinal swing angle indicators and
load cell transducers.

These instruments were placed at the

attachment of the insulator to the Tower 16/4 (see Figure


8).

The conductors are suspended at the bottoms of the

insulators.

The load cells measure axial load in the

insulator and the swing angle indicators measure swing of


the insulator from the vertical.

Appropriate combinations

of values of load cells and swing angle indicators provide


vertical, longitudinal and transverse components of loads
applied on the insulators by the conductors.

Interest in

this study is restricted to transverse load components since


the primary influence of wind is in the transverse
direction.

The transverse load component at any instant is

calculated by the following ecjuation:


F = P cos (p sin G
where

(4.7)

P = axial load measured by the load cell,


(p = longitudinal swing angle, and
9 = transverse swing angle.
Twenty-one records of response for each west, east, and

central twin conductor are available.

Record N14 is used as

a reference zero wind speed record to initialize all

75
records.

Static loads are transferred from the conductors

to the tower structure even before being subjected to wind


loads.

These initial static loads are due to the self

weight of the conductors and the inclined orientation of the


insulators.

The inclined orientation of the insulators is

caused by the differences in elevation of the supporting


tower structures and by the different configuration of the
tower (refer to Chapter III). Load cell and swing angle
readings of record N14 give the weight of the conductors and
initial transverse loads.

These initial transverse loads

are subtracted from the transverse loads obtained in each


record.

Initialization with the zero wind speed record is

necessary to obtain conductor response due to wind only.


A time history plot of the response of the west
conductor for record NOl is shown in Figure 14.

Similar to

wind records, each response record has 7200 points, which


corresponds to data collection at the rate of 10 sps for a
time duration of 12 minutes.

The time history plot in

Figure 14 uses averages of 10 consecutive points; thus the


points are at one second intervals.

Mean, standard

deviation and one second interval peak values are shown on


the time history plot.

The purpose of these plots is to

provide a graphical display of conductor response versus


time, thereby illustrating the overall quality and trends in
the data.

76

- Imer\-al Peak - 4.61

-, Mean+Sigma 3.43
Mean - 2.97
Mean-Sigma = 2.51

Time (minutes)

Figure 14:

Time History Plot of West Conductor Response for


Record NOl

Similar to the wind data, stationarity checks are done


for the load cell and conductor swing angle data.

Most of

the conductor response data (load cell and swing angle data)
are found to be stationary with 95% confidence limits as
checked by the run and trend tests (Bendat, 1966).

The

response spectra are plotted (in subsecjuent section) to


check for noise in the data.

In general, the conductor

response data are valid and can be used for analysis.

77
The mean and RMS values of response for the three
conductors (west, east, and central) are shown in Table 9.
The negative values in Table 9 indicate wind from the east.
The choice of sign for conductor response is arbitrary.

The

values in the table show that response recorded at the


central conductor is higher than that for the west and east
conductors.

This is expected since the central conductor is

suspended at 26.7 m above ground level, which is 8.4 m above


the west and east conductors at Tower 16/4 (see Figure 8).
The response of the west conductor is slightly higher
than that of the east conductor in many records, even though
they are suspended at the same height above ground.

The

range of the difference in values is from -0.08 to 0.33 kN;


less than 10% of the recorded values.

There could be one of

several reasons or a combination of reasons for this


difference.

These reasons and observations based on the

data are given below.


One set of twin conductors (west or east) could shield
or intensify wind load effect on the other set of twin
conductors.

This is not likely because the west and east

conductors are 13.4 m apart at Tower 16/4.

In addition,

data do not show a specific trend in response for winds


coming from east or west.

78
TABLE 9
Mean and RMS Values of Conductor Response (Transverse Load
Component)

Record
Number

West Conductor
Mean
RMS
(kN)

East Conductor
Mean
RMS
(kN)

NOl
N02
N03

2.97
1.36
2.48

N04

Zero Wind reco rd

0.46
0.43
0.21

2.82
1.34
2.43

Central Conductor
Mean
RMS
(kN)

0.41
0.40
0.19

3.13
1.34
2.56

0.53
0.43
0.20

N05
NO 6
N07

-2.84*
2.25
2.82

0.13
0.20
0.89

-2.62
2.19
2.80

0.09
0.18
0.82

-2.77
2.32
2.94

0.10
0.22
1.00

NOB
N09
NIO

5.14
-1.10
2.81

0.76
0.10
0.52

4.83
-1.13
2.76

0.76
0.11
0.46

5.47
-1.16
2.91

0.79
0.14
0.54

Nil
N12
N13

3.35
1.27
3.70

0.61
0.41
0.66

3.28
1.34
3.51

0.56
0.41
0.59

3.53
1.35
3.85

0.66
0.39
0.69

N14

Reference Zero Wind Speed Recc>rd

N15
N16
N17

2.61
0.72
3.32

0.31
0.12
0.61

2.47
0.80
3.10

0.28
0.14
0.57

2.68
0.81
3.34

0.34
0.12
0.67

N18
N19
N20

4.67
2.48
-2.74

0.78
0.52
0.12

4.27
2.33
-2.78

0.72
0.17
0.02

4.94
2.58
-3.00

0.80
0.55
0.14

N21
N22
N23

2.99
3.09
2.39

0.51
0.60
0.50

2.79
2.90
2.20

0.44
0.53
0.43

3.12
3.22
2.44

0.57
0.65
0.52

* negative sign indicates swinging to the west

79
Measurement of weight of the conductor during reference
zero wind condition can cause discrepancy in calculation of
transverse loads.

The west conductor weight is recorded

higher than the east conductor by 2.4 kN (a difference of


9%).

The information obtained from the field indicates that

detuner dampers are added to the west conductors.

These

dampers account for approximately 0.9 kN of extra weight.


Discrepancy in dead weight of 1.5 kN in west conductor is
not reconciled.

This extra dead weight could be a reason

for higher response of west conductor compared to east


conductor in most of the records (see Table 9).
Accuracy in swing angle measurements is important for
estimation of the transverse load component.

The swing

angle indicators have a resolution of 0.2 degrees.


Discrepancy in swing angle reading of 0.2 degree could cause
error in response by 2-4% for the range of readings obtained
in these data.

Accuracy of the swing angle indicator can be

one of the factors causing a difference in recorded


response.

Notwithstanding the differences in recorded

responses of conductors mentioned above, the data for the


three conductors are fairly consistent.

Some variations and

scattering in the field data are inherent; they cannot be


avoided.

In general, conductor response data are considered

valid for use in analysis.

80
Effective Conductor Force
Coefficient
As discussed in Chapter II, the mean response of the
conductor is related to a nondimensional force coefficient.
To calculate the force coefficient based on field measured
data, it is necessary to use measured values of the mean
wind speed, V, and the mean transverse load component, F.
The field measured effective conductor drag coefficient is
obtained by ecjuating the measured mean transverse load
component to the 'stagnation pressure' load.

4 pV
2

L d

where C^ = conductor effective force coefficient,


F

= mean transverse load measured in kN,

= mass density of air in kN m

-3

( = 12.02x10"^ kN m"^)
V = mean wind speed at conductor effective
height in mps,
L = effective conductor span in meters, and
d = conductor diameter in meters
(= 2 times the diameter; for present study)

81
The diameter is multiplied by 2 because of twin
conductors in each conductor bundle.

The effective span of

the conductor depends on the attachment heights, and the


conductor span.

In addition, temperature, and conductor

tension have some effect on the effective span.

As

indicated in Figure 9, effective spans are determined to be


376 m for west and east conductors, and 402 m for the
central conductor.

For determination of effective spans,

the horizontal tension in the conductor is taken to be 45 kN


at 60 F, based on data obtained from the BPA.
The conductor effective force coefficient is a function
of wind direction (Potter, 1981).

Computing the transverse

wind speed component from the wind speed and yaw angle to
relate to the transverse load component, the C^ values are
believed to be realistic.

Wind tunnel tests show that the

force coefficient essentially remains the same for a yaw


angle up to 22 degrees (Potter, 1981).

Therefore the

effective force coefficients computed from the data are


limited to records with yaw angles of 22 degrees or less.
Calculated effective conductor force coefficient values for
eleven records are shown in Table 10.
The effective force coefficient values are scattered
between 0.47 and 0.74.

The values for the west conductor

are higher than those for the east conductor because the

82
TABLE 10
Field Measured Conductor Effective Force Coefficients

Record
Number

Yaw
Angle
(degrees)

NOl
N02
N03

35
53
22

N04

West
Conductor
0.74

Central
Conductor

0.72

0.63

0.68
0.60

0.60
0.53

Zero Wind record

N05
NO 6
N07

11
2
47

0 .74
0 .61

NOB
N09
NIO

1
29
34

0 .73

Nil
N12
N13

35
42
25

N14

East
Conductor

0.68

0.64

Reference Zero Wind Speed Record

N15
N16
N17

3
42
27

0.69
-

0.65

0.59

N18
N19
N20

16
14
8

0.64
0.65
0.52

0.58
0.61
0.53

0.57
0.56
0.48

N21
N22
N23

12
18
16

0.57
0.54
0.68

0.53
0.51
0.63

0.50
0.47
0.58

83
mean response of the west conductor is larger than that for
the

east conductor as discussed in the previous section (see

Table 9 ) .
There are three possible reasons for the scatter in the
values of effective force coefficient.

The first reason may

be the accuracy with which transverse loads are measured.


Field measured transverse load components could vary by
about 10%, (see the previous section on validity of
conductor response data).

The second reason may be the

modification of the wind speed for different heights above


the ground.

In calculating the effective height of each

conductor, the conductor shape is assumed to be parabolic


and the site topography is taken into account.

More

specifically, each effective height is calculated as the


average of the vertical distances between the ground and the
conductor at all points along the half spans on both sides.
The resulting values are 15.7 m for the east and west
conductors, and 23.8 m for the central conductor.

The wind

profile is taken as having the same properties with respect


to the ground no matter how the ground elevation varies.
This raises some uncertainties, especially with regard to
the valley shown between Towers 16/4 and 16/5 in Figure 9.
Wind speeds are modified from the 34.7 m height to the 23.8
m effective height for the central conductor and to 15.7 m

84
effective height for west and east conductors using a wind
profile exponent value of a = 0.14.

The modification in

wind speed is smaller for the central conductor than for the
west and east conductors.

This could be the reason for the

force coefficients for the central conductor being smaller


than

for the west and east conductors (refer to Table 10).

The third reason may be the wind characteristics of


turbulence and fluctuations in wind direction.

As shown in

Table 5, many records show RMS of wind direction


fluctuations to be in the neighborhood of ten degrees.
These data in Table 5 suggest that the wind direction may
have fluctuated as much as 40 degrees within a record.

In

addition, the turbulence intensity values shown in Table 8


are different between the records.

Wind characteristics may

have significant effects on effective force coefficients.


The values of effective force coefficient obtained from
the field data are fairly consistent with the values shown
in Figure 3.

The Reynolds Number for the field data is in


4

the range of 3x10 to 6x10 . For this range of Reynolds


Number, Figure 3 suggest effective force coefficients in the
neighborhood of 0.6.

85
Response Spectrum
The conductor response spectrum represents the
fluctuating response about the mean response in the
frecjuency domain.

Response spectra of west, east, and

central conductors for all records were obtained.

A typical

response spectrum for the west conductor for record NOl is


shown in Figure 15.

The response spectra were obtained

using the same procedure as for the gust spectra discussed


earlier.

Similar to the gust spectra, the lower and upper

limits of the frecjuency range in spectral calculation are


0.0025 and 5 Hz, respectively.

The response spectrum in

Figure 15 shows fluctuations in frecjuencies below 1 Hz, but


does not show a spike at a specific frecjuency.

The range of

frecjuencies of interest for the present study is based on


natural frecjuencies of conductor vibration.

The conductor's

natural frecjuencies are in the range of 0.1-0.4 Hz,


depending on its configuration.

At frecjuencies between 2

and 4 Hz, the spectrum shows several very high peaks.

These

peaks in the figure are somewhat misleading because the


spectral density function in the plot is multiplied by the
frecjuency.

Generally, the spectral density function is

multiplied by frecjuency to get enhanced results in the


higher frecjuency ranges.

86

f - Frequency (in Hertz)


S(f) - Spectral Density at f
SIGSQ - Mean Sqaure Value

3*

n
CO
a
O
teiH

CD

CO

"^

>,'
CO

VM
1

Zone

Zone

Zone III

CM

O
3

3
>

CO

ii
CU

.1
f

a
o

^^Vr-n..
l.O

O.l

G.OOL

Frequenc:y, f (Hertz)

Figure 15:

Response Spectrum Plot for West Conductor


Response for Record NOl

As noted in Chapter II, the area under the response


spectrum gives the mean scjuare of response.

The response

spectrum is viewed in terms of area representing extent of


response.

Peaks in the spectrum represent response at

specific frecjuencies.

In order to discuss background and

resonant response, the spectrum is divided into three zones


for determination of relative response:

(1) frequencies

87
less than 0.1 Hz, (2) frequencies between 0.1 Hz and 1.0 Hz,
and (3) frequencies greater than 1 Hz.

Each zone of the

spectrum is reviewed in light of the conductor response to


wind.
The spectral area in the zone of frecjuencies less than
0.1 Hz is in the neighborhood of 75% for all records.

This

response is primarily due to the background turbulence where


wind has significant turbulent energy (see Figure 13). The
peaks in the spectra in this zone do not represent dynamic
amplification of the response.
The spectral area in the zone between frecjuencies of
0.1 and 1.0 Hz is close to 15% for all records.

Fundamental

transverse frecjuency of the conductor, f , in Hz for a


parabolic profile can be obtained using ecjuation 2.27.

The

conductor sag depends on conductor span, horizontal tension


and temperature.

The conductor span on the south side of

the Tower 16/4 is 450 m and toward the north it is 252 m


(see Figure 9).

The sags, calculated using conductor

horizontal tension of 45 kN at 60F, are 6.3 m and 21.0 m


corresponding to the 252 m and 450 m spans, respectively (a
parabolic profile is assumed).

Using ecjuation 2.27 with

estimated sags, the calculated natural frecjuencies in the


transverse direction are 0.12 Hz and 0.22 Hz.

The spectrum

in Figure 15 shows small but distinguishable peaks in the

88
neighborhood of these frequencies.

The two unequal spans

are expected to respond in the transverse direction


independent of each other due to the low stiffness of the
conductor.

The two peaks are judged to be due to resonant

response at natural transverse frecjuencies of the conductor.


At frecjuencies close to the conductor natural
frecjuencies the gust spectrum in Figure 13 shows that wind
has a fair amount of energy.

Presence of gust energy at

natural frecjuencies of the conductor can cause significant


resonant amplification of the conductor response.
The spectral areas in the zone of frecjuencies greater
than 1 Hz are less than 15% in most of the records.

High

spectral peaks (Figure 15) in the 2-4 Hz frecjuency range are


believed to be due to vibration of the tower and the
frecjuency of the swing angle indicators.

These frecjuencies

of the tower and instruments enter into the conductor


response records since the conductor is connected to the
tower through load cells and swing angle indicators.
Natural frequencies of Tower 16/4 are determined using
MSC/NASTRAN version 63 software.

The tower structure is

modelled as a space frame, without conductors and overhead


ground wires.

Natural frecjuencies of the tower are 2.88,

3.01 and 4.92 Hz, corresponding to longitudinal, transverse,


and torsional modes of vibration, respectively.

In

89
addition, the frequency of vibration of the swing angle
indicator is 3.2 Hz (Kempner, 1980).
the

These frecjuencies of

tower and the swing angle indicator are believed to

cause high peaks in the response spectrum in the frecjuency


range of 2 to 4 Hz.

Even though spectral peaks are high in

the frecjuency range above 1 Hz, the amount of energy is


relatively low.
Since the amount of gust spectral energy above 1 Hz in
Figure 13 is very small and since peaks in the response
spectra can be justified as above, it is believed that the
response spectral energy in the frecjuency range above 1 Hz
is not due to response of the conductor to wind.

This

spectral energy is neglected in further consideration of the


response of the conductors.

The background and resonant

responses assessed from the response spectra are compared


with the analytical procedure in the next chapter.

CHAPTER V
COMPARISON AND REFINEMENT OF THE
ANALYTICAL MODEL

The goal of the present study is to compare and to


refine the analytical model using the results of field data.
The analytical model proposed by Davenport (1980) to
determine peak response of transmission line structures was
presented in Chapter II. The key elements in determining
the fluctuating response of conductors in the model are the
background response and the resonant response (see ecjuation
2.23).

In addition, the model recjuires establishment a

value for the peak factor to determine peak response (see


ecjuation 2.1).
The field data analysis yields the peak response as
well as the fluctuating response.

Background and resonant

responses are determined from response spectra of field data


utilizing the procedure indicated in Chapter IV.

Peak

factors are obtained from the field peak responses utilizing


the upcrossing rate procedure.
Background and resonant responses assessed from the
field data are compared with values calculated using the
analytical model.

Joint acceptance function coefficients

90

91
are obtained from the field data to refine the background
response of the analytical model.

Also, aerodynamic damping

ratios are recovered from the field data to improve the


resonant response of the analytical model.

Since the

fluctuating response depends on many parameters, total


fluctuating responses from field data are not compared with
total fluctuating response predicted using the Davenport
model (1980).

Comparison of Analytically Predicted


Mean Square Response With Field
Measured Values
One of the two parts of the fluctuating response
component in ecjuation 2.1 is the mean scjuare value.

In the

frecjuency domain analysis, the mean scjuare value of response


is computed as the area under the response spectrum.

The

mean scjuare value can be considered as a summation of


background response, B , and resonant response, R
2.23).

(ecjuation

Background response is due to the wind turbulence at

low frecjuencies, and can be considered as quasi-static


response.

Resonant response is due to coincidence of

conductor natural frecjuencies with gust frecjuencies.

This

resonant response is the area under the response spectrum at


frecjuencies close to conductor natural frecjuencies.

92
Field Measured Mean Scjuare Response
The gust spectrum of Figure 13 shows that the wind
turbulence has energy up to 1 Hz, and energy beyond 1 Hz is
negligible.

Conductor natural frecjuencies are in the range

of 0.1 to 0.4 Hz, hence the resonant peaks should dominate


above the background response in this frecjuency range.

It

is difficult to separate background and resonant responses


in the field data.

In Figure 16, which is the same response

spectrum as Figure 15, there are peaks at the natural


frecjuencies of the conductors:
these peaks are diffused.

0.12 and 0.22 Hz. However,

As discussed in Chapter IV, the

spectral areas between the frecjuencies of 0.1 and 1.0 Hz for


most records were less than 15% of the total area under the
response spectrum (total mean scjuare value).

At the risk of

being on the high side, the total spectral areas between


frecjuencies of 0.1 and 1.0 Hz are assumed to be resonant
response.

The error introduced by this assumption is small

because the response in this frecjuency range is a small


portion of the total fluctuating response.
response is indicated by R

This resonant

in Figure 16.

It is reasonable to assume that the area under the


response spectrum below 0.1 Hz is the background response.
This area is close to 75% of the total area (total mean
scjuare value) in most of the

records as discussed in Chapter

93

f - Frequency (in Hertz)


S(f) - Spectral Density at f
SIGSQ - Mean Sqaure Value

a
3*

CJ

a
CO

o
en

CO

CO
M

CM

>
1

CO

Q.OOL

0.1

10.0

Frequency, f (Hertz)

Figure 16:

IV.

West Conductor Response Spectrum Plot for Record


NOl

The area designating background response is shown as B

in Figure 16.
The area under the response spectrum for frecjuencies
above 1 Hz is not considered to be response due to extreme
wind effects.
IV

This was discussed in some detail in Chapter

94
Delineation of background and resonant responses in the
field response spectra permits assessment of responses in
each of the three conductors, west, east, and central
conductors, for all twenty-one field records.

Field

measured values for background and resonant responses for


west, east, and central conductors are tabulated in Tables
11 through 13, along with mean scjuare values, for each
conductor.

These values are compared with values predicted

by the analytical model as described in the next section.


The mean scjuare values of the three conductors are
fairly consistent, but have scatter for a given record.
This is not surprising, since there was variation (by about
10%) in the mean response of the three conductors (refer to
Table 9).

It is reasonable to expect larger variation in

fluctuating response between the conductors.

The variation

in the field data suggests that results will have scatter,


and that it is important to use an ensemble of data for
appropriate interpretation of the results.
Analytical Model Predicted Mean
Scjuare Value
The analytical model used to predict the background and
resonant response contains a number of wind and conductor
related parameters (see ecjuations 2.24 and 2.25).

These

parameters can be separated into fixed and variable

95
TABLE 11
West Conductor Response Spectrum Data Analysis

Background Response
Record
Number
NOl
N02
N03

0.211
0.184
0.043

Analytical
Field Analytical Field
Model*
Measured
Measured
Model*
0.141
0.132
0.028

0.201
0.038
0.052

0.033
0.024
0.009

0.254
0.043
0.056

Zero Wind Record


0.017
0.039
0.790

0.003
0.028
0.609

0.005
0.043
0.143

0.008
0.006
0.087

0.006
0.048
0.203

0.582
0.009
0.270

0.447
0.007
0.199

0.365
0.014
0.161

0.057
0.001
0.037

0.502
0.012
0.200

Nil
N12
N13

0.366
0.165
0.441

0.269
0.151
0.300

0.256
0.033
0.217

0.043
0.010
0.077

0.348
0.034
0.288

22 2

N05
NO 6
N07
MOO
O vD 00

N04

Mean
Scjuare

Resonant. Response

Reference Zero Wind Speed Record


0.098
0.013
0.377

0.077
0.011
0.309

0.069
0.005
0.152

0.015
0.002
0.041

0.077
0.004
0.192

N18
N19
N20

0.606
0.273
0.014

0.486
0.236
0.005

0.186
0.125
0.002

0.066
0.024
0.002

0.259
0.140
0.003

0.258
0.361
0.250

0.190
0.243
0.194

0.076
0.151
0.177

0.042
0.060
0.036

0.095
0.194
0.194

2 2 2

N15
N16
N17

to to to
CA) to M

N14

* Davenport, 1980

96
TABLE 12
East Conductor Response Spectrum Data Analysis

Background Response
Record
Number

Mean
Scjuare

rH CM CO

2 2 2

ooo

0.168
0.162
0.037

N04

Resonant Response

Analytical
Field Analytical Field
Measured
Model*
Model*
Measured
0.115
0.128
0.023

0.181
0.037
0.050

0.034
0.024
0.011

0.229
0.042
0.054

N05
NO 6
N07

0.009
0.033
0.677

0.002
0.021
0.541

0.004
0.041
0.141

0.002
0.006
0.080

0.005
0.045
0.200

22 2

MOO
O vD 00

Zero Wind Record

0.572
0.013
0.207

0.450
0.009
0.157

0.322
0.015
0.155

0.061
0.001
0.034

0.443
0.012
0.193

Nil
N12
N13

0.310
0.165
0.348

0.231
0.153
0.248

0.245
0.037
0.195

0.047
0.011
0.062

0.333
0.038
0.259

Reference Zero Wind Speed Record


0.078
0.020
0.321

0.059
0.017
0.268

0.062
0.007
0.133

0.016
0.003
0.041

0.069
0.005
0.167

to M M
O VD 00

0.512
0.205
0.008

0.420
0.178
0.002

0.155
0.110
0.002

0.067
0.023
0.002

0.216
0.124
0.003

2 2 2
to to to

N14

0.193
0.276
0.187

0.147
0.199
0.153

0.066
0.133
0.150

0.036
0.055
0.027

0.082
0.171
0.164

CA)

to M

22 2

N15
N16
N17

* Davenport, 1980

97
TABLE 13
Central Conductor Response Spectrum Data Analysis

Background Response
Record
Number
NOl
N02
N03
NO 4

Mean
Scjuare
0.284
0.183
0.040

Resonant Response

Field Analytical Field


Analytical
Measured
Model*
Measured
Model*
0.175
0.119
0.026

0.211
0.035
0.053

0.038
0.021
0.008

0.207
0.031
0.044

Zero Wind Record

NO 5
NO 6
N07

0.011
0.047
1.004

0.004
0.036
0.714

0.005
0.043
0.147

0.005
0.007
0.085

0.004
0.037
0.162

NOB
NO 9
NIO

0.618
0.020
0.287

0.467
0.009
0.206

0.391
0.015
0.163

0.046
0.010
0.035

0.417
0.009
0.158

Nil
N12
N13

0.432
0.148
0.471

0.294
0.136
0.308

0.269
0.035
0.222

0.038
0.009
0.065

0.283
0.028
0.228

N14

Reference Zero Wind Speed Record

N15
N16
N17

0.117
0.014
0.443

0.094
0.012
0.365

0.069
0.006
0.146

0.016
0.001
0.038

0.060
0.004
0.142

N18
N19
N20

0.638
0.300
0.019

0.524
0.261
0.005

0.197
0.128
0.002

0.054
0.024
0.013

0.212
0.111
0.002

N21
N22
N23

0.320
0.420
0.269

0.236
0.271
0.210

0.078
0.155
0.175

0.049
0.061
0.032

0.076
0.155
0.148

* Davenport, 1980

98
parameters.

Some of the parameters depend on the geometry

and physical characteristics of the conductors.

These

parameters such as conductor sag, diameter of the conductor,


effective height, etc. are fixed parameters.

On the other

hand, other parameters such as mean wind speed, turbulence


intensity and aerodynamic damping ratio vary with each wind
record; these parameters are considered as variable
parameters.
The fixed and assumed parameters used in the analytical
model to calculate background and resonant responses are
tabulated in Table 14. Each conductor bundle consists of
two Chukar conductors, hence the effective diameter used in
the model is twice the diameter of the individual conductor.
The mean wind speed at the effective conductor height is
determined using the recorded mean wind speed at 34.7 m and
the power-law exponent, a = 0.14.
Table 15 shows the mean wind speeds calculated for the
west, east, and central conductors at their effective
heights.

The exposure factor, E, in ecjuations 2.24 and 2.25

is twice the turbulence intensity.

The exposure factor

values in Table 15 are obtained from the turbulence


intensity recorded at 34.7 m on Tower 16/4 (refer to Table
7).

The exposure factors used in the model at the effective

heights of the conductors are assumed to be the same as at

99
TABLE 14
Fixed and Assumed Parameters Used in the Analytical Model

Parameters

Typical Values Used

Values based on physical characteristics


(1) Conductor diameter (d)

0.08 m

(2) Effective span (L)


east and west conductors
central conductor

376 m
402 m

(3) Effective height (h)


east and west conductors
central conductor

15.7 m
23.8 m

(4) Conductor fundamental frecjuency (f ) 0.12 Hz


Assumed values*
(5) Conductor force coefficient (C^)

1.0

(6) Coherence exponent (c)

(7)
Scale of turbulence (L s)
^ '
(8) Kaimal's gust spectrum constant (A)

65 m

(9) Kaimal's gust spectrum constant (n)

0.67

0.28

*assumed values are recommended by Davenport (1980)

34.7 m on Tower 16/4. The conductor aerodynamic damping


ratio is calculated using ecjuation 2.26.

Since the

aerodynamic damping ratio depends on the mean wind speed, it


is different for each record.

100
TABLE 15
Variable Parameters Used in the Analytical Model

Mean Wind Speed (mps)


Record
Number
NOl
N02
N03

@ 15.7 m*

0.36
0.34
0.22

@ 23.8 m**

16.9
14.7
13.3

17.9
15.5
14.1

Ze ro Wind Rec<ord
0.06
0.22
0.32

14.2
13.9
20.1

15.0
14.7
21.2

0.28
0.26
0.34

19.3
8.6
16.6

20.4
9.1
17.5

Nil
N12
N13

0.36
0.34
0.30

18.9
12.6
18.2

19.9
13.2
19.2

222

N05
NO 6
NO 7
MOO
O vD 00

N04

Exposure
Factor

Reference Zero Wind Speed Record


0.24
0.24
0.28

14.1
9.0
16.9

14.9
9.5
17.8

0.22
0.34
0.04

19.6
14.2
16.6

20.7
15.0
17.5

0.22
0.30
0.42

16.6
17.3
13.6

17.5
18.3
14.4

CO to M

2 2 2

to to to

to M M
O VD 00

N15
N16
N17

22 2

N14

* effective height for west and east conductors


** effective height for central conductor

101
Instead of calculating the mean wind pressure, P, and
the influence coefficient, 0^ (which translates the pressure
to response), the field measured mean transverse load
components are used for each record (refer to Table 9 ) .
Calculated background and resonant responses using the
analytical model are shown for the three conductors in
Tables 11 through 13.

The values are obtained using

ecjuations 2.24 and 2.25.

The majority of the records show

that the background response calculated from the analytical


model is smaller than the background response measured in
the field.

Field measured values versus analytical model

values of background response of the three conductors are


plotted in Figure 17.

The figure shows that the background

response predicted by the model is an underestimation of the


measured value.

Since the background response accounts for

75% of the mean scjuare response value, refinement of the


analytical model of this part is desirable.
The analytical model predicts higher resonant responses
than the field measured values (refer to Tables 11 to 13).
Field measured values versus analytical model values of
resonant response of the three conductors are plotted in
Figure 18.

The figure clearly shows that the predicted

values are very much higher than the field measured values.
One of the significant variables in the analytical model is

102
0.8
o

West Cond.
East Cond.
Central Cond

0.6U

a
>

0.4-

a
a

0.4

0.6

0.8

Field Measured Values

Figure 17:

Analytical Model Background Response Versus


Field Measured Background Response

the aerodynamic damping ratio.

If the damping ratio is

higher than predicted by equation 2.26, the calculated


resonant response will be smaller.

Field measured data are

used to assess a possible damping ratio for each record.

103
Even though a large scatter in evaluation of the damping
ratio is expected, it can lead to a better prediction of
resonant response.

0.60

(/)
3

" /

0.45-

>

0.30-

y/

Q n

c
<

- s" v y v

0.15-

y^

Q / y y^
a ^9
/yy

West Cond.
East Cond.
Central Cond.

0.00 \
0.00

0.15

0.30

0.45

0.60

Field Measured Values

Figure 18: Analytical Model Resonant Response Versus Field


Measured Resonant Response

104
Refinement of the Analytical Model
As noted in previous sections, the analytical model
underestimates the background response and overestimates the
resonant response.

Refinement of the analytical background

expression using the field measured data is attempted.

The

resonant response prediction is improved by recovering a


damping ratio for the conductor from the field data.

Background Response
The mean scjuare value of fluctuating response can be
calculated as the area under the response spectrum,
al

= J Sj^(f) df
0

(5.1)

2
where a = mean scjuare value of response,
So(f) = spectral density value of response, and
f

= frecjuency.

Utilizing ecjuations 2.10 and 2.11, equation 5.1 can be


expressed as
_2
2 = iL- J y^^(f)
0
V

|H(f)|2 S^(f) df

where S (f) = gust spectral density,


F

= mean transverse force,

= mean wind speed.

(5.2)

105
2
Z (f) = aerodynamic admittance function, and
2
|H(f)| = mechanical admittance function.
The area under the response spectrum is a summation of
background and resonant responses.

Ecjuation 5.2 can be

written in a simple form as:


2

_2
4F

^R =

T B

^ Aj,|

(5.3)

V
where Ag accounts for background response, and A accounts
for resonant response.

Davenport (1977) developed ecjuations

for Ag and Ap as

Ag = j x^(4r) S^(f) ^^
0

(5.4)

and
f L
AR = X ^ ( - ^ ) S (f ) J |H(f)|^ df
V
0
where f

(5.5)

is the fundamental frecjuency of the structure and

other terms are defined above.

The background response due

to wind turbulence can be obtained using ecjuations 5.3 and


5.4, if the gust spectrum is defined by some appropriate
analytical function.
The aerodynamic admittance function is a relationship
between the gust spectral density function and the force

106
spectral density function in the frequency domain.
measure of the

It is a

effect that the wind turbulence has on the

transverse forces.

The shape of the conductor and the sizes

of gusts relative to the size of the conductor influence


this function.

A large gust, totally enveloping the

structure, is well correlated, while a small gust, acting on


a portion of the conductor, is uncorrelated.
The aerodynamic admittance function is usually
2 fL
expressed in a nondimensional form as x (-^-^)/ where L is
V
the conductor span, and the ratio -F ^^ designated as the
scale of turbulence (L ). The aerodynamic admittance
function is termed as a 'joint acceptance function (JAF),'
if it is modified to account for the mode shape.

In other

words, the important link between the gust fluctuations,


(described by the gust spectrum) and the modal force
fluctuations is provided by the the JAF.

This function

depends on the mode shape and the velocity field, which


varies widely from structure to structure.

Davenport (1977)

reduced the JAF to a simple form as below:


|JAF|2 = -^^
(5.6)
'
'
1 + m (p
where m is a constant to account for the mode shape and

107
<p = cfL/V.

The quantities c, f, L, and V represent the

coherence exponent, frequency, conductor span and mean wind


speed, respectively.
The theory described for computing conductor response
in Chapter II is based on the conventional assumption of a
constant force coefficient.

For conductors with a

cylindrical shape the force coefficient, C^, depends on the


Reynolds Number (refer to Figure 3).

This change in force

coefficient affects the fluctuating component of response.


The analytical model of fluctuating response should account
for changes in the force coefficient at Reynolds Numbers
corresponding to the mean wind speed (Davenport, 1980).

To

account for this effect the numerator of ecjuation 5.6 is


replaced with an unknown constant Q.

The resluting

ecjuation, which is a product of JAF and Q is simply termed


as JAF in this study, as shown below:
|JAF|2 =

2_^_.
1 + M(-^)
V

(5.7)

In addition to introduction of coefficient Q in ecjuation


5.7, the coefficient M is used to account for mode shape and
the coherence exponent, c (transverse correlation of
turbulence).

The available data are not able to provide

separate coefficients for the mode shape and correlation of

108
turbulence.

Equation 5.7 is in the same form as equation

2.24 given in the analytical model developed by Davenport


(1980).

In 1:he model, Davenport uses approximate values of

1 and 0.81 for the coefficients Q and M, respectively.

Here

the field response data are used to evaluate these two JAF
coefficients.

Determinincr the JAF Coefficients


The frecjuency transfer function (FTF) is a transfer
function between the spectral densities of fluctuating wind
turbulence and conductor response.

The FTF can be

considered as the product of the aerodynamic admittance


function and mechanical admittance function.

The FTF can be

written as
^SR(f)
5:
_2

v^
2
1-^- = 4 |H(f)r
fS

(f)

I V

/I

2
IJAFr.

(5.8)

F
The FTF can be obtained by plotting the ratios of the
nondimensionalized response spectral values to the
nondimensionalized gust spectral values (refer to equation
5.8).

As explained in Chapter IV, the IMSL program FTFREQ

is used to compute the spectral density values of wind


turbulence and conductor response fluctuations.

FTF plots

for all 21 records of west, east, and central conductors

109
were obtained.

A typical FTF plot for the west conductor

for record NOl is shown in Figure 19.

The spectral density

values of wind turbulence above 1 Hz are very small (refer


to Figure 13), and use of very small values in the
denominator of the FTF would be inappropriate.
FTF values are plotted up to 1 Hz only.

Hence, the

It was noted in

Chapter IV that wind gust and response spectra show cjuite a


bit of fluctuation because of the computational procedures
used in obtaining the spectral densities.

The FTF plot in

Figure 19 is obtained from the ratios of two spectra, so


large fluctuations in ordinates are not surprising.
The mechanical admittance function, also known as the
dynamic amplification factor, depends on the structural
dynamic properties such as frecjuencies and damping ratios.
This factor amplifies the response spectrum (resonant
response) at the natural frecjuencies of the conductor.

The

JAF related to background response can be obtained by


removing the resonant peaks at the natural frecjuencies of
the conductor from the FTF plot.

Equation 5.7 was fitted to

the field measured JAF plot by regression analysis to obtain


values of the coefficients Q and M.

Equation 5.7 is a

nonlinear ecjuation, hence nonlinear regression needed to be


applied.

The SAS procedure NLIN (SAS, 1982) was used to fit

the nonlinear equation to the computed JAF field response

110

28.0 -,

I I1 1 1 1 '

10.0

0.001

Frequency (Hertz)

Figure 19:

data.

Frecjuency Transfer Function of West Conductor


Response for Record NOl

The procedure NLIN is used to fit ecjuation 5.7 to the

field data of all 21 records of west, east, and central


conductors.
Procedure NLIN implements iterative methods that
attempt to find least squares estimates for the nonlinear
equations.

Parameter names and starting values, expressions

for the model, and expressions for derivatives of the model

Ill
with respect to the parameters need to be specified.

Based

on expectations, the specified ranges for the coefficients Q


and M were 0.4-1.0 and 0.1-0.4, respectively.

The NLIN

procedure first examined the starting value specifications


of the parameters in the specified search grid.

The NLIN

procedure then evaluated the residual sum of scjuares at each


combination of values to determine the best values to start
the iterative algorithm.

A modified Gauss-Newton iterative

method (SAS, 1982) was used, which involved regressing the


residuals on the partial derivatives of the model with
respect to the parameters until the iterations converged.
Some variation in data was expected, since the data were
measured in the field.
To find the best coefficient values, which in general
satisfied most of the records, a contour of lowest residual
sum of scjuares was plotted in the specified search grid.
Any combination of coefficients Q and M within the lowest
residual sum of squares contour is acceptable.

A typical

contour plot for west conductor response for record NOl is


shown in Figure 20.

Combining all plots of three conductors

(west, east, and central), there are 63 contour plots of


residual sum of scjuares.

The sixty three contour plots show

some degree of dispersion of lowest residual sum of scjuare


contours over the search grid.

The contour plots were

112

e
(J

0.1

0.2

0.3

0.4

Coefficient, M

Figure 20:

West Conductor JAF Coefficients Contour Plot for


Record N15

113
overlapped to find the best values of the coefficients Q and
M, which were 0.45 and 0.2, respectively.

These values are

significantly lower than the values of 1.0 and 0.81 used in


tihe Davenport model (equation 2.24).

For better

visualization the JAF with Davenport model values and


refined values are plotted on the same graph as shown in
Figure 21.
The low value of coefficient M obtained from the field
data may be because of a low value of coherence exponent, c,
due to the

long span of the conductors.

Also, when

-=- >> 1, the joint acceptance function is independent of


^s
the mode shape and is proportional to the ratio of the
correlation length to the conductor span.

These comments

are based on the results of wind tunnel experiments


conducted on a rod (Blevins, 1977).
The expression for background response of the
conductor, ecjuation 2.24, with the new coefficients is
^2 2
0.45
B^ = P Ol E^ 1 + 0.2(-^)
^s

(5.9)

The parameters in ecjuation 5.9 are defined below equation


2.24.

The background response calculated using the refined

analytical model for all 21 records of west, east, and


central conductors are tabulated in Tables 16 through 18.

114
o

1 - Davenport Model
2 - Refined Model

CM

O
O

o
CJ

o
c

o
LT)

oo
in

o
CJ
<

c
'o

o
o
o

-r

0.001

1II

0.0]

-I

1Irill

-I

1II I I

0.

"T

1.0

1II I I I

10.c

Reduced Frequency - fLA^

Figure 21:

Joint Acceptance Function Plot

These tables also show the ratios of the refined analytical


model values to the field measured values and the ratios of
the analytical model values to the field measured values.
In general, these nondimensional ratios show a comparison
between the analytical model and the field response values.
For better visualization the field measured values versus

115
the refined analytical model values are plotted in Figure
22.

The refined analytical model gave slightly better

predicted than the analytical model when Figures 22 and 17


are compared.

In Tables 16 through 18, means and

coefficient of variations (COV) of the ensemble of the


ratios are shown.

In each table, the mean of the ratio is

closer to 1 for the refined analytical model.


COV for each conductor did not change.

However, the

This improvement in

mean value of the ensemble and insignificant change in COV


value are due to inherent scatter in the field data.
Resonant Response
As noted earlier the analytical model overestimates the
resonant response.

One of the reasons may be the use of low

damping ratio values as determined by ecjuation 2.26.

Here

field measured resonant response data are used to estimate


damping ratios for the conductors.
Determining the Aerodynamic Damping
Ratio
As noted in Chapter II, three types of damping are
noted for conductor response, namely, material, structural,
and aerodynamic damping.

For conductors aerodynamic damping

is very much higher than material or structural damping.


Therefore, both material and structural dampings are

116
TABLE 16
Background Response of West Conductor

Record
Number

Mean
Scjuare

rH CM CO

O O O

Ratio
(1)*

0.239
0.037
0.062

1.695
0.280
2.214

1.426
0.288
1.857

Ratio
(2)*

N05
NO 6
NO 7

0.017
0.039
0.790

0.003
0.028
0.609

0.006
0.051
0.170

2.000
1.821
0.279

1.667
1.536
0.235

0.582
0.009
0.270

0.447
0.007
0.199

0.432
0.017
0.191

0.966
2.429
0.960

0.817
2.000
0.809

Nil
N12
N13

0.366
0.165
0.441

0.269
0.151
0.300

0.304
0.039
,0.257

1.130
0.258
0.857

0.952
0.219
0.723

N14

Reference Zero Wind Speed Rec ord

N15
N16
N17

0.098
0.013
0.377

0.077
0.011
0.309

0.082
0.006
0.180

1.065
0.546
0.583

0.896
0.455
0.492

0.606
0.273
0.014

0.486
0.236
0.005

0.220
0.148
0.003

0.453
0.627
0.600

0.383
0.530
0.400

0.258
0.361
0.250

0.190
0.243
0.194

0.090
0.179
0.210

0.474
0.739
1.083

0.400
0.621
0.912

1.003
65.5%

0.839
65.3%

to M M
O VD 00

Zero Wind Record

222

N04

MOO
O vD 00

0.141
0.132
0.028

Refined
Model

222

22 2

0.211
0.184
0.043

Field
Measured

2 2 2

rH CM CO
CM CM CM

mean value
coefficient of variation

(1)* ratio of refined model value to the measured value


(2)* ratio of analytical model value to the measured value

117
TABLE 17
Background Response of East Conductor

Record
Number

Mean
Scjuare

Field
Measured

Refined
Model

Ratio

0.215
0.043
0.060

1.870
0.336
2.609

1.574
0.289
2.174

(D*

Ratio
(2)*

NOl
N02
N03

0.168
0.162
0.037

N04

Zero Wind Record

N05
NO 6
NO 7

0.009
0.033
0.677

0.002
0.021
0.541

0.005
0.022
0.168

2.500
1.048
0.311

2.000
1.952
0.261

N08
N09
NIC

0.572
0.013
0.207

0.450
0.009
0.157

0.382
0.018
0.184

0.849
2.000
1.172

0.716
1.667
0.987

Nil
N12
N13

0.310
0.165
0.348

0.231
0.153
0.248

0.291
0.043
0.231

1.260
0.281
0.932

1.061
0.242
0.786

N14

Reference Zero Wind Speed Record

N15
N16
N17

0.078
0.020
0.321

0.059
0.017
0.268

0.073
0.008
0.157

1.237
0.471
0.586

1.051
0.412
0.496

N18
N19
N20

0.512
0.205
0.008

0.420
0.178
0.002

0.184
0.131
0.003

0.438
0.736
1.500

0.369
0.618
1.000

N21
N22
N23

0.193
0.276
0.187

0.147
0.199
0.153

0.079
0.158
0.178

0.537
0.794
1.163

0.449
0.668
0.980

1.078
63.8%

0.940
64.3%

0.115
0.128
0.023

mean value
coefficient of variation

(1)* ratio of refined model value to the measured value


(2)* ratio of analytical model value to the measured value

118
TABLE 18
Background Response of Central Conductor

Record
Number

Mean
Scjuare

NOl
N02
N03

0.284
0.183
0.040

N04

Zero Wind Record

NO 5
NO 6
NO 7

0.011
0.047
1.004

NOB
N09
NIO

Field
Measured

Refined
Model

Ratio

0.256
0.042
0.064

1.463
0.356
2.462

1.206
0.294
2.039

0.004
0.036
0.714

0.006
0.052
0.178

1.500
1.444
0.249

1.250
1.194
0.206

0.618
0.020
0.287

0.467
0.009
0.206

0.472
0.018
0.197

1.011
2.000
0.956

0.837
1.667
0.791

Nil
N12
N13

0.432
0.148
0.471

0.294
0.136
0.308

0.325
0.042
0.269

1.105
0.309
0.873

0.915
0.257
0.721

N14

Reference Zero Wind Speed Record

N15
N16
N17

0.117
0.014
0.443

0.094
0.012
0.365

0.083
0.008
0.176

0.883
0.667
0.482

0.734
0.500
0.400

N18
N19
N20

0.638
0.300
0.019

0.524
0.261
0.005

0.238
0.155
0.003

0.454
0.594
0.600

0.376
0.490
0.400

N21
N22
N23

0.320
0.420
0.269

0.236
0.271
0.210

0.095
0.188
0.211

0.403
0.694
1.005

0.331
0.572
0.833

0.929
62.0%

0.763
63.4%

0.175
0.119
0.026

mean value
coefficient of variation

(D*

Ratio
(2)*

(1)* ratio of refined model value to the measured value


(2)* ratio of analytical model value to the measured value

119
0.8
"
D

West Cond.
East Cond.
Central Cond

0.6_3
>

0.4s
c

0.2-

0.0

0.2

0.4

0.6

0.

Field Measured Values


Figure 22:

Refined Model Background Response Versus Field


Measured Values

neglected in this study, and the computed total damping is


assumed to be aerodynamic damping.
The response of the conductor at its natural frequency
of vibration is a function of excitation force and damping.

120
The magnification factor method is used here to estimate the
aerodynamic damping ratio.

For a single degree of freedom

system subjected to wind turbulence, the resonant peak is


amplified at a fundamental frecjuency of the conductor.

The

height of this peak is controlled by the damping for the


conductor.
The frecjuency transfer function (Figure 19) described
in the previous section is used to establish damping.
noted earlier, the

As

FTF is a combination of aerodynamic

admittance function and the mechanical admittance function.


The mechanical admittance function amplifies the response at
the natural frecjuency of vibration of the conductor.

The

expression for the mechanical admittance function is defined


in ecjuation 2.13.

At the fundamental frecjuency of the

conductor, the ecjuation for the mechanical admittance


function simplifies as follows:

|H(f^)|^ = - V

(^-^^^

4 C
where C
f

= aerodynamic damping ratio, and


= fundamental frecjuency of conductor.

The damping ratios of the conductors are determined


using the heights of the resonant peaks in the FTF plots.
As noted in Chapter III, the conductor spans on two sides of
Tower 16/4 are different.

The natural frecjuencies of

121
vibration of the conductors are 0.12 Hz and 0.22 Hz,
corresponding to spans of 450 m and 252 m, respectively.
The aerodynamic damping ratios estimated to be related to
these two natural frecjuencies of the west, east, and central
conductors from the FFT are tabulated in Table 19.
Conductor aerodynamic damping ratios for east wind records
(N05, N09 and N20) are not calculated because the peaks in
the FTF plots for these records are highly erratic.

The

east wind records give poor results for the FTF because of
low turbulence intensities in tihe records.

The estimated

aerodynamic damping ratios in Table 19 vary between 18 and


91%.

This large scatter in establishing damping ratios is

expected because computational technicjues used to obtain the


spectra cause large fluctuations in FTF.

In addition only

two specific peak values, closest to the frecjuencies of 0.12


and 0.22 Hz, are used in each FTF plot.

The peak values in

the FTF plots are not expected to be highly accurate.

Most

of the estimated damping values in Table 19 fall between 30%


and 60%.
The aerodynamic damping values predicted by ecjuation
2.26 are between 5% and 11% based on the mean winds recorded
in the field.

These values used in the analytical model are

significantly smaller than the ones estimated from the field


data.

In recognition of this discrepancy, a conservative

122
TABLE 19
Estimated Aerodynamic Damping Ratios in Percentages

Record
Number
NOl
N02
N03
N04
N05
NO 6
NO 7

West
Conductor

(1)*

(2)*

45
66
43

43
66
91

East
Conductor
(1)*
(2)*
43
32
38

41
34
67

Central
Conductor
(1)*
(2)*
46
27
62

44
34
66

Zero Wind Record


_

50
18

30
33

41
18

32
33

65
47

37
22

52

35

50

35

55

35

NOB
N09
NIO

34

38

27

45

30

41

Nil
N12
N13

32
50
39

58
52
67

32
22
44

58
54
60

38
28
47

50
50
63

N14

Reference Zero Wind Speed Record

N15
N16
N17

50
88
47

41
75
69

45
29
44

40
28
75

50
50
54

37
50
52

N18
N19
N20

33
42

35
46

30
45

41
42

37
54

51
42

N21
N22
N23

45
58
58

29
33
29

45
60
60

29
32
34

41
56
67

(1)* corresponding to resonant peak at 0.12 Hz


(2)* corresponding to resonant peak at 0.22 Hz

27
33
56

123
ensemble average value of 40% aerodynamic damping ratio is
suggested for conductors.

Resonant responses are calculated

with this suggested aerodynamic damping for comparison


purposes.
Resonant Response with Suggested Damping Ratio
Resonant response values for all three (west, east, and
central) conductors are calculated using an aerodynamic
damping ratio of 40% in the analytical model.
are tabulated in Tables 20 through 22.

The values

The tables also show

the field measured resonant response and the total mean


scjuare value for each record.
In addition, ratios of resonant responses obtained from
the analytical model with 40% damping to field measured
values and from the analytical model with damping from
ecjuation 2.26 to field measured values are shown in the
tables.

Use of 40% damping improves the prediction of

resonant response significantly.

Mean values of the ratios

for 40% damping are close to unity.

The COV of the ratios

in the tables are not effected significantly, though this is


misleading.

The mean values of the ratios of responses from

the analytical model with damping from ecjuation 2.26 are a


little more than 4; hence associated COV values of 45%
reflect a large variation.

124
TABLE 20
West Conductor Resonant Response With 40% Damping

Record
Number
NOl
N02
N03
NO 4

Mean
Scjuare
0.211
0.184
0.043

Field
Analytical
Measured
Model
0.033
0.024
0.009

0.053
0.008
0.009

Ratio
(1)*

Ratio
(2)*

1.606
0.333
1.000

7.697
1.792
6.222

0.008
0.051

1.333
0.586

8.000
2.333
8.807

Zero Wind Record

NO 5
N06
NO 7

0.017
0.039
0.790

0.008
0.006
0.087

NOB
N09
NIO

0.582
0.009
0.270

0.057
0.001
0.037

0.119

2.088

0.041

1.108

5.405

Nil
N12
N13

0.366
0.165
0.441

0.043
0.010
0.077

0.081
0.006
0.064

1.884
0.600
0.831

8.093
3.400
3.740

N14

Reference Zero Wind Speed Reco rd

N15
N16
N17

0.098
0.013
0.377

0.015
0.002
0.041

0.014
0.001
0.039

0.933
0.500
0.950

5.133
2.000
4.683

N18
N19
N20

0.606
0.273
0.014

0.066
0.024
0.002

0.062
0.025

0.939
1.042

3.924
5.833

N21
N22
N23

0.258
0.361
0.250

0.042
0.060
0.036

0.019
0.042
0.033

0.452
0.700
0.917

2.262
3.233
5.389

0.989
48.5%

4.886
45.8%

mean value
coefficient of variation

(1)* ratio of analytical model values with 40%


damping ratio to the field measured values
(2)* ratio of analytical model values with damping ratio
from ecjuation 2.26 to the field measured values

125
TABLE 21
East Conductor Resonant Response With 40% Damping

Record
Number
NOl
N02
N03
N04

Mean
Scjuare
0.168
0.162
0.037

Field
Analytical
Measured
Model
0.034
0.024
0.011

Ratio
(1)*

Ratio
(2)*

0.047
0.008
0.009

1.382
0.333
0.818

0.008
0.050

1.333
0.625

7.500
2.500
7.262

6.735
1.750
4.909

Zero Wind Record

N05
NO 6
N07

0.009
0.033
0.677

0.002
0.006
0.080

NOB
N09
NIC

0.572
0.013
0.207

0.061
0.001
0.034

0.105

1.721

0.039

1.147

5.676

Nil
N12
N13

0.310
0.165
0.348

0.047
0.011
0.062

0.078
0.006
0.057

1.660
0.550
0.919

7.085
3.455
4.177

N14

Reference Zero Wind Speed Record

N15
N16
N17

0.078
0.020
0.321

0.016
0.003
0.041

0.012
0.001
0.035

0.750
0.333
0.854

4.313.
1.667
4.073

N18
N19
N20

0.512
0.205
0.008

0.067
0.023
0.002

0.052
0.022

0.776
0.957

3.224
5.391

N21
N22
N23

0.193
0.276
0.187

0.036
0.055
0.027

0.017
0.036
0.027

mean value
coefficient of variation

0.472
0.665
1.000

2.278
3.109
6.074

0.905
45.4%

4.510
42.6%

(1)* ratio of analytical model values with 40%


damping ratio to the field measured values
(2)* ratio of analytical model values with damping ratio
from ecjuation 2.26 to the field measured values

126
TABLE 22
Central Conductor Resonant Response With 40% Damping

Record
Number

Mean
Scjuare

0.038
0.021
0.008

Ratio

(D*

Ratio
(2)*

0.045
0.006
0.008

1.184
0.286
1.000

5.447
1.476
5.500

0.007
0.043

1.000
0.506

5.286
1.906

Zero Wind Record

N05
NO 6
NO 7

0.011
0.047
1.004

0.005
0.007
0.085

0.618
0.020
0.287

0.046
0.010
0.035

0.105

2.283

9.065

0.034

0.971

4.514

Nil
N12
N13

0.432
0.148
0.471

0.038
0,009
0.065

0.070
0.005
0.054

1.842
0.556
0.831

7.447
3.111
3.508

N14

Reference Zero Wind Speed Rec ord

N15
N16
N17

0.117
0.014
0.443

0.016
0.001
0.038

0.011
0.001
0.032

0.688
1.000
0.842

3.750
4.000
3.737

0.638
0.300
0.019

0.054
0.024
0.013

0.054
0.020

1.000
0.833

3.926
4.625

0.320
0.420
0.269

0.049
0.061
0.032

0.017
0.035
0.026

0.347
0.574
0.813

1.551
2.541
4.625

0.920
53.0%

4.223
45.8%

22 2

to M M
O VD 00

MOO
O vD 00

N04

22 2

rH CM CO

O O O

22 2

0.284
0.183
0.040

Field
Analytical
Measured
Model

rH CM CO
CM CM CM

2 2 2

mean value
coefficient of variation

(1)* ratio of analytical model values with 40%


damping ratio to the field measured values
(2)* ratio of analytical model values with damping ratio
from equation 2.26 to the field measured values

127
For better visualization, a plot of the resonant
response predicted by the analytical model with 40% damping
versus the field measured resonant values is shown in Figure
23.

This figure, when compared with Figure 18, shows that

the analytical model predicted better resonant values with


40% aerodynamic damping ratio.

Figure 23 also illustrates

the inherent scatter in the field data.


Peak Factors
Another important component of the analytical model for
fluctuating response is the peak factor, g.

It is used to

predict the peak response value that can occur in a time


segment.

The peak factor is defined as the number of root

mean scjuare values by which the peak value exceeds the mean
value.

The peak factor for field measured response data is

calculated using the ecjuation

g =

(5.11)

JLJL_R

^R
where ft = peak response value,
R = mean response value, and
<Tp = root mean scjuare of response.
The peak factor values vary depending on the averaging
time interval; the smaller the peak averaging time interval
the higher the peak factor value.

The peak factors from the

128
0.12

West Cond.
East Cond.
Central Cond

0.09u
3

a
>

"8

:z

0.06-

"a
u
c
c
<

0.03-

0.00
0.00

0.03

0.06

0.09

0.12

Field Measured Values


Figure 23:

Analytical Model Resonant Response With 40%


Damping Versus Field Measured Values

field data for response of the three conductors (west, east,


and central) are calculated and tabulated in Table 23.
The values computed are based on 0.1 and 1 second time
averaged peak values.

As expected, peak factors calculated

129
TABLE 23
Peak Factors for Conductor Response

Record
Number
NOl
N02
N03
NO 4

West
Conductor
5.62*
3.97
4.54

4.00**
3.03
3.94

East
Conductor

Central
Conductor

6.60
3.99
5.12

4.54
3.16
4.32

6.91
5.37
5.53

3.26
2.69
4.54

Zero Wind Record

N05
NO 6
NO 7

4.77
4.07
5.14

2.07
3.30
3.20

4.18
4.02
3.74

2.31
3.00
3.12

3.71
4.75
6.40

2.80
4.28
3.04

NOB
N09
NIC

3.48
3.28
4.29

2.74
2.88
3.34

3.42
3.30
4.33

2.72
2.73
3.38

3.59
3.52
4.32

2.52
3.07
3.43

Nil
N12
N13

6.67
3.76
5.45

4.52
3.60
4.20

6.78
3.37
7.09

5.11
3.15
4.69

6.06
3.95
4.45

4.02
3.58
3.36

N14

Reference Zero Wind Speed Record

N15
N16
N17

4.15
3.88
5.68

3.75
3.69
4.61

4.65
3.69
5.71

3.85
3.34
4.90

4.76
4.12
4.67

3.81
3.68
3.86

NIB
N19
N20

3.89
4.01
3.75

3.18
3.52
2.01

4.19
4.63
4.37

3.05
3.88
2.15

3.94
4.41
4.02

3.13
3.51
3.04

N21
N22
N23

4.02
5.62
3.32

3.29
4.43
2.42

4.83
5.99
3.16

3.85
5.01
2.51

4.01
4.10
4.09

3.27
3.34
2.47

* based on 0.1 second peak values (instant peaks)


** based on 1 second average peak values

130
for 0.1 second peaks are higher than those calculated for 1
second time average.

The peak factors for 0.1 second time

averages range between 3.16 and 6.78, and for 1 second time
averages they range between 2.01 and 5.11.

The suggested

range for peak factors in the analytical model is 3.5 to 4.0


(Davenport, 1980).

Many records show that the peak factors

measured in the field are higher than 4.0.

This is not

surprising in view of the fact that wind speed fluctuations


tend to be Gaussian, while the response fluctuations
associated with flow separation are often highly
intermittent, thus giving rise to large peak factors.

The

peak factors vary from record to record and their


distribution functions are recjuired in order to establish
peak values with a specified probability of being exceeded.
Peak factors as function of the probability distribution of
upcrossings, or ecjuivalently, a specified number of
occurrences in a given interval of time, are obtained from
the field data.

Probabilistic Peak Factors from


Field Data
For a stationary Gaussian process the cumulative
probability distribution in terms of upcrossings can be
stated as (refer to ecjuation 2.21).

131
2
P(>x) = exp - { - ^ 1

(5.12)

2 G^
X

where

P(>x) = probability of upcrossing.


X

- threshold level specified (=gCT ), and

CT

- root mean scjuare.

The Rayleigh distribution function in ecjuation 5.12 can


be expressed in Weibull distribution form as
k
P(>x) = exp -(-2.)
where
and

g
c, k

(5.13)

X
/

= constants.

Ecjuation 5 . 1 3 can be expanded a s


I n ( - I n P(>x)

) = k ln(g) - k ln(c).

(5.14)

A graph of P(>x) versus x on an appropriate log scale


will yield k as the slope of the straight line and c as the
zero intercept.
Upcrossing rates were calculated for different
threshold values (multiples of RMS) of response of all three
conductors for each record.

The field data used were the

0.1 second time interval responses.

A linear regression

line was fitted for each record to assess trends with

132
respect to wind speed, wind direction, and turbulence
intensity.

The conductor response data did not indicate any

specific trend for wind speed, wind direction or turbulence


intensity (terrain roughness).

The upcrossing rates for

eighteen west wind records are plotted on the same graph, as


shown in Figure 24.

Even though values have scatter, there

is a specific trend.

A linear regression line is fitted to

the ensemble of data, as shown in Figure 24.


line has a correlation coefficient of 0.93.

The regression
This

correlation coefficient lends credence to the use of data as


an ensemble.

Values for k and c in ecjuation 5.14 are 0.580

and 0.136, respectively.


For the Rayleigh distribution, values of k and c are 2
and 1.414, respectively.

A line representing a Rayleigh

distribution is shown in Figure 24.

The regression line

fitted to the field data is quite different from the


Rayleigh distribution line, indicating that the conductor
response data has a non-Gaussian distribution.

It is

observed in Figure 24, that a Rayleigh distribution


underestimates upcrossing rate as compared to the field
data.

The upcrossing rate plot can be used to determine the

peak factor for a desirable probability of upcrossings.

The

peak factors obtained from the plot are based on 0.1 second
peak values.

133

0.000000001 T

0.000005

o
JO

0.0006

CO
CO

9
D
"o
!

0.0111 r

(0

0.066"

0.2

0.4

1.7

Figure 24:

2.7
Peak Factor, g

4.5

Cumulative P r o b a b i l i t y D i s t r i b u t i o n of
U p c r o s s i n g s f o r Conductor Response

7.4

CHAPTER VI
CONCLUSIONS

The purpose of this study was to compare and refine the


analytical model to predict dynamic responses of electrical
transmission line conductors to extreme winds using field
data.

Wind and conductor response field data were obtained

from a full-scale field experiment.

The field data were

collected by the Bonneville Power Administration (BPA) from


an instrumented single circuit 500 kV lattice tower on the
John Day-Grizzly line 2, which is located at the Moro site
in northern Oregon.

The conductor spans 252 m and 450 m on

two sides of the instrumented tower.

A total of

twenty-three twelve-minute duration records were utilized in


thi s study.
Based on the analysis of field data and the refinement
of the analytical model originally developed by Davenport
(1980), the following conclusions are made:
(1)

The field measured wind and conductor response data


were found to be valid.

The mean -responses of three

conductors were within 10% of each other for all


records.

Fluctuating responses of the conductors

showed a significant amount of scatter.


134

135
(2)

Winds traversing over the valley showed a wide


variation in profile and turbulence.

Winds coming from

similar terrains of valleys and hills have a power-law


exponent range from 0.11 to 0.18 and a turbulence
intensity range from 0.11 to 0.21.
(3)

The wind spectra showed 99% of spectral energy in the


frecjuency range below 1 Hz.

The amplitude constant A

for Kaimal's gust spectrum was found to be within the


suggested range of 0.15 to 0.60; however, the exponent
constant n from the field data was much higher than the
suggested range of 0.33 to 0.67.
(4)

The field measured effective conductor force


coefficient was found to vary between 0.48 and 0.75.

(5)

Noticeable resonant peaks occurred in the frecjuency


range from 0.1 to 0.4 Hz. in the conductor response
spectra.

Two of these peaks close to 0.12 and 0.22 Hz

were identified as corresponding to natural transverse


frecjuencies of the conductors associated with the two
unecjual spans.

The resonant response energy level was

found to be low, less than 15% of the total energy in


most records.
(6)

The majority of the records showed that the field


measured background turbulence response of the
conductors accounted for 75% of the fluctuating
response.

136
(7)

The analytical model for background response was


refined by determining the joint acceptance function
(JAF) coefficients from the field measured data.

The

best values for the coefficients (ecjuation 5.7) are


judged to be Q=0.45 and M=0.20.
(8)

Damping of the conductors, found from the field


measured data, was much higher than the theoretical
aerodynamic damping ratio.

A damping value of 40% is

suggested for the conductors.


(9)

The refinement of JAF coefficients and use of


aerodynamic damping factor of 40% in the analytical
model gave a significant improvement in prediction of
background and resonant responses when results were
compared with the field measured data.

However, ratios

of refined analytical model values to field measured


values showed a large scatter.
(10)

Many records showed response peak factors (for 0.1


second response) to be higher than the range of 3.5-4.0
suggested in the analytical model.

The upcrossing rate

principle was used to determine the peak factors on a


probabilistic basis.

The Weibull distribution

satisfactorily describes the probability distribution


of the upcrossings rate.

137
It is recommended that additional field data be
obtained, particularly at reasonably predictable sites, to
further verify and refine the analytical model. The
computational procedures presented here are general and are
applicable to additional field data.

REFERENCES
ANSI, 1982: "Minimum Design Loads for Building and Other
Structures, ANSI 58.1-1982," American National
Standards Institute, Inc., ANSI, New York, NY.
ASCE, 1984: "Guidelines for Transmission Line Structural
Loading," American Society of Civil Engineers, New
York, NY.
Bendat, J.S., and Piersol, A.G., 1966: "Measurements and
Analysis of Random Data," John Wiley and Sons, Inc.,
New York, NY.
Bendat, J.S., and Piersol, A.G., 1980: "Engineering
Applications of Correlation and Spectral Analysis,"
John Wiley and Sons, Inc., New York, NY.
Blevins, R.D., 1977: "Flow-Induced Vibration," Von Nostrand
Reinhold Company, New York, NY.
Brunt,. S.D., 1952: "Physical and Dynamical Meteorology,"
Second Edition, Cambridge University Press, England.
Chatfield, C , 1975: "The Analysis of Time Series: Theory
and Practice," Chapman and Hall, London.
Criswell, M.E., and Vanderbilt, M.D., 1987:
"Reliability-Based Design of Transmission Line
Structures: Final Report," EPRI EL-4793, Project
1352-2, Vol. I, EPRI, Palo Alto, California, March.
Davenport, A.G., 1960: "Wind Loads on Structures,"
Technical paper No. 88 of the Division of Building
Research, National Research Council of Canada, March.
Davenport, A.G., 1961: "The Application of Statistical
Concepts to the Wind Loading of Structures,"
Proceedings of Institute of Civil Engineers, Vol. 19,
pp. 449-471, August.
Davenport, A.G., 1967: "Gust Loading Factors," Journal of
the Structural Division, Vol. 93, ST3, ASCE, New York,
NY, pp. 11-34, June.

138

139
Davenport, A.G., 1972: "Approaches to the Design of Tall
Buildings Against Wind," Theme Report, Proceedings of
the ASCE-IABSE International Conference on Planning and
Design of Tall Buildings, Lehigh University, pp. 1-22,
August.
Davenport, A.G., 1977: "The Prediction of the Response of
Structures to Gusty Wind," Proceedings of International
Research Seminar, Safety of Structures Under Dynamic
Loading, Vol. I, Norwegian Institute of Technology,
Trondheim, Norway, pp. 257-284, June.
Davenport, A.G., 1980: "Gust Response Factors for
Transmission Line Loading." Wind Engineering, Proc. of
the Fifth International Conference on Wind Engineering,
J.E. Cermak, Ed. (Fort Collins, CO; July 1979),
Pergamon Press, New York, NY.
Ferraro, V., 1983: "Transverse Response of Transmission
Lines to Turbulent Winds." Masters Thesis, The
University of Western Ontario, London, Ontario, Canada,
July.
GAI Consultants, Inc., 1981: "Transmission Line Wind
Loading Research," EPRI Interim Report, RP 1277, EPRI,
Palo Alto, California, April.
Ghiocel, D., and Lungu, D., 1975: "Wind Snow and
Temperature Effects on Structures based on
Probability," Abacus Press, Kent, England.
Gould, P.L., and Abu-Sitta, S.H., 1980: "Dynamic Response
of Structures to Wind and Earthcjuake Loading, " John
Wiley and Sons, Inc., New York, NY.
IMSL, 1982: International Mathematical and Statistical
Libraries, Inc., Vol. 2.
Jan, Con-Lin, 1982: "Analysis of Data for the Response of
Full-Scale Transmission Tower Systems Real Winds,"
Masters thesis, Texas Tech University, Lubbock, Texas,
December.
Jenkins, G.M, and Watts, D.G., 1968: "Spectral Analysis and
its Applications," Holden-Day, San Fransisco,
California.
Kaimal, J . C , 1978: "Horizontal Velocity Spectra in an
Unstable Surface Layer," Journal of the Atmospheric
Sciences, Vol. 35, pp. 18-24.

140
Kancharla, V.S., 1987: "Analysis of Wind Characteristics
from Field Wind Data," Masters thesis, Texas Tech
University, Lubbock, Texas, May.
Kempner, L., Jr., 1979: "Wind Loading on a Latticed
Transmission Tower," Report No. ETKA-79-1, Bonneville
Power Administration, U.S. Dept. of Energy, November.
Kempner, L., Jr., and Laursen, H.I., 1977: "Response of a
Latticed Transmission Tower," Report No. ETB-77-1,
Bonneville Power Administration, U.S. Dept. of Energy,
November.
Kempner, L., Jr., and Laursen, H.I., 1981: "Measured
Dynamic Response of a Latticed Transmission Tower and
Conductors to Wind Loading," Report No. ME-81-2,
Bonneville Power Administration, U.S. Dept. of Energy,
July.
Kempner, L., Jr., and Thorkildson, R.M., 1982: "Moro Test
Site Wind Parameter Study." Report No. ME-82-3,
Bonneville Power Administration, U.S. Dept. of Energy,
December.
Kempner, L., Jr., Volpe, H.W., and Thorkildson, R.M., 1985:
"Wind Loading Research on Transmission Line,"
Proceedings of the Fifth International Conference on
Wind Engineering, Texas Tech University, Lubbock,
Texas, November.
Kim, Soo-Il, 1977: "Wind Loads on Flat Roof Area Through
Full-Scale Experiment," Doctoral Dissertation, Texas
Tech University, Lubbock, Texas.
Lawson, T.V., 1980: "Wind Effects on Buildings: Vol. 1,
Design Applications," Applied Science Publishers, Ltd.,
London, England.
Levitan, M., 1987: Unpublished Data, Civil Engineering
Department, Texas Tech University, Lubbock, Texas.
Mehta, K.C., Norville, H.S., and Kempner, L., Jr., 1986:
"Electric Transmission Structure Response to Wind,"
Proc. International Symposium on Probabilistic Methods
Applied to Electric Power Systems, Samy Krishnasamy,
Ed., (Ontario Hydro, Toronto, Canada; July 1986),
Pergamon Press, New York, NY.

141
Melbourne, W.H., 1975: "Peak Factors for Structures
Oscillating Under Wind Action." P r o c , Fourth
International Conference on Wind Effects on Buildings
and Structures, Cambridge University Press, London,
England.
Miller, I., and Freund, J.E., 1977: "Probability and
Statistics for Engineers," Second ed., Prentice-Hall,
Inc., Englewood Cliffs, New Jersey.
NESC, 1984: "National Electric Safety Code," ANSI C2,
Institute of Electrical and Electronics Engineers,
Inc., Wiley-Interscience, New York, NY.
Nigam, N.C., 1983: "Introduction to Random Vibrations," The
MIT Press, Cambridge, Massachusetts, London, England.
Norville, H. S., K.C. Mehta, andA.F. Farwagi, 1985: "500
kV Transmission Tower/Conductor Wind Response: phase
I," Institute for Disaster Research, Texas Tech
University, Lubbock, Texas, May.
NRCC, 1980: "National Building Code of Canada," NRCC No.
17724, Associate Committee on the National Building
Code, National Research Council of Canada, Ottawa,
Canada.
Potter, J.L., and Boylan, D.E. 1981: "Aerodynamic Force
Coefficients for the Prediction of Wind Loads on
Electrical Transmission Lines," Sverdrup Corporation,
Report submitted to EPRI, Palo Alto, California.
Pries, R.A., 1981: "Moro 1200 kV Structural/Mechanical Test
Program," Report No. ME-81-1, Bonneville Power
Administration, U.S. Dept. of Energy, March.
Racicot, R.L., 1969: "Random Vibration Analysis-Application
to Wind Loaded Structures," Report No. 30, School of
Engineering, Case Western Reserve University,
Cleveland, Ohio, February.
Rice, S.O., 1945: "Mathematical Analysis of Random Noise,"
Bell System Technical Journal, Vol. 24, pp. 46-156.
Sachs, P., 1978: "Wind Forces in Engineering," Second
Edition, Pergamon Press, New York.
Simiu, E., and Scanlan, R.H., 1985: "Wind Effects on
Structures: An Introduction to Wind Engineering," John
Wiley and Sons, Inc., New York, NY.

142
Statistical Analysis System (SAS), 1985: "SAS User Guide:
Statistics," SAS Institute, Inc., Gary, North Carolina.
Symon, R.K., 1961: "Mechanics," Second Ed., Addison-Wesley
Publishing Co., Inc., Reading, MA, May.
Twu, S.R., 1983: "Gust Response Factor for Transmission
Line Structures," Masters Thesis, Texas Tech
University, Lubbock, Texas, August.

You might also like